Email: steven.brady@yale.edu
Office: Greeley Laboratory, Room 119
Phone: (203) 432-5321
Fax: (203) 432-3929

EDUCATION

RESEARCH INTERESTS

Rapid Evolution

Over large landscapes, genotypic frequency of meta-populations may exhibit a normal distribution.  However, local scale investigation may reveal important patterns of adaptations that shift the distribution of alleles away from the population mean.  Recent experiments conducted in the Skelly lab have demonstrated that development rate variation within local populations of wood frogs is tantamount to that seen across global populations (e.g. temperate to arctic).  Wood frog larvae originating from ponds just meters apart exhibit patterns of development rate countergradient variation that can be predicted by wetland canopy cover.  This finding suggests that wood frogs may evolve rapidly in response to breeding pond changes, and that evolution is an important driver of patterns of distribution and abundance of wetland animals living in dynamic and heterogeneous environments. 

Host parasite ecology

Echinostomes are multi-lifecycle trematodes known to parasitize hosts (two intermediate and one definitive) across multiple trophic levels.  In this study system, freshwater snails play the role of primary intermediate host, while frogs (or even other snails) act as secondary intermediate hosts; mammals and birds comprise the category of definitive hosts.  Members of the Skelly lab recently conducted a study of echinostome infection in green frogs (Rana clamitans), which demonstrated a positive correlation between urbanization gradient and infection rates: 99% of total infections were found in frogs originating from urban wetlands.  This study indicates that snail density increases with urbanization and predicts echinostome infection rates.  However, the question remains: what mechanisms associated with urbanization generate the observed infection gradient?    

In order to address this question, I am currently conducting research to identify definitive echinostome hosts acting in Connecticut permanent ponds.  Using a suite of methodology, including frog and snail sampling, avian point-counts, mammal detection stations, and GIS analysis, I hope to better understand species assemblage across an infection and urbanization gradient.  Following these results, I will be screening candidate hosts for infection in order to further our understanding of trematode infection patterns.

Predicting marbled salamander (Ambystoma opacum) distributions through remotely sensed imagery

Annual springtime densities of marbled salamanders (Abystoma opacum) have been empirically documented for a number of autumnal ponds at the Yale-Myers forest for the years 2000-2006.  Reproductive life history traits of A. opacum are unique among the genus.  Rather than breeding in vernal ponds, A. opacum breed on land adjoining reduced ponds or in dried depressions, where fertilized eggs remain relatively dormant until fall rains transform the egg repository into an autumnal wetland.  During the winter months, little development occurs, and juvenile A. opacum seek refuge in the unfrozen waters and mud below the frozen surface of the pond.  Unlike many salamander species that possess the capacity for withstanding sub-freezing temperatures, A. opacum cannot survive in a pond that freezes to depth; A. opacum require unfrozen water or mud in order to persist through the winter. 

Two parameters govern the degree to which a pond will freeze: temperature regime and hydrology.  Given cold enough temperatures, pond freezing is contingent upon the absence of ground or water infiltration; ponds receiving ground water in sub-freezing temperatures typically do not freeze across the entire surface due to mixing.  Conversely, shallow ponds exhibiting complete surface ice typically freeze to depth.  In this study, surface ice was used as a surrogate for ice to depth.

I conducted supervised and unsupervised classifications on a multi-spectral IKONOS image of the Yale-Myers Forest Union, CT captured on 18 March 2003. The four-meter resolution spectral bands (RGB and NIR) covered a swath size of 8 x 10 km.  I quantified pond attributes such as ice coverage and frozen wetland vegetation to generate a logistic regression predicting A. opacum presence.

While preliminary results suggest this approach will prove useful, the inference of this project was limited by small-scale spatial heterogeneity abstracted by the four-meter spatial and four-band spectral resolution imagery.  I am interested in further exploring the capacity remotely sensed high-resolution hyper-spectral imagery offers for predicting species distributions in heterogeneous environments.