Maintaining habitat connectivity is crucial for maintaining genetic diversity and reducing the risk of inbreeding depression in numerous species. Human-modified habitats pose a threat to population persistence, particularly for amphibians that require both aquatic and terrestrial habitats and often exhibit low dispersal capacity. In this study, we investigated the genetic structure and variation of two sympatric newt species, Triturus cristatus and Lissotriton vulgaris, in an urbanized and agricultural landscape. Using a multiscale resistance modeling approach, we evaluated how landscape features affect gene flow in both species. Our results confirmed that the less abundant T. cristatus exhibited more genetically distinct demes than the more common L. vulgaris, which showed a more subtle structure. Additionally, levels of genetic variation were lower in T. cristatus than in L. vulgaris, but with a similar spatial distribution pattern across shared ponds. We found that the proportion of managed grassland in the study area played a major role in reducing the level of connectivity in both species. Furthermore, considering multiple spatial scales proved to be effective in improving the fit of the landscape resistance models, with the largest scale (900 m) providing the best fit. In T. cristatus, areas with low proportions of trees or high proportions of crop fields negatively affected genetic connectivity. Notably, resistance surfaces optimized for each species were highly correlated. The intensively managed grassland lacked the structural heterogeneity necessary for newts. Therefore, switching to a more traditional management approach in the landscape to increase spatial patchiness at the scale of dispersal would benefit both species.

The study area, of approximately 130 km², is located in the municipalities Temse and Kruibeke (Belgium). In the period March-May 2018, during the newts’ breeding season, 55 ponds were sampled using one to five Dewsbury Newt Traps and four ponds using three to four traps type ‘Vermandel’, with the number of traps varying with pond size. Most ponds were visited three times (12 ponds were sampled twice and three ponds once. The sex of each adult newt was identified on the basis of the sexually dimorphic cloaca and on the presence of a crest in male newts. A cloacal swab was taken from each individual newt with cotton swabs. We aimed to collect 30 samples per species and sex in each pond. In June 2018, larvae were caught using dip nets, and tail clips were taken. We aimed at sampling 30 larvae per species and pond. In three ponds, no newts were found. Swabs were air-dried for several hours and stored at room temperature in the dark for a maximum period of one month. Tail clips were taken, preserved in absolute ethanol and stored at -20°C. 

Following DNA extraction, T. critatus was genotyped at 31 microsatellite markers, and L. vulgaris at 14 microsatellite markers via multiplex PCR and genotyping analysis on an ABI 3500 Genetic Analyzer. Allele calls were scored using the GeneMapper v4.3 software with fragment sizes based on GeneScan 600 LIZ Size Standard. Negative controls were included in each 96‐well PCR to allow for detection of reagent contamination. Reproducibility was evaluated using 3 % blindly replicated samples, two to five times within and across well plates. One reference sample was further added to each well plate. Samples with genotypes for less than 50 % of the loci were discarded. The overall error rate per locus was less than 1% for both species. Genotypes of recaptured great crested newts, poor quality genotypes and identified full sibs among larvae within ponds using the software COLONY 2.0.6.5 following a maximum likelihood approach were excluded. One monomorphic locus (TRCR421) in T. cristatus was removed. Three markers used for L. vulgaris were removed: LVG-542, which regularly produced more than two alleles, Tv3Ca19, which showed 33 % missing values, and LVG-398, which deviated from Hardy-Weinberg equilibrium and showed high proportions of null alleles in a substantial number of demes. After removing low-quality genotypes, recaptures and full siblings per family except one, 667 genotypes of T. cristatus (per pond: mean = 23; range = 1–50) and 2096 of L. vulgaris (per pond: mean = 40; range = 1–77) remained, collected in 29 and 52 ponds, respectively.

raw_Tcrist_genotypes

The CSV file shows the microsatellite genotypes of T. cristatus for all 31 microsatellite markers without low-quality genotypes, and with recaptures and full siblings still included. In the first eight columns, the following information is provided: the sample ID, the ID of the pond/deme, the life stage of the individual, the trap number indicating if (sub)adult newts were caught in the same or different trap on a particular day, the sex (m: male; f: female), sampling date (yyyy-mm-dd) and the longitude followed by the latitude of the pond (WGS84). The next 31 columns represent the different microsatellite markers with the loci names mentioned in the column headers. Alleles are separated by a forward slash. Unavailable or unknown information is denoted with “NA”, missing genotypes with “NA/NA”.

filtered_Tcrist_genotypes

The CSV file shows the microsatellite genotypes of T. cristatus without low-quality genotypes, recaptures and full siblings. The same information is provided in the first 7 columns and genotypes in the following 30 columns as in raw_Tcrist_genotypes.csv, except for trap number and the genotypes for TRCR421.

raw_Tcrist_genotypes

The CSV file shows microsatellite genotypes of L. vulgaris for 13 microsatellite markers without low-quality genotypes, and with recaptures and full siblings still included. Genotypes for locus LVG-542 are not included. The same information is provided as in raw_Tcrist_genotypes.csv

filtered_Lvulg_genotypes

The CSV file shows the microsatellite genotypes of L. vulgaris without low-quality genotypes, recaptures and full siblings. The same information is provided in the first 7 columns and genotypes in the following 11 columns as in raw_Lvulg_genotypes.csv, except for trap number and the genotypes for Tv3Ca19 and LVG-398.

xxxx_30m_b2km.asc

Rasters in ASCII format used in landscape genetic analysis. Each raster has a 30 m resolution and is in the Belgian Lambert 72 projection.

• arable_30m_b2km.asc: raster of proportion of arable land, derived from LifeWatch Wallonia-Brussels ecotope database (2 m resolution raster layer for Belgium for 2015).
• grassland_30m_b2km.asc: raster of proportion of managed grassland, derived from LifeWatch Wallonia-Brussels ecotope.
• ndvi_30m_b2km.asc: raster of normalized difference vegetation index derived from the color-infrared (CIR) orthophotos of Flanders.
• openwater_30m_b2km.asc: raster of proportion of open water, derived from LifeWatch Wallonia-Brussels ecotope database.
• ponds_30m_b2km.asc: raster with proportion of land covered by ponds derived from the Watervlakken version 1.1 layer and layers of known ponds in the area provided by the NGO ‘Regionaal Landschap Schelde-Durme’. We filtered the combined pond layers, retaining only ponds established before 2018 and with a surface of less than 1,350 m².
• prop_cat_wlasexclOpenwater_30m_b2km.asc: raster of proportion of land covered with waterways derived from the Flemish Hydrographic Atlas (VHA) for waterways of GDI-Vlaanderen edition 2018. Weights were first assigned to waterways according to their category. Non-classified systems were given a weight of 1, while the 3rd and 2nd categories received a weight of 50. First-category waterways and navigable waterways were assigned a weight of 100. Open water was masked from the waterway raster.
• tpi_30m_b2km.asc: raster with topographic position index derived from the 1 m digital terrain model raster DTM II, which is part of the digital elevation model Digitaal Hoogtemodel Vlaanderen II 2013-2015.
• trees_30m_b2km.asc: raster of proportion of land covered with trees, derived from LifeWatch Wallonia-Brussels ecotope database.
• urban_30m_b2km.asc: raster of proportion of urban land cover (which includes buildings and artificialized impervious surfaces), derived from LifeWatch Wallonia-Brussels ecotope database.