Spectroscopy-based analysis of
SOM-soil fauna relationships
Juan J. Jiménez
22 January 2019
Tartu (Estonia)
2
Soil ecosystem engineering – Non-trophic based interactions
Ecosystem engineers(sensu Jones et al. 1994) OM
Aggregates
(Decaëns et al. 2008)
??
3
Soil aggregation
iPOM
Fresh
residues
Macro-
aggregate
<250 µm
Micro-aggregate
<250 µm
Binding agents
+ bacteria and
fungi
Aggregate size (µm)
Cementing agents
Process
Abiotic Biotic
0.2 Clay plates, Oxides, Amorphous alluminosillicates
Organic polymers sorbed onto clay surfaces
Electrostatic bonding; flocculation
2-20 Clay particles (packages)
Microbial, fungal and plant debris
Encrustation
200 Particles or aggregates
Roots, hyphae
Binding
2,000 Aggregates Soil fauna Selection of particlesFormation of biogenic structures
Aggregation processes in soil
Grouping of primary soil particles into larger units = aggregates, surrounded by pores (diff. size)
Physical or biological origin: angular-blocky, prismatic forms, rounded shapes.
(Jiménez and Lal 2006)
4
Soil aggregate’s turnover affecting
SOM dynamics (soil fauna)
Soil structure
iPOM
Free
aggregates
Macro-
aggregates
(>250 μm)
Micro-
aggregates
(<250 μm)
Fresh residues, litter
Macro-aggregate
Soil ecosystem engineers
(compact vs. decompact)
Micro-aggregate
Increased
mineralization
CO2 flux
Reduced
mineralization
CO2 flux
Larger macro-aggregates (casts, fecal pellets)
iPOM-clay
interaction
Fresh residues-micro-
organisms’ interaction +
biogenic structures
Conceptual model of aggregate formation
by soil ecosystem engineering activities
Jiménez and Lal (2006)
5
LitterBiogenic
structures
CO2 CO2 CO2
Aggregatedetachment Free macro-
and micro-aggregates
Wind, watererosion
[DOC]L
[DOC]BS
[DOC]F
Bel
ow
-gro
un
dA
bo
ve-g
rou
nd
Litter and soil organisms
StreamsAquatic systems
Anthropic practices
C-Clay interaction
Aggregate formation
Bioturbation
Incorporation to soil matrix
Macro-aggregates
Micro-aggregates
Tillage
ASML ASMBS ASMF
BioturbationSpatial and temporal
scales
Physical protection of C
DOC = Dissolved organic C; L = litter, BS = biogenic structure; F = free aggregates; ASM = Activation of soil micro-organisms (Jiménez and Lal, 2006)
Conceptual framework of the role of soil ecosystem engineering activities on C accumulation at different
spatio-temporal scales
Water run-off
6
Glass fiber net
Metal plate
4O cm
3O cm2
4Chemical control area
7O cm
15 cm
1
Sampling(0, 6 and 18 months):
• Vegetation• Soil• Earthworms
3
15 cm
12 unitsw/wt
Loss of beneficial functions when soil ecosystem engineers are absent
(Decaëns et al. 1999)
Soil organism-driven impacts are measurable
6m
18m
A
B+
-
0m
bSamplingdate
cGrasslandtype d
Ecosystemengineerpresence
Axis1.
Axis2
Axis1.Axis1.
Axis2
GB LB
WB
%L
LB RB
BD
MWD
pH
%C
%N
TP
Al
CaMg
K
PR
STermAnt
Gs
Ay
Mc
Oc
Axis 1.(29.3%)
Axis 2.(16.5%)
A
a
Axis2
> soil compaction< plant biomass> weed incidence< C in soil
Soil degradation
PCA of soil and plant properties with and without earthworms
7
Each biogenic structure has its own
NIRS signal
Case 1
8
Near Infrared (region) Spectroscopy – a region of the electromagnetic
spectrum from 700 to 2,500 nm;
Most equipments also obtain the visible región (VNIRS);
Efficient, rapid and non-destructive analytical technique to analyse soil
samples to estimate soil properties, i.e. C and N contents;
Useful in SOM studies linking species responsible for the production of
biogenic aggregates in the soil matrix;
Also MIRS (mid-infrared spectroscopy), more sensitive;
What is NIRS?
