Wildfires and mass effects of dispersal disrupt the local uniformity of type I songs of Hermit Warblers in California

Abstract

Hermit Warblers (Setophaga occidentalis) sing a formulaic, type I song to attract mates, in contrast to a repertoire of more complex, type II songs to defend territories. A single, dominant type I song, or a low diversity of type I songs, often occur within a geographic area. We provide the first comprehensive description of Hermit Warbler type I song variants throughout California, USA. We recorded type I songs from 1,588 males across 101 study sites in the state from April through July 2009–2014. Using those locations and a pre-existing range map of the species, we created a maximum entropy-based breeding habitat suitability map and classified the songs into 35 variants using a typological rubric. We validated consistent classification of songs for 87.5% of the birds. We then modeled the effects of recent fire history at the local scale (10 yr, 315 km2), the amount of breeding habitat at the regional scale (8,000 km2), and the distance between territories to examine factors involved in song sameness at the local scale. We found that the probability of different birds singing the same form declined with the amount of local fire, regional habitat, and distance, and that these findings were robust to uncertainty in our song classification rubric. Using a longitudinal analysis including additional data from 10 study areas revisited in 2019, we showed that song structure within forms had drifted since our initial visits 5–10 yr earlier, and that the evenness (e.g., Simpson’s measure) of song forms increased at locations that had been burned by wildfire between visits. Taken together, the results suggest that wildfires and the mass effects of dispersal of birds singing rival song forms disrupt the uniformity of type I songs locally. The results demonstrate how species traits, such as birdsong, can be used to disentangle the ecological processes that regulate observed patterns in biodiversity. Further investigation is recommended to determine whether song pattern dynamics reflect underlying genetic differences and habitat specializations among subpopulations.

Can the Hermit Warbler ( Setophaga occidentalis ) serve as an old-forest indicator species in the Sierra Nevada?

“…Finally, the bioacoustic approach (passive acoustic surveys and machine learning-based species observations) is not without error, even with our highly conservative threshold (pr[true positive] ≥ 0.99 required for inclusion). In particular, detection accuracy could vary by dialect type, which itself varies systematically with the phenomena in question-large, severe fires (Furnas et al 2020). Our customization of BirdNET to this ecosystem and the large quantity of data collected could mitigate these issues.…”

“…We analyzed the audio with the BirdNET algorithm, a deep artificial neural network capable of identifying the vocalizations of > 95% of Sierra Nevada birds, including the Hermit Warbler and Spotted Owl (Kahl et al 2021). We used a customized version of BirdNET that incorporated training data from our study area, an important step because the Hermit Warbler is known to have diverse song dialects across California (Furnas et al 2020). We manually validated a random selection of Hermit Warbler predictions and used logistic regression to relate BirdNET's unitless confidence score to the probability that any given prediction was correct.…”

Recurrent Convolutional Neural Networks for Large Scale Bird Species Classification

“…In Figure 5(b),(c) we use (8) to show the projection coefficient variations over time with the maximum degree of 64. Directly evaluating the projections at each time-step t using ( 8) is not computationally feasible, especially when the time-horizon is large.…”

“…Further, visual monitoring is difficult for many small and elusive birds, for cryptic species 5 , and for species found in ecosystems difficult to reach for ecologists 6 . Besides, acoustic monitoring of birds is also helpful for other conservation activities, such as measuring forest restoration 7 , and studying the impact of wild fires 8 .…”

Comparing recurrent convolutional neural networks for large scale bird species classification

“…Further, visual monitoring is difficult for many small and elusive birds, for cryptic species 5 , and for species found in ecosystems difficult to reach for ecologists 6 . Acoustic monitoring of birds is also helpful for other conservation activities, such as measuring forest restoration 7 , and studying the impact of wild fires 8 .…”