Distribution of amphibians and reptiles in agricultural landscape across Europe

Evaluating herpetofauna presence and the species-specific and species richness patterns in response to agricultural landscape features is essential for understanding the herpetofauna decline in agricultural landscapes. This work aimed to explore how different categories, extent and heterogeneity of crops affect herpetofauna distribution and diversity patterns at different spatial resolutions: UTM 50 km2, UTM 10 km2 and GPS coordinates. Using presence-only data from online repositories we documented the occurrence of European amphibians and reptiles in crops and quantified crop extent and heterogeneity in 50 and 10 km2 grid cells, and in the recorded presence locations. We used logistic regressions to test the effect of crop extent on species occurrence and calculated the proportion of species showing a significant response to each crop category. We analysed species richness patterns with generalized additive models against crop extent, crop heterogeneity and crop categories extracted at GPS locations, as fixed effects. We recorded 71 amphibian and 143 reptile species at 50 and 10 km2 spatial resolutions, and 58 amphibian and 108 reptile species at GPS resolution. Our results showed that amphibian and reptile species presence and richness are influenced by crop category, extent and heterogeneity and that spatial patterns were scale dependent. Species richness of both amphibians and reptiles was generally negatively correlated with crop extent but was enhanced by crop heterogeneity. Our results provide useful information for future risk assessment of herpetofauna and conservation efforts to restore or sustain herpetofauna biodiversity in agricultural systems across Europe. We stress that the scale of landscape management may lead to contrasting outcomes.


Introduction
Herpetofauna (amphibians and reptiles) is the most threatened group of vertebrates in the world (Gibbons et al. 2000).Pollution, often pesticide-related, is recognised as a major driver leading to population declines in both amphibians and reptiles (Mann et al. 2009;Todd et al. 2010).This is particularly evident in agricultural areas, that have become one of the largest terrestrial biomes on Earth, occupying more than 40% of the land surface (Foley et al. 2005).For example, croplands receive three quarters of all pesticides which are employed in agricultural production.Agricultural areas include essential habitats for amphibians and reptiles.Thus, pesticide exposure could play a key role in the observed amphibian and reptile declines in agricultural landscapes (Todd et al. 2010; Mingo et al. 2016; Arntzen et al. 2017).Moreover, both amphibians and reptiles are also exposed to pesticides in habitats adjacent to agricultural landscapes.For example, amphibians are exposed to pesticide during migrations to and from spawning waters (Fryday and Thompson 2012).Pesticides can be transported as well via spray drift or run-off to water bodies used by amphibians for breeding or to sites with habitats suitable for reptiles (Ockleford et al. 2018;Adams et al. 2021).Yet, neither amphibians nor reptiles have been included in regulations concerning the environmental risk assessment (ERA) of pesticides and the available data for their risk evaluation is limited (Brühl et al. 2011).
To ll this knowledge gap and provide scienti cally sound and robust information for including amphibians and reptiles within the ERA, it is necessary to gather solid data on herpetofauna occurrence in agricultural landscapes and on amphibian and reptile species-speci c responses to their features (i.e.crop type and crop diversity).Moreover, the substitution of natural areas by croplands and the agriculture intensi cation (Tscharntke et al. 2012) affects species richness (Scales and Marsden 2008) and cause shifts in species composition (Newbold et al. 2016).It is acknowledged that less is known about the response of amphibian and reptile diversity to land use change than other groups, i.e. mammals and birds (Palacios et al. 2013;Newbold et al. 2014).Thus, it becomes evident that documenting how agricultural systems impact amphibian and reptile diversity patterns should receive priority.
Finally, the spatial pattern of biodiversity is scale dependent (Levin 1992).The resolution and the overall size of the study area can affect the observed biodiversity spatial pattern (Rahbek 2005).Few studies have investigated the agro-biodiversity at a continental scale (Billeter et al. 2008;Wilson et al. 2020;Mupepele et al. 2021).Small-scale studies may lead to suboptimal management of the agricultural landscape, if applied more generally (Billeter et al. 2008;Ekroos et al. 2016).In Europe, agricultural landscapes host a wide range of land use types, different sizes and shapes, and semi-natural elements that vary in abundance and pattern.These agricultural landscapes, but in particular those with a negrained landscape mosaics and low-intensity forms of agricultural production were formerly characterised by high biodiversity (Edwards et al. 1999).However, in the last decade, the biodiversity in agricultural landscapes has markedly decreased with the decline in semi-natural elements and the change of production systems towards a predominantly intensive agricultural mode (Krebs et al. 1999;Robinson and Sutherland 2002).Thus, updated scienti c knowledge on land use change and species diversity in agricultural landscapes across Europe, especially for endangered species such as amphibians and reptiles, is urgently needed.
In this study we document the distribution of amphibian and reptile species in agricultural landscapes across Europe to: (i) investigate their occurrence in crops, (ii) quantify the proportion of crops extent in 50km 2 (UTM 50km × 50km) and in 10km 2 (UTM 10km × 10km) grid cells, and (iii) explore how different crop categories affect herpetofauna distribution and diversity patterns at two different landscape resolutions, i.e. 50km 2 and 10km 2 .