9
Granular casts produced by Prosellodriluspsammophilus Qiu & Bouché, 1998 from “Los Monegros” at the Ebro basin (Spain)
10
Soil habitat transformation
Indirect modification of resources via
production of physical structures
Autogenic engineers "change the environment via
their own physical structures, i.e. their living and dead
tissues."
Allogenic engineers "change the environment by
transforming living or nonliving materials from one
physical state to another, via mechanical or other
means."
• In soil: earthworms, ants, termites,
and roots
Pheidole sp1
Atta laevigata
Trachymyrmex sp.
Atta cephalotes
Velocitermes sp.
Nasutitermes sp.
Microcerotermes sp.
Termitinae
Ruptitermes sp.
Spinitermes sp.
Globular
Martiodrilus sp.
Hyperiodrilus africanus
Millsonia anomala
Colonization andRoot biomass
Granular
Prosellodrilus sp.
earthworms
ants
termites
Soil ecosystem engineers
11
Pasteboard-like
termite mounds
Earthworm
casts
Organo-mineral
termite moundsTermite
sheathings
Ant
deposits
Control
soil
F1 (64.1%)
F2 (
12.8
%)
(Decaëns et al. 2001; Hedde et al. 2005;
Velasquez et al. 2007; Zhang et al., 2009;
Zangerlé et al. 2011)
Near Infrarred Spectroscopy (NIRS) spectra of
14 biogenic structures (Colombian «Llanos»)
species-specific organic
fingerprints in their respective
above-ground biostructures
1. Accumulators of protected OM
2. Soil compactors
3. Soil decompactors
12
(Jiménez et al. 2006)
Biogenic
structure 50
100
20
40
60
80
Spatial C and N dynamics in biogenic structures
1 m
0-5
5-10
Sample location
1 2 3 4 5 6
NH
4 (
g d
ry s
oil
-1)
0
100
200
300
400
500
600
700
800
b b b b
b
b
a aa a
7(0-5 cm)
8 7 8(5-10 cm)
Sample location
1 2 3 4 5 6
Co
rg (
g k
g-1
)
0
10
20
30
40
50
60
70
ns
7(0-5 cm)
8 7 8(5-10 cm)
13
45 m
45
m
Earthwormsampling
Soil sampling (Metal cylinders):
1 Size-class aggregates
2 Bulk density, hydraulicconductivity and soilcompaction
3 C, N and P determinations
4 Root length and biomass (fine and coarse)
1
10
91
100 50
Attalea maripa
Nectandra membranacea
Cecropia sp.
Didymopanax morototoni
Virola sp.
Tree species, > 5cm DBH
Sampling protocol
Study site
SouthAmerica
Eastern Plains of ColombiaCIAT Carimagua researchstation (22,000 ha.)
Well drained isohyperthermic savannasOxisols of low fertility, low pH and high Al saturation (>90%)
14
Species pool hypothesis –
species filtering process
Decaëns et al. (2006)
Anthropogenic changes
Global change, N2
deposition
Soil erosion
Land use changes
Agricultural practices
Exotic introduction (invasive species)
Pictures © Pavel Krasensky; Tree of life web
Spatial scale
REGION
PATCH
LANDSCAPE
ECOSYSTEM
GLOBAL
COMMUNITY
Climate
Landscape structure
Spatial competitive mechanisms
Soil abiotic factors
Land use practices
Biotic interactions
15
Sem
i-v
aria
nce
Lag distance (m)
0 5 10 15 20
0.005
0
0.015
0.010
0.020
0.030
0.025
0.035
335224
251268447161
178
0.0106067 Nug(0) + 0.0220092 Sph(8.11863)1
20
30
40
50
60
70
45
0 4515 30
15
30
C concentration (g kg-1)
Spatial distribution of selectedsoil variables
(Jiménez et al. 2011)
Lag distance (m)
0 5 10 15 20
Sem
i-v
aria
nce
1.46103 Nug(0) + 4.41099 Sph(721.71)
1
178
161447
335
224
251
268
0
1
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
0
20
40
60
80
100
120
140
160
180
0 4515 30
45
15
30
Soil penetration resistance (kg cm-3)
16
Interestructure(dates)
0
1.4
0 4
F1=44.9%
