Research ArticleSemideciduous Seasonal Forest Production of Leaves andDeciduousness in Function of the Water Balance LAI and NDVI
Thomaz Correa e Castro da Costa1 Joatildeo Herbert Moreira Viana1 and Juliana Leite Ribeiro2
1 Embrapa Milho e Sorgo Road MG 424 km 45 35701-970 Sete Lagoas MG Brazil2 Universidade Federal de Sao Joao Del Rei Campus Sete Lagoas Road MG 424 km 47 35701-970 Sete Lagoas MG Brazil
Correspondence should be addressed toThomaz Correa e Castro da Costa thomazcostaembrapabr
Received 28 August 2013 Revised 19 December 2013 Accepted 4 February 2014 Published 31 March 2014
Academic Editor Panos V Petrakis
Copyright copy 2014 Thomaz Correa e Castro da Costa et al This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited
This study investigated the relationship between leaf production litterfall water balance Leaf Area Index (LAI) and NormalizedDifference Vegetation Index (NDVI) in semideciduous forests The goal was to model this phenomenon to obtain the estimates ofthis component as an additional compartment of the ecosystem carbon sinkThe tests were conducted in eight semideciduous forestfragments Twenty-four permanent plots were monitored monthly and LAI measurements and weighing of litterfall deposited innets were conducted for a period of thirteenmonths In this period Landsat 5 and IRS satellite images were obtained and processedfor generation of NDVI The water balance was calculated for each day The relationship among the variables ldquoleaf dry weightrdquoldquoLAIrdquo ldquoNDVIrdquo and ldquowater balancerdquo was verified and a regression model was built and evaluated The deciduous phenomenon canbe explained by hydric balance and LAI and NDVI are ancillary variablesThe tendency of the variables in the period of 13 monthswas explained by quadratic functionsThe varied behavior among themonitoring sites helped to know differences in the depositionof leaves This study showed that only the leaf component of the litterfall of a semideciduous forest in tropical climate can capture4 to 8Mgsdothaminus1sdotyrminus1 of CO
2
and this amount can be estimated using climate biophysics and vegetation index variables
1 Introduction
The sprouting development blooming fruit bearing andsenescence phases determine the phenology of plant speciesIn forest ecosystems the sprouting and leaf growth thesenescence and the leaf fall are crucial for their maintenanceand for survival through the nutrient cycling The fall ofleaves branches flowers and fruits supplies organic materialto the surface layer of the soil nourishing the plant speciesBy means of this process nutrients are deposited and min-eralized maintaining the soil fertility in these ecosystems [1ndash3] The type of vegetation and the environmental conditionsinfluence the distribution quantity and quality of thesematerials which form the litterfall [4 5]
The accumulated litterfall is all the deposited material onthe surface of the forest soil for a determined period whichmay be measured by the deposition in collectors of prefixedsizes during preestablished time intervals [6]
The deciduousness phenomenon is governed by thespecies occurring in each ecosystem Such phenomenon isinduced by environmental conditions especially by tem-perature and water stress [7] according to the regionalclimateTherefore Seasonal Semideciduous Forest is a type ofvegetation conditioned by the climatic seasonality whichmaybe tropical defined by the rainfall regime with a rainy seasonand a dry season with dynamics that depend on the soil waterstatus or subtropical determined by lowwinter temperatures[8] In temperate climates the seasonality is severe and thephenology of the species is forcefully dependent on thetemperature [7 9]
In ecosystems driven by water or temperature deficiencythe quantity of material that falls from the canopy formingthe litterfall reaches a rate of tons per hectareyear thenthe forest starts producing leaves again by the beginning ofthe rainy season or with the increase of the temperaturerenewing the lost biomass
Hindawi Publishing CorporationInternational Journal of EcologyVolume 2014 Article ID 923027 15 pageshttpdxdoiorg1011552014923027
2 International Journal of Ecology
One of the purposes of litterfall deposition estimation isto present the efficiency of the natural ecosystems in provid-ing the soil with the necessary nutrients for its maintenance[10] Another purpose is to evaluate its role as a carbonsink which may be an important environmental service thatjustifies its preservation
There are rare cases in the many studies about decid-uousness and nutrient cycling monitoring that relate thismonitoring with biophysical and orbital variables whichis the way of estimating these phenomena Identifying thebehavior patterns in these processes will make its modelingpossible
On the other hand many attempts to detect the vege-tation phenological patterns by remote sensing in order tounderstand the year-to-year cycling pattern of the carbonin the terrestrial ecosystems have been made especiallyafter the release of the MODIS sensor [11] and for thecalibration quality and the products provided such as LAIand fAPAR [12] Zhang et al [13] using MODIS data on acontinental scale were able to detect phenological leaf growthand dormancy stages which are related to temperatureoccurrence periods and the latitude
The relationship of remote sensing vegetation index to theleaf area is well known especially in natural vegetation [14ndash24]The leaf area is a key biphasic variable which is related tothe physiologic processes of the plant such as the productionand the consumption of the plant phytomass through theabsorbed electromagnetic radiation photosynthesis respira-tion and transpiration One of the forms to estimate the leafarea in the field is by themeasurement of the electromagneticenergy transmission rate using specific instruments like LAI2200 [25ndash27] or through direct methods such as decline ofleaf litter component [28]
The empiric relation between LAI (Leaf Area Index) andNDVI (Normalized Difference Vegetation Index) is affectedby NDVI saturation in dense plant coverage [21] whichoccurs specially in ombrophilous typologies
One of the experiments that approached the orbital indexrelation with the deciduousness was done byWang et al [16]who related NDVI and LAI in deciduous forest sites takinginto consideration that LAI was acquired from two methodsOne was the direct method using the weight of the period-ically collected leaves per unit of surface area with the leafarea obtained by calibration with the simple measurementof small samples and the indirect method using the inversemodel of the Beer-Bouguer-Lambert Law from the effect ofthe zenithal angle in the extinction coefficient and in theclumping index having the global daily radiation above andbelow the canopy as input
The deciduous phenomenon studied is controlled by thetropical climatic seasonality which is defined by the rainfallregime The climatic variations especially precipitation andevapotranspiration determine along with the physiologicalbehavior of the deciduous species and the edaphic conditionsthe moment when the deciduousness increases and the leafproduction is reduced and the period inwhich the deciduous-ness is reduced and the leaf production is increasedThewaterbalance (periods with water excess or deficiency) is the mainconditioning of this phenomenon Potithep et al [29] and
Figure 1 Location of the study sites in each fragment of theEmbrapa Experimental Farm Sete Lagoas MG Brazil
Kale et al [30] established two stages in deciduous tropicalforests leaf growth and senescence In deciduous temperateforests three stages were observed whereas in the summer aperiod of leaf permanence occurs [16]
In the litterfall deposition and leaf production the leafarea may be a descriptor of this phenomenon as it has adirect relationship with the leaf loss (LAI reduction) and withseeding and leaf growth (LAI increasing) Both processesdeciduousness and leaf production show little temporalintersection under these climatic conditions
The objective of this work is to study the deciduousnessrelationships and the leaf production water balance leaf areaand the NDVI with the intention of modeling the annuallitterfall production rate in a semideciduous seasonal forest Itwill be therefore possible to estimate the nutrient depositionand the carbon fixation through this phenomenon
2 Materials and Methods
21 Study Site The study was conducted in forest fragmentswhere a phytosociological inventory was doneThe inventorydesign had twenty-four 20 times 20 meter permanent plots ineight fragments of deciduous seasonal forest in a CerradoBioma at the Experimental Farm of Embrapa Maize andSorghum which is located in the city of Sete Lagoas MGBrazil (Figure 1) characterized by the parameters of Table 1
The forest fragments have been monitored for the nutri-ent balance and carbon fixation including the litterfalldecomposition rate soil classification by profiles fertilityanalyses soil granulometry water retention curves infil-tration tests and microbial activity The bush and arborealspecies with Diameter at Breast Height (DBH) greater than5 cm were measured and identified
Fragment 1 is a forest after regenerated pasture whichwas verified with the aid of an aerial photo from 1949
International Journal of Ecology 3
Table 1 Coordinates (119883119884) UTM 23 ZoneWGS84 of the 1st permanent plot in the fragments geometric altitude (119885) declination in degrees(Decl) number of permanent plots (Parc) individual density (indsdotmminus2) basal area (119861) volume (Vol) average tree height (119867) Shannon index(1198671015840) and Simpson dominance index (119862)
Table 2 Collection dates of the Litterfall LAI measurement and Landsat 5 images (082011ndash092111) and IRS (080212ndash091112) andrespective days of the period (010111ndash123112)
Dlw (gsdotmminus2) LAI (m2sdotmminus2) NDVI
Date Julian day Gather Period Date Julian day Date Julian day200811 232
with a small diversity of species The soils in this area areof the Haplohumult and Kandiudox classes of USA SoilTaxonomy [32] Fragment 2 is transition between the Savannaand Semideciduous Seasonal Forest on the Haplustox soilclass Fragment 3 is on the Inceptic Hapludox soil classand a water Table 2 meters deep Fragments 4 5 and 6 areremnants of the Semideciduous Seasonal Forest on IncepticHaplustox Inceptic Eutrustox and Humic Haplustox soilsThe permanent plot in Fragment 7 has a great occurrence ofbush vegetation low tree density and the predominance ofindividuals of great size on an Inceptic Haplustox soil classFragment 8 is close to water bodies on Inceptic Haplustoxand on Kanhaplustalf soil classes All the soils present highto very high saturation by aluminum (Al) and low to verylow base saturation (119881) in the subsurface horizons with theexception of Site 2 in Fragment 8The horizonA of those soilsis the only eutrophic horizon
22 Water Balance In 2011 the average temperature for theyear was 2224∘C with an annual precipitation of 14463mmand potential evapotranspiration of 10732mm
The water balance of Thornthwaite was calculated from2011 to October 2012 with daily data of PET (PotentialEvapotranspiration) calculated by Penman-Monteith [33]and the precipitation with data obtained from the INMETClimatologic Station which is installed at the EmbrapaExperimental Farm In order to calculate the daily waterexcess and the water deficiency an available water capacity of150mm was used and the methodology described in Pereiraet al [34] was carried out by the development of a VBAroutine (Visual Basic for Application)
23 Deciduousness Measurement In order to measure thedeciduousness data 24 permanent plots of the phytosocio-logical inventory were used The gathering of the depositionof litterfall in the nets was done in the intermediate days ofeach month After the installation of the 1m times 25m nets inthe field the length width and diagonals of each net weremeasured to calculate their exact area since their installationproduced small variations in the original dimensions There-fore each net had its own area to measure the weight of theleaves in gsdotmminus2
4 International Journal of Ecology
