Ecological niche variation in the Wilson's warbler Cardellina pusilla complex

Abstract

Wilson's warbler comprises three subspecies separated into two geographic groups: C. p. pusilla that breeds in eastern North America; and C. p. pileolata and C. p. chryseola that breed in western North America. Given the differences between the groups in genetics, morphology, habitat use, and population decline, we tested for ecological niche similarity in both their breeding and wintering distribution using niche modeling based on temperature and precipitation data. We first conducted an inter-prediction approach considering the percent of summer and winter localities of one group that are predicted by the potential distribution of the alternate group. We also applied a null model approach that compares self-predictions and pseudoreplicates of each group to indicate similarity, divergence, or indeterminate niche overlap. Finally, we compared ecological distances between and within groups using the Gower similarity equation. We found that the western group had an ecological niche of broader climatic conditions, while the eastern group had a narrower ecological niche. The interprediction approach showed that, for both summering and wintering ranges, ecological niche models of the western group predicted ∼50% of the observed distribution of the eastern group, whereas eastern group models predicted  18% of the western group distribution. The null model approach found that similarity in ecological niches was indeterminate, possibly due to the large area occupied by the two groups; but it suggests a more restricted set of climatic conditions of the eastern group distribution. However, the Gower coefficients demonstrated that the ecological distance between the two geographic groups was larger than the ecological distance within groups, indicating distinct ecological niches. Overall, our results support the hypothesis that the eastern and western groups of Wilson's warbler are two cryptic species; this should be taken into consideration for future analyses, particularly with respect to vulnerability categorization and conservation efforts.

Deciphering ecology from statistical artefacts: Competing influence of sample size, prevalence and habitat specialization on species distribution models and how small evaluation datasets can inflate metrics of performance

“…Model performance for partial ROC compared well with literature reported values (Adelino, Anjos, & Lima, ; Martínez‐Gutiérrez et al, ; Ruiz‐Sánchez et al, ; Yañez‐Arenas et al, ). Calculated with the independent evaluation data, the effects of sample size (Figure c,d) and habitat specialization (Figure b) on partial ROC matched results for AUC and literature expectations (Bean et al, ; Soultan & Safi, ).…”

“…These metrics measure model discrimination capacity, an important component of model performance, and are herein referred to as metrics of model performance. Partial ROC was calculated with an omission error threshold of 5% (Martínez‐Gutiérrez, Martínez‐Meyer, Palomares, & Fernández, ; Ruiz‐Sánchez, Renton, Landgrave‐Ramírez, Mora‐Aguilar, & Rojas‐Soto, ). To create confidence intervals and compare partial ROC from our models to random models, we resampled 50% of each dataset with replacement for 500 iterations.…”

“…Prevalence appeared to have little effect on model performance as measured by AUC. As landscape prevalence is an unknown latent Ruiz-Sánchez et al, 2015;Yañez-Arenas et al, 2018).…”

Climatic patterns in the establishment of wintering areas by North American migratory birds

“…; Ruiz‐Sánchez et al. ). This would incur confusion in any predictions regarding their winter distribution.…”

“…Notwithstanding the previous results, we found two exceptions to the general selective establishment hypothesis pattern, namely Cardellina pusilla and Oreothylpis ruficapilla, but this may have been the result of taxonomic issues. For instance, recent studies suggest that breeding populations of these species may actually constitute independent lineages and therefore represent cryptic species (Kimura et al 2002;Clegg et al 2003;Ruiz-S anchez et al 2015). This would incur confusion in any predictions regarding their winter distribution.…”