at SciVerse ScienceDirect

Acta Oecologica 50 (2013) 1e9

Contents lists available

Acta Oecologica

journal homepage: www.elsevier .com/locate/actoec

Original article

Regeneration of Rhizophora mucronata (Lamk.) in degraded mangroveforest: Lessons from point pattern analyses of local tree interactions

Adewole O. Olagoke a,c,*, Jared O. Bosire b, Uta Berger c

a Federal University of Technology, Department of Forestry and Wood Technology, P.M.B. 704, Akure, NigeriabKenya Marine and Fisheries Research Institute (KMFRI), P.O. Box 81651, Mombasa, KenyacDepartment of Forest Biometry and Systems Analysis, Institute of Forest Growth and Forest Computer Sciences, Technische Universitaet Dresden,Postfach 1117, 01735 Tharandt, Germany

a r t i c l e i n f o

Article history:Received 16 November 2012Accepted 2 April 2013Available online

Keywords:Degraded mangrovePoint pattern analysisPropagule dispersalSpatial structureRegeneration

* Corresponding author. Federal University ofForestry and Wood Technology, P.M.B. 704, Akure, Nig

E-mail addresses: adewoleolagoke@gmail.comkmfri.co.ke (J.O. Bosire), uta.berger@tu-dresden.de (U

1146-609X/$ e see front matter � 2013 Elsevier Mashttp://dx.doi.org/10.1016/j.actao.2013.04.001

a b s t r a c t

Spatial structural patterns emerging from local tree interactions influence growth, mortality andregeneration processes in forest ecosystems, and decoding them enhance the understanding of ecologicalmechanisms affecting forest regeneration. Point-Patterns analysis was applied for the very first time tomangrove ecology to explore the spatial structure of Rhizophora mucronata regeneration in a disturbedmangrove forest; and the pattern of associations of juvenileeadult trees. R. mucronata trees weremapped in plots of 50 m � 10 m located at the seaward, central and landward edge along 50 m widetransect in the forest, and the mapped patterns were analysed with pair correlation and mark-connectionfunctions. The population density of R. mucronata differed along the tidal gradient with the highestdensity in the central region, and the least near the shoreline. The study revealed that short distancepropagule dispersal, resulting in the establishment of juveniles in closed distance to the mother trees,might not be the driving force for distribution of this species. The spatial structural pattern of R.mucronata population along tidal gradient showed a characteristic spatial aggregation at small scale, butrandomly distributed as the distances become larger. There was a distinct spatial segregation betweenrecruits and adult trees, and hence spatially independent. Though, adulteadult trees associations did notshow a clear spatial segregation pattern; the recruiterecruit species associations exhibited significantclustering in space. Although habitat heterogeneity might be responsible for the local scale aggregationin this population, the effect of planteplant conspecific interactions is more probable to inform the long-term structure and dynamics of the population of R. mucronata, and ditto for the entire forest.

� 2013 Elsevier Masson SAS. All rights reserved.

1. Introduction

Mangrove forest species mostly thrive in harsh environment atthe landesea interface; and it is suggestive that environmentalconditions vary along landesea gradients, informing the speciesspecific natural constellation in optimally favourable zones basedon their habit and tolerance to prevailing conditions. Species dis-tribution and spatial structure in mangrove forests are informed bythe synergistic and/or antagonistic influences of temperature,salinity, tidal inundation, soil texture, pH, geomorphology, propa-gule predation, among others (Smith 1992). Coalescing with thesefactors in determining the forest spatial structure are recurrent

Technology, Department oferia. Tel.: þ234 8102919537.(A.O. Olagoke), jbosire@

. Berger).

son SAS. All rights reserved.

natural disturbances (wind, cyclones, flooding, tsunamis, etc.) andhuman-induced pressure (including purposeful forest manage-ment). Albeit, mangrove ecosystems continue to undergo structuraland compositional changes (Reddy and Roy, 2008); and the spatialpattern of the current vegetation provides a good indicator for theprocesses underlying such changes.

