Remnant patches and paddocks
We found that some frog species occurred more frequently in either the
remnants or the paddocks. Attributes of remnant woodland vegetation, such as canopy cover and understorey characteristics, have previously been linked with amphibian richness and abundance in modified landscapes (Hazell et al. 2001; Russildi et al. 2016). For example, we found that Uperoleia laevigata was positively associated with remnant woodland patches. Remnant vegetation can provide habitat, reduce desiccation risk, provide calling sites, and link water bodies (Hazell et al. 2001; Parris and
McCarthy 1999).
We also found that other terrestrial habitat types were important for frogs in agricultural landscapes. For example, in addition to remnant vegetation providing habitat, paddocks appeared to be suitable habitat for two species. C. parinsignifera was significantly more likely to occur in paddocks than remnants, and L. tasmaniensis was likewise more likely to occur in paddocks than remnants with the exception of planting transects. Higher capture rates in paddocks can, in some cases, indicate higher rates of movement rather than habitat suitability. Our recapture findings, however did not support this as we found no difference in probability of movement between recapture instances for remnants and paddocks (in prep). These results also highlight the contrasting habitat requirements amongst species in the frog assemblage.
The influence of habitat patch size and isolation on populations has long occupied ecological attention (Andrén 1994; Prevedello and Vieira 2010). Few studies have examined the influence of vegetation patch size on amphibians, although Russildi
et al. (2016) described a positive association between amphibian abundance and patch size. Our findings indicate that patch size did not influence the frog assemblage,
suggesting these species are not responding to discrete land cover type differences such as patch vs matrix (sensu Forman 1995; MacArthur and Wilson 1967) with many types of land cover in human-modified landscapes providing habitat for some species
(Kennedy et al. 2011; Prevedello and Vieira 2010).
Grazing regime
Species richness was more strongly associated with rainfall in continuously grazed properties than in rotationally grazed properties. This difference between grazing regimes may be due to the positive relationship between rainfall and ground cover height (62% correlation), and ground cover height had a positive influence on several species in this study. Continuously grazed properties had shorter ground cover on
average and the taller ground cover associated with higher rainfall may have brought the ground cover up to a threshold or height that better provides shelter and reduces
desiccation risk (Parris and McCarthy 1999). Some other studies have found that grazing effects on amphibians can be both species-specific and dependent on other factors such as interacting disturbances and habitat type (Badillo-Saldana et al. 2016; Buckley et al. 2014; Burton et al. 2009).
Paddock types
At the outset of this study we expected that land use type would influence frog species occurrence (Irizarry et al. 2016; Perfecto and Vandermeer 2010). In particular, we assumed that pasture would provide the lowest quality habitat of the four paddock types. However, rare species and L. tasmaniensis were the most abundant in open pasture. Open pasture was the most structurally simple paddock type, which could allow frogs to have an increased perceptual range relative to the other paddock types that had more complex structures. Increased perceptual range can reduce mortality by improving an animal’s ability to navigate and move through the landscape effectively (Zollner and Lima 1997). However, reduced mortality from greater perceptual range may come at a cost of increased desiccation risk in shorter vegetation (Parris and McCarthy 1999). The idea that increased perceptual range allows greater ability to move in open pasture is supported by Kay et al. (2016a), who found that shorter, more open agricultural matrix types resulted in increased perceptual range and movement of an arboreal gecko.
Environmental variables
We found that most frogs in our study responded to interactions between, or additive effects of, land management types and environmental variables, especially water body variables, climatic variables and ground cover height. We found frog
abundance and richness decreased as distance to water increased, a finding supported by several other studies (e.g. Russildi et al. 2016; Westgate et al. 2012). All of our study species require water bodies to breed and many amphibian species travel only a few hundred metres between breeding sites and terrestrial habitat (Becker et al. 2010; Semlitsch 2008). Intriguingly, total abundance exhibited a more nuanced response to distance to water. In paddocks, frog abundance reduced only a small amount as distance to water increased, but in remnants, abundance started higher and dropped more steeply as distance increased (Fig 3a). Frogs present in paddocks may be different species or have different traits to the individuals found in the remnants. These individuals in the paddocks may have had greater movement ability, had behaviours that reduce
desiccation risk, or were more tolerant to open human land use (Gascon et al. 1999; Russildi et al. 2016; Tracy et al. 2010),which resulted in the limited change in abundance in paddocks.
We found a reduction in frog abundance and richness in sites with higher average temperatures, consistent with the continental negative trend of amphibian species richness with temperature (Powney et al. 2010). Additionally, total abundance was higher in sites with higher average rainfall, which is known to influence amphibian reproduction events, population dynamics, and movement (Davis and Roberts 2011; Wassens et al. 2013). The effects of temperature and rainfall likely moderate the effects of the land management on the frog assemblage. For example, Rotem et al. (2016) found that grazing impact on reptiles in Israel differed according to the climate of the region, and Urbina-Cardona et al. (2006) showed the difference in rainfall across the wet and dry seasons influenced amphibian and reptile responses in a tropical
agricultural/forest environment.
Taller vegetative ground cover can provide greater protection from predation for small ground-dwelling animals like frogs. For example, Sato et al. (2014) found greater rates of predation of lizard models in mown ski runs than taller undisturbed grassland. Similarly, Larson (2014) found that adult amphibians were positively associated with tall grass height and high percentage grass cover. In our study, the most abundant
species (L. tasmaniensis) and rare species richness increased with taller ground cover in remnants but there was no relationship in paddocks. This difference may be due to different types of predation risk occurring in the more treed remnants compared to open paddocks, with other studies showing that vegetation structure can be an important determinant of predation risk (Arthur et al. 2004; Posa et al. 2007). Frogs may be more likely to be predated in remnants where there is little cover, or actively seek greater cover in remnants more than in paddocks. This would mean that the height of the ground cover is a more important determinant of shelter from the types of predation occurring in the remnants. Alternatively, the differing response to vegetation risk could be related to ease of movement, as taller and more complex ground vegetation can impede movement compared to more open cover (e.g. Prevedello et al. 2010).