Gearing up to start the Wood Thrush report for the PA Game Commission and digging into the literature. The two papers I just read were:

Kellner, K. F., et al. (2018). “Local-scale Habitat Components Driving Bird Abundance in Eastern Deciduous Forests.” The American midland naturalist 180(1): 52-65.

Ruhl, P. J., et al. (2018). “Characterization of Worm-eating Warbler (Helmitheros vermivorum) breeding habitat at the landscape level and nest scale.” Avian Conservation and Ecology 13(1).

The Ruhl paper on Worm-eating Warblers looks to be a subset of the data presented in Kellner et al. Both are out of Purdue, which is pumping out great ecological papers.

Both use the latest-greatest in analyzing point count data (where you stop at particular distances and count all birds – in their case every 150 m) by using Bayesian methods and incorporate detection probabilities (the probability that you discover a bird that is actually there).  They throw a number of landscape variables at abundance data. They used some sophisticated (for me) methods to extract landscape data that I will need to learn. I have used R to many analyses but not really get landscape information from maps. It will be an adventure.

The detection probabilities ranged from 0.08 to 0.83.  The detection probability of 0.08 was associated with Brown-headed Cowbird (BHCO), indicating a bird that is more widespread than what encounters would indicate. BHCO is a brood parasite and not territorial so I wonder how that affects the estimates. Interesting compared to my point counts which had many BHCO. On the other were Eastern Wood-Pewee and Ovenbird – which sing consistently during the breeding season.

Canopy height , elevation, and slope were the most important variables for many species – but not Wood Thrush (WOTH). Midstory stem density and slope were most important for WOTH . This species preferred sites with steeper slopes and more open midstory.

Just a word of caution. I worry in using the word “important” for variables and using random sites to find what’s important. If a has a strong relationship to an environmental variable but the value of that preference is centered on the average then that variable might be interpreted as not being important at all. The average value of a variable is entirely dependent on the choice of habitats selected.  So, for example, this study only included forested sites but if they had included clearcuts or floodplains then the average would shift and, perhaps, other variables would become important.

• Uncategorized