During the gathering a screening was made to discardthe branches and the seeds and fruits were separated for thegermination tests The leaves were weighted on an analyticalscale to obtain the freshmass moisture () and the drymass(g) after the sample-weight stabilization in an oven at 65∘C
The data for the dry leaves in gsdotmminus2sdotnetminus1 were summa-rized by fragment which is an average of the installed nets inthe permanent plots of each fragment
To obtain the deciduousness variation rate by fragmentthe observed data of the dry leaf weight (gsdotmminus2) were adjustedto the number of days by quadratic equations and the firstderivative of each equation was taken which correspondsto the variation rate between the recurrent periods Thusrelationships of deciduousness variation rates were obtainedwith the variables water excess and deficiency (mm)
24 LAI Measurement In order to measure the leaf area anLAI 2200 Plant Canopy Analyzer [25] was used an indirectmeasurer which uses transmittance for the LAI calculation(m2 of leafsdotmminus2 of soil) The probability of noninterceptionof light by the canopy 119875(120579) is the function of the length ofthe coursed path 119878(120579) canopy density 120583 (m2 foliage per m3canopy volume) and 119866(120579) the fraction of foliage projectedtoward 120579 given by [25]
119875 (120579) = 119890
minus119866(120579)120583119878(120579)
(1)
And the exact solution for the density (120583) proven by [35] isquoted by Li-Cor [25] as follows
120583 = 2int
1205872
0
minus
ln119875 (120579)119878 (120579)
sin 120579 d120579 (2)
In continuous and homogeneous forest ecosystems
119878 (120579) =
ℎ
cos (120579) (3)
Where ℎ is the height of the canopyThe relation between thedensity of the leaf area by the canopy volume (120583) and LAI is
119871 = 120583 lowast ℎ (4)
Substituting (3) and (4) in (2) yields
L = 2int1205872
0
minuslnP (120579) cos 120579 sin 120579d120579 (5)
To procedure of the Lai 2200 it is need the apparent clumpingfactor
120596app =2 int
1205872
0
minus (ln P (120579)S (120579)) sin 120579d120579
2 int
1205872
0
minus (lnP (120579)S (120579)) sin 120579d120579(6)
Where 120579 = 7 23 38 53 68 zenithal angles delimited by eachring in the fisheye lens of the sensor
119875 (120579) =
119861
119860
(7)
Expression (7) is the transmittance or probability of alight beam passing over the foliage without being interceptedmeasured by the ratio between the B readings obtained belowthe canopy and the A readings before being intercepted bythe canopy
The effective leaf index L119890
is finality computed by
Le = 119871 lowast 120596119886119901119901 (8)
To obtain precise measurements with LAI-2200 it isnecessary to capture only the diffuse energy To avoid or atleast strongly reduce the direct energy captured by the sensorthe readings were made using view cap 45∘ under cloudysky or at dawn or twilight verifying the luminous stabilitythroughout and after the measurements which was madeby the data tendency analysis after the measurement Thedata with tendency were discardedThe readings were alwaysperformed on intermediate days of each month and theystarted inOctober 2011 with a reading in each of the collectornets always in the same position and direction There were 5nets in each permanent plot in a total of 120 netsThe LAI wascalculated for each permanent plot
25 NDVI Data Processing To generate NDVI an atmo-sphere correction model was used which would allow theuse of atmospheric pressure and relative humidity aimingto obtain the removal of the atmosphere interference with agreater local precision Three Landsat 5 images were selected(httpwwwinpebr) as they coincide with their final recep-tion complementing the period with 7 images of the IndianRemote Sensing Satellite (IRS) a sensor with characteristicscloser to LandsatThemain differences between them are thespatial (30m and 24m pixel) and radiometric (8 and 7 bits)resolution and a slim difference in the range of the spectralbands All images arewith visibility above 10 km (informationfromAir Traffic Control Service bulletin of Confins Airport)
The atmospheric correction was performed with anATMOSC module of Idrisi Taiga using the Cost Model
The Dn Raze [36] was obtained through a procedurewhich uses a band frequency table which is available onthe electronic spreadsheet at httpwwwdsrinpebrdown-loadshtm based on the atmosphere correction method ofChaves [37]
The atmosphere optical dimension is the sum of thecomponents Rayleigh scattering aerosols water vapor andtypical ozone Aerosols and typical ozone were not computeddue to lack of data The Rayleigh scattering was adjustedto local height with atmospheric pressure data using thefollowing expression
where 119886 = 005 for band 119877 and 119886 = 001 for band NIR ([38]quoted by [39])
The water vapor component was obtained for the nearinfrared (NIR) in function of the relative humidity by thelinear relation [39]
Water vapor = 00012 lowast RH + 0016 (10)
International Journal of Ecology 5
Table 3 Coefficients of determination (1198772 adjusted) for polynomial equations in the observed dry leaf weight (gsdotmminus2) LAI (m2sdotmminus2) and
NDVI data and respective minimum and maximum points by fragment (sequential day from 01012011)
The geometrical correction was performed using theGeotiff Examiner software as well as the assistance of thegraphic software Inkscape using a reference point to dislocateeach image As the images contain deformation correctionsthis form of correction was more precise compared to thepolynomial corrections even with RMS smaller than 12pixel In order to extract NDVI in each permanent plot arectangle of 9 pixels centralized in the central point of thepermanent plot was digitalized
26 Monitoring Period The three variables dry leaveweightsdotmminus2 LAI (m2sdotmminus2) and NDVI were analyzed in thefunction of the day of the period from 01012011 to 2012The analyses period started in August 2011 (when the leavesstart to accumulate) until September 2012 (Table 2) Fromthe 1560 visits to the nets in this period 58 correspondto nets damaged by thick fallen branches or damaged byanimals Fourteen were discarded because they correspondto outliers So 1488 leaf weight data by period were usedOf the 289 LAI values calculated per permanent plot 7were discarded as outliers And of the 276 NDVI extractedvalues 3 were classified as outliers During the collectionperiods the accumulation of leaveswas not exactly of 30 dayswith a maximum fluctuation of 36 days between Octoberand November 2011 because of the unexpected rains in thecollection period The decision was made in order not tointerpolate the data
3 Results
31 LAI and Litterfall LAI has a strong relationship with thedeciduousness in this forest typology as its value is the conse-quence of the quantity of leaves in the canopy so alterationsin this quantity directly reflect in the LAI Figure 2 presentsthe LAI behavior with the dry leaf weight (gsdotmminus2sdotperiodminus1)Except for Fragment 3 the tendency was similar to thecompared fragments The increase of LAI indicates that thesprouting and growth of leaves increases until a maximumsynchronic with the reduction of deciduousness when the
latter reaches a minimum and the reduction of leaf areastarts synchronic with the increase of the deciduousness Inthis stage the deciduousness drives the reduction of LAIbecause in this period the leaf production practically stops
32 LAI and NDVI The NDVI starts its increase beforeLAI (Figure 3) probably because of the influence of theunderstory which promptly responds to the first rain beforethe canopy stars sproutingThis understory does not count onLAI computation as the sensor is positioned above this veg-etation for the readings The NDVI also decreases before theLAI because before the leaf falls a gradual depigmentation ofchlorophyll takes place and the loss of cellular turgidity in thepalisade parenchyma of the leaves increases the reflection ofthe absorption band by the chlorophyll (red) and decreasesthe reflection band due to the reduction of the intracellularspaces and the turgidity of the cells (near infrared) Thisphenomenon reflects on the reduction of the NDVI beforethe deciduous response detected by the LAI Maki et al[17] conducted experiments in a cool-temperate deciduousforest with Modis data and LAI field measurements Theyhad similar results indicating that the understory influencesthe comparisons between LAI and NDVI and that in the leafsenescence period the discoloration of the leaves interferesin the NDVI results before it interferes in the LAI response
In Fragment 3 due to the subsurfacewater availability theNDVI results presented here have an intermediate tendencybetween the seasonal forest and the ombrophilous forest inthe study by Soudani et al [7] who observed the constancyof the NDVI in an ombrophilous forest of Ghana throughoutthe year and abrupt changes in the NDVI in temperateclimate forests in the north of Europe driven by the climaticseasonality by temperature
It is known that the behavior of the deciduousness ofthe LAI and NDVI is sigmoidal during a period beyond oneyear but in one year these behaviors can be fit in polynomialmodels The best adjustment for the deciduous behavior LAIand NDVI in the whole period of 13 months (September2011 to September 2012) was through the second-degree
6 International Journal of Ecology
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
One of the purposes of litterfall deposition estimation isto present the efficiency of the natural ecosystems in provid-ing the soil with the necessary nutrients for its maintenance[10] Another purpose is to evaluate its role as a carbonsink which may be an important environmental service thatjustifies its preservation
There are rare cases in the many studies about decid-uousness and nutrient cycling monitoring that relate thismonitoring with biophysical and orbital variables whichis the way of estimating these phenomena Identifying thebehavior patterns in these processes will make its modelingpossible
On the other hand many attempts to detect the vege-tation phenological patterns by remote sensing in order tounderstand the year-to-year cycling pattern of the carbonin the terrestrial ecosystems have been made especiallyafter the release of the MODIS sensor [11] and for thecalibration quality and the products provided such as LAIand fAPAR [12] Zhang et al [13] using MODIS data on acontinental scale were able to detect phenological leaf growthand dormancy stages which are related to temperatureoccurrence periods and the latitude
The relationship of remote sensing vegetation index to theleaf area is well known especially in natural vegetation [14ndash24]The leaf area is a key biphasic variable which is related tothe physiologic processes of the plant such as the productionand the consumption of the plant phytomass through theabsorbed electromagnetic radiation photosynthesis respira-tion and transpiration One of the forms to estimate the leafarea in the field is by themeasurement of the electromagneticenergy transmission rate using specific instruments like LAI2200 [25ndash27] or through direct methods such as decline ofleaf litter component [28]
The empiric relation between LAI (Leaf Area Index) andNDVI (Normalized Difference Vegetation Index) is affectedby NDVI saturation in dense plant coverage [21] whichoccurs specially in ombrophilous typologies
One of the experiments that approached the orbital indexrelation with the deciduousness was done byWang et al [16]who related NDVI and LAI in deciduous forest sites takinginto consideration that LAI was acquired from two methodsOne was the direct method using the weight of the period-ically collected leaves per unit of surface area with the leafarea obtained by calibration with the simple measurementof small samples and the indirect method using the inversemodel of the Beer-Bouguer-Lambert Law from the effect ofthe zenithal angle in the extinction coefficient and in theclumping index having the global daily radiation above andbelow the canopy as input
