The probability that individuals initiated t-phase within the 30-day experimental period was investigated by fitting whether (1) or not (0) individuals entered t-phase as the response term using two GLMs with binomial error structures with logit link functions.
Differences in population prevalence between species were tested with a logistic regression model with binomial error structure applying the GENMOD procedure in SAS/enterprise while estimating the scaling parameter by the square root of DEVIANCE/DOF.
In the first analysis, we conducted a generalized linear mixed model (GLMM) with binomial error structure to determine factors associated with female aggression.
We used GLMMs with binomial error structures and a logit-link function.
To weigh these proportions in relation to total number of leaves per seedlings, we used GLMMs with distance and plateau as fixed effects and maternal tree as random effect, and with binomial error structures.
Infection prevalence at both oocyst and sporozoite stages and gametocyte sex ratios were analysed using generalized linear models with binomial error structures.
To test the null hypothesis that RS was similar for hatchery and naturally spawned fish, we used a generalized linear model (GLM) with negative binomial error structure and a log link function fit separately for each sex using the package MASS in the program R (Venables and Ripley 2002).
For analyses of immune parameters as a response variable, linear models were used for immune parameters that could be normalised, and generalised linear models (GLMs) with a binomial error structure were used for parameters as presence/absence data.
For analyses of intestinal worm burden and per capita fecundity as response variables, GLMs with a negative binomial error structure were used.
The rate of female transmission was affected by the mother's genotype (GLM with quasi-binomial error structure, t30 = 2.096,, p = 0.04; figure 1), with transmission being on average 28.1 per cent lower in females that carried the resistant allele.
Male transmission, on the other hand, was not affected by the father's genotype (GLM with quasi-binomial error structure, t48 = 0.156,,, p = 0.87, Δtransmission = 2.3%; figure 1).
