Multi-state, multi-stage modeling of nest-success suggests interaction between weather and land-use

Many factors may affect daily nest survival. We present a novel multi-state, multi-stage model to estimate daily survival for each nest stage, daily hatching probability and probability that a failed nest died during a specific stage when stage of failure is unknown. The model does not require that hatching date be known. We used data from a large citizen science dataset to demonstrate the application of this approach, exploring the impact of laying dates, weather conditions, conserved soil moisture, soil carbon, habitat type and urbanization on failure rates of common blackbird (Turdus merula) nests. Models selected and estimates of nest success were similar to those of the simpler logistic exposure method, but accounted for additional uncertainty. Simulations suggest the multi-state approach performs better when incubation mortality is affected by nest age, but not when incubation mortality is assumed constant. Both approaches worked best when date of incubation initiation was known for all nests first visited during the incubation stage. Daily blackbird survival probabilities were higher in human rural habitat than in urban or countryside habitats supporting the hypothesis that these intermediate habitats offer a better balance between low food availability in urban areas and high predation rates in the wider countryside. Nest success was influenced more by recent precipitation in urban habitats, but by a longer-term measure of water availability, soil moisture, in non-human dominated habitats, indicating that climatic change is likely to alter relationships between habitat and breeding success (and their temporal scale) by influencing the trade-off between food availability and predation rates. The multi-state, multi-stage model developed here may be helpful to other researchers modelling ecological processes in which transition probabilities between multiple stages are of interest.