The relative importance of biotic and abiotic determinants of temporal occupancy for avian species in North America

Abstract

Aim We examined the relative importance of competitor abundance and environmental variables in determining the species distributions of 175 bird species across North America. Unlike previous studies, which tend to model distributions in terms of presence and absence, we take advantage of a geographically extensive dataset of community time series to model the temporal occupancy of species at sites throughout their expected range. Location North America. Time period 2001–2015. Major taxa studied One hundred and seventy‐five bird species. Methods We calculated variation in temporal occupancy across species’ geographic ranges and used variance partitioning and Bayesian hierarchical models to evaluate the relative importance of (a) the abundance of potential competitors and (b) the environment (elevation, temperature, precipitation, vegetation index) for determining temporal occupancy. We also created a null model to test whether designated competitor species predicted variation in temporal occupancy better than non‐competitor species. Results On average, the environment explained more variance in temporal occupancy than competitor abundance, but this varied by species. For certain species, competitor abundance explained more variance than the environment. Migrant species with smaller range sizes and greater range overlap with competitors had a higher proportion of variance explained by competitor abundance than the environment. The abundance of competitor species had a stronger effect on focal species temporal occupancy than non‐competitor species in the null model. Main conclusions Temporal occupancy represents an underutilized method for describing species distributions that is complementary to presence/absence or abundance. Geographic variation in temporal occupancy was explained by both biotic and abiotic drivers, and abiotic drivers explained more variation in temporal occupancy than abundance on average. Species traits also play a role in determining whether variation in temporal occupancy is best explained by biotic or abiotic drivers. The results of our study can improve species distribution models, particularly by accounting for competitive interactions.

A simulation study of the use of temporal occupancy for identifying core and transient species

“…However, despite such a potential bias, past work has shown that there is typically a positive relationship between either temporal occupancy or species richness and remotely sensed proxies for productivity, meaning the observed patterns were opposite what would be predicted purely from a detectability effect (7,17,23). As such, these patterns of occupancy and richness were observed despite, and not because of, geographic variation in detectability.…”

“…reproductive rate, dispersal distance), the inferences that can be made from this simulation model are more broadly generalizable. We chose to focus on birds because they are highly mobile, can disperse widely, and have been studied empirically in this core-transient context (3,7,23). These first two attributes make temporal occupancy particularly useful for identifying core and transient birds in communities, but also potentially more prone to errors due to sourcesink dynamics.…”

A simulation study of the use of temporal occupancy for identifying core and transient species

“…Several studies have used temporal occupancy to infer the persistence of populations over time and the degree to which a species can be considered a core member of a community in a particular location [ 3 , 7 – 9 , 23 ]. Our simulations showed that in many realistic scenarios, this is a valid approach, but also confirmed that temporal occupancy is subject to misclassification errors where core species are inferred to be transient and transient species are inferred to be core.…”

“…Geographic patterns in the relative prevalence of core and transient species can influence our understanding of ecological communities when failing to recognize this distinction [ 7 , 23 ], especially if the probability of misclassification varies geographically. One likely source for this is detection probability, which is thought to vary along environmental gradients.…”

“…We ran simulations for 200 time steps, which was long enough for species richness to achieve equilibrium in the landscape, and used the last 15 time steps to calculate temporal occupancy. Fifteen time steps represented an ecological dataset with a 15-year time series, a sampling period used in several previous studies which provides a reasonably high resolution estimate of temporal occupancy [ 7 , 23 ]. Additionally, we calculated landscape-wide abundances for each species at the end of the simulation.…”