Compromise(species)
0
0.35
0 7
Axis I Axis II
F1=28.1%
F1=2
2.5
%
Spatio-temporal stability (PTA)
(Jiménez et al. 2006)
17
(Jiménez et al. 2012)
Use of trophic and space resources
Number of inds. found in each variable (range)
C0-5
0-25 25-50 >50 N
Glo 9 819 18 846
Aym 0 630 18 648
And 18 288 0 306
Epi 0 477 477 954
Anr 0 9 0 9
Ocn 36 2313 54 2403
Mrt 9 972 45 1026
C5-10
0-15 15-30 30-45 >45 >60
0 819 27 0 0
0 585 18 45 0
9 297 0 0 0
477 477 0 0 0
0 9 0 0 0
36 2331 18 18 0
9 963 45 9 0
BD
0.7-1.0 1.0-1.2 1.2-1.4 >1.4
0 225 549 72
0 315 333 0
0 45 216 45
513 27 414 0
0 9 0 0
18 513 1827 45
9 225 783 9
PR2.5
0-1 1-3 3-5 5-7 >7
315 18 9 261 243
495 0 18 108 27
72 72 18 90 54
180 9 36 720 9
9 0 0 0 0
936 153 81 783 450
594 36 9 288 99
Trophic (nutrients) Space (physical)
Also for assemblages CA1+, CA1-, CA2+, CA2- ……
Pianka Ojk index
18
ESSEM COST Action ES1406
Glossodrilus
AymaraAndiodrilus
Andiorrhinus
OcnerodrilidaeMartiodrilus
Litter
N0-5
N5-10
P0-5
P5-10
C0-5
C5-10
C:N0-5
C:N5-10
FiRL
CoRL
FiRW
CoRW
PR5-20
0.053-0.125
0.125-0.250.25-0.5
0.5-1
1-2
2-5
5-10
>10Agg
BulkDensity
Soil compaction
Cond
Hum(%)
-0.43
0.36-0.36 0.3
F I (64.1%)
F II
(1
7.7
%)
p<0.0001
MedAgg
SmallAgg
LargeAgg
VLargeAgg
VSmallAggRoot Length
Root Biomass
Montecarlo rand.
Co-Inertia analysis (fauna$tab and vars$tab)
Ew effect
Nutrient resources
Species 1(Jiménez et al. 2012)
Challenges on modelling SOM – soil fauna
A huge quantity of empirical data needed?
Many mechanisms are involved (comminution, bioturbation, “priming effect”, …. )
Several spatial and temporal scales are involved (processes may vary)
Many organisms are involved
Many OM compartments are involved
The impact of soil fauna on the dynamics of dead organic matter has been studied for ages,
Yes but …
19
Correlogram with the factorial
coordinates of axis 1 (square)
and axis 2 (triangle) extracted
in the CoIA
Mo
ran
's I
2 7 11 16 20 25 30 34 39 43 48 52-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
Lag distance
CA1+
CA1-
CA2+
CA2-
CA3+
CA3-
Species
Species
assemblages
Heterogeneity of soil resources allow
co-occurrence of competing species
(Jiménez et al. 2012, 2014)
PCNM1PCNM3PCNM5PCNM8PCNM12PCNM33PCNM51
10 20 30 40 50 60
0.0
02
0.0
06
0.0
10
0.0
14
Distance (m)
Sem
ivari
an
cePrincipal Component of Neighbouring Matrices (Dray et al. 2006)
Spatial analysis of biotic interactions (competition)
20
(Jiménez et al. 2014)
Environmental contribution to observed spatial pattern
(>30 m) (10-20 m) (<10 m)
21
NIRS signals differ with the age of the
in-soil biogenic structure… until a
certain time-lag!
Case 2
22
Lab protocol
EW removal 1d
Soil and ew casts Incubation
1 – 32 (64) days
3 reps /3 species
casts retrieval
sieved at <200 µm
NIR spectral readings
QualitySpec® spectrophotometer
C and N determinations with dry combustion method (Variomax CN Analyzer, Germany)
Diffuse reflectance(Stenberg et al. 2010)
<2 mm sieved soil
+
23
Background
Endogeic earthworm Aporrectodea caliginosa (Savigny 1826)
(Zangerlé et al. 2014)
NIRS and age of aggregates
24
Background
2. Is it possible to identify the origin and age of a single macroaggregatefrom its NIR spectral signature?
3. Explore, in natural field conditions of mountain ecosystems, the capacity of NIRS to indicate which ecosystem engineer was responsible for the formation of a given macroaggregate.
