The barred grass snake, Natrix helvetica (Lacepède, 1789), is a common and widely distributed semiaquatic snake species of Western Europe (Kindler et al. 2017; Schultze et al. 2020). Recent molecular genetic research clarified that it comprises four subspecies (Schultze et al. 2020). Two subspecies are island endemics and occur on Corsica and Sardinia, respectively. These two taxa are morphologically highly distinctive (Thorpe 1979; Kabisch 1999; Di Nicola et al. 2023). The other two subspecies N. h. helvetica and N. h. sicula (Cuvier, 1829) have only recently been genetically delimited (Schultze et al. 2020). Their distribution ranges are parapatric. Natrix helvetica sicula occurs mostly in the Italian Peninsula, the Po Plain and Sicily (Schultze et al. 2020). However, it also extends across the Alps to southwestern Switzerland (Schild et al. 2024) and has spread along the Inn River to Austria and southernmost Bavaria (Asztalos et al. 2021). Natrix helvetica helvetica is distributed in northern Switzerland, France, the Benelux countries, western Germany and Great Britain (Kindler et al. 2017; Asztalos et al. 2020; Schultze et al. 2020), with the Alps constituting the main divide between the two subspecies, with the exceptions mentioned above. Admixture between the two continental subspecies occurs in southern Switzerland, as well as in northwestern Italy and adjacent southern France (Schultze et al. 2020; Schild et al. 2024). Even though many previous attempts have been made to capture the considerable phenotypic variation of grass snakes for taxonomic purposes (e.g., Hecht 1930; Mertens 1947; Thorpe 1979; Lanza 1983), the current molecular genetic delineation of N. h. helvetica and N. h. sicula requires a reexamination of their external morphology to determine whether the subspecies can be morphologically told apart. In the present study, we lay the foundation for doing so by using approximately 3800 photographic records, primarily from the Citizen Science platforms iNaturalist and ornitho.it. Based on this large dataset, we describe and compare coloration and pattern across the whole distribution range of N. h. helvetica and N. h. sicula.

Using georeferenced photographic records, we examined coloration and pattern of 2973 Natrix helvetica sicula from Italy. This number included all informative photos available on the iNaturalist platform by 31 December 2025 (2854 individuals) and all photos from Basilicata, Calabria, Apulia and Sicily available up to the same date on ornitho.it (66 individuals), i.e., we used all photos in which at least one of the relevant traits could be scored. We also added our own unpublished geotagged images of 53 snakes. For comparison with the nominotypical subspecies we used photos of 800 individuals. For doing so, we randomly selected iNaturalist photos of 100 N. h. helvetica each from Germany and Great Britain, as well as 20 barred grass snakes from Luxembourg and 40 each from Belgium and the Netherlands. Since admixture between the two subspecies is known to occur in southern Switzerland and southern France (Schultze et al. 2020; Schild et al. 2024), additional data for these countries were divided into subsamples corresponding to southern and northern Switzerland (n = 45 and n = 55, respectively), as well as southern and northern France (n = 189 and n = 211, respectively). The southern group in Switzerland corresponded to observations from the cantons of Geneva, Valais, Vaud and Ticino, where N. h. sicula occurs or the two subspecies admix. There is also evidence for single genetic signatures of N. h. sicula in the northern cantons of Bern and Schwyz (Schild et al. 2024). However, the vast majority of northern records in Switzerland genetically represent the nominotypical subspecies (Kindler et al. 2017; Schild et al. 2024). The divide for northern and southern records from France was a straight line connecting La Rochelle (France) and Basel (Switzerland). This accounts for the possibility that admixture could also extend from southwestern Switzerland into France. The sizes of the individual subsamples, ranging from 20 for Luxembourg to 400 for France, reflected both differences in the size of the covered geographic areas and the expected higher variation in regions where the two subspecies meet (France, Switzerland). For the 800 comparative photos, we selected only such images that showed all or at least nearly all traits. We disregarded for the present study the peripheral occurrences of barred grass snakes in Bavaria (Germany) and western Austria, mostly representing N. h. sicula (Glaw et al. 2019; Asztalos et al. 2021; Neumann et al. 2024). We also ignored morphologically identified hybrids from the contact zones with the common grass snake, Natrix natrix (Linnaeus, 1758), in Germany, Switzerland and northeastern Italy. To assess potential latitudinal variation, we arbitrarily assigned photos from Italy to three geographic groups. The northern subsample comprised all barred grass snakes from the northern border southward to the latitude of Grosseto, Tuscany (42.7667); the central subsample, all records south of the latitude of Grosseto to the latitude of Naples (40.8333); and the southern one, all observations south of the latitude of Naples (Fig. 1a). The northern subsample most likely also included some snakes with genetic impact from N. h. helvetica (see Schultze et al. 2020).