Data sources
We compiled distributional data for amphibian species and reptile species native to Europe from the Atlas of European amphibians and reptiles (Sillero et al. 2014), and species occurrence data sources including Global Information Facility (GBIF: www.gbif.org),iNaturalist (inat: www.inaturalist.org),and VertNet (vertnet.org).We homogenised the databases by deleting all other information except species names, coordinates, and data source.We used the standard geographic coordinate system WGS84.We cleaned the data by removing records with coordinates accuracy below 100 m and duplicated records.To avoid mismatches due to differences in the species nomenclature between the databases we followed the taxonomy from Sillero et

Data analysis
We calculated the proportion of each of the ve crop categories within UTM 50 km 2 and UTM 10 km 2 grid cells.These datasets were used to derive maps of proportion of each crop category, of the most abundant eight species, and of amphibian and reptile species richness.To test the effect of crop category on the occurrence of amphibian and reptile species, we modelled the presence/absence of each species using a logistic regression (General Linear Model, GLM), with binary response and binomial distribution of errors and a logit link function.Then, we calculated the proportion of species for each crop category with a signi cant effect.To explain species richness patterns, we used a generalized additive model (GAM), with a Gaussian error distribution and an identity link function (Zuur et al. 2009) with regression splines (Wood 2003).We ran both GLM and GAM using the proportion of a crop category within 50 km 2 and 10 km 2 grid cell, respectively, as continuous xed effects and subject to smoothing for GAM.Models for UTM 10 km 2 resolution included a purely spatial term to account for spatial autocorrelation.The normality of residuals was tested using Q-Q plots (Sokal and Rohlf 1995).We report total deviance (%) and adjusted R 2 for the coe cient of determination as indicators of the explained variation of the full model.We performed all analyses using R 4.1.1(R Core Team 2021) software.

Distribution patterns of amphibians and reptiles and crops
We recorded 72 amphibian and 142 reptile species at UTM 50 km 2 and UTM 10 km 2 grids (Fig. 1).At 50 km 2 spatial resolution, amphibian species richness in crops was higher in Western-Central Europe, while for reptiles it was higher in the southern peninsulas, particularly the Iberian Peninsula and Greece.At 10 km 2 spatial resolution, the pattern is less continuous.Presence/absence maps of the eight most abundant amphibian and reptile species in agricultural landscapes across Europe at UTM 50km 2 and UTM 10km 2 resolutions are presented in the Online Resource 1 (EMS1_1, EMS1_2, EMS1_3 and EMS1_4).

Crops extents in grid cells occupied by amphibians and reptiles
At UTM 50km 2 , the most extended crop category with occurrences of amphibians and reptiles was the dry crop (Fig. 3a and 3b) while at UTM 10km 2 , the most extended one was agroforestry crop (Fig. 3c and 3d).
However, the differences between crop categories at 10km 2 presented less variability (Fig. 3).The GLMs on the presence/absence of each species indicated that at both resolutions all crop categories had a signi cant effect on more than 20% of the species: thus, both amphibian and reptile species are selecting speci c crop categories (Online Resource 2, EMS2_1).The species presence in agroforestry crops was signi cantly more likely than would be expected by chance for more than 60% of species at UTM 50km 2 spatial resolution, whilst at UTM 10km 2 , woody crops had a signi cant effect on more than 45% of the species (Online Resource 2, EMS2_1).When considering separately the proportion of species positively or negatively affected by crop categories, agroforestry crops affected positively the highest proportion of species at UTM 50km 2 spatial resolution (Fig. 4).Woody crops positively affected the highest proportion of species at both UTM 10km 2 (Fig. 4) while pasture negatively affected the highest proportion of species at both UTM 50km 2 and UTM 10km 2 spatial resolutions (Fig. 4).
The species richness of amphibians and reptiles was signi cantly correlated with the extent of crop categories at UTM 50km 2 and 10km 2 spatial resolutions, except with pasture and woody extent at UTM 10km 2 spatial resolution for amphibians (Table 1).Species richness generally decreased with the increase in the crop extent both at UTM 50km 2 and 10km 2 spatial resolutions, with some exceptions (Fig. 5).Amphibian species richness increased with the increase of agroforestry crop extent at UTM 50km 2 , while reptile species richness showed a signi cant positive relationship with the extent of irrigated crops both at UTM 50km 2 and 10km 2 resolution, and the extent of woody crops at UTM 50km 2 .Also, reptile species richness showed an initial decline with the increase of agroforestry crop extent to 0.3%, followed by an abrupt decrease to higher agroforestry crop extent at UTM 50km 2 (Fig. 5).Species richness increased signi cantly with crop heterogeneity for amphibians (at UTM 50km 2 spatial resolution, edf = 8.246, F = 296.6,P < 0.001; at UTM 10km 2 spatial resolution, edf = 5.996, F = 26.61,P < 0.001) and reptile (at UTM 50km 2 spatial resolution, edf = 7.884, F = 475.3,P < 0.001; at UTM 10km 2 spatial resolution, edf = 8.743, F = 3.16, P < 0.001).

