Supporting Information: Appendix S1 Vegetation structure We

advertisement
Supporting Information: Appendix S1
Vegetation structure
We measured woody plant species density and average heights to ensure the habitats surveyed were comparable at different sites along the aridity gradient.
We used a line transect through each 1km transect in which the birds were sampled. We identified points every 20m, and within a 10m radius of that point,
we counted and measured the height of each woody plant. This was then used to calculate average height and density of woody plants in each transect.
Results
Trees within riparian habitats were significantly taller than the other two habitats (K-W Chi-squared = 516.5, df = 2, p < 0.001, confirmed using multiple
comparison test after Kruskal-Wallis); thickened habitats contained significantly more low woody cover than riparian, and both thickened and riparian habitats
had greater low woody cover than open habitats (K-W Chi-squared = 1563.7, df = 2, p < 0.0001, Figure S1). Habitat types were comparable between sites
(i.e., all riparian, open and thickened habitats did not differ significantly in woody plant density and height between sites along the aridity gradient).
Figure S1. Summary of the (a) density, and (b) maximum height of woody plants at each of the habitat types at each of the sites along the aridity gradient.
Sites are increasingly arid from left to right.
Ok
Cl
We
Ro
100
80
0
20
40
60
80
0
20
40
60
80
60
40
% woody cover
20
0
Ne
Ne
Ok
Cl
We
Ro
Cl
We
Ro
Cl
We
Ro
15
0
5
10
15
10
0
Ok
Ok
Thickened
5
10
5
0
Ne
Ne
Riparian
15
Open
Woody plant height (m)
Thickened
100
Riparian
100
Open
Ne
Ok
Cl
We
Ro
Ne
Ok
Cl
We
Ro
Appendix S2
200
150
100
50
Chao 1 Mean
Chao1Mean[Habitat == "Riparian"]
Species accumulation curves for sampling in riparian, thickened and open habitats, based on Chao1 species richness estimator.
0
Riparian 20
Number of transects
40
Thickened 60
Sample[Habitat == "Riparian"]
80
Open
Appendix S4
Models Fit for FRic, Using Environmental Correlates that Best Explained FD and sesFD
FRic
Rainfall:
Habitat (R)
Habitat (T)
Prev 6
month
Rainfall:
Prev year
Season
Habitat (R)
Habitat (T)
x Season
x Season
d.f.
AIC
Akaike
weighting
Model 1
n.s.
n.s.
-
P < 0.0001
P < 0.05
P < 0.001
n.s.
9
1566.7
0.99
Model 2
n.s.
n.s.
P < 0.001
-
P < 0.01
P < 0.01
n.s.
9
1581.7
0.0005
Null model
-
-
-
-
-
-
-
3
1649.2
~0
(Best
model)
Best model (Model 1): FRic = 1.72 (±0.78) Wet Season + 0.0088 (±0.002) Rainfall over previous year + 3.77 (±1.11) Riparian Habitat in Wet Season
Model 2 : FRic = 2.32 (±0.81) Wet Season + 0.0074 (±0.002) Rainfall over previous year + 3.81 (±1.43) Riparian Habitat in Wet Season
Based on SES calculated assuming chance of each species occurring in random assemblages is linked to its abundance in the dataset (that
is, constrained)
Model 1
(Best
model)
Model 2
Null Model
Habitat
(O)
Habitat
(R)
Habitat
(T)
Rain:
prev 6
mo (mm)
Rainfall:
prev y
Habitat
(R) x
Season
Habitat
(T) x
Season
Season
d.f.
AIC
Akaike
weighting
P < 0.05
P<
0.0001
P < 0.05
-
-
P < 0.05
n.s.
n.s.
8
656.9
0.99
n.s.
P<
0.0001
n.s.
n.s.
-
P < 0.05
n.s.
n.s.
9
672.8
0.0004
-
-
-
-
-
-
-
-
3
699.8
~0
Best model (Model 1): sesFRic = -0.27 (±0.12) Open – 1.4 (±0.12) Riparian -0.31 (±0.13) Thickened + 0.54 (±0.21) Riparian Habitat in Wet Season
Model 2 : Same explanatory variables as those for sesFD constrained, but not all with the same levels of significance, so given in tables below in more
detail
Based on SES calculated assuming chance of each species occurring in random assemblages is not linked to its abundance in the dataset
(that is, not constrained)
Rain:
Habitat
Habitat
Habitat
Habitat
Habitat
Rainfall:
Akaike
prev 6
(R) x
(T) x
Season
d.f.
AIC
(O)
(R)
(T)
prev y
weighting
mo (mm)
Season
Season
Model 1
(Best
model)
Model 2
Null Model
P<
0.0001
P<
0.0001
P<
0.0001
P<
0.0001
-
P<
0.0001
-
P<
0.0001
-
-
-
P < 0.001
n.s.
n.s.
8
610.1
0.99
n.s.
-
P < 0.001
n.s.
n.s.
9
620.4
0.008
-
-
-
-
-
3
626.5
~0
Best model (Model 1): sesFRic = -0.74 (±0.11) Open – 1.4 (±0.11) Riparian -0.94 (±0.12) Thickened + 0.71 (±0.20 Riparian Habitat in Wet Season
Model 2 : Same explanatory variables as those for sesFD not constrained, so given in tables below in more detail
sesFRic constrained by prevalence:
Habitat Open
Habitat Riparian
Habitat Thickened
Season Wet
Rainfall.6mo.prev
Habitat Riparian:Season Wet
Habitat Thickened:Season Wet
Value
-0.25
-1.39
-0.29
0.11
-0.00006
0.53
0.18
Std.Error
0.155
0.162
0.176
0.150
0.00036
0.211
0.220
t-value
-1.65
-8.57
-1.67
0.76
-0.15
2.55
0.82
p-value
0.11
0.0000***
0.10
0.45
0.88
0.0115*
0.41
sesFRic not constrained by prevalence:
Value
Std.Error
t-value
p-value
Habitat Open
-0.672
0.140
-4.81
0.0000***
Habitat Riparian
-1.32
0.146
-9.00
0.0000***
Habitat Thickened
-0.866
0.159
-5.43
0.0000***
Season Wet
-0.082
0.139
-0.59
0.55
Rainfall.6mo.prev
-0.00025
0.00033
-0.75
0.45
Habitat Riparian:Season Wet
0.72
0.195
3.66
0.0003***
Habitat Thickened:Season Wet
0.30
0.202
1.48
0.14
Note that the levels of significance for each term in the mixed models using Fric and sesFric differ somewhat to those for FD and sesFD, but the direction of
effects (t-value) are identical. Differences in the levels of significance are likely due to the “randomness” of null model sampling and the behaviour of convex
hull volumes (Fric) relative to functional dendrograms (FD). Convex hull volumes are vulnerable to overinflation when a species that greatly differs to the
community on a single axis is added, because the volume of the additional multidimensional shape that is added to the index may be as large or larger than
the volume for the rest of the community.
Download