Table 2. Cumulative-link mixed effect ordinal regression (a) testing the effect of environmental, urban and human variables on the ordinal intensity of offspring defense; and (b) linear mixed effect model testing the effect of intensity of offspring defense in incubation on eventual fledgling production.
Variable | ß ± SE | Z-test | P- value |
---|---|---|---|
a. Dependent variable: Intensity of defense (N = 657) a,b | |||
Breeding Stage (incubation) | 1.11 ± 0.8 | 1.39 | 0.164 |
Breeding Stage (small nestling) | 3.11 ± 0.82 | 3.81 | < 0.001 |
Breeding Stage (large nestling) | 3.83 ± 0.8 | 4.78 | < 0.0001 |
Breeding Stage (pre-fledging) | 2.34 ± 0.76 | 3.07 | 0.002 |
Breeding Stage (post-fledging) | 0.71 ± 0.97 | 0.73 | 0.463 |
Offspring number | 1.61 ± 0.33 | 4.81 | < 0.0001 |
Access to Muslim subsidies | -6.23 ± 2.01 | -3.1 | 0.001 |
Hygiene score | 3.25 ± 1.01 | 3.24 | 0.001 |
Access to Muslim subsidies * Hygiene score | 4.5 ± 2.02 | 2.22 | 0.026 |
Green cover | -1.65 ± 0.68 | -2.4 | 0.016 |
Urban cover | -3.69 ± 1.62 | -2.28 | 0.022 |
Urban cover ^2 | 3.20 ± 1.67 | 1.9 | 0.057 |
b. Dependent variable: fledglings produced c (N = 103) | |||
Intensity of defense (during incubation) | 0.28 ± 0.12 | 2.27 | 0.023 |
Intercept | -0.36 ± 0.17 | - | - |
a Cumulative link mixed model with a logit link function, based on N = 657 defense trials from 101 independent territories. The dependent variable is the ordinal score of offspring defense intensity. Territory-identity nested within plot-identity and year was fitted as a random factor.
b Variables presented to the model: Julian date, Team size, Number of onlookers, Previous visits, Breeding Stage, Offspring number, NND5, Territories within 200 m, Tree arrangement, Balcony, Index of road density, Urban cover, Green cover, Hygiene score, Human density, Access to Muslim subsidies, Access to Muslim subsidies*Hygiene score, Access to Muslim subsidies*Urban cover, Access to Muslim subsidies*Green cover (the rationale for fitting interactions can be found in the Methods).
c Generalised linear mixed model with Poisson errors and a logit link function, based on N = 103 defense trials from 60 independent territories sampled during incubation. The dependent variable is the number of young raised to fledging stage. Territory-identity nested within plot-identity and year was fitted as a random factor.