Discussion
Our study indicated that the species presence and richness of amphibians and reptiles are in uenced by crop extent and crop category in agricultural landscapes across Europe.Moreover, we found that the observed spatial patterns of species richness and of crop extent were scale dependent.More speci cally, although the number of species occurring in crops did not differed between scales, the geographic location of peaks and troughs of species richness and of crop extent at coarse scale (grain size 50km 2 ) did not coincide with those at ner scale (10km 2 ).At coarse spatial resolution, amphibian species richness in crops was higher in Western-Central Europe, while for reptiles it was higher in the southern peninsulas.This was in general agreement with the distribution and geographic patterns of amphibians and reptiles of Europe (Sillero et al. 2014).The analysis at a ner scale led to less obvious hotspots and cold spots of species richness, due to the lack of homogeneous distribution data at that spatial resolution.In the case of crop extent, the smaller the areas, the lesser the crop categories, thus increasing the dominance of one crop category, especially on islands.
Our results showed that species presence may be affected by crop extent and crop category.Although all crop categories contributed signi cantly to explain species presence, our results revealed some scaledependent contrasts in crop in uence on species presence.For example, while agroforestry crop extent had a signi cant positive effect on most species at UTM 50km  (Kotliar and Wiens 1990).Habitat patches vary depending on an organism's perception (Wiens and Milne 1989) which involves the spatial resolution and the concept of contrast (i.e."the magnitude of differences in measures across a given boundary between adjacent patch types", Wiens and Milne 1989).The response of species to a mosaic of habitat patches depends on both spatial scale and contrast levels (Chust et al. 2003).Thus, a multi-contrast levels and scales approach would help to evaluate properly effect of different crop categories on species occurrence and biodiversity and explain the different patterns we found at different resolutions.
Our results provided evidence that the species richness of amphibians and reptiles is in general negatively correlated with crop extent.However, the results also highlighted scale-and group-dependent richness patterns in response to crop extent and crop categories.This emphasises the importance of considering the context speci city of scale-and taxa-dependent responses to crop types when investigating these patterns.Previous studies have indicated that agricultural lands are often not optimal habitats for amphibians and reptiles (e.g.Loman andLardner 2006, 2009;Ribeiro et al. 2009), both groups being particular sensitive to agricultural activities (Dürr et al. 1999), such as the use of pesticides (Brühl et al. 2013).Further, agricultural lands may even serve as ecological traps (Rotem et al. 2013).
The extent of dry crops may be the main threat to amphibian species diversity in agricultural landscape.
Both at coarse and ner spatial resolutions, amphibian species richness was negatively affected by the extent of dry crops.Pond-breeding amphibians are dependent on availability of aquatic habitats, being strongly related to aquatic habitats characteristics (e.g.surface, hydroperiod, water chemistry, predators and cover vegetation) (Skelly et al. 1999;Skelly 2001) and vegetation around them for migratory events and other terrestrial activities (Mazerolle and Desrochers 2005; Prevedello and Vieira 2010).The low spatial and temporal availability of aquatic habitats in drylands, subject to human activities altering their characteristics and the vegetation around them, most likely affect negatively the amphibian species diversity (Gardner et al. 2007).
Species richness of amphibians was also negatively affected both at coarse and ner spatial scale by irrigated crops, possibly due to higher use of pesticides or alteration of aquatic habitats in this crop category.Similarly, at the coarse spatial scale, agroforestry crops had also a positive effect on amphibians richness, probably due to the higher landscape heterogeneity in those crop categories than in monoculture lands (Brüning et

