2.6 Correlations between prey items and relative specialization
For each scat, prey item proportions (number of sequences per prey species / total number of sequences generated) were lumped into orders and summed. We then performed correlations between the proportion of the diet that each order comprised in each sample and the \(\text{PS}_{i}\) for that sample. To examine if similar patterns occurred in both sexes, the analysis above was completed for male and females separately. We used the Bonferroni method to correct alpha for multiple comparisons (alpha values are specified in table captions). Additionally, to determine if benthic species were associated with a more specialist diet, a correlation was run between the proportion benthic prey and specialization value for each scat. Due to the heteroscedasticity of the dataset, we used Spearman’s rank method for all correlations (Sokal & Rohlf, 2012). All correlation analysis was conducted in R 3.3.1. Because smaller \(\text{PS}_{i}\) values indicate higher levels of specialization, a negative correlation value suggests a positive relationship with specialization.