The deciduous phenomenon studied is controlled by thetropical climatic seasonality which is defined by the rainfallregime The climatic variations especially precipitation andevapotranspiration determine along with the physiologicalbehavior of the deciduous species and the edaphic conditionsthe moment when the deciduousness increases and the leafproduction is reduced and the period inwhich the deciduous-ness is reduced and the leaf production is increasedThewaterbalance (periods with water excess or deficiency) is the mainconditioning of this phenomenon Potithep et al [29] and
Figure 1 Location of the study sites in each fragment of theEmbrapa Experimental Farm Sete Lagoas MG Brazil
Kale et al [30] established two stages in deciduous tropicalforests leaf growth and senescence In deciduous temperateforests three stages were observed whereas in the summer aperiod of leaf permanence occurs [16]
In the litterfall deposition and leaf production the leafarea may be a descriptor of this phenomenon as it has adirect relationship with the leaf loss (LAI reduction) and withseeding and leaf growth (LAI increasing) Both processesdeciduousness and leaf production show little temporalintersection under these climatic conditions
The objective of this work is to study the deciduousnessrelationships and the leaf production water balance leaf areaand the NDVI with the intention of modeling the annuallitterfall production rate in a semideciduous seasonal forest Itwill be therefore possible to estimate the nutrient depositionand the carbon fixation through this phenomenon
2 Materials and Methods
21 Study Site The study was conducted in forest fragmentswhere a phytosociological inventory was doneThe inventorydesign had twenty-four 20 times 20 meter permanent plots ineight fragments of deciduous seasonal forest in a CerradoBioma at the Experimental Farm of Embrapa Maize andSorghum which is located in the city of Sete Lagoas MGBrazil (Figure 1) characterized by the parameters of Table 1
The forest fragments have been monitored for the nutri-ent balance and carbon fixation including the litterfalldecomposition rate soil classification by profiles fertilityanalyses soil granulometry water retention curves infil-tration tests and microbial activity The bush and arborealspecies with Diameter at Breast Height (DBH) greater than5 cm were measured and identified
Fragment 1 is a forest after regenerated pasture whichwas verified with the aid of an aerial photo from 1949
International Journal of Ecology 3
Table 1 Coordinates (119883119884) UTM 23 ZoneWGS84 of the 1st permanent plot in the fragments geometric altitude (119885) declination in degrees(Decl) number of permanent plots (Parc) individual density (indsdotmminus2) basal area (119861) volume (Vol) average tree height (119867) Shannon index(1198671015840) and Simpson dominance index (119862)
Table 2 Collection dates of the Litterfall LAI measurement and Landsat 5 images (082011ndash092111) and IRS (080212ndash091112) andrespective days of the period (010111ndash123112)
Dlw (gsdotmminus2) LAI (m2sdotmminus2) NDVI
Date Julian day Gather Period Date Julian day Date Julian day200811 232
with a small diversity of species The soils in this area areof the Haplohumult and Kandiudox classes of USA SoilTaxonomy [32] Fragment 2 is transition between the Savannaand Semideciduous Seasonal Forest on the Haplustox soilclass Fragment 3 is on the Inceptic Hapludox soil classand a water Table 2 meters deep Fragments 4 5 and 6 areremnants of the Semideciduous Seasonal Forest on IncepticHaplustox Inceptic Eutrustox and Humic Haplustox soilsThe permanent plot in Fragment 7 has a great occurrence ofbush vegetation low tree density and the predominance ofindividuals of great size on an Inceptic Haplustox soil classFragment 8 is close to water bodies on Inceptic Haplustoxand on Kanhaplustalf soil classes All the soils present highto very high saturation by aluminum (Al) and low to verylow base saturation (119881) in the subsurface horizons with theexception of Site 2 in Fragment 8The horizonA of those soilsis the only eutrophic horizon
22 Water Balance In 2011 the average temperature for theyear was 2224∘C with an annual precipitation of 14463mmand potential evapotranspiration of 10732mm
The water balance of Thornthwaite was calculated from2011 to October 2012 with daily data of PET (PotentialEvapotranspiration) calculated by Penman-Monteith [33]and the precipitation with data obtained from the INMETClimatologic Station which is installed at the EmbrapaExperimental Farm In order to calculate the daily waterexcess and the water deficiency an available water capacity of150mm was used and the methodology described in Pereiraet al [34] was carried out by the development of a VBAroutine (Visual Basic for Application)
23 Deciduousness Measurement In order to measure thedeciduousness data 24 permanent plots of the phytosocio-logical inventory were used The gathering of the depositionof litterfall in the nets was done in the intermediate days ofeach month After the installation of the 1m times 25m nets inthe field the length width and diagonals of each net weremeasured to calculate their exact area since their installationproduced small variations in the original dimensions There-fore each net had its own area to measure the weight of theleaves in gsdotmminus2
4 International Journal of Ecology
During the gathering a screening was made to discardthe branches and the seeds and fruits were separated for thegermination tests The leaves were weighted on an analyticalscale to obtain the freshmass moisture () and the drymass(g) after the sample-weight stabilization in an oven at 65∘C
The data for the dry leaves in gsdotmminus2sdotnetminus1 were summa-rized by fragment which is an average of the installed nets inthe permanent plots of each fragment
To obtain the deciduousness variation rate by fragmentthe observed data of the dry leaf weight (gsdotmminus2) were adjustedto the number of days by quadratic equations and the firstderivative of each equation was taken which correspondsto the variation rate between the recurrent periods Thusrelationships of deciduousness variation rates were obtainedwith the variables water excess and deficiency (mm)
24 LAI Measurement In order to measure the leaf area anLAI 2200 Plant Canopy Analyzer [25] was used an indirectmeasurer which uses transmittance for the LAI calculation(m2 of leafsdotmminus2 of soil) The probability of noninterceptionof light by the canopy 119875(120579) is the function of the length ofthe coursed path 119878(120579) canopy density 120583 (m2 foliage per m3canopy volume) and 119866(120579) the fraction of foliage projectedtoward 120579 given by [25]
119875 (120579) = 119890
minus119866(120579)120583119878(120579)
(1)
And the exact solution for the density (120583) proven by [35] isquoted by Li-Cor [25] as follows
120583 = 2int
1205872
0
minus
ln119875 (120579)119878 (120579)
sin 120579 d120579 (2)
In continuous and homogeneous forest ecosystems
119878 (120579) =
ℎ
cos (120579) (3)
Where ℎ is the height of the canopyThe relation between thedensity of the leaf area by the canopy volume (120583) and LAI is
119871 = 120583 lowast ℎ (4)
Substituting (3) and (4) in (2) yields
L = 2int1205872
0
minuslnP (120579) cos 120579 sin 120579d120579 (5)
To procedure of the Lai 2200 it is need the apparent clumpingfactor
120596app =2 int
1205872
0
minus (ln P (120579)S (120579)) sin 120579d120579
2 int
1205872
0
minus (lnP (120579)S (120579)) sin 120579d120579(6)
Where 120579 = 7 23 38 53 68 zenithal angles delimited by eachring in the fisheye lens of the sensor
119875 (120579) =
119861
119860
(7)
Expression (7) is the transmittance or probability of alight beam passing over the foliage without being interceptedmeasured by the ratio between the B readings obtained belowthe canopy and the A readings before being intercepted bythe canopy
The effective leaf index L119890
is finality computed by
Le = 119871 lowast 120596119886119901119901 (8)
To obtain precise measurements with LAI-2200 it isnecessary to capture only the diffuse energy To avoid or atleast strongly reduce the direct energy captured by the sensorthe readings were made using view cap 45∘ under cloudysky or at dawn or twilight verifying the luminous stabilitythroughout and after the measurements which was madeby the data tendency analysis after the measurement Thedata with tendency were discardedThe readings were alwaysperformed on intermediate days of each month and theystarted inOctober 2011 with a reading in each of the collectornets always in the same position and direction There were 5nets in each permanent plot in a total of 120 netsThe LAI wascalculated for each permanent plot
25 NDVI Data Processing To generate NDVI an atmo-sphere correction model was used which would allow theuse of atmospheric pressure and relative humidity aimingto obtain the removal of the atmosphere interference with agreater local precision Three Landsat 5 images were selected(httpwwwinpebr) as they coincide with their final recep-tion complementing the period with 7 images of the IndianRemote Sensing Satellite (IRS) a sensor with characteristicscloser to LandsatThemain differences between them are thespatial (30m and 24m pixel) and radiometric (8 and 7 bits)resolution and a slim difference in the range of the spectralbands All images arewith visibility above 10 km (informationfromAir Traffic Control Service bulletin of Confins Airport)
The atmospheric correction was performed with anATMOSC module of Idrisi Taiga using the Cost Model
The Dn Raze [36] was obtained through a procedurewhich uses a band frequency table which is available onthe electronic spreadsheet at httpwwwdsrinpebrdown-loadshtm based on the atmosphere correction method ofChaves [37]
The atmosphere optical dimension is the sum of thecomponents Rayleigh scattering aerosols water vapor andtypical ozone Aerosols and typical ozone were not computeddue to lack of data The Rayleigh scattering was adjustedto local height with atmospheric pressure data using thefollowing expression
where 119886 = 005 for band 119877 and 119886 = 001 for band NIR ([38]quoted by [39])
The water vapor component was obtained for the nearinfrared (NIR) in function of the relative humidity by thelinear relation [39]
Water vapor = 00012 lowast RH + 0016 (10)
International Journal of Ecology 5
Table 3 Coefficients of determination (1198772 adjusted) for polynomial equations in the observed dry leaf weight (gsdotmminus2) LAI (m2sdotmminus2) and
NDVI data and respective minimum and maximum points by fragment (sequential day from 01012011)
The geometrical correction was performed using theGeotiff Examiner software as well as the assistance of thegraphic software Inkscape using a reference point to dislocateeach image As the images contain deformation correctionsthis form of correction was more precise compared to thepolynomial corrections even with RMS smaller than 12pixel In order to extract NDVI in each permanent plot arectangle of 9 pixels centralized in the central point of thepermanent plot was digitalized
26 Monitoring Period The three variables dry leaveweightsdotmminus2 LAI (m2sdotmminus2) and NDVI were analyzed in thefunction of the day of the period from 01012011 to 2012The analyses period started in August 2011 (when the leavesstart to accumulate) until September 2012 (Table 2) Fromthe 1560 visits to the nets in this period 58 correspondto nets damaged by thick fallen branches or damaged byanimals Fourteen were discarded because they correspondto outliers So 1488 leaf weight data by period were usedOf the 289 LAI values calculated per permanent plot 7were discarded as outliers And of the 276 NDVI extractedvalues 3 were classified as outliers During the collectionperiods the accumulation of leaveswas not exactly of 30 dayswith a maximum fluctuation of 36 days between Octoberand November 2011 because of the unexpected rains in thecollection period The decision was made in order not tointerpolate the data
3 Results
31 LAI and Litterfall LAI has a strong relationship with thedeciduousness in this forest typology as its value is the conse-quence of the quantity of leaves in the canopy so alterationsin this quantity directly reflect in the LAI Figure 2 presentsthe LAI behavior with the dry leaf weight (gsdotmminus2sdotperiodminus1)Except for Fragment 3 the tendency was similar to thecompared fragments The increase of LAI indicates that thesprouting and growth of leaves increases until a maximumsynchronic with the reduction of deciduousness when the