1. Changes in SOM quantity and quality that occur in biostructures produced by ecosystem engineers result in species-specific OM fingerprints.
25
Study site
Surroundings of Ordesa and Monte PerdidoNational Park, UNESCO Heritage and LTER network site.
Alpine grasslands grazed by domestic cattleduring summer.
Climate alpine, 5 ◦C and 1,720 mm
Earthworm species:
Aporrectodea rosea Savigny 1826
Lumbricus friendi Cognetti, 1904
Prosellodrilus pyrenaicus (Cognetti, 1904)
(Bueno and Jiménez 2014)
26
Sampling2
0 m
20 m
Eartworm hand-sorting area for
lab rearing and NIRS analysis.
Soil invertebrates hand-sorting
(25x25x20 cm)
Nine soil blocks (10x10 cm)
Morphology analysis
Morphology analysis10x10 cm and 10 cm depth soil blocks were takenbiogenic aggregates, root-aggregates, and non-aggregated soil in order to compare the signals withthe temporal reference signatures of earthwormcasts from the lab experiment.
Elena Velásquez Patrick Lavelle
Yamileth Domínguez
Silvia Gutiérrez
Mountain Team
27
A =
log(
1/R
)
(Unscrambler X 10.2, CAMO software)
Data pretreatment needed
Reflectance (R) is converted to absorbance (A) 10 scans per sample
NIR spectra from 1,100 to2,400 nm (10 nm intervals)further transformed toSavitzky-Golay 2nd deriv(noise reduction).(Savitzky and Golay 1964; Guthrie and Walsh, 1997; Cécillon et al. 2010)
PCAs and Partial Least SquareRegression are normally used
As H2O has a strong absorbance in the NIRS due to complicated hydrogen-bonding interactions
Dry samples
28
Savitzky-Golay 2nd deriv
Derivatives are used to:
– Correct for various baseline effects
– Reveal “hidden” spectral features in
overlapping peaks
29
Data analysis
NIRS raw data
NIRS y’’ 2nd deriv
30
2nd Derivative of NIR-data
31
Data pretreatment
Noise reduction
WinISI software (Shenk and Westerhaus, 1991), MixOmics, Ade4, Prospectr package in R
Matlab, Unscrambler (CAMO software),
d1 <- t(diff(t(NIRsoil$spc), differences = 1)) # first
derivative
d2 <- t(diff(t(NIRsoil$spc), differences = 2)) #
second derivative
1200 1400 1600 1800 2000 2200 2400
-0.0
010.
000
0.00
10.
002
0.00
30.
004
0.00
5
Wavelength
1st der
gap-segment 1st der
Prospectr package in R
32
ResultsFor the correlation circle, variables that had respective weights on axes 1 and 2 higher than 50% of total variance explained are selected.
In total 89 columns, i.e. number of variables = 2nd deriv. absorbance and 130 rows, i.e. number of objects = samples.
33
A. rosea
RMSECV: 52.6%
d = 5
s1_1
s1_16 s1_2
s1_32
s1_4
s1_8
d = 5
s2_1
s2_16
s2_2
s2_32
s2_4
s2_8
L. friendi
RMSECV: 57.3%
Root mean square error of cross validation = prediction ability calculated by the model based on the NIR data of the reference library (Shenk and Westerhaus 1993)
d = 2
s3_1
s3_16
s3_2
s3_32 s3_4
s3_8
P. pyrenaicus
RMSECV: 49.8%
35
RMSECV: 43.7%Spectral data of fresh casts for 3 species (reference library data)
A. rosea
1 day-old
L. friendi
1 day-old
P. pyrenaicus
1 day-old
PLS-DA modelPLS factors are determined with the root
mean squared error of prediction (RMSEP) d = 5
s1_1
s2_1
s3_1
Not identified: 44.4% (4 casts) !!