Earlier scientists articulated that the spatial patterns of treeconfiguration describing the structural characteristics of forestecosystems often reveal the pattern of successional development inplant communities (Levin, 1992; Getzin et al., 2008; Eichhorn,2010; etc.); and can provide insight into the intrinsic ecosystemprocesses and functional diversity in the system (Luo et al., 2012).The competition or facilitation processes resulting from tree asso-ciation and interactions do result in specific spatial patterns(Grabarnik and Sarkka, 2009). Getzin et al. (2006) also opined thatdifferent spatial patterns may reflect species abilities to surviveintra and inter-specific competition during succession. In essence,

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e92

forest spatial patterns, in combination with other factors, influencegrowth, mortality and regeneration processes, and thereby playsignificant role in forest ecosystem dynamics (Dieckmann et al.,2000; Grabarnik and Sarkka, 2009). Invariably, analysis of thespatial structural pattern of a given forest ecosystemmakes way forincreasing understanding of the history, ongoing ecological pro-cesses and functions, and future trajectories of forest development(Moorcroft et al., 2001).

The total area covered by mangroves in Kenya was once esti-mated to be 55,280 ha, but now 45,590 ha with an estimate coverloss of about 18% between 1985 and 2010 (Kirui et al., 2012). Withincreasing population, economic development, agriculturalexpansion, urbanization and technology and concomitant over-exploitation of mangrove resources, clear-cutting, hydrologicmodification and pollution among others, impacts of human ac-tivities on the mangrove forests become more pronounced (Bosireet al., 2008; Duke et al., 2007). Also, climate change anomalies (e.g.fluctuation in sea surface temperature and sea level rise, extensivesiltation and sedimentation, altered biogeochemical cycles, etc)work synergistically with anthropogenic pressure to furthercompromise the ecological integrity of mangroves and the intrinsicecosystem goods and services (Olagoke, 2012).

In the case of Tudor creek, the mangrove forest has sufferedsignificantly from combined impacts of recurrent human pressureand disturbances from climate-driven Indian Ocean Dipole (IOD) of1997/98 and 2006, as evidenced by the reduction in areal extent,change in species distribution, biomass loss, canopy damage andgap creations, and irrefutable alteration to stand structure; but alsowith evidence of viable natural regeneration (Mohamed et al.,2008; Olagoke, 2012). Surveys on vegetation structure indicatedthat Rhizophora mucronata was the dominant species in the forest(Olagoke, 2012 and literature cited therein). It is ipso factoworth toexamine the spatial pattern of the most abundant species,R. mucronata in this rejuvenating mangrove forest to decipher thepattern of regeneration, and the association of juvenile to adulttrees, while advancing possible explanation on the processes un-derlying local to large scale distribution of this species. With TidalSorting Hypothesis (TSH) in mangrove systems (Rabinowitz, 1978),or the general opinion that mangrove zonation results from speciesphysiologically aligning to edaphic and habitat conditions, and theexpected environmental heterogeneity along tidal gradient, avarying population distribution and spatial structural pattern alongtidal gradient was hypothesized in this disturbed forest.

This study therefore aimed at investigating and comparingspatial distribution pattern of R. mucronata population along tidalgradients, and determine whether: (1) the species demonstrates aconsistent regeneration and distribution pattern, or the spatialstructure is formed by tidal gradient; and (2) juvenile positivelybenefit in the neighbourhood of adult or co-juvenile tree, at fine tocoarse scale, irrespective of the position along the tidal gradient.

2. Methodology

2.1. Description of the investigated site

The study was conducted in a rejuvenating degraded mangroveforest in Tudor Creek (4�0200400S; 39�4002700E), located at thenorthwest ofMombasa Island in the coastal province of Kenya. Tudorcreek extends some 10 km inland with two main seasonal rivers,Kombeni and Tsalu, draining over 45,000 and 10,000 ha respectively.It is characterized by a 20mmean depth single narrow sinuous inletthat widens inland to a central 5 m depth basin, covering an area of637 ha and 2235 ha at low- and high water spring tides respectively(Mohamed et al., 2008). Sediments covering the forest mainlycomprise mud, and sand in some parts (ibid). Mangrove forest is

distributed over an area of 1465 ha, composed of R. mucronata, Avi-cennia marina and Sonneratia alba with no distinctive display ofspecies zonation (Mohamed et al., 2008; Olagoke, 2012). Map ofTudor Creek showing the study location and image of the typicalappearance of investigated site is shown in Fig. 1

The climate of the study is under the influence of semi annualpassage of the inter-tropical convergence zone (ITCZ) and themonsoons in two distinct seasons (Mohamed et al., 2008). TheNorthern Easterly Monsoon (NEM) and the Southern EasterlyMonsoon (SEM) manifest between December and March, and Mayand October respectively. The pattern of average monthly rainfalland temperature distribution from 2004 to 2011 in Mombasawhere the study area resides is presented in Fig. 2.