latter reaches a minimum and the reduction of leaf areastarts synchronic with the increase of the deciduousness Inthis stage the deciduousness drives the reduction of LAIbecause in this period the leaf production practically stops
32 LAI and NDVI The NDVI starts its increase beforeLAI (Figure 3) probably because of the influence of theunderstory which promptly responds to the first rain beforethe canopy stars sproutingThis understory does not count onLAI computation as the sensor is positioned above this veg-etation for the readings The NDVI also decreases before theLAI because before the leaf falls a gradual depigmentation ofchlorophyll takes place and the loss of cellular turgidity in thepalisade parenchyma of the leaves increases the reflection ofthe absorption band by the chlorophyll (red) and decreasesthe reflection band due to the reduction of the intracellularspaces and the turgidity of the cells (near infrared) Thisphenomenon reflects on the reduction of the NDVI beforethe deciduous response detected by the LAI Maki et al[17] conducted experiments in a cool-temperate deciduousforest with Modis data and LAI field measurements Theyhad similar results indicating that the understory influencesthe comparisons between LAI and NDVI and that in the leafsenescence period the discoloration of the leaves interferesin the NDVI results before it interferes in the LAI response
In Fragment 3 due to the subsurfacewater availability theNDVI results presented here have an intermediate tendencybetween the seasonal forest and the ombrophilous forest inthe study by Soudani et al [7] who observed the constancyof the NDVI in an ombrophilous forest of Ghana throughoutthe year and abrupt changes in the NDVI in temperateclimate forests in the north of Europe driven by the climaticseasonality by temperature
It is known that the behavior of the deciduousness ofthe LAI and NDVI is sigmoidal during a period beyond oneyear but in one year these behaviors can be fit in polynomialmodels The best adjustment for the deciduous behavior LAIand NDVI in the whole period of 13 months (September2011 to September 2012) was through the second-degree
6 International Journal of Ecology
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Table 1 Coordinates (119883119884) UTM 23 ZoneWGS84 of the 1st permanent plot in the fragments geometric altitude (119885) declination in degrees(Decl) number of permanent plots (Parc) individual density (indsdotmminus2) basal area (119861) volume (Vol) average tree height (119867) Shannon index(1198671015840) and Simpson dominance index (119862)
Table 2 Collection dates of the Litterfall LAI measurement and Landsat 5 images (082011ndash092111) and IRS (080212ndash091112) andrespective days of the period (010111ndash123112)
Dlw (gsdotmminus2) LAI (m2sdotmminus2) NDVI
Date Julian day Gather Period Date Julian day Date Julian day200811 232
with a small diversity of species The soils in this area areof the Haplohumult and Kandiudox classes of USA SoilTaxonomy [32] Fragment 2 is transition between the Savannaand Semideciduous Seasonal Forest on the Haplustox soilclass Fragment 3 is on the Inceptic Hapludox soil classand a water Table 2 meters deep Fragments 4 5 and 6 areremnants of the Semideciduous Seasonal Forest on IncepticHaplustox Inceptic Eutrustox and Humic Haplustox soilsThe permanent plot in Fragment 7 has a great occurrence ofbush vegetation low tree density and the predominance ofindividuals of great size on an Inceptic Haplustox soil classFragment 8 is close to water bodies on Inceptic Haplustoxand on Kanhaplustalf soil classes All the soils present highto very high saturation by aluminum (Al) and low to verylow base saturation (119881) in the subsurface horizons with theexception of Site 2 in Fragment 8The horizonA of those soilsis the only eutrophic horizon
22 Water Balance In 2011 the average temperature for theyear was 2224∘C with an annual precipitation of 14463mmand potential evapotranspiration of 10732mm
The water balance of Thornthwaite was calculated from2011 to October 2012 with daily data of PET (PotentialEvapotranspiration) calculated by Penman-Monteith [33]and the precipitation with data obtained from the INMETClimatologic Station which is installed at the EmbrapaExperimental Farm In order to calculate the daily waterexcess and the water deficiency an available water capacity of150mm was used and the methodology described in Pereiraet al [34] was carried out by the development of a VBAroutine (Visual Basic for Application)
23 Deciduousness Measurement In order to measure thedeciduousness data 24 permanent plots of the phytosocio-logical inventory were used The gathering of the depositionof litterfall in the nets was done in the intermediate days ofeach month After the installation of the 1m times 25m nets inthe field the length width and diagonals of each net weremeasured to calculate their exact area since their installationproduced small variations in the original dimensions There-fore each net had its own area to measure the weight of theleaves in gsdotmminus2
4 International Journal of Ecology
During the gathering a screening was made to discardthe branches and the seeds and fruits were separated for thegermination tests The leaves were weighted on an analyticalscale to obtain the freshmass moisture () and the drymass(g) after the sample-weight stabilization in an oven at 65∘C
The data for the dry leaves in gsdotmminus2sdotnetminus1 were summa-rized by fragment which is an average of the installed nets inthe permanent plots of each fragment
To obtain the deciduousness variation rate by fragmentthe observed data of the dry leaf weight (gsdotmminus2) were adjustedto the number of days by quadratic equations and the firstderivative of each equation was taken which correspondsto the variation rate between the recurrent periods Thusrelationships of deciduousness variation rates were obtainedwith the variables water excess and deficiency (mm)
24 LAI Measurement In order to measure the leaf area anLAI 2200 Plant Canopy Analyzer [25] was used an indirectmeasurer which uses transmittance for the LAI calculation(m2 of leafsdotmminus2 of soil) The probability of noninterceptionof light by the canopy 119875(120579) is the function of the length ofthe coursed path 119878(120579) canopy density 120583 (m2 foliage per m3canopy volume) and 119866(120579) the fraction of foliage projectedtoward 120579 given by [25]
119875 (120579) = 119890
minus119866(120579)120583119878(120579)
(1)
And the exact solution for the density (120583) proven by [35] isquoted by Li-Cor [25] as follows
120583 = 2int
1205872
0
minus
ln119875 (120579)119878 (120579)
sin 120579 d120579 (2)
In continuous and homogeneous forest ecosystems
119878 (120579) =
ℎ
cos (120579) (3)
Where ℎ is the height of the canopyThe relation between thedensity of the leaf area by the canopy volume (120583) and LAI is
119871 = 120583 lowast ℎ (4)
Substituting (3) and (4) in (2) yields
L = 2int1205872
0
minuslnP (120579) cos 120579 sin 120579d120579 (5)
To procedure of the Lai 2200 it is need the apparent clumpingfactor
120596app =2 int
1205872
0
minus (ln P (120579)S (120579)) sin 120579d120579
2 int
1205872
0
minus (lnP (120579)S (120579)) sin 120579d120579(6)
Where 120579 = 7 23 38 53 68 zenithal angles delimited by eachring in the fisheye lens of the sensor
119875 (120579) =
119861
119860
(7)
Expression (7) is the transmittance or probability of alight beam passing over the foliage without being interceptedmeasured by the ratio between the B readings obtained belowthe canopy and the A readings before being intercepted bythe canopy
The effective leaf index L119890
is finality computed by
Le = 119871 lowast 120596119886119901119901 (8)
To obtain precise measurements with LAI-2200 it isnecessary to capture only the diffuse energy To avoid or atleast strongly reduce the direct energy captured by the sensorthe readings were made using view cap 45∘ under cloudysky or at dawn or twilight verifying the luminous stabilitythroughout and after the measurements which was madeby the data tendency analysis after the measurement Thedata with tendency were discardedThe readings were alwaysperformed on intermediate days of each month and theystarted inOctober 2011 with a reading in each of the collectornets always in the same position and direction There were 5nets in each permanent plot in a total of 120 netsThe LAI wascalculated for each permanent plot
25 NDVI Data Processing To generate NDVI an atmo-sphere correction model was used which would allow theuse of atmospheric pressure and relative humidity aimingto obtain the removal of the atmosphere interference with agreater local precision Three Landsat 5 images were selected(httpwwwinpebr) as they coincide with their final recep-tion complementing the period with 7 images of the IndianRemote Sensing Satellite (IRS) a sensor with characteristicscloser to LandsatThemain differences between them are thespatial (30m and 24m pixel) and radiometric (8 and 7 bits)resolution and a slim difference in the range of the spectralbands All images arewith visibility above 10 km (informationfromAir Traffic Control Service bulletin of Confins Airport)
The atmospheric correction was performed with anATMOSC module of Idrisi Taiga using the Cost Model
The Dn Raze [36] was obtained through a procedurewhich uses a band frequency table which is available onthe electronic spreadsheet at httpwwwdsrinpebrdown-loadshtm based on the atmosphere correction method ofChaves [37]
The atmosphere optical dimension is the sum of thecomponents Rayleigh scattering aerosols water vapor andtypical ozone Aerosols and typical ozone were not computeddue to lack of data The Rayleigh scattering was adjustedto local height with atmospheric pressure data using thefollowing expression
where 119886 = 005 for band 119877 and 119886 = 001 for band NIR ([38]quoted by [39])
The water vapor component was obtained for the nearinfrared (NIR) in function of the relative humidity by thelinear relation [39]
Water vapor = 00012 lowast RH + 0016 (10)
International Journal of Ecology 5
Table 3 Coefficients of determination (1198772 adjusted) for polynomial equations in the observed dry leaf weight (gsdotmminus2) LAI (m2sdotmminus2) and
NDVI data and respective minimum and maximum points by fragment (sequential day from 01012011)
The geometrical correction was performed using theGeotiff Examiner software as well as the assistance of thegraphic software Inkscape using a reference point to dislocateeach image As the images contain deformation correctionsthis form of correction was more precise compared to thepolynomial corrections even with RMS smaller than 12pixel In order to extract NDVI in each permanent plot arectangle of 9 pixels centralized in the central point of thepermanent plot was digitalized
26 Monitoring Period The three variables dry leaveweightsdotmminus2 LAI (m2sdotmminus2) and NDVI were analyzed in thefunction of the day of the period from 01012011 to 2012The analyses period started in August 2011 (when the leavesstart to accumulate) until September 2012 (Table 2) Fromthe 1560 visits to the nets in this period 58 correspondto nets damaged by thick fallen branches or damaged byanimals Fourteen were discarded because they correspondto outliers So 1488 leaf weight data by period were usedOf the 289 LAI values calculated per permanent plot 7were discarded as outliers And of the 276 NDVI extractedvalues 3 were classified as outliers During the collectionperiods the accumulation of leaveswas not exactly of 30 dayswith a maximum fluctuation of 36 days between Octoberand November 2011 because of the unexpected rains in thecollection period The decision was made in order not tointerpolate the data
3 Results
