Ecological Archives E096-192-A6
María C. Rodríguez-Rodríguez, Pedro Jordano, and Alfredo Valido. 2015. Hotspots of damage by anatgonists shape the spatial structure of plant–pollinator interactions. Ecology 96:2181–2191. http://dx.doi.org/10.1890/14-2467.1
Appendix F. Further details on methods and results from generalized linear mixed-effects models.
Methods
We compared non-spatial and spatial generalized linear mixed-effects models (GLMMs) maintaining the same regressors (two functional groups or six animal guilds) to test if the addition of a spatial correlation term significantly improved model fit. In the spatial GLMMs, plant coordinates were included in an exponential correlation function. This procedure allows the accommodation of plot differences in autocorrelation distances, and assumes autocorrelation only between plants within the same plot (e.g., Dormann et al. 2007). The choice of the exponential over other spatial covariance structures was based on inspection of semi-variograms of non-spatial GLMM residuals. Although the addition of the spatial correlation term did not improve model fit in any case (anova test via analysis of variance, P > 0.05 for all cases), we included in the article and the present appendix the results from the spatial GLMMs.
Table F1. Estimated effects of the interaction strengths with mutualists and antagonists on plant reproductive success (PRS) of Isoplexis canariensis, analyzed with spatially explicit GLMM (df = 115). Independent variables were standardized to mean 0 and variance 1 and the dependent variable was relative PRS. Regression coefficients (ß) and their associated standard errors (SE) are given. Notes: * P, < 0.05; ** P, < 0.01; *** P, < 0.001.
Source of variation |
ß coefficient |
SE |
P value |
|
Interaction strength with mutualists |
0.035 |
0.016 |
0.032 |
* |
Interaction strength with antagonists |
-0.086 |
0.016 |
< 0.001 |
*** |
Table F2. Estimated effects of the interaction strengths with mutualistic and antagonistic animal guilds on PRS of Isoplexis canariensis, analyzed with spatially explicit GLMM (df = 111). Independent variables were standardized to mean 0 and variance 1 and the dependent variable was relative PRS. Regression coefficients (ß) and their associated standard errors (SE) are given. Notes: * P, < 0.05; ** P, < 0.01; *** P, < 0.001. See Appendix A for the taxonomic composition of animal guilds.
Source of variation |
ß coefficient |
SE |
P value |
|
Interaction strength with legitimate bird pollinators |
0.025 |
0.016 |
0.119 |
|
Interaction strength with facultative bird pollinators |
0.043 |
0.015 |
0.006 |
** |
Interaction strength with legitimate lizard pollinators |
0.028 |
0.015 |
0.074 |
|
Interaction strength with floral herbivores |
-0.065 |
0.016 |
< 0.001 |
*** |
Interaction strength with nectar larcenists |
-0.034 |
0.016 |
0.032 |
* |
Interaction strength with predispersal seed predators |
-0.029 |
0.016 |
0.067 |
|
Literature cited
Dormann, C. F., et al. 2007. Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30:609–628.