2.2. Plot location and field measurement

We carried out field survey and forest measurements to assessmangrove structure and spatial pattern across this site betweenFebruary and April 2012. For this purpose, we relied on the resultsof classified satellite SPOT imagery data of 2009 and informationfrom vegetation analyses of our field survey (Olagoke, 2012) inselecting a representative study location for entire forest.

For tree measurement, three rectangular plots of 50 m � 10 mwere selected in a 50 m wide transect along the tidal landeseagradient; the first (Seaward; 3�5805700S, 39�3604500E) and the thirdplots (Landward; 3�5900600S, 39�3604100E) located near the shorelineand landward side respectively with consideration for edge effects,and the second plot (Central; 3�5900100S, 39�3604300E) at the centre.The plots were partitioned into grids of 10 � 10 m to facilitate ac-curate and efficient measurement of tree coordinates. Assessmentof diameter at breast height (DBH,1.3m) and height, and the x and ycoordinates (reference to the right hand corner of the plot) of theposition of all R. mucronata trees having DBH �2.5 cm was donewithin the 10m� 10m sub-plots. Juvenile with height greater than40 cm and DBH lower than 2.5 cmwere also mapped based on theirrespective x and y coordinates in each quadrat. Measurement wasrestricted to this juvenile size class as sampling of tiny seedlingsmay lower the accuracy and efficiency in tree mapping. All mappedstems were systematically classified into two DBH size classes,including (1) recruits e all regeneration class with no measuredDBH and saplings below 5 cm DBH, and (2) adult trees with DBH�5 cm. This classification is based on subjective and arbitrary cut-off chosen to allow for comparisons between size classes.

2.3. Data analysis

2.3.1. Methodological approach for spatial statisticsSpatial patterns describing the general distribution of

R. mucronata, and the intra-specific associations at varying distancescales r (m) were explored with both univariate and bivariate pair-correlation functions (pcf), and mark connection functions (mcf)(Perry et al., 2006; Getzin et al., 2008). The pcf is a normalizeddistance-dependent density function that describes spatial re-lationships of neighbouring points, defined by the x, y position oftheir stems, or point types, defined by stem diameter, species type,growth stage, etc (Stoyan and Stoyan, 1994; Wiegand and Moloney,2004). It describes the probability of observing a pair of pointsseparated by a distance r, divided by the corresponding probabilityfor a Complete Spatial Randomness (CSR); and it is expressed as:

gðrÞ ¼ K 0ðrÞ2pr

for r � 0 (1)

where K 0ðrÞis the derivative of Ripley’s K-function (Ripley, 1977;Wiegand and Moloney, 2004; Wiegand et al., 2007; Law et al.,

Fig. 1. Map of the study location with imagery showing typical appearance of the investigated site (Modified fromMohamed et al., 2008). Source of imagery: Google Earth, Imagerydate: 21.03.2011, Access date: 23.10.2012.

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e9 3

2009). The shape or value of gðrÞ ¼ 1 denotes complete random-ness; gðrÞ > 1 denotes clustering and gðrÞ < 1 means regularity(Baddeley, 2010). The bivariate (second order statistic) pcf, g12(r)measures the strength of the relationship between point type “1”and type “2” (in this study, we refer to type “1” as the adult treesand type “2” as the juveniles). It is also related to the bivariate Kfunction K12(r) (Ripley, 1977); and it can be expressed as:

g12ðrÞ ¼ K 012ðrÞ2pr

(2)

Fig. 2. Climate information of Mombasa, based on historical data from 2004 to 2011.

The bivariate pcf thus will provide information about the spatialattraction or repulsion between the mother trees and the recruits.

Also, the mark connection functions (mcf) p12ðrÞmeasures thespatial correlation between discrete marks (e.g. species identity,size class, etc) of a point pattern (Illian et al., 2008). It represents theconditional probability to find two arbitrarily chosen points asjoined case (recruit, Adult) and control (Adult), at positions sepa-rated by distance r (Getzin et al., 2008; Illian et al., 2008). This teststatistic is expressed as:

p12 ¼ l1l2g12ðrÞðl1 þ l2Þ2gðrÞ

(3)

where l1 and l2 are the intensity of case and control (marks)respectively. As a plus, the mcf compares only the effect of pointepoint interactions to random labelling, and hence excluding thepossible large-scale heterogeneity. Random labelling assumes thatthe case and control are conditionally independent and identicallydistributed in given locations. Higher value of pij(r) than the ex-pected value of mcf interprets that large proportion of ij-pairs existsbeyond expected at random labelling and hence spatial aggregationof recruiterecruit, recruiteadult or adulteadult, as the case may be(Getzin et al., 2008; Illian et al., 2008).

