Nest survival modelling using a multi-species approach in forests managed for timber and biofuel feedstock.
Switchgrass (Panicum virgatum) intercropping is a novel forest management practice for biomass production intended to generate cellulosic feedstocks within intensively managed loblolly pine-dominated landscapes. These pine plantations are important for early-successional bird species, as short rotation times continually maintain early-successional habitat. We tested the efficacy of using community models compared to individual surrogate species models in understanding influences on nest survival. We analysed nest data to test for differences in habitat use for 14 bird species in plots managed for switchgrass intercropping and controls within loblolly pine (Pinus taeda) plantations in Mississippi, USA. We adapted hierarchical models using hyper-parameters to incorporate information from both common and rare species to understand community-level nest survival. This approach incorporates rare species that are often discarded due to low sample sizes, but can inform community-level demographic parameter estimates. We illustrate use of this approach in generating both species-level and community-wide estimates of daily survival rates for songbird nests. We were able to include rare species with low sample size (minimum n=5) to inform a hyper-prior, allowing us to estimate effects of covariates on daily survival at the community level, then compare this with a single-species approach using surrogate species. Using single-species models, we were unable to generate estimates below a sample size of 21 nests per species. Community model species-level survival and parameter estimates were similar to those generated by five single-species models, with improved precision in community model parameters. Covariates of nest placement indicated that switchgrass at the nest site (<4 m) reduced daily nest survival, although intercropping at the forest stand level increased daily nest survival. Synthesis and applications. Community models represent a viable method for estimating community nest survival rates and effects of covariates while incorporating limited data for rarely detected species. Intercropping switchgrass in loblolly pine plantations slightly increased daily nest survival at the research plot scale (0.1 km2), although at a local scale (50 m2) switchgrass negatively influenced nest survival. A likely explanation is intercropping shifted community composition, favouring species with greater disturbance tolerance.