Distribution and calling phenology of generalist frog species along a climate gradient

Knowledge of the factors influencing frog species occupancy and breeding phenology provides important ecosystem insight for natural resource management and preparing for ecosystem change in response to anticipated future climate warming. Frogs are considered to be particularly vulnerable to climate change, with warmer temperatures and altered hydrology expected to have major impacts on the distribution of species and their breeding phenology at the global scale. The primary objective of this thesis was to gain insight into the spatial and temporal niches of frog species, including their responses to short term and long term variability in temperate south eastern Australia, while carefully considering detectability. A secondary research objective was to develop a novel methodology to expedite accurate quantification of error associated with automated frog call recognition.

In this thesis, patterns of frog occupancy are assessed in the upper-mid Lachlan River catchment in temperate New South Wales. The sites selected followed a mean annual temperature gradient of approximately 4.5°C. Knowledge of the resident frog communities was very limited for the study area, thus baseline occupancy data are initially described. Ten frog species were recorded, of which, six generalist species displayed the highest levels of occupancy, L. tasmaniensis, L. latopalmata, C. parinsignifera, L. peronii, L. fletcheri and C. signifera. The non-detection of an additional 16 species, previously known from the region was explicitly considered in regards to their life history traits and the survey method employed. Species distribution modelling, accounting for imperfect detection, was used to statistically link habitat characteristics with species occupancy for the six most commonly encountered species. Mean annual temperature was identified as a limiting factor in explaining the distribution of two of the species, L. fletcheri and C. signifera. Increased species occupancy was associated with the availability of certain vegetation types (varying between species). More complex habitats were more likely to foster a higher diversity of species. Detection probabilities, according to both season and survey method, were calculated for each of the species using the software program Presence.

Call recorders were established along the climate gradient of the study area and the call phenology of the same six frog species was examined over a 15 month period (July 2014 until October 2015). Frog calling was continuously recorded for 5 minutes every hour at nine sites and automated audio recognition software was used to extract species-specific calling data. A novel methodology was developed to quantify the accuracy of automated audio recognition models to extract species call data and expedite this process for the large audio dataset. An accuracy metric was continuously calculated until the cumulative margin of error reached a threshold. Given that accuracy tended to decrease during periods of very low calling activity, data were temporally verified on days when detection counts were consistently low. Application of the procedure varied according to whether the species had a long (year-long) or short (seasonal) calling period. The procedure thus enabled time series data describing calling activity to be produced in a time efficient manner that includes valuable qualifying metrics describing statistical confidence.

Using the calling time series data, considerable intra- and inter-specific variation in diel and seasonal calling peaks were observed across the frog assemblages at the nine waterbodies. Distinct conspecific relationships between calling, temperature, rainfall and hydrology were identified using generalised mixed effects regression modelling. Ambient air temperature had a strong and disparate influence on calling by the six sympatric species observed. The four Litoria and Limnodynastes species called in response to warmer temperatures while the two Crinia species responded to cooler temperatures. Positive and negative associations were also made between calling and rainfall for all species except for L. fletcheri, which instead responded to the availability of aquatic habitat over time (water level). Temperature was likely a principal driver of calling phenology for these species, whereas rainfall and hydrology were more likely to affect population dynamics through the limited availability of aquatic breeding habitat.

Overall, this thesis identifies highly idiosyncratic habitat and breeding preferences for even widespread and opportunistic frog species for local habitat and climatic conditions. The disparate species relationships with ambient air temperature in particular has important implications for individual species and communities in response to changing habitat and climatic conditions in the future. Based on the findings, even subtle changes to short term and long term climatic variability in temperate south eastern Australia are likely to disrupt the temporal partitioning of frog species calling in the Lachlan catchment. The environmental triggers for species calling activity (taken as a proxy for breeding) identified in this study can therefore provide guidance to the management of water resources in order to maximise frog breeding in current and future climate change scenarios. The novel error quantification method also contributes to our capacity to continuously monitor ecological responses and thereby refine natural resource management actions.