Bias due to “edge effects” may arise because the point patternobservations were made within restricted spatial window, and wetherefore applied ‘‘border’’ and “isotropic” edge correction ap-proaches to counteract such effects (Diggle, 2003; Baddeley, 2010).The use of envelopes constructed from null model simulations isadopted to test if spatial patterns differ significantly from complete

Table 1Distribution of R. mucronata population and nearest neighbour (NN) distances be-tween individuals in the sampled plots.

Location Density (stems/ha) Max NNdistance (m)

Median NNdistance (m)

Adult Recruit Total Adult Recruit Adult Recruit

Seaward 15 150 165 1.70 10.70 0.50 0.22Central 19 472 491 5.60 1.80 0.51 0.20Landward e 344 344 e 1.90 e 0.22

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e94

spatial randomness (CSR), heterogeneous Poisson models, randomlabelling etc, as it may apply in each case (Eichhorn, 2010).Following edge correction for all spatial pattern analyses, envelopeswere calculated with 99 Monte Carlo simulations for the nullmodels and the confidence limits were obtained from the func-tions’ highest and lowest values. All cases were tested for signifi-cance using simulation envelope method (Grabarnik et al., 2011).

2.3.2. Scheme of spatial patterns analysesThe intensity function (density, sigma ¼ 15) of R. mucronata in

each of the mapped plots was plotted and the nearest neighbour(NN) statistic was conducted, to examine the population distribu-tion pattern. The contour plots were visually interpreted to connoteheterogeneity when the pattern has inconsistent intensity (Lawtonand Lawton, 2010).

Following the visualization of the intensity function, spatialdistribution patterns of the species’ population in each of theselected plots were analysed to determine whether the spatialagglomeration pattern of the species is consistent along the tidalgradient. Univariate homogenous pcf was applied to plots in thethree locationse seaward, central and landward, and the shape andg(r) values resulting curves were compared at varying scales. Thetest was conducted in 3 batches: (1) the entire population, (2) thepopulation of recruits, and (3) the population of adult in Seawardand Central plots, as no individual falls within the adult category in

Fig. 3. Spatial pattern of R. mucronata distribution (A) and co

the landward plot. Similarly, batches 2 and 3 were examined forconspecific spatial correlation of trees at different growth stages,taking null models of CSR for all plots. We analysed the recruit toadult relationship, considering the hypothesis that large treessuppress the growth of the recruits (Getzin et al. 2008; Lawton andLawton, 2010). This assumption was tested in seaward and centralplots with consideration for all pairs (Adult: recruit; Adult: Adultand recruit: recruit) using mark connection function. All analyseswere executed using in R 2.14.1 environment for statisticalcomputing (R Development Core Team, 2011), using the Spatstatpackage (Baddeley and Turner, 2005).

3. Results

The abundance of R. mucronata, classified into two growthstages, in each plot (adult and recruits) is presented in Table 1. Thepopulation densities of recruits were substantially higher than thatof adults in all locations, and no large size individual adult wasencountered in the landward plot where the population was pre-dominantly composed of recruits. The central plot had the highestjuvenile population and that of young adult trees, while theseaward plot which was characterized by the presence of old adultand large trees had the least juvenile population density within thesample plot. Harvested stump density varied significantly, with35 stumps/500 m2 at the seaward edge, 283 stumps/500 m2 at thecentral zone and the highest 319 stumps/500 m2 at the landwardedge.

Average distances between nearest neighbour (NN) for recruitswere 0.42 m, 0.25 m and 0.32 m for seaward, central and landwardplots, respectively; and 0.74m and 1.30 m for adults in seaward andcentral plots, respectively. The NN distances for the recruits differedsignificantly from the seaward to the landward edge (KruskalWallis test, P¼ 0.005). Also, the adult NN distance was significantlydifferent between the central and seaward plots (ManneWhitneyU-test, P¼ 0.025) within R. mucronata population. The NN distances

rresponding density plot based on intensity function (B).

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e9 5

differed significantly between adult and recruits in both seawardand central plots (P < 0.05).