31 LAI and Litterfall LAI has a strong relationship with thedeciduousness in this forest typology as its value is the conse-quence of the quantity of leaves in the canopy so alterationsin this quantity directly reflect in the LAI Figure 2 presentsthe LAI behavior with the dry leaf weight (gsdotmminus2sdotperiodminus1)Except for Fragment 3 the tendency was similar to thecompared fragments The increase of LAI indicates that thesprouting and growth of leaves increases until a maximumsynchronic with the reduction of deciduousness when the
latter reaches a minimum and the reduction of leaf areastarts synchronic with the increase of the deciduousness Inthis stage the deciduousness drives the reduction of LAIbecause in this period the leaf production practically stops
32 LAI and NDVI The NDVI starts its increase beforeLAI (Figure 3) probably because of the influence of theunderstory which promptly responds to the first rain beforethe canopy stars sproutingThis understory does not count onLAI computation as the sensor is positioned above this veg-etation for the readings The NDVI also decreases before theLAI because before the leaf falls a gradual depigmentation ofchlorophyll takes place and the loss of cellular turgidity in thepalisade parenchyma of the leaves increases the reflection ofthe absorption band by the chlorophyll (red) and decreasesthe reflection band due to the reduction of the intracellularspaces and the turgidity of the cells (near infrared) Thisphenomenon reflects on the reduction of the NDVI beforethe deciduous response detected by the LAI Maki et al[17] conducted experiments in a cool-temperate deciduousforest with Modis data and LAI field measurements Theyhad similar results indicating that the understory influencesthe comparisons between LAI and NDVI and that in the leafsenescence period the discoloration of the leaves interferesin the NDVI results before it interferes in the LAI response
In Fragment 3 due to the subsurfacewater availability theNDVI results presented here have an intermediate tendencybetween the seasonal forest and the ombrophilous forest inthe study by Soudani et al [7] who observed the constancyof the NDVI in an ombrophilous forest of Ghana throughoutthe year and abrupt changes in the NDVI in temperateclimate forests in the north of Europe driven by the climaticseasonality by temperature
It is known that the behavior of the deciduousness ofthe LAI and NDVI is sigmoidal during a period beyond oneyear but in one year these behaviors can be fit in polynomialmodels The best adjustment for the deciduous behavior LAIand NDVI in the whole period of 13 months (September2011 to September 2012) was through the second-degree
6 International Journal of Ecology
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
During the gathering a screening was made to discardthe branches and the seeds and fruits were separated for thegermination tests The leaves were weighted on an analyticalscale to obtain the freshmass moisture () and the drymass(g) after the sample-weight stabilization in an oven at 65∘C
The data for the dry leaves in gsdotmminus2sdotnetminus1 were summa-rized by fragment which is an average of the installed nets inthe permanent plots of each fragment
To obtain the deciduousness variation rate by fragmentthe observed data of the dry leaf weight (gsdotmminus2) were adjustedto the number of days by quadratic equations and the firstderivative of each equation was taken which correspondsto the variation rate between the recurrent periods Thusrelationships of deciduousness variation rates were obtainedwith the variables water excess and deficiency (mm)
24 LAI Measurement In order to measure the leaf area anLAI 2200 Plant Canopy Analyzer [25] was used an indirectmeasurer which uses transmittance for the LAI calculation(m2 of leafsdotmminus2 of soil) The probability of noninterceptionof light by the canopy 119875(120579) is the function of the length ofthe coursed path 119878(120579) canopy density 120583 (m2 foliage per m3canopy volume) and 119866(120579) the fraction of foliage projectedtoward 120579 given by [25]
119875 (120579) = 119890
minus119866(120579)120583119878(120579)
(1)
And the exact solution for the density (120583) proven by [35] isquoted by Li-Cor [25] as follows
120583 = 2int
1205872
0
minus
ln119875 (120579)119878 (120579)
sin 120579 d120579 (2)
In continuous and homogeneous forest ecosystems
119878 (120579) =
ℎ
cos (120579) (3)
Where ℎ is the height of the canopyThe relation between thedensity of the leaf area by the canopy volume (120583) and LAI is
119871 = 120583 lowast ℎ (4)
Substituting (3) and (4) in (2) yields
L = 2int1205872
0
minuslnP (120579) cos 120579 sin 120579d120579 (5)
To procedure of the Lai 2200 it is need the apparent clumpingfactor
120596app =2 int
1205872
0
minus (ln P (120579)S (120579)) sin 120579d120579
2 int
1205872
0
minus (lnP (120579)S (120579)) sin 120579d120579(6)
Where 120579 = 7 23 38 53 68 zenithal angles delimited by eachring in the fisheye lens of the sensor
119875 (120579) =
119861
119860
(7)
Expression (7) is the transmittance or probability of alight beam passing over the foliage without being interceptedmeasured by the ratio between the B readings obtained belowthe canopy and the A readings before being intercepted bythe canopy
The effective leaf index L119890
is finality computed by
Le = 119871 lowast 120596119886119901119901 (8)
To obtain precise measurements with LAI-2200 it isnecessary to capture only the diffuse energy To avoid or atleast strongly reduce the direct energy captured by the sensorthe readings were made using view cap 45∘ under cloudysky or at dawn or twilight verifying the luminous stabilitythroughout and after the measurements which was madeby the data tendency analysis after the measurement Thedata with tendency were discardedThe readings were alwaysperformed on intermediate days of each month and theystarted inOctober 2011 with a reading in each of the collectornets always in the same position and direction There were 5nets in each permanent plot in a total of 120 netsThe LAI wascalculated for each permanent plot
25 NDVI Data Processing To generate NDVI an atmo-sphere correction model was used which would allow theuse of atmospheric pressure and relative humidity aimingto obtain the removal of the atmosphere interference with agreater local precision Three Landsat 5 images were selected(httpwwwinpebr) as they coincide with their final recep-tion complementing the period with 7 images of the IndianRemote Sensing Satellite (IRS) a sensor with characteristicscloser to LandsatThemain differences between them are thespatial (30m and 24m pixel) and radiometric (8 and 7 bits)resolution and a slim difference in the range of the spectralbands All images arewith visibility above 10 km (informationfromAir Traffic Control Service bulletin of Confins Airport)
The atmospheric correction was performed with anATMOSC module of Idrisi Taiga using the Cost Model
The Dn Raze [36] was obtained through a procedurewhich uses a band frequency table which is available onthe electronic spreadsheet at httpwwwdsrinpebrdown-loadshtm based on the atmosphere correction method ofChaves [37]
The atmosphere optical dimension is the sum of thecomponents Rayleigh scattering aerosols water vapor andtypical ozone Aerosols and typical ozone were not computeddue to lack of data The Rayleigh scattering was adjustedto local height with atmospheric pressure data using thefollowing expression
where 119886 = 005 for band 119877 and 119886 = 001 for band NIR ([38]quoted by [39])
The water vapor component was obtained for the nearinfrared (NIR) in function of the relative humidity by thelinear relation [39]
Water vapor = 00012 lowast RH + 0016 (10)
International Journal of Ecology 5
Table 3 Coefficients of determination (1198772 adjusted) for polynomial equations in the observed dry leaf weight (gsdotmminus2) LAI (m2sdotmminus2) and
NDVI data and respective minimum and maximum points by fragment (sequential day from 01012011)
The geometrical correction was performed using theGeotiff Examiner software as well as the assistance of thegraphic software Inkscape using a reference point to dislocateeach image As the images contain deformation correctionsthis form of correction was more precise compared to thepolynomial corrections even with RMS smaller than 12pixel In order to extract NDVI in each permanent plot arectangle of 9 pixels centralized in the central point of thepermanent plot was digitalized
26 Monitoring Period The three variables dry leaveweightsdotmminus2 LAI (m2sdotmminus2) and NDVI were analyzed in thefunction of the day of the period from 01012011 to 2012The analyses period started in August 2011 (when the leavesstart to accumulate) until September 2012 (Table 2) Fromthe 1560 visits to the nets in this period 58 correspondto nets damaged by thick fallen branches or damaged byanimals Fourteen were discarded because they correspondto outliers So 1488 leaf weight data by period were usedOf the 289 LAI values calculated per permanent plot 7were discarded as outliers And of the 276 NDVI extractedvalues 3 were classified as outliers During the collectionperiods the accumulation of leaveswas not exactly of 30 dayswith a maximum fluctuation of 36 days between Octoberand November 2011 because of the unexpected rains in thecollection period The decision was made in order not tointerpolate the data
3 Results
31 LAI and Litterfall LAI has a strong relationship with thedeciduousness in this forest typology as its value is the conse-quence of the quantity of leaves in the canopy so alterationsin this quantity directly reflect in the LAI Figure 2 presentsthe LAI behavior with the dry leaf weight (gsdotmminus2sdotperiodminus1)Except for Fragment 3 the tendency was similar to thecompared fragments The increase of LAI indicates that thesprouting and growth of leaves increases until a maximumsynchronic with the reduction of deciduousness when the
latter reaches a minimum and the reduction of leaf areastarts synchronic with the increase of the deciduousness Inthis stage the deciduousness drives the reduction of LAIbecause in this period the leaf production practically stops
32 LAI and NDVI The NDVI starts its increase beforeLAI (Figure 3) probably because of the influence of theunderstory which promptly responds to the first rain beforethe canopy stars sproutingThis understory does not count onLAI computation as the sensor is positioned above this veg-etation for the readings The NDVI also decreases before theLAI because before the leaf falls a gradual depigmentation ofchlorophyll takes place and the loss of cellular turgidity in thepalisade parenchyma of the leaves increases the reflection ofthe absorption band by the chlorophyll (red) and decreasesthe reflection band due to the reduction of the intracellularspaces and the turgidity of the cells (near infrared) Thisphenomenon reflects on the reduction of the NDVI beforethe deciduous response detected by the LAI Maki et al[17] conducted experiments in a cool-temperate deciduousforest with Modis data and LAI field measurements Theyhad similar results indicating that the understory influencesthe comparisons between LAI and NDVI and that in the leafsenescence period the discoloration of the leaves interferesin the NDVI results before it interferes in the LAI response
In Fragment 3 due to the subsurfacewater availability theNDVI results presented here have an intermediate tendencybetween the seasonal forest and the ombrophilous forest inthe study by Soudani et al [7] who observed the constancyof the NDVI in an ombrophilous forest of Ghana throughoutthe year and abrupt changes in the NDVI in temperateclimate forests in the north of Europe driven by the climaticseasonality by temperature
It is known that the behavior of the deciduousness ofthe LAI and NDVI is sigmoidal during a period beyond oneyear but in one year these behaviors can be fit in polynomialmodels The best adjustment for the deciduous behavior LAIand NDVI in the whole period of 13 months (September2011 to September 2012) was through the second-degree
6 International Journal of Ecology
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Table 3 Coefficients of determination (1198772 adjusted) for polynomial equations in the observed dry leaf weight (gsdotmminus2) LAI (m2sdotmminus2) and
NDVI data and respective minimum and maximum points by fragment (sequential day from 01012011)
