Appendix B. Analysis of deviance tables.
TABLE B1. Analysis of deviance table for a generalized linear model with binomial response. Explanatory variables are observation day and isolation (distance to third nearest neighbor). The response is bee visitation (yes/no) within 20 minutes. Parameter estimates for the additive model are shown in Fig. 1a.
Explanatory 
Model df 
Model deviance 
Test df 
Test deviance 
Test P value 
offset^{†} 
728 
865.01 



obsday 
699 
750.14 
29 
114.86 
<0.0001 
obsday + log(iso3) 
698 
746.14 
1 
4.00 
0.05 
obsday × log(iso3) 
669 
718.16 
29 
27.98 
0.52 
^{†} Note: all models include log(minutes observed) as a known linear predictor (offset).
TABLE B2. Analysis of deviance table for a generalized linear model with binomial response. Explanatory variables are observation day and population size. The response is bee visitation (yes/no) within 20 minutes. Parameter estimates for the additive model are shown in Fig. 1b.
Explanatory 
Model df 
Model deviance 
Test df 
Test deviance 
Test P value 
offset^{†} 
768 
899.22 



obsday 
739 
775.10 
29 
124.12 
<0.0001 
obsday + log(popSize) 
738 
772.89 
1 
2.21 
0.14 
obsday × log(popSize) 
710 
740.57 
28 
32.32 
0.26 
^{†} Note: all models include log(minutes observed) as a known linear predictor (offset).