Modelling the relationship between native amphibian species and the non-native marsh frog

Due to increasing globalization the rate of non-native introductions has rapidly increased. This is likely to continue as climate change leads to increases in range of some species. In some cases, the effect of non-native species on biodiversity has been very severe. However, due to the complex nature of ecosystems it is sometimes difficult to determine if a non-native species is having a negative impact. This can be the case if interactions with other species are involved or native species are declining due to other threats such as habitat loss. This thesis investigates whether the marsh frog (Pelophylax ridibundus), a non-native species introduced to the UK in 1935, is affecting the distribution and abundance of the common frog (Rana temporaria) in Kent, south east England.

Species distribution modelling predicted marsh frogs to be present in areas where common frog presence was low. Many of these areas were in coastal regions with lots of watercourses that have higher salinity levels. These conditions are more suited to marsh frogs than common frogs, thereby explaining the predicted distributions. However, an area in the centre of Kent with high pond density was also predicted as less suitable for common frogs. This prediction fitted the hypothesis that in areas of high pond density common frog numbers have been reduced by the combined presence of marsh frogs, and great crested newts (Triturus cristatus). To test this hypothesis, a local level study compared the presence of common frogs in this high pond density area in Kent with ponds in Sussex where marsh frogs were absent. To make the comparison more meaningful, propensity modelling was used to match the ponds to be compared in the two areas by other characteristics as far as possible. Occupancy modelling was used to determine probability of detection for great crested newts and marsh frogs from survey data collected in ponds in Kent. This showed that accounting for variation in detectability did not increase the predicted occupancy of the survey ponds for these species. The presence of common frogs was found to be much higher in the ponds in Sussex. Logistic regression showed that common frogs were positively associated with shaded ponds, which marsh frogs tended to avoid. This suggested predation and/or competition by

v

marsh frogs on common frogs was unlikely because of these different habitat preferences. Therefore, a higher proportion of great crested newts in the survey ponds in Kent may be the main cause of the lower number of common frogs in that area.

Results from an eDNA metabarcoding analysis of water samples taken in 2014 from ponds in central Kent were obtained. The data provided the presence/absences of common frogs, great crested newts, and marsh frogs. Logistic regression showed that common frogs were not negatively associated with great crested newt presence. However, there was a much higher proportion of ponds with great crested newts compared to ponds with common frogs. In contrast, the proportion of ponds occupied by marsh frogs was very low. This supported the hypothesis that marsh frogs are unlikely to be the cause of lower common frog presence in the area. Common frog spawn surveys were conducted in 2017 on a subset of the same ponds analysed by the DNA metabarcoding in 2014. These showed a change in pond occupancy between 2014 and 2017. This could be due to natural changes in occupancy or metabarcoding could be detecting non-breeding common frog ponds and missing some breeding ponds.

Both landscape and local level studies have indicated that common frog presence is lower in an area of high pond density in Kent. This is unlikely to be caused by the presence of marsh frogs because of a difference in pond preference between common frogs and marsh frogs reducing the risk of predation or competition. There was also a relatively low presence of marsh frogs in areas that were showing lower proportions of common frogs compared to great crested newts. The high proportion of ponds occupied by great crested newts is more likely to be the reason for lower common frog presence. Therefore, more active measures to control the spread of marsh frogs is not required when considering conservation measures to protect common frogs.