Data, Walker, 1862

Ramos, Yan Gabriel Celli, Aximoff, Izar Araujo & Rosa, Clarissa Alves da, 2018, Capybaras (Rodentia: Cavidae) in highlands: environmental variables related to distribution, and herbivory effects on a common plant species, Journal of Natural History 52 (27 - 28), pp. 1801-1815 : 1806

Data analysis

We used generalised linear mixed models (GLMM) with a binomial distribution ( Bolker et al. 2008; Zuur et al. 2009) to evaluate the environmental variables that affect capybara occurrence in the high-altitude grasslands of INP. We considered each site a sample unit, and the response variable was the presence/absence of capybaras during each field campaign. To define the candidate models, we used environmental variables as fixed factors and the field campaigns (seasonal variation) as random effects to control pseudoreplication ( Millar and Anderson 2004). Since the seasonal variation of capybaras is not the focus of our study, the control of these random effects is important because we lack of knowledge of the history of the sampled areas. In addition, capybaras can alter their habitat use throughout the year ( Corriale and Herrera 2014). We considered the models with small Akaike's Information Criterion (ΔAIC ≤ 2) to have the best descriptive capacity ( Burnham and Anderson 2002) regarding capybara habitat use. We averaged the values of the variables that comprised the best models in terms of the Akaike's Information Criterion weight (AICω) values of each model. Modelling was performed using the function ‘glmer’ in the ‘lme4’ suite of the R software environment (R Development Core Team 2009; Bates 2010).

To evaluate the effects of capybara herbivory on A. polyanthus individuals, we calculated the proportions of damaged and intact rosettes and compared them using a chi-square test. To evaluate the effects of capybara herbivory on the reproductive stages of A. polyanthus , we tested two approaches using chi-square: (1) we compared the proportions of both damaged and intact rosettes that flourished and those that did not flourish; (2) for damaged rosettes, we compared the proportions of anomalous rosettes and non-anomalous rosettes. We performed the analysis in the Bioestat program ( Ayres et al. 2007).