Integrating age structured and landscape resistance models to disentangle invasion dynamics of a pond-breeding anuran

Modelling population dynamics of invasive species may help to propose effective management countermeasures. Invasion dynamics generally show recursive patterns across species and regions, where initial lag is followed by spread and eventual dominance phases. However, timing and modes of these phases are highly variable, emerging from the interplay between traits of the invader and characteristics of the invaded landscape. Disentangling this interplay is particularly arduous in species with complex life-histories, where an individual passes through different life stages that alter physiology, behaviour and interactions with the environment. Here, we describe an age structured model that can be utilized to simulate population dynamics of invasive pond-breeding anurans. The model follows a spatially structured population approach, each pond representing a discrete habitat patch that exchanges individuals with other similar patches, and simulates change in survival and dispersal behaviour as a function of age. It also integrates dispersal with landscape complexity through landscape resistance modelling to depict functional connectivity across the pond network. Then we apply the model to a case study, the invasion of the guttural toad Sclerophrys gutturalis in Cape Town, first detected in 2000. Age-structured demographic and spatial dynamics of the focal population are reconstructed in a network of 415 ponds embedded in a heterogeneous landscape. Parameterization is conducted through field and laboratory surveys, a literature review and data collected during an ongoing extirpation from 2010. We use the model to explore: i) occurrence and duration of lag phase; ii) whether the spatial spread fits an accelerating or a linear trend; iii) how simulated dynamics match field observations. Additionally we test model sensitivity to demographic and behavioural traits. We found a lag phase in both demographic and spatial dynamics; however the lag duration of these dynamics does not coincide, where invaders start to spread across the pond network five years before the demographic explosion. Also, we found that the spatial spread fits an accelerating trend that causes complete invasion of the network in six years. Such dynamics noticeably match field observations and confirmed patterns previously detected in other invaders characterized by high dispersal abilities. Sensitivity analysis suggests that it would have been preferable to quantify initial propagule size and post-metamorphic survival in the field; both timing and modes of invasion are particularly sensitive to these parameters. We conclude that the model has potential to forecast amphibian invasion dynamics and test management countermeasures.