The geometrical correction was performed using theGeotiff Examiner software as well as the assistance of thegraphic software Inkscape using a reference point to dislocateeach image As the images contain deformation correctionsthis form of correction was more precise compared to thepolynomial corrections even with RMS smaller than 12pixel In order to extract NDVI in each permanent plot arectangle of 9 pixels centralized in the central point of thepermanent plot was digitalized
26 Monitoring Period The three variables dry leaveweightsdotmminus2 LAI (m2sdotmminus2) and NDVI were analyzed in thefunction of the day of the period from 01012011 to 2012The analyses period started in August 2011 (when the leavesstart to accumulate) until September 2012 (Table 2) Fromthe 1560 visits to the nets in this period 58 correspondto nets damaged by thick fallen branches or damaged byanimals Fourteen were discarded because they correspondto outliers So 1488 leaf weight data by period were usedOf the 289 LAI values calculated per permanent plot 7were discarded as outliers And of the 276 NDVI extractedvalues 3 were classified as outliers During the collectionperiods the accumulation of leaveswas not exactly of 30 dayswith a maximum fluctuation of 36 days between Octoberand November 2011 because of the unexpected rains in thecollection period The decision was made in order not tointerpolate the data
3 Results
31 LAI and Litterfall LAI has a strong relationship with thedeciduousness in this forest typology as its value is the conse-quence of the quantity of leaves in the canopy so alterationsin this quantity directly reflect in the LAI Figure 2 presentsthe LAI behavior with the dry leaf weight (gsdotmminus2sdotperiodminus1)Except for Fragment 3 the tendency was similar to thecompared fragments The increase of LAI indicates that thesprouting and growth of leaves increases until a maximumsynchronic with the reduction of deciduousness when the
latter reaches a minimum and the reduction of leaf areastarts synchronic with the increase of the deciduousness Inthis stage the deciduousness drives the reduction of LAIbecause in this period the leaf production practically stops
32 LAI and NDVI The NDVI starts its increase beforeLAI (Figure 3) probably because of the influence of theunderstory which promptly responds to the first rain beforethe canopy stars sproutingThis understory does not count onLAI computation as the sensor is positioned above this veg-etation for the readings The NDVI also decreases before theLAI because before the leaf falls a gradual depigmentation ofchlorophyll takes place and the loss of cellular turgidity in thepalisade parenchyma of the leaves increases the reflection ofthe absorption band by the chlorophyll (red) and decreasesthe reflection band due to the reduction of the intracellularspaces and the turgidity of the cells (near infrared) Thisphenomenon reflects on the reduction of the NDVI beforethe deciduous response detected by the LAI Maki et al[17] conducted experiments in a cool-temperate deciduousforest with Modis data and LAI field measurements Theyhad similar results indicating that the understory influencesthe comparisons between LAI and NDVI and that in the leafsenescence period the discoloration of the leaves interferesin the NDVI results before it interferes in the LAI response
In Fragment 3 due to the subsurfacewater availability theNDVI results presented here have an intermediate tendencybetween the seasonal forest and the ombrophilous forest inthe study by Soudani et al [7] who observed the constancyof the NDVI in an ombrophilous forest of Ghana throughoutthe year and abrupt changes in the NDVI in temperateclimate forests in the north of Europe driven by the climaticseasonality by temperature
It is known that the behavior of the deciduousness ofthe LAI and NDVI is sigmoidal during a period beyond oneyear but in one year these behaviors can be fit in polynomialmodels The best adjustment for the deciduous behavior LAIand NDVI in the whole period of 13 months (September2011 to September 2012) was through the second-degree
6 International Journal of Ecology
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
polynomial model The coefficients of determination andtheir minimum and maximum points which are the periodsof curves inversion are shown in Table 3
An average displacement between the NDVI and the LAImay be estimated using its maximum points around 70 dayswhich means that the NDVI started its decrease about twomonths and 10 days before the LAI in this period of theexperiment (considering all the fragments with exceptionof Fragment 3 which presents a deciduous behavior with adifferent pattern) It could be verified that the LAI and theNDVI in seasonal forests present a temporal displacementas the LAI will only be affected later when the leaves fallThus this is one of the reasons for the temporal relationshipof two stages between LAI and NDVI (Figure 4)The greatestcomplexity in the temporal relation between the NDVI andthe LAIwas also noticed byGupta et al [40] in the cultivationof onions and wheat who suggested polynomial models ofgreater order to explain this behavior
33 Litterfall LAI and NDVI between the Fragments Anevaluation of consistency between the variables may beobtained by comparing the position of the curves among thefragments (Figures 7 8 and 9) The NDVI is the descriptorvariable with lower precision due to limitations of spatial andradiometric resolution cloudiness and especially the atmo-spheric interferences when the analyses are temporal Its useis justified by the acquisition by remote sensing with no needfor field acquisition For the NDVI in Figure 9 a behaviorsimilar to the curves obtained with LAI is observed howeverone inconsistency was observed The NDVI for Fragment1 was one of the highest indexes and is close to that ofFragment 3 which is inconsistent with its expected leaf areaThe other fragments presented compatible correspondencebetween deciduous LAI and NDVI values
The deciduousness relationshipwith each descriptor vari-able (water balance LAI and NDVI) is linear according tothe results presented in Table 4 Its behavior depends on thewater balance and it changes the leaf area and the NDVI forseasonal forest sites in the analyzed period
34 Water Balance and Litterfall The region where thisexperiment was carried out has a tropical climatic seasonalityIts rainy season ranges from October to March and the dryseason ranges from April to September Figure 5 presentsthe water balance for the study period The water deficitstarted on 05142011 and reached its maximum of minus156mmon 09132011 recovering its water excess on 12142011 Themaximum excess occurred on 01162012 574mm resumingthe water deficit on 05142012
The deciduousness responded to the water reserve in thesoil with the inversion of the rate of average change in thefall on day 427 030212 (Table 3) The middle of the rainyperiod from the start of the excess (day 348) (2142011) untilthe start of the deficit (day 500) on 05142012 was on day424 This indicates that the resuming of the deciduousnessincrease occurs from the middle of the water excess periodand presents small variation among the fragments with the
exception of Fragment 3 This site does not express thisbehavior due to a shallow water table which makes it littlesensitive to the rain reduction Therefore it has the tendencyto present a perennial forest with a greater LAI (Figure 9)and an irregular litterfall compared to the other fragments(Figure 2)
The Pearsonrsquos meaningful correlations of the water bal-ance variables with the deciduous are presented in Table 4In order to demonstrate deceleration and acceleration of leafdeposition in function of the soil water reserves the graphicsof the leaf loss variation rate (gsdotmminus2sdotperiodminus1) are presentedin Figure 6 The leaf loss variation rate was obtained by theapplication of the 1st derivative to the polynomial functionincluding the water deficit and excess When the rainy seasonstarts (October) the water deficit is quickly reduced andthe fall rate deceleration takes place In the water excessstage the tendency change corresponds with a nonmodeledvariation of the phenomenon Small water deficit intervalsin the rainy period appreciably increase the deciduousnesswhich is not explained by the polynomial function andwhichis also observed in the leaf fall graphic in the LAI function(Figure 2)
The behavior is again explained by the end of the waterexcess when the maximum point of the function is reachedThe deciduousness accelerates almost until its stabilizationat the end of the dry period in September when the waterdeficiency is at its maximum There is a content of water inthe soil between the field capacity and the permanent wiltingpoint which starts this inversion of the curve and that mayvary according to the soil class and depth
35 Regression Model Although there are variations in thewater deficit in each site determined by the soil classsoil depth and topographic position among others themeasurement of the water deficit site by site was not possibleand would not have a practical effect on the usefulness of themodel for other sites Similarly climatological variables suchas the wind speed which influence the deciduousness werenot included since the integration of daily speed data wouldbe another factor to complicate the data acquisition for theestimation The model must be precise and also practical
For the linear regression model fitness tests the data setof Fragment 3 was excluded To reduce the heterogeneity ofthe variance of the dry leaf weight variable the data weretransformed to the logarithmic form ln(Dlw) The modelbelow was reached after the following steps analysis ofTable 4 execution of the regression through the stepwisebackward and forward procedures with the explanatoryvariables (LAI NDVI def exc ppt and etr) and removalof the excess water (exc) and the real evatranspiration (etr)due to its high correlation with precipitation (ppt) and waterdeficiency (def) respectively and to avoid multicollinearityproblems Consider the following
Dlw (g sdotmminus2 sdotmonthminus1)
= exp (853 + 017 lowast LAI minus 75 lowastNDVI
minus001 lowast def + 000048 lowast ppt)
(11)
International Journal of Ecology 7
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
222426283032343638404244464850
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
1015202530354045
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2030405060708090
100
464850525456586062646668
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
100
3436384042444648505254565860
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
102030405060708090
3035404550556065
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
0102030405060708090
100
2530354045505560
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(f)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
010203040506070
202224262830323436384042
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
020406080
100120140
4042444648505254565860626466
Dry
leaf
wei
ght (
gmiddotmminus2)
LAI (
m2middotm
minus2)
(h)
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Figure 2 Relation between dry leaf weight (gsdotmminus2sdotperiodminus1) and LAI (m2sdotmminus2) for the fragments
8 International Journal of Ecology
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
222426283032343638404244464850
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(a)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
1015202530354045
072074076078080082084086088090092
ND
VI
LAI (
m2middotm
minus2)
(b)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
464850525456586062646668
084
086
088
090
092
094
096
ND
VI
LAI (
m2middotm
minus2)
(c)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3436384042444648505254565860
078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(d)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
3035404550556065
080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(e)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
2530354045505560
074076078080082084086088090092094
ND
VI
LAI (
m2middotm
minus2)
(f)
LAI (
m2middotm
minus2)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
202224262830323436384042