Mapping of the spatial patterns of the population distributionwas done and the contour diagram was plotted based on the in-tensity function (Fig. 3A and B). The intensity function showed afairly evenness in trend in the seaward side. In contrast, the densityintensity functions varied in different parts of the other plots.Examining the general spatial distribution pattern of R. mucronatain Tudor creek, the results of the pcf generally showed a completedeviation from randomness (P < 0.05) but a clumping pattern atsmall scale (ca. 1.6 m), which approached randomness withincreasing scale. The univariate pcf for the seaward plot showed asignificant aggregation pattern up to the scale of 2.5 m (Fig. 4) incomparison to the central and landward plots, where the shape

Fig. 4. Pair correlation functions describing the spatial pattern of R. mucronata population insolid black line; CSR benchmark ¼ red dash line; Highest to lowest 95% confidence interval ¼referred to the web version of this article.)

showed randomness from less than 1.8 m distances (Fig. 4). The pcfof the adult trees showed slight deviation from the pattern com-mon to others, and exhibited spatial clustering which approachedrandomness only at the scale r of ca. 1.1e1.3 m in the seaward andcentral plots (Fig. 5).

Mark connection functions of the R. mucronata population inSeaward and Central plots were examined to establish the patternof seedling to adult relationships. As shown in Fig. 6, the pairs ofadulterecruit and recruiterecruit showed substantial deviationfrom the expected probability Pij(r) at random labelling while thepair of adulteadult only shows a fluctuation around a line corre-sponded to the random labelling in the seaward plot. The proba-bility that a recruit is found next to recruit for any random pair ishigher than expected, and this showed high degree of clustering at

Tudor creek [Seaward plot ¼ top, Central plot ¼middle and Landward plot ¼ base; g(r):grey]. (For interpretation of the references to colour in this figure legend, the reader is

Fig. 5. Pair correlation functions for the spatial pattern of R. mucronata adult trees. [g(r): solid black line; CSR benchmark ¼ red dash line; Highest to lowest 95% confidenceinterval ¼ grey.]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e96

the juvenile stage. There seems to be lower than expected proba-bility that of a chosen pair of trees, one is an adult and the other is arecruit (indication for segregation). Deviation from expectedprobability at random labelling was obtained in all pairs of treeassociations in the central plot (Fig. 7). The trend is similar to theresults obtained in the seaward plot except that the spatial aggre-gation in a pair of adult was higher than the expected at randomlabelling up to a scale r of 1.3e1.4 m. Recruits showed no spatialdependency on adult tree (Figs. 6 and 7). Albeit, some measures ofconsistency were generally revealed in the spatial patterns ofconspecific tree distribution, structure and interactions along tidalgradient in the study site.

4. Discussion

There is generally lack of information on the spatial pattern ofpopulation distribution and local interactions of R. mucronataspecies despite the ecological and economic importance of thisspecies especially in the ecosystem where it occurs. This studytherefore attempted to address this gap, and hence forms the firstcomprehensive species-specific study on the spatial structure ofmangrove species. Current analyses examined the population dis-tribution and spatial structure of R. mucronata along a tidal gradientto provide complementary information for mangrove forest struc-tural dynamics from the perspective of point pattern analysis ofconspecific interactions. This study is thus a valuable contributionto the understanding of the processes informing the forest struc-tural dynamics, and brings to the fore application of the method-ology of spatial point patterns analyses in mangrove ecology.

The results revealed a substantial difference in the populationdensity of R. mucronata in the studied plots along tidal gradient,

with the highest density in the central region. The density of ju-venile in the inland was also greater than that of the shoreline.Being a rejuvenating forest, the adult trees are very active in seedproduction (Olagoke, 2012), and the species benefits from regularhigh-energy tidal flow that disperse the seed farther away from themother trees. Sherman et al. (2000) stressed the importance ofcanopy disturbance in the re-colonization and subsequent dy-namics of a disturbed mangrove forest. Extensive canopy distur-bance that resulted in gap creations in the central and landwardplots could be the possible reason for massive establishment ofrecruits in the area.

A general low population of adult treewas observed because theforest is currently rejuvenating from a combination of past intenseanthropogenic pressure (owing to the proximity of the landwardedge to contiguous villages which enhances accessibility for treeharvesting and charcoal burning), and dieback that resulted fromthe impact of Indian Ocean dipole (IOD). It is noteworthy that largetrees (potential mother trees as seed source for re-colonization)that pre-dated the IOD events in 1997/98 and 2006 only lined theshore in the seaward plot; but it is clear the adult trees in otherplots were established after the last major disaster (Olagoke, 2012).In this case, the colonization of the central to landward sidesseemingly opposed the tidal sorting hypothesis (TSH) as empha-sized by (Jimenez and Sauter, 1991) being the key process con-trolling vertical distribution of mangrove species along a tidalflooding gradient. By the TSH, Rabinowitz (1978) put forward thatspecies with large propagules are sorted near the shoreline and arephysiologically adapted to establish at higher inundation region.Howbeit, the distribution of R. mucronata across the entire sitemight have benefitted from high spring tidal level of ca. 4 m; beingof the key factor shaping the forest structure (Smith 1992).