Figures
Figures

Figure 1 Maps
Figure 1

Figure 2 Maps
Figure 2

Table 1
Generalized Additive Model results for effect of crop categories on species richness of amphibians (SR amp) and reptiles (SR rep) at UTM 50 km 2 and UTM 10 km 2 spatial resolutions; edf = the effective degrees of freedom; test statistics are either chi-square (for models with Poisson distribution) or F (for models with Gaussian distribution); agfr = agroforestry crops; dry = dry crops, irr = irrigated crops; past = pastures; wood = woody crops.Models for UTM 10 km 2 resolution included a spatial term to account for spatial autocorrelation.
(Carpio et al. 2017)ated olive plantations host an exceptionally high proportion of specialist reptiles (Kazes et al. 2020, but seeCarpio et al. 2016).Furthermore, when herbaceous cover exist woody crops may harbour a diverse community of reptiles(Carpio et al. 2017).All above explanations assume that landscape has a hierarchical structure, i.e. regions consist of a mosaic of smaller habitat patches which occur within larger habitat patches (Biaggini and Corti 2021)rop category positively affected the highest number of species.Agroforestry crop in our study includes land principally occupied by agriculture with signi cant areas of natural vegetation and agroforestry areas (like the typical montados and dehesas in the Iberian Peninsula; CORINE Land Cover 2018, version v.2020_20u1).Previous studies have shown that the combination of agriculture and forestry enhances the persistence of amphibian and reptile species (Brüning et al. 2018; Warren-Thomas et al. 2020; Fulgence et al. 2021), being in general bene cial for biodiversity (Hartley 2002; Harvey et al. 2007; Torralba et al. 2016).Agroforests are important for biodiversity as they provide a more diverse habitat than a conventional agricultural system and can serve as corridors between habitats(Harvey et al. 2007).Woody crops in our study included vineyards, fruit trees and berry plantations, olive and chestnut groves and walnut groves shrub orchards.Rptile species, such as lizards, can be widespread in vineyards(Biaggini and Corti 2021).
(Fahrig et al. 2011)tein 2016).Agroforestry crops have less intensive agricultural modes and contain patches of natural vegetation and hedges, providing shelter and food resources.Indeed, ponds are very abundant in the Spanish dehesas and Portuguese montados agroforests, for providing water to cattle.Reptile species richness responded signi cantly to all ve crop categories, decreasing with the extent of all crop categories except irrigated and woody crops at coarse spatial resolution.Previous studies have shown that non-irrigated crops adversely affect reptiles (Atauri and de Lucio 2001).Grazing is an important anthropogenic disturbance on pastures and can have negative and positive effects on biodiversity(Schieltz and Rubenstein 2016).The response of reptile species to grazing has been found to be in uenced by arboreality(Neilly et al. 2018).Terrestrial reptiles were generally negatively affected by increasing grazing pressure, unlike arboreal reptiles (Howland et al. both coarse and ner spatial resolutions.However, we measured the compositional heterogeneity(Fahrig et al. 2011), that describes the diversity of crops grown in a landscape, as the Shannon diversity index of crop categories.Con gurational heterogeneity(Fahrig et al. 2011) describing the spatial arrangement of elds and measured as mean eld size or density of eld borders, might play an important role on amphibian and reptile species richness.Therefore, further studies are needed to understand whether crop heterogeneity could be an effective way to increase the diversity of amphibians and reptiles in agricultural landscape.5 .Sokal RR, Rohlf FJ (1995) Biometry: the principles and practice of statistics in biological research, 2nd edn.Freeman, New York 57.Wanger TC, Iskandar DT, Motzke I, Brook BW, Sodhi NS, Clough Y, Tscharntke T (2010) Effects of land-use change on community composition of tropical amphibians and reptiles in Sulawesi, Indonesia.Conserv Biol 24:795-802.https:// doi.org/10.1111/j.1523-1739.2009.01434.xWiens JA, Milne BT (1989) Scaling of 'landscapes' in landscape ecology, or, landscape ecology from a beetle's perspective.Landsc Ecol 3:87-96.https://doi.org/10.1007/BF001311720. Wilson S, Alavi N, Pouliot D, Mitchell GW (2020) Similarity between agricultural and natural land covers shapes how biodiversity responds to agricultural expansion at landscape scales.Agric Ecosyst Environ 301:107052.https://doi.org/10.1016/j.agee.2020.107052 1. Wood SN (2003) Thin plate regression splines.J R Stat Soc: Series B (Statistical Methodology) 65:95-114.https://doi.org/10.1111/1467-9868.003742. Zuur A, Ieno EN, Walker N, Saveliev AA, Smith GM (2009) Mixed effects models and extensions in ecology with R. Springer Science and Business Media (Collins and Fahrig 2017ces species richness of both amphibians and reptiles.Crop heterogeneity affects positively a wide range of taxa, including amphibians(Collins and Fahrig 2017).Yet, evidence of biodiversity bene ts from crop diversity was claimed to strongly depend on spatial resolution.Our results provided support for an increase in the species richness of amphibians and reptiles with crop heterogeneity at