080082084086088090092094096
ND
VI
(g)
726
11
914
11
113
11
122
311
211
12
41
125
211
27
101
28
291
210
18
1212
71
2
Data
4042444648505254565860626466
084085086087088089090091092093
ND
VI
LAI (
m2middotm
minus2)
(h)
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Figure 3 Relations between LAI (m2sdotmminus2) and NDVI for the fragments
International Journal of Ecology 9
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Table 4 Pearson correlation (nsnonsignificant at 005 of probability 119899 = 85) Correlation of the LAI and the NDVI estimated data for the leafcollection dates using the polynomial equations adjusted to the LAI and the NDVI per fragment Fragment 3 was removed from this analysisbecause of a different tendency in relation to the other fragments
Dry leaf weight(gsdotmminus2periodminus1)
LAI(m2sdotmminus2) (119903 minus nir)(119903 + nir) Water deficiency
(mm)Water excess
(mm) Precipitation
psflh LAI NDVI DEF EXC ppt (mm)LAI
minus050NDVI
minus078 036DEF
minus081 065 052EXC
minus032 017
ns 038 043Ppt
minus042 021
ns 050 052 096ETR
minus075 061 059 085 021
ns 038
082083084085086087088089
09091092
0 2 4 6
ND
VI
LAI (m2middotmminus2)
(a)
076
078
08
082
084
086
088
09
092
0 1 2 3 4 5
ND
VI
LAI (m2middotmminus2)
(b)
Figure 4 Temporal relation between LAI and NDVI with data adjusted by polynomial curves for Fragments 1 and 2
0100200300400500600700
130
920
11
111
020
11
161
120
11
141
220
11
160
120
12
140
220
12
150
320
12
150
420
12
140
520
12
150
620
12
150
720
12
140
820
12
130
920
12
Date to collect leaves
minus200
minus100
Hyd
ric d
efici
ency
Hyd
ric ex
cess
(mm
)
Figure 5 Climatologic water balance of Thornthwaite FromSeptember 2011 to September 2012 Observe that the accumulationperiod is the same as the litterfall collection
The equation parameters are shown in Table 5 The 1198772adjustment was 802
The scatterplots are shown in Figures 10 11 and 12They indicate the model adjustment quality through thecomparison between the predicted and observed values the
residuals in function of the predicted data and the predictorvariables and the normality of the residuals
In Figure 10(a) most of the data is out of the 95 con-fidence interval however the data points present dispersionwithout bias along the 45∘ line The precision is raised by thecompensation within the yearly integration of the monthlypredictions In Figure 10(b) the residuals did not presentany tendency with the predicted variable value increaseindicating that there is not a serial correlation and theheterogeneity of the variances was reduced It is also verifiedthat the inserted variables are enough to predict the variableDlw In Figure 11 the normal distribution of residuals wasverified Figure 12 shows that for most of the values thereis no tendency in the plot of the predicted variables and theresiduals indicating that the proposed model is consistent
36 CO2
Fixation Assessment The predictions of the CO2
capture by the adjusted deciduousness dynamics equationfor a period of 12 months between 2011 and 2012 were com-pared to themeasured data for the fragments under study Bymeans of the sprouting process and the seasonal leaf growthby the leaf biomass production via photosynthesis and theposterior deposition of this biomass by the deciduousnessprocess an annual quantity of carbon is deposited in the soil
10 International Journal of Ecology
0 02 04 06 08
minus9
minus8
minus7
minus6
minus5
minus4
minus3
minus2
minus1
0
minus08 minus06 minus04 minus02
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
Hyd
ric d
efici
ency
(mm
)
(a)
0102030405060708090
0 02 04 06 08minus08 minus06 minus04 minus02
Hyd
ric ex
cess
(mm
)
Variation rate of the deciduous (gmiddotmminus2middotmonthminus1)
(b)
Figure 6 Relationship between the variation rate of the deciduousness with water deficiency and excess in Fragment 1
913
11
101
111
111
611
121
411
116
12
214
12
315
12
415
12
514
12
615
12
715
12
814
12
913
120
20
40
60
80
100
120
140
F1F2F3F4
F5F6F7F8
Dry
leaf
wei
ght (
gmiddotmminus2)
Data to collect leaves
Figure 7 Dry leaf weight (gsdotmminus2) with the second-degree polyno-mial tendency for each fragment
by the decaying leaves Thus for each period of 12 monthsincluding the sequence of rainy and water deficit seasons ameasurable amount ofCO
2
is captured by this forest typologyadding to the biomass increase in the trunks branches androots determined by the growth of the vegetation
The amount of deciduousness may vary in a year toyear basis according to the climatic conditions and to theecosystem characteristics Despite the continental events asEl Nino-Southern Oscillation (ENSO) that has influencedBrazilrsquos climatic regime [41] especially the pluviometricdistribution the water balance is expected to be the variableresponsible for the year to year variations
The estimation of annual leaf deposition was accurateaccording to the totalization of the monthly data (Table 6)The biggest error 25 in Fragment 8 may be verifiedby observing the dispersion of the LAI and the NDVI inFigure 3 Concerning the fixation ofCO
2
the smallest capturewas in Fragment 7 with 37Mgsdothaminus1sdotyrminus1 and the biggest was77Mgsdothaminus1sdotyrminus1 in Fragment 8 The estimated data consider
191
011
161
111
121
211
150
112
100
212
130
312
170
412
080
512
200
612
130
712
140
812
130
912
171
012
1
2
3
4
5
6
7
F1F2F3F4
F5F6F7F8
LAI (
m2middotm
minus2)
Data
Figure 8 LAI (m2sdotmminus2) with second-degree polynomial tendencyfor each fragment
Table 5 Values (coef) standard errors of coef (EP) statisticsof Studentrsquos test (119905) probability of significances (119875-level) of theequation coefficients for the water deficiency in mm (def) NDVILAI and precipitation in mm (ppt) variables
Fragment 7 with the smallest capture of 37Mgsdothaminus1sdotyrminus1also but the largest are Fragments 4 and 6 with 64 and63Mgsdothaminus1sdotyrminus1 respectively Fragment 8 was not so easy tomodel because it has a riparian forest characteristic thereforeit presented more error
International Journal of Ecology 11
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Table 6 Observed and estimated data of dry leaf weight (Dlw) error of estimation () and fixed carbon in the leaves (C) considering 423of the biomass (average value of contents in leaves of forest species is obtained byWatzlawick et al 2011 [31]) capturedCO2 by the equivalenceof atomic weight between C (12 g) and CO2 (44 g) with the observed data and that estimated by the model
Figure 9 NDVI with second-degree polynomial tendency for eachfragment
4 Discussion
In plant species the phenology is divided into sproutingblooming fructification senescence and leaf fall stages Thesprouting or leaf renewal and the senescence or leaf fallreflect in the LAI and in the NDVI in seasonal forests Somedeciduous forest typology has three stagesThe first is the leafproduction period followed by the second with a stable leafarea period and the third is the senescence (leaf fall) Wanget al [16] observed this three stage pattern In this presentwork the stable leaf area period was not observed
Another point is that there is divergence between authorswith respect to the linear or nonlinear relationships betweenLAI and NDVI [12 16 21 23] In this study the relationshipis nonlinear and the tendency changes according to the stage(Figure 4)
Soudani et al [7] verified the annual NDVI behavior withhigh precision which was obtained on a daily basis in the
field in seasonal forest sites (European beech)They observedtwo stages Firstly it was the leaf season from mid springuntil October during which the sprouting development andmaturation take place The second stage is the dormancystage from the end of October until the winter This is whenthe yellowing the senescence and the leaf fall take placeThe leaf season starts with a NDVI jump from minimumto maximum in 26 to 30 days followed by a slight decreaseduring the summer period when another brusque changeoccurs the reduction is in the beginning of the fall Thebehavior presented by Soudani et al [7] is different fromthat predicted by us but in their work did not there test ofdynamics of the South American forest sites On the otherhand our NDVI data follow the LAI measured in the fieldwith a displacement explained by the distance between theyellowing of the leaves and their posterior fall
A future investigation with a spectroradiometer and agreater temporal resolution would confirm the tendency inforest sites with seasonality by water regime
By modeling the deciduous dynamics behavior LAI andNDVI good adjustments were obtained with the polynomialequations for a period of 13 months The relationship of LAIand NDVI adjusted by these curves shows a behavior similarto that found in Figure 6 inWang et al [16] referring to patchAB and CD with the leaf production stage with a faster LAIascension and relatively slowNDVI and the deciduous stagewith gradual reduction in both variables
By calculating the water balance for the leaf accumulationperiods a good correlation was achieved with the deciduous-ness change rates It was verified that when the rainy seasonends the deciduousness rate presents a strong correlationwith the water deficit increase as shown in Figure 6 Thisevidence was a key to choose the variable ldquowater deficitrdquo toexplain deciduous dynamics The water excess is a variablewith little explanatory capability in this process since thefluctuation of thewater excess above the soil water availabilitydoes not interfere in the plants response and in the waythe deciduousness is driven by the water deficiency In theconstruction of a predictive model for the deciduousnessdynamics in the seasonal forest driven by the rainfall regimethe LAI and the NDVI complement themselves as predictivevariables It is not possible to eliminate the LAI in this model
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Figure 10 (a) Observed data in function of the predicted data for the adjusted equation (b) distribution of residuals in function of thepredicted data PS logarithmic data
00 02 04 06 08Residuals
0
1
2
3
Expe
cted
nor
mal
val
ue
minus10 minus08 minus06 minus04 minus02
minus3
minus2
minus1
Figure 11 Presupposition of the normality of errors
The use of medium resolution orbital images for suchresearch succeeds for many forest fragment sizes Howeverthey demand a great effort to reach consistent atmosphericcorrections The products of the MODIS sensor have incom-patible resolution with many forest fragments In this exper-iment it was not possible to use them However in largerforest areas its applicationmay go without LAImeasurementin the field
Another observation refers to the LAI calculation fromNDVIMODIS based on the results presented in Figure 5 inPotithep et al [29] The displacement between LAI based onVIMODIS and the LAI in situ presented in the graphic maybe due to the calculation manner of the variable that hasorigin in vegetation index LAI is the leaf area detector withfew variations in relation to its physiologic state senescencevigor and is more sensitive to leaf fall while VI is sensitiveto the chlorophyll degradation with the yellowing anddiscoloration of the leaf before its fall Another issue is that
the startup of the LAIMODIS anticipates itself to LAI in situPotithep et al [29] attributed this to the forest soil interferencedue to herbaceous sprouting justifying the early LAI growthvalues generated byMODIS first than the real LAI valueThesame delay behavior between LAIMODIS and LAI in situ wasobserved by Ahl et al [42]
One of the advantages of making predictions by periods(the period adopted was monthly) using a model withacceptable dispersion however without bias is the compen-sation of errors in the totalization of the results for a longerperiod This raises the precision which was confirmed bythe annual CO
2
fixation estimate in each fragment Smallerperiods of 15 days and longer continuous monitoring longerthan a year are recommended to validate this model
The equation was developed for semideciduous seasonalforests of the Atlantic Forest with deciduousness driven bythe tropical climatic seasonality with similar parameters tothose measured in these sites Its objective is to estimate theannual production of the litterfall leaves which constitutesanother CO