Fig. 6. Mark connection functions for the spatial interactions between pairs of R. mucronata recruit and adult trees in seaward plot. [P12(r) ¼ solid black line; Expected probability ofrandom labelling ¼ red dash line]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e9 7

If the idea that Rhizophora species can only disperse over a shortrange thus hold (Sousa et al., 2007), the regeneration ofR. mucronata in this forest might have followed a cascade pattern ofcolonization in the landward direction since no mature adult tree isfound in the landward plot. An alternative explanation is the pos-sibility that mature adult trees are extracted by thewood cutters fordomestic use or charcoal burning from the inland area. Accessibilityof this site to the ambient villages, and the presence of a number ofharvested stumps with decreasing intensity towards the sea in thestudy area are supportive of this argument. Characterizing thepopulation distribution pattern of R. mucronata in this forest is ageneral local scale spatial aggregation which assumes randomnesswith increasing scale. Thaxton et al. (2007) similarly found a non-random dispersion pattern in a 5-year post-hurricane recoveringforest, and they attributed such population behaviour in mangroveforest to the possible influence of the interaction of locally het-erogeneous edaphic conditions, dispersal variables, and localizedgap formations, among other factors.

The population structure in the three plots however demon-strated varying nearest neighbour (NN) distances among adulteadult and recruiterecruit in the forest. Farthest NN distance(about six fold higher compare to other plots) was noted amongthe recruits in the seaward area. This might be interpreted as thepresence of possible inhibition for seedling establishment nearthe shoreline which experiences higher tidal currents, and hinder

propagules from settling and establishment or possibly most ofthe propagules drift away from the sea-side in contrast to find-ings of Sousa et al. (2007). Another probable explanation is thefact that generally R. mucronata is not an inundation class I col-oniser due to long periods of submergence and infestation bybarnacles (Bosire et al., 2008). The results of the current studyhowever align with the thought that seedlings establish betterfarther away from the zone dominated by the adult of the samespecies (Rabinowitz, 1978); though the extent to which thisassertion holds in the absence of human disturbance was notascertained.

Although Baldwin et al. (2001) advanced that adult trees mayfacilitate the establishment and growth of new recruits positivelyby providing seed sources, they contrarily remarked that such treesmay have negative effect on the recruits by competing for limitedresources or rather occupying the potentially available space forcolonization. The results of this study show a distinct spatialsegregation between recruits and adult trees of R. mucronata.Contrastingly, Thaxton et al. (2007) found strong intensity of spatialaggregation between seedlings and surviving adult trees of Rhizo-phora species in a post-hurricane recovering forest in South Florida.This pattern however show a slight difference in the adulteadultrelationship, but a rather completely opposite in the recruiterecruitassociations. It is therefore plausible that there might be possiblecompetition for resources and in most cases adult trees out-

Fig. 7. Mark connection functions for the spatial interactions between pairs of R. mucronata recruit and adult trees in central plot. [P12(r) ¼ solid black line; Expected probability ofrandom labelling ¼ red dash line]. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e98

compete the juvenile, or simply put a negative influence ofshadow-effect.

We would like to indicate a need for caution in the interpreta-tion of our results, given to some identified limitations in the scopeof our field survey, and widely defined limitations of inferringprocesses from patterns (for instance, a pattern may be correlatedto multiple processes e Lin et al., 2011). We also recognized thatlack of adequate replications could debar the drawing of outrightconclusions based on this current study. Howbeit, our approachbring to the fore the feasibility of a small-scale approach to thestudy of spatial patterns of tree regeneration in disturbedmangrove forests, and emphasize the importance of point patternsand processes analysis to regeneration study.