A. rosea
L. friendi
P. pyrenaicus
11.1% identified (1 cast)
33.3% identified (3 casts)
11.1% identified (1 cast)
Projection of field signals
36
In a highly heterogenous soil (i.e. mine)
NIRS is able to distinguish biogenic and
non-biogenic structures
Case 3
37
La Guajira, Colombia
Restored sites at Cerrejón coal mine
38
6-yr
39
Biogenic aggregates (BA) produced by macroinvertebrates (mainly earthworms, ants, termites, Coleopteran larvae, and Diplopoda)
Rhizosphere aggregates (RHIZ) are made of soil stuck to the roots
Physical aggregates (PHYS) produced by physical processes (e.g., drying and rewetting of the soil
Non-macroaggregated fraction (NON): this corresponds to smaller soil aggregates that are difficult to identify
90 – 120 minutes/sample
40
Projection of barycenters of NIRS spectra grouped by site and type of aggregates in the plane defined by the first two factors of PCA – between. BA = Biogenic aggregates, NAS= non-agreggrate soil and PA = Physical aggregates. 1, 5, 8, 9, 16, 20-y = Age of rehabilitation. F = Forest.
g
BA
NAS
PA
Axis
II
= 3
1%
Axis I = 68%
P < 0.01
Projection of NIRS spectra of macroaggregates type from forests and rehabilitated areas in the plane defined
by the first two factors of PCA.
(Domínguez et al. 2018)
Between-class PCA , 39%, p < 0.01
BA-16-y
BA-20-y
BA-5-y
BA-8-y BA-9-y
BA-F-1 BA-F-2
NAS-1-y
NAS-16-y
NAS-20-y NAS-5-y
NAS-8-y NAS-9-y
NAS-F-1 NAS-F-2
PA-1-y
Axis 1 = 29%
Axis 2 = 24%
41
KEYSOM protocol 5
Case 4
42
Germany (Hind Khalili)
Italy (A. Sofo)
Croatia (D. Hackenberger)
Poland (Lidia Sas)
Poland, site 2 (Lidia Oktaba)
Finland (V. Nuutinen)
Portugal (J. Nunes)
Romania (M. Iamandei)
Russia (A. Tiunov)
Switzerland (A. Frossaard)
UK (D. Jones)
KEYSOM protocol 5
List of countries
43P<0.001
Montecarlo rand
DE_Meadow
HR_Meadow
Olive
Forest
Histogram of sim
sim
Fre
quency
0.02 0.06 0.10
0100
300
Axis I = 76.5%
Ax
is I
I =
13
.7%
NE Spain
SE Italy
Germany
Croatia
Preliminary results
44
Preliminary results
Second derivative
NIR spectra
1,000 – 2,500 nm
CROATIA FINLAND
GERMANY GERMANY2
ITALY2 POLAND2
PORTUGAL
ROMANIA RUSSIA
SWITZERLAND
TURKEY UK
Histogram of sim
sim
Fre
qu
en
cy
0.02 0.08 0.14
05
01
50
F1: 45.3%
F2:
17.1
%
P<0.001
Montecarlo rand
45
P<0.001
Montecarlo rand
Preliminary results
Axis I = 78.0%
Ax
is I
I =
17
.5%
Biogenic
NON
Rhizospheric
Histogram of sim
sim
Fre
quency
0.05 0.15 0.25
0100
200
300
Physicogenic
Russia – mixed forest
Aggregate types
46
Physical aggregates (18.69 g)Biogenic aggregates (15.60 g)Non-macroaggregated fraction (19.4 g)Rhizosphere aggregates (16.70 g)
Thanks Alexei Tiunov!!
47
The likelihood to distinguish spectral values of aggregates from different specieswas demonstrated, at least during the first 30 days after being produced.
Spectroscopy analyses are useful to predict and distinguish aggregates andstructures produced by large invertebrates from many sites and ecosystems.
Take-home message
Many systems are inherently non-linear, which limits our ability to make ecological predictions. NIRS offers the potential for a shift towards whole-system empirical modeling in several areas of ecology.
NIRS is not just a tool to build an empirical model but indicated an underlying mechanism that could be further investigated.
These analyses are necessary to help implement the role of soil fauna into SOM models for land use and natural resource management.
Models were not robust enough to predict soil aggregates age. Only fresh casts were be predicted.
Thanks!!
Spectroscopy-based analysis of
SOM-soil fauna relationships
Juan J. Jiménez
22 January 2019
Tartu (Estonia)
1
2
3
Soil monolith
extraction25x25 x 30 cm3
30 c
mHand-sorting- Litter
- 0-10 cm
- 10-20 cm
- 20-30 cm
Laboratory:
- Identification (S)
- Abundance (N)
- Biomass (B)
Biogenic structures
identificacion
25 cm
Método TSBFISO 23611-5:2011
(Anderson e Ingram 1993)