2
capture mechanism that is not yet accountedfor in the carbon sequestration projects additional to thecontinuous increase of the aerial and root biomass
5 Conclusions
The deciduousness phenomenon in semideciduous seasonalforest was measured and relationships between the climaticbiophysics and orbital variables were found allowing themodeling and acquisition of leaf fall and annual CO
2
captureestimates The possible use of the quadratic model for thedeciduousness behavior was confirmed for short periodsthose of 13 months For longer periods the sigmoidal modelsare adequateThe hypotheses of the detection andmodeler ofthe inverted relation of the LAI and NDVI with the decid-uousness were confirmed Temporal relationships betweenNDVI and LAI were revealed indicating that the correlation
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Figure 12 Scatter of the residuals with LAI NDVI water deficiency and ppt
is nonlinear The NDVI is showed to be correlated with thedeciduousness as an explanatory variable For the perennialforest site this perspective of prediction was not successfuldue to the subsurface water availability This procedure toestimate the annual production of renewed leaves with orbitaland climatological data is not yet viable on a large scalefor a large part of the forest fragments due to the necessityof the LAI measurement In the MODIS use hypothesis inthese situations there is a limitation to the fragments withareas noncompatible with their resolution The modelingof the litterfall production with the intention to contributeto the study of nutrient cycling carbon sequestration andforest fragment biomass productive capacity is possible ona small scale However it still needs validation Deciduousseasonal forest fragments are able to capture between 4 to8 tons of CO
2
per hectareyear only due to the deciduous-ness phenomenon which will depend on the regenerationstage and forest conservation besides the other biophysicalfactors
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This study is a result of Project CNPq 5618642010-1 ldquoParam-eters of forest fragmentation as subsidy for the environmentquality and recovery of degraded environmentsrdquo which is asponsor of this research The authors would like to thankthe intern team who participated in the acquisition ofthe data and made the monitoring possible Marielle RaidGabriel Miranda Samara E F Carvalho Marcos Chamonand the botanists Andrea Fonseca Silva and Morgana FlaviaRodrigues Rabelo of Epamig Herbario for the identificationof the species Douglas Soares Felipe Guimaraes and DaulerGomes that helped me on instalation and some mensura-ments of this forest inventory and Antonio Claudio da Silva
14 International Journal of Ecology
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
Barros and Ross Thomas for reviewing the language in thismanuscript
References
[1] E J Brun M V Schumacher and S Vaccaro ldquoProducao deserapilheira e devolucao de nutrientes em tres fases sucessionaisde uma Floresta Estacional Decidual no municıpio de SantaTereza (RS)rdquo in Anais do Simposio de Fertilizacao e NutricaoFlorestal Piracicaba p 348 364ESALQ Piracicaba Brazil 1999
[2] K H Feger and S Raspe ldquoOkosystemforschung im chwar-zwald auswirkungen von atmogenen eintragen und Restabil-isierungsmassnahmen auf den Wasser-und Stoffhaushalt vonFichtenwaldernrdquo in Verbundprojekt Arinus S Raspe K HFeger and H W Zotil Eds pp 1ndash18 Umweltforschung inBaden-Wurttenberg Landsberg Germany 1998
[3] G C Cunha Aspectos da ciclagem de nutrientes em diferentesfases sucessionais de uma floresta estacional do Rio Grande doSul [Dissertation] Escola Superior de Agricultura ldquoLuiz deQueirozrdquo Piracicaba Brazil 1997
[4] F M S Moreira and J O Siqueira Microbiologia e bioquımicado solo UFLA Lavras Brazil 2002
[5] L R Barichello M V Schumacher H L M Vogel and J SDallago ldquoQuantificacao dos nutrientes no solo e serapilheirade diferentes estagios sucessionais em um sistema de agricul-tura migratoriardquo 2000 Resumos expandidos da Reuniao SulBrasileira de Ciencia do Solo Pelotas (CD-ROM)
[6] F Poggiani and M V Schumacher ldquoCiclagem de nutrientesem florestas nativasrdquo in Nutricao e fertilizacao florestal J LM Goncalves and V Benedetti Eds pp 287ndash308 Instituto dePesquisas Florestais Piracicaba Brazil 2000
[7] K Soudani G Hmimina N Delpierre et al ldquoGround-basedNetwork ofNDVImeasurements for tracking temporal dynam-ics of canopy structure and vegetation phenology in differentbiomesrdquo Remote Sensing of Environment vol 123 pp 234ndash2452012
[8] F G Konig M V Schumacher E J Brun and I SelingldquoAvaliacao da sazonalidade da producao de serapilheira numafloresta Estacional Decidual no municıpio de Santa Maria-RSrdquoRevista Arvore vol 26 no 4 pp 429ndash435 2002
[9] M Ludeke A Janecek and G H Kohlmaier ldquoModellingthe seasonal CO
2
uptake by land vegetation using the globalvegetation indexrdquo Tellus B vol 43 no 2 pp 188ndash196 1991
[10] C J Silva F A Lobo M E Bleich and L SanchesldquoContribuicao de folhas na formacao da serrapilheira e noretorno de nutrientes em floresta de transicao no norte deMatoGrossordquo Acta Amazonica vol 39 no 3 pp 591ndash600 2009
[11] X Zhang M A Friedl C B Schaaf et al ldquoMonitoring vegeta-tion phenology usingMODISrdquo Remote Sensing of Environmentvol 84 no 3 pp 471ndash475 2003
[12] R Fensholt I Sandholt and M S Rasmussen ldquoEvaluationof MODIS LAI fAPAR and the relation between fAPAR andNDVI in a semi-arid environment using in situ measurementsrdquoRemote Sensing of Environment vol 91 no 3-4 pp 490ndash5072004
[13] X ZhangMA Friedl C B Schaaf andAH Strahler ldquoClimatecontrols on vegetation phenological patterns in northern mid-and high latitudes inferred from MODIS datardquo Global ChangeBiology vol 10 no 7 pp 1133ndash1145 2004
[14] MDoiron P Legagneux G Gauthier and E Levesque ldquoBroad-scale satellite Normalized Difference Vegetation Index data
predict plant biomass and peak date of nitrogen concentrationin Arctic tundra vegetationrdquo Applied Vegetation Science vol 16pp 343ndash351 2013
[15] L Fan Y Gao H Bruck and C Bernhofer ldquoInvestigating therelationship between NDVI and LAI in semi-arid grassland inInner Mongolia using in-situ measurementsrdquo Theoretical andApplied Climatology vol 95 no 1-2 pp 151ndash156 2009
[16] Q Wang S Adiku J Tenhunen and A Granier ldquoOn therelationship of NDVI with leaf area index in a deciduous forestsiterdquo Remote Sensing of Environment vol 94 no 2 pp 244ndash2552005
[17] M Maki K Nishida N Saigusa and T Akiyama ldquoEvaluationof the relationship between NDVI and LAI in cool-temperatedeciduous forestrdquo in Asian Association on Remote SensingProceedings of the Asian Conference on Remote Sensing andAsian Space Conference Ha Noi Vietnam 2005
[18] A C Xavier and C A Vettorazzi ldquoMonitoring leaf area indexat watershed level through NDVI from landsat-7ETM+ datardquoScientia Agricola vol 61 no 3 pp 243ndash252 2004
[19] T C C Costa L J O Accioly M A J Oliveira N Burgos andF H B Silva ldquoPhytomass mapping of the ldquoSerido Caatingardquovegetation by the plant area and the normalized differencevegetation indecesrdquo Scientia Agricola vol 59 no 4 pp 707ndash7152002
[20] D P Turner W B Cohen R E Kennedy K S Fassnachtand J M Briggs ldquoRelationships between leaf area index andLandsat TM spectral vegetation indices across three temperatezone sitesrdquo Remote Sensing of Environment vol 70 no 1 pp52ndash68 1999
[21] T N Carlson andD A Ripley ldquoOn the relation between NDVIfractional vegetation cover and leaf area indexrdquo Remote Sensingof Environment vol 62 no 3 pp 241ndash252 1997
[22] S Amaral J V Soares D S Alves et al ldquoRelacoes entre Indicede Area Foliar (LAI) Area Basal e Indice de Vegetacao (NDVI)em relacao a diferentes estagios de crescimento secundario naFloresta Amazonica em Rondoniardquo in Anais do VIII SimposioBrasileiro de Sensoriamento Remoto pp 485ndash489 INPE SaoJose dos Campos 1996
[23] M A Friedl F W Davis J Michaelsen and M A MoritzldquoScaling and uncertainty in the relationship between the NDVIand land surface biophysical variables an analysis using ascene simulation model and data from FIFErdquo Remote Sensingof Environment vol 54 no 3 pp 233ndash246 1995
[24] M A Spanner L L Pierce S W Running and D L PetersonldquoThe seasonality of AVHRR data of temperate coniferousforests relationship with leaf area indexrdquo Remote Sensing ofEnvironment vol 33 no 2 pp 97ndash112 1990
[25] Li-Cor LAI-2200 Plant Canopy Analyzer Instruction ManualLi-Cor Biosciences Lincoln UK 3rd edition 2011
[26] G Zheng and L M Moskal ldquoRetrieving leaf area index (LAI)using remote sensing theories methods and sensorsrdquo Sensorsvol 9 pp 2719ndash2745 2009
[27] J M Welles and S Cohen ldquoCanopy structure measurementby gap fraction analysis using commercial instrumentationrdquoJournal of Experimental Botany vol 47 no 302 pp 1335ndash13421996
[28] S T Gower C J Kucharik and J M Norman ldquoDirectand indirect estimation of leaf area index f(APAR) and netprimary production of terrestrial ecosystemsrdquo Remote Sensingof Environment vol 70 no 1 pp 29ndash51 1999
[29] S Potithep S Nagai K N Nasahara H Moraoka and RSuzuki ldquoTwo separate periods of the LAI-Vis relationships
International Journal of Ecology 15
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
using in situ measurements in a deciduous broadleaf forestrdquoAgricultural and Forest Meteorology vol 169 pp 148ndash155 2013
[30] M Kale S Singh and P S Roy ldquoEstimation of Leaf AreaIndex in dry deciduous forests from IRS-WiFS in central IndiardquoInternational Journal of Remote Sensing vol 26 no 21 pp 4855ndash4867 2005
[31] L F Watzlawick A A Ebling A L Rodrigues Q J I Veresand A M Lima ldquoVariacao nos teores de carbono organicoem especies arboreas da floresta ombrofila mistardquo Floresta eAmbiente vol 18 no 3 pp 248ndash258 2011
[32] USDA ldquoSoil taxonomy a basic system of soil classification formaking and interpreting soil surveysrdquo inAgriculture Handbook1999
[33] R G Allen L S Pereira D Raes and M Smith ldquoCrop evap-otranspiration guidelines for computing crop water require-mentsrdquo Irrigation and Drainage Paper 56 FAO Rome Italy1998
[34] A R Pereira L R Angelocci and P C Sentelhas Agrom-eteorologia fundamentos e aplicacoes praticas AgropecuariaGuaıba Brazil 2002
[35] J B Miller ldquoA formula for average foliage densityrdquo AustralianJournal of Botany vol 15 pp 141ndash144 1967
[36] P S Chavez Jr ldquoAn improveddark-object subtraction techniquefor atmospheric scattering correction of multispectral datardquoRemote Sensing of Environment vol 24 no 3 pp 459ndash479 1988
[37] S Gurtler J C N Epiphanio A J B Luiz and A R FormaggioldquoPlanilha Eletronica para o calculo da reflectancia em imagensTM e ETM+ Landsatrdquo Revista Brasileira De Cartografia vol 57no 2 pp 162ndash167 2005
[38] B C Forster ldquoDerivation of atmospheric correction proceduresfor LANDSAT MSS with particular reference to urban datardquoInternational Journal of Remote Sensing vol 5 no 5 pp 799ndash817 1984
[39] J R Eastman ldquoATMOSCrdquo IDRISI Help System Accessed inIDRISI Taiga Clark University Worcester UK 2009
[40] R K Gupta T S Prasad andD Vijayan ldquoRelationship betweenLAI and NDVI for IRS LISS and LANDSAT TM bandsrdquoAdvances in Space Research vol 26 no 7 pp 1047ndash1050 2000
[41] L M T Oliveira G B Franca R M Nicacio et al ldquoA studyof the El Nino-Southern oscillation influence on vegetationindices in Brazil using time series analysis from 1995 to 1999rdquoInternational Journal of Remote Sensing vol 31 no 2 pp 423ndash437 2010
[42] D E Ahl S T Gower S N Burrows N V Shabanov RB Myneni and Y Knyazikhin ldquoMonitoring spring canopyphenology of a deciduous broadleaf forest using MODISrdquoRemote Sensing of Environment vol 104 no 1 pp 88ndash95 2006
![GeneticVariationinPastandCurrentLandscapes:Conservation ...downloads.hindawi.com/journals/ijecol/2010/503759.pdf · For the endemic-rich Florida scrub [35, 36], such genetic patterns](https://static.documents.pub/doc/80x56/5fa234cd23dde9423a799d3c/geneticvariationinpastandcurrentlandscapesconservation-for-the-endemic-rich.jpg)