5. Conclusion

Spatial patterns of R. mucronata were analysed in three zonesalong the tidal gradient in this study, with a view to apply pointpatterns analyses to explore local plant interactions and broadenthe understanding of the processes determining the forest struc-tural dynamics in mangrove ecology. The result shows distinctvariation in the population distribution, especially with higher ju-venile population in the central and landward zones where theircolonization and establishment might have benefitted from canopydisturbance. This paper presented contradictory findings to the idea

of mangrove ecological sorting, and physiological adaptation ofmangrove species to cluster in a localized region along tidalgradient, though scope of the current study seemed not to providesufficient data for outright contest of the hypotheses. It howeverwould require further studies, with sufficient replicates in differentregions to achieve a viable generalization. The spatial patterns of theentire population show a fairly consistent, local scale clusteringwhich change to random pattern at a larger scale, trend in all thestudied plots. Mature pre-IOD surviving trees were found to alignthe shoreline, but negatively spatially correlatedwith the recruits. Itwas shown that recruits are spatially independent of the adult trees.Although habitat heterogeneity might be responsible for the localscale aggregation in this population, the effect of planteplantconspecific is more probable to inform the long-term structure anddynamics of the population and the entire forest.

Acknowledgement

Adewole Olagoke is very grateful to the Ramsar Section of theSociety of Wetland Scientists for granting him Student ResearchAward of 2012, and the financial provision in support of thisresearch project. Field support was obtained through the MASMARegional Project on “Resilience of mangroves and dependentcommunities in the WIO region to climate change”, Grant No:MASMA/CC/2010/08.

A.O. Olagoke et al. / Acta Oecologica 50 (2013) 1e9 9

References

Baddeley, A., 2010. Analysing Spatial Point Patterns in R. Workshop Notes. Version4.1. CSIRO Online Technical Publication. www.csiro.au/resources/pf16h.html.

Baddeley, A., Turner, R., 2005. Spatstat: an R package for analyzing spatial pointpatterns. Journal of Statistical Software. ISSN: 1548-7660 12 (6), 1e42. URL:www.jstatsoft.org.

Baldwin, A., Egnotovich, M., Ford, M., Platt, W., 2001. Regeneration in fringemangrove forests damaged by Hurricane Andrew. Plant Ecology 157, 149e162.

Bosire, J.O., Kairo, J.G., Kazungu, J., Koedam, N., Dahdouh-Guebas, F., 2008. Spatialand temporal regeneration dynamics in Ceriops tagal (Perr.) C.B. Rob. (Rhizo-phoraceae) mangrove forests in Kenya. Western Indian Ocean Journal of MarineScience 7 (1), 69e80.

Dieckmann, U., Law, R., Metz, J.A., 2000. The Geometry of Ecological Interactions.Simplifying Spatial Complexity. Cambridge University Press, Cambridge, UK.

Diggle, P.J., 2003. Statistical Analysis of Spatial Point Patterns, second ed. Arnold,London.

Duke, N.C., Meynecke, J.O., Dittmann, A.M., Ellison, A.M., Aanger, K., Berger, U.,Cannicci, S., Diele, K., Ewel, K.C., Field, C.D., Koedam, N., Lee, S.Y., Marchand, C.,Nordhaus, I., Dahdouh-Guebas, F., 2007. A world without mangroves? Science317, 41e42.

Eichhorn, M.P., 2010. Pattern reveals process: spatial organisation of a Kamchatkanstone birch forest. Plant Ecology and Diversity 3 (3), 281e288.

Grabarnik, P., Myllymäki, M., Stoyan, D., 2011. Correct testing of mark independencefor marked point patterns. Ecological Modelling 222, 3888e3894.

Grabarnik, P., Sarkka, A., 2009. Modelling the spatial structure of forest stands bymultivariate point processes with hierarchical interactions. Ecological Model-ling 220, 1232e1240.

Getzin, S., Dean, C., He, F., Trofymow, T., Wiegand, K., Wiegand, T., 2006. Spatialpatterns and competition of tree species in a Douglas-fir chronosequence onVancouver Island. Ecography 29, 671e682.

Getzin, S., Wiegand, T., Wiegand, K., He, F., 2008. Heterogeneity influences spatialpatterns and demographics in forest stands. Journal of Ecology 96, 807e820.

Illian, J., Penttinen, A., Stoyan, H., Stoyan, D., 2008. Statistical Analysis and Modellingof Spatial Point Patterns. John Wiley & Sons, Chichester.

Jimenez, J.A., Sauter, K., 1991. Structure and dynamics of mangrove forest along aflooding gradient. Estuaries 14, 49e56.

Kirui, K.B., Kairo, J.G., Bosire, J., Viergever, K.M., Rudra, S., Huxham, M., Briers, R.A.,2012. Mapping of mangrove forest land cover change along the Kenya coastlineusing Landsat imagery. Ocean and Coastal Management. http://dx.doi.org/10.1016/j.ocecoaman.2011.12.004.

