A frog's-eye view of the landscape : quantifying connectivity for fragmented amphibian populations

The spatial habitat requirements are studied for two amphibian species: the tree frog ( Hyla arborea ) and the moor frog ( Rana arvalis ). Fragmentation, the destruction of suitable habitat, results in small fragments that are separated by unsuitable habitat or barriers. Metapopulation theory implies that a species can survive on a regional level if local extinctions are compensated for by recolonisations. For nature conservation it is relevant whether species can cope with habitat destruction or have crossed the viability threshold. Connectivity for ground-dwelling species with low dispersal capacity, such as amphibians, depends on the distance and the relative resistance of the landscape between suitable habitat patches.

In chapter 2 the distribution pattern of the tree frog in Zealand Flanders is analysed. The occurrence of apparently suitable but unoccupied patches and the fact that the occupation probability increases with higher connectivity are indications that the distribution pattern is affected by habitat fragmentation.

Chapter 3 focuses on the distribution pattern of the moor frog in south-west Drenthe. The positive effect of pond size and the negative effect of road density on the probability of occupation indicate negative effects of habitat fragmentation.

In chapter 4 the genetic distance between populations of the moor frog is studied, as an indirect indicator of dispersal. The genetic distance correlates positively with geographical distance between populations, especially when distance is weighted for (rail)road density. This points to some exchange between local populations and that the connectivity between ponds is reduced by (rail)roads.

In chapter 5 it is concluded that the tree frog in Zealand Flanders functions as a metapopulation. Firstly, extinctions take place regularly and extinction probability decreases with patch size. Secondly, empty patches are recolonised and colonisation probability increases for large patches with high connectivity. Finally, the observed dispersal distances and the recolonisations show that the habitat network is still connected by dispersing individuals.

It is discussed that the tree frog and the moor frog differ in their sensitivity to habitat fragmentation. The tree frog, a species of unpredictable habitat, is better adjusted to habitat fragmentation. However, the tree frog is probably no longer viable in Zealand Flanders and the degree and speed of fragmentation in the Dutch agricultural landscapes may have exceeded a critical threshold. The habitat of the moor frog is less fragmented and the species seems to be still viable. However, the present distribution may be still linked to the former continuous landscape.

Chapter 6 further focuses on connectivity. Radio-tracking experiments with displaced tree frogs show a preference for hedgerows and an avoidance of arable land. The calibration of the movement model SmallSteps shows the best fit between observed and simulated movement paths, when transition probabilities are strongly biased towards hedgerows. It is concluded that the resistance of the landscape is an important element to determine connectivity for the tree frog and that hedgerows may function as corridors that will enhance dispersal.

In chapter 7 a general framework is developed to predict viability for a broad array of species. Empirical data and model simulations illustrate that species have different scale-dependent responses to habitat fragmentation. The concept of ecological profiles is developed to group species according to characteristics that are important in responses to landscape fragmentation. The ecological profiles are transformed into two ecologically-scaled landscape indices (ESLI): 'average patch carrying capacity' and 'average patch connectivity'. The empirical data show a positive correlation between the fraction of occupied habitat patches (an important predictor of metapopulation viability) and the ESLI.

Several tools and their potentials for application in landscape management are given. Statistical models give guidelines for the design of optimal amphibian landscapes. Simulation models render standards for viable population networks and optimal corridor design that can play a role in landscape evaluation and the comparison of planning scenarios.