Scientists traditionally monitor bird nests with two outcomes in mind, either the nest is successful, meaning one or more chicks survive to fledge; or it fails.
In the case of the coastal plain swamp sparrow, a subspecies of the swamp sparrow that nests in mid-Atlantic tidal marshes, the failure of a nest might be due to contradictory factors: place the nest too low and it is vulnerable to flooding, place it too high and it becomes visible to predators.
A new method of analysis, Markov chain nest-failure models, was used to look at the different reasons for nest failure at a study site in Woodland Beach, Delaware.
The time of year, and the particular year, were found to be the best explanations for a nest's fate. The swamp sparrow nesting season is long, from early May to the end of August. Earlier nests fare better.
The amount of vegetation obscuring the nest was also important, but the surprise was that the more cover, the worse the nest fate. This may be due to the fact that the marshes the birds nest in are slow to leaf out in the spring, there is more vegetation by the late summer, which is when the nests are more likely to fail. It is unknown why late season nests would be more likely to fail.
This article summarizes the information in this publication:
Etterson, Matthew A., Olsen, Brian and Greenberg, Russell S. 2007. The Analysis of Covariates in Multi-Fate Markov Chain Nest-Failure Models. Studies in Avian Biology, 34: 55-64.
In this manuscript we show how covariates may be included in Markov chain nest-failure models and illustrate this method using nest-monitoring data for Coastal Plain Swamp Sparrows (Melospiza georgiana nigrescens) from Woodland Beach Wildlife Area, Delaware. First, we explore hypotheses for nest failure as a single event class, which is the converse of modeling covariates to survival. We then generalize to consider separate covariates to two classes of nest failure - predation and flooding. Temporal variability, both within and between years, was the most important factor for describing daily nest failure probabilities, though percent cover around the nest also received strong support. The Markov chain estimators for a single class of failure are likely to be similar to other generalizations of the original Mayfield estimator. The estimators for modeling two or more classes of failure should prove useful, but must be employed with caution. They are sensitive to nest-fate classification errors and they can lead to a proliferation of models, which could result in over-fitting.
Teachers, Standards of Learning, as they apply to these articles, are available for each state.