Climate change has the potential to alter the strength of species interactions, but not only do we lack sufficient information about possible species responses, we also lack the tools to monitor and document these changes over meaningful spatial scales. Due to their physiology and life history, ectotherms such as amphibians, are particularly sensitive to changes in prevailing environmental conditions. Amphibians breed in a variety of standing water body types and one of the main structuring variables is the amount of time the water body holds water, the hydroperiod. The province of New Brunswick in Canada is expected to warm by ~4.5°C and see anywhere from a 5-50% increase in precipitation over the next 100 years and, as a result, it is anticipated that pond hydroperiods could change. I conducted translocation experiments with wood frogs (Lithobates sylvaticus) to determine the effect of different hydroperiods on embryo and larval survivorship. Permanent ponds that contained overwintered green frog larvae (Lithobates clamitans) had lower embryo and larval survival than both permanent ponds without green frog larvae and ephemeral ponds. To investigate the population-level impacts I conducted simulations using a population projection model of different climate change and hydroperiod scenarios. I found that wetter conditions (i.e., longer hydroperiods) had a negative effect while drier conditions (i.e., shorter hydroperiods) had a positive effect, at least initially, on wood frog populations. In order to facilitate monitoring improvements for these and other population-level impacts, I used automated recorders and automated sound recognition models. I found that choice of variable settings had a greater impact on recognizer performance than the amount and type of training data, but increases in the number of sites used did improve performance slightly, and that by optimizing variable settings on new data it was possible to create reliably transferable models that minimized false negative (Type II) errors but not false positive (Type I) errors. I then used these bioacoustic monitoring tools to predict abundance in wood frog populations based on vocalization activity. The best model predicted egg mass abundance with an average absolute error of 16 masses and a relative error of 59%. Overall, I have provided evidence of an additional mechanism by which climate change can alter amphibian communities, as well as developed and evaluated a methodology to detect those alterations.
Zircon - This is a contributing Drupal Theme Design by WeebPal.