Law, R., Illian, J., Burslem, D.F.R.P., Gratzer, G., Gunatilleke, C.V.S.,Gunatilleke, I.A.U.N., 2009. Ecological information from spatial patterns ofplants: insights from point process theory. Journal of Ecology 97, 616e628.

Lawton, R.M., Lawton, R.O., 2010. Complex spatial structure in a population ofDidymopanax pittieri, a tree of wind-exposed lower montane rain forest. PlantEcology 210, 125e135.

Levin, S.A., 1992. The problem of pattern and scale in Ecology. Ecology 73, 1943e1967.

Lin, Y.C., Chang, L.W., Yang, K.C., Wang, H.H., Sun, I.F., 2011. Point patterns of treedistribution determined by habitat heterogeneity and dispersal limitation.Oecologia 165, 175e184. http://dx.doi.org/10.1007/s00442-101-1718-x.

Luo, Z.R., Yu, M.J., Chen, De L., Wu, Y.G., Ding, B.Y., 2012. Spatial associations of treespecies in a subtropical evergreen broad-leaved forest. Journal of Plant Ecology.http://dx.doi.org/10.1093/jpe/rtr048.

Mohamed, M.O.S., Neukermans, G., Kairo, J.G., Dahdouh-Guebas, F., Koedam, N.,2008. Mangrove forests in a peri-urban setting: the case of Mombasa (Kenya).Wetlands Ecology and Management 17 (3), 243e255.

Moorcroft, P.R., Hurtt, G.C., Pacala, S.W., 2001. A method for scaling vegetationdynamics: the ecosystem demography model (ed). Ecological Monographs 74,557e586.

Olagoke, A.O., 2012. Vegetation Dynamics and Carbon Distribution in a HeavilyImpacted Peri-urban Mangrove Forest in Tudor Creek, Kenya. MSc thesis sub-mitted to the Institute of International Forestry and Forest Products. TechnischeUniversitaet Dresden, Germany.

Perry, G.L.W., Miller, B.P., Enright, N.J., 2006. A comparison of methods for thestatistical analysis of spatial point patterns in plant ecology. Plant Ecology 187,59e82.

R Development Core Team, 2011. R: a Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria, ISBN 3-900051-07-0. URL: http://www.R-project.org/.

Rabinowitz, D., 1978. Early growth of mangrove seedlings in Panama and an hy-pothesis concerning the relationship of dispersal and zonation. Journal ofBiogeography 5, 113e133.

Reddy, C.S., Roy, A., 2008. Assessment of three decade vegetation dynamics inmangroves of Godavari Delta, India using multi-temporal satellite data and GIS.Research Journal of Environmental Sciences 2, 108e115.

Ripley, B., 1977. Modelling spatial patterns (with discussion). Journal of the RoyalStatistical Society, Series B 39, 172e212.

Sherman, E.S., Fahey, J.T., Battles, J.J., 2000. Small-scale disturbance and regenera-tion dynamics in a neotropical mangrove forest. Journal of Ecology 88, 165e178.

Smith III, T.J., 1992. Forest structure. In: Robertson, A.I., Alongi, D.M. (Eds.), TropicalMangrove Ecosystems. Coastal Estuarine Stud., vol. 41. AGU, Washington, DC,pp. 101e136. http://dx.doi.org/10.1029/CE041p0101.

Sousa, W.P., Kennedy, P.G., Mitchell, B.J., Ordonez, L.B.M., 2007. Supply-side ecologyin mangroves: do propagule dispersal and seedling establishment explain theforest structure? Ecological Monographs 77, 53e76.

Stoyan, D., Stoyan, H., 1994. Fractals, Random Shapes and Point Fields: Methods ofGeometrical Statistics. John Wiley and Sons, Chichester, UK.

Thaxton, J.M., Dewalt, S.J., Platt, W.J., 2007. Spatial patterns of regeneration afterHurricane Andrew in two south Florida fringe mangrove forests. Florida Sci-entist 70 (2), 148e156.

Wiegand, T., Moloney, K.A., 2004. Rings, circles, and null-models for point patternanalysis in ecology. Oikos 104, 209e229.

Wiegand, T., Gunatilleke, S., Gunatilleke, N., 2007. Species associations in aheterogeneous Sri Lankan dipterocarp forest. American Naturalist 170,E67eE95.