Appendix 1 - Oikos Journal

Oikos
OIK-03229
Derroire, G., Balvanera, P., Castellanos-Castro, C., Decocq,
G., Kennard, D. K., Lebrija-Trejos, E., Leiva, J. A., Odén, P.C., Powers, J. S., Rico-Gray, V., Tigabu, M. and Healey, J.
R. 2016. Resilience of tropical dry forests – a meta-analysis
of changes in species diversity and composition during
secondary succession. – Oikos doi: 10.1111/oik.03229.
Appendix 1
Figure A1. Fitted local polynomial regressions of each index of species richness, evenness and
composition similarity for tree and shrub species with age of succession in tropical dry forest
datasets. The red line is the fitted regression (using function LOESS in R ver. 3.1.3). The grey
shading is the area demarcated by the 95% confidence intervals. The dashed line is the mean value
for old-growth forest plots, when available. Dots are observations. Castellanos stands for
Castellanos-Castro.
(a) Species richness
(b) Shannon evenness index
(c) Sørensen similarity index
(d) Chao–Sørensen similarity index
Appendix 2
Figure A2. Forest plot of meta-analysis of the slope of linear regression of the considered index
against successional age (effect size) for each index of species richness, evenness and composition
of tree and shrub species in successional tropical dry forest. The dashed line represents a slope of
zero (no correlation between the considered index and successional age). The size of boxes
represents the weight of the study in the analysis. Horizontal lines represent 95% confidence
intervals. Positive effect sizes indicate a positive change with successional age. The position of the
centre of the diamond represents the overall summary effect (obtained with a random-effect
weighted meta-analysis model) and its horizontal extent represents the positive and negative 95%
confidence intervals. Castellanos stands for Castellanos-Castro.
(a) Species richness
(b) Shannon evenness index
(c) Sørensen similarity index
(d) Chao–Sørensen similarity index
Appendix 3
Table A3. Goodness of fit (R2) of the linear regressions fitted for each index and datasets.
Goodness of fit (R2) for
Dataset
Castellanos-Castro and Newton 2015 – Ceibal - pasture
Castellanos-Castro and Newton 2015 – Rosales - pasture
Castellanos-Castro and Newton 2015 – Rosales - shifting
Kennard 2002
Lebrija-Trejos et al. 2008
Leiva et al. 2009
Mora et al. 2015
Powers et al. 2009 – Palo Verde TDF
Powers et al. 2009 – Santa Rosa oak
Powers et al. 2009 – Santa Rosa TDF
Rico-Gray and Garcia-Franco 1992
Sovu et al. 2009
Spittler 2001
Species
richness
Shannon
evenness
Sørensen
similarity
0.102
0.261
0.001
0.089
0.684
0.795
0.574
0.010
0.605
0.472
0.083
0.001
0.110
0.113
0.474
0.020
0.014
0.608
0.176
0.041
0.014
0.038
0.349
0.188
0.014
0.040
0.424
0.894
0.841
0.421
0.972
0.378
Chao–
Sørensen
similarity
0.297
0.910
0.842
0.224
0.080
0.432
Appendix 4
Table A4. Results of meta-analyses (all datasets) and meta-regressions of the slope of linear
regression of the considered index against successional age (effect size) for each index of species
richness, evenness and composition of tree and shrub species in successional tropical dry forest.
MAP is mean annual precipitation and PET/MAP is the ratio between annual potential
evapotranspiration and MAP. The estimates of effect size in bold are significant (the 95%
confidence interval excludes 0). For the meta-regressions on categorical moderators (previous land
use and duration of succession studied), p-values give the significance of the difference between the
two categories. For the meta-regressions on continuous moderators (MAP, number of dry months
and PET/MAP), p-values give the significance of the relationship between the effect size and the
continuous moderators.
All datasets
Previous land use
Duration of succession
studied
MAP
Number of dry months
PET/MAP
Species
richness
Shannon
evenness
Sørensen
similarity
effect size
effect size pasture
effect size shifting
0.088
0.118
0.044
0.002
0.002
0.008
0.011
0.005
Chao–
Sørensen
similarity
0.009
0.012
0.003
p-value
effect size ≤ 30 years
0.108
0.145
0.504
0.045
effect size > 30 years
p-value
slope of meta-regression
0.070
0.171
–0.0004
0.002
0.049
0.623
0.038
0.003
<0.001
0.003
<0.001
–0.0000
0.810
0.001
0.983
0.019
0.749
–0.0000
0.207
0.0004
0.674
0.002
0.130
–0.0000
0.143
0.005
0.363
0.012
0.203
–0.0000
0.213
0.006
0.449
0.014
0.268
p-value
slope of meta-regression
p-value
slope of meta-regression
p-value
–0.0002
0.003
Appendix 5
Figure A5. Results of the meta-regressions of the slope of linear regression of the considered index
against successional age (effect size) for each index of species richness, evenness and composition
of tree and shrub species and each continuous moderator, in successional tropical dry forest.
PET/MAP is the ratio between annual potential evapotranspiration and the mean annual
precipitation. The red line is the meta-regression, the dashed lines the 95% confidence interval and
dots are the effect size for the datasets. p-values give the significance of the relationship between
the effect size and the continuous moderators.
(a) Species richness
(b) Shannon evenness index
(c) Sørensen similarity index
(d) Chao–Sørensen similarity index
Appendix 6
Figure A6. Funnel plots for the overall analysis (without moderators) of the slope of linear
regression of the considered index against successional age and for the meta-regressions with the
moderators previous land use and duration of succession studied, for each index of species richness,
evenness and composition of tree and shrub species in successional tropical dry forest. Dashed lines
represent 95% pseudo-confidence intervals.
(a) Species richness
(b) Shannon evenness index
(c) Sørensen similarity index
(d) Chao–Sørensen similarity index