The Specifics of Avian Natural History and Their Relevance to Conservation of Declining Species.
Kamm, Matthew.
2019
-
This dissertation comprises a series of studies relating avian natural history to conservation challenges and outcomes. I applied a combination of fieldwork, remote sensing, geospatial analysis, and modeling of both short- and long-term ecological datasets to illustrate the importance of understanding specifics of natural history when designing ecological experiments and interpreting ecological ... read moredata. The first chapter is a broad study of migratory songbird species banded at Manomet on the southeastern coast of Massachusetts, USA, from 1969-2017. I used a Bayesian state-space modeling approach to derive abundance trends for migratory cohorts of 70+ passerine species in spring and fall. This work demonstrated that, while categories such as "Neotropical migrant" and "aerial insectivore" can be useful, they tell us remarkably little about what conservation status to expect in a given species. Broad life-history groupings proved to be a poor predictor of trend shape and direction, while migratory timing (especially in the spring) proved a more reliable predictor. The specifics of where and when each individual species migrates, and what breeding and wintering populations are sampled at Manomet, proved to be key considerations in interpreting the multitude of trends. The remainder of the dissertation deals with the American Kestrel (Falco sparverius sparverius), a small, widespread raptor species that has been declining in many areas for several decades. The reasons behind the decline remain unclear, but habitat loss has been theorized to play an important role. Accordingly, I set out to understand American Kestrel habitat requirements from a multiscale perspective. I installed light and temperature loggers inside nest boxes and evaluated the (often conflicting) findings and recommendations about the importance of interior light, temperature, and cavity orientation from four decades of previous research into kestrel microhabitat. My work demonstrated that, in Massachusetts, interior light and temperature are remarkably uncorrelated, and cavity orientation has no effect on box occupancy. Openness of the immediate surrounding habitat turned out to be a much better predictor of occupancy by kestrels. Moving up to the scale of the habitat surrounding the nest box, I pioneered the use of an off-the-shelf consumer grade camera drone with a visible spectrum camera to collect useful and specific data on habitat surrounding several nest boxes. The aerial photos were classified with good (Kappa > 0.6) or better accuracy, and the supervised classification methodology proved robust to application by users who had not visited the site being classified. Finally, I collaborated with another nest box program in the Madison, WI, USA area to develop multiscale predictive habitat models for kestrel occupancy. We integrated data from the local habitat patch (100m-400m from the nest box), the territory level (1000m-1500m), and the landscape level (5000m) to parameterize a true multiscale model for each region. Both models predicted their own region's occupancy with an AUC of the ROC curve > 0.7, but fared much worse at classifying the other region's occupancy. These models also identified new variables of potential interest to kestrel managers, such as the importance of emergent wetlands to nest box occupancy.
Thesis (Ph.D.)--Tufts University, 2019.
Submitted to the Dept. of Biology.
Advisor: Michael Reed.
Committee: Michael Romero, Elizabeth Crone, and Morgan Tingley.
Keyword: Ecology.read less - ID:
- rx914262s
- To Cite:
- TARC Citation Guide EndNote