For iNaturalist observations with obscured coordinates, we used the displayed modified coordinates, i.e., each true site lies within a box measuring 0.2 × 0.2 degrees (approximately 20 × 20 km) shared with the shown locality. This inaccuracy is irrelevant for the present study. In total, we examined pictures of 3773 barred grass snakes (see complete list in File S1).

Based on our expert knowledge of the variation of coloration and pattern in barred grass snakes, we scored for each individual the following traits to quantify geographic differences:

(1) Body coloration: light/dark/melanistic/picturata morph and presence of very dark individuals

(2) Side bars: large/small/absent

(3) Additional dorsal dark body pattern: large elements/small elements/absent

(4) Tessellated body pattern: yes/no

(5) Side bars and dorsal dark pattern form a pattern resembling asp vipers (Vipera aspis): yes/no

(6) Snout color orange or red: strong/weak/no

(7) Occipital dark spots ring-shaped: yes/no/uniform grey (absent)

(8) Anterior side bars form closed body rings: yes/no

(9) Occipital dark spot: moderately elongated/much elongated/uniform grey (absent)

(10) Occipital light crescents (lunar marks): strong/fading/absent

(11) Occipital light crescents: widely separated/nearly touching/closed ring.

Figures 2, 3, 4, 5, 6, 7 illustrate the variation of the scored character states and phenotypic variation. It should be noted that many traits are not visible in melanistic snakes and representatives of the picturata morph. Melanistic snakes are defined here as completely black individuals that show no trace of the regular body coloration, or as partially black individuals that retain remnants of the regular coloration pattern (Figs 2, 6). Thus, our concept of melanistic snakes also includes the charcoal form of Bruni et al. (2022). The picturata phenotype corresponds to black snakes with many conspicuous light speckles or spots (Fig. 6; see also Kabisch 1999; Bruni et al. 2022). Considerable variation also occurs with respect to coloration. Some barred grass snakes have an olive grey, silvery grey, bluish, yellowish or ocher skin color. However, these colors may depend much on light conditions of the photos, which is why we did not quantify these and assigned the photos only to the categories light/dark/very dark, melanistic or picturata morph.

We scored for all snakes each trait as far as possible, except for trait (7). We examined this trait only for those 212 individuals from southern Italy in which it was visible, 1016 snakes from the more northern parts of Italy and for N. h. helvetica. Although age-dependent variation is known for some traits (e.g., Thorpe 1979; Fritz et al. 2023; Di Nicola et al. 2025), we could not reliably assign many snakes to age classes. Therefore, we processed all the data together for each subsample. We present the results of our scorings as percentages in maps, either by country or for each of the above-defined geographic subsamples. However, we lumped together the data from Belgium, the Netherlands and Luxembourg into a single subsample (“Benelux”). For selected traits, we also show maps with the individual records. For maps, we used the WGS 84 geographic coordinate reference system (EPSG:4326) in QGIS 3.28.10; shaded-relief backgrounds were based on the Esri World Shaded Relief basemap.

Representatives of Natrix helvetica sicula are much more variably colored and patterned than those of the nominotypical subspecies (Figs 2, 3, 4, 5, 6, 7). This is reflected in all scored traits. Within Italy, barred grass snakes from the south are more variable than those from the north of the country. In contrast, the subsamples of pure N. h. helvetica showed little variation; the two subsamples from regions where the distribution ranges of the two subspecies abut and admixture is known (southern France, southern Switzerland) were mostly intermediate in the studied traits (Figs 1, 8; File S2; for individual sample sizes and exact percentages, see Files S3 and S4).

We did not attempt to capture the many diverse skin colors, but differentiated between generally light, dark or very dark snakes, melanistic individuals and representatives of the picturata phenotype (Fig. 1b; File S2a). When the percentages for the individual subsamples are compared, it is obvious that lighter skin colors were more frequent in N. h. sicula (42.5–48.4%) than in N. h. helvetica (19.0–35.0%), with the subsample from southern France (42.9%) resembling N. h. sicula and that from southern Switzerland (26.7%), the nominotypical subspecies. The rare picturata morphotype was only recorded for N. h. sicula but not for the nominotypical subspecies or the subsamples from southern France and southern Switzerland. In N. h. helvetica, the highest percentage of dark and very dark snakes was recorded for northern Switzerland (25.0%, very dark plus melanistic individuals). In N. h. sicula, fewer very dark snakes occurred. The highest percentage (13.6%, very dark and melanistic individuals plus picturata phenotype) was found in southern Italy (File S2a). Even though not quantified, the variety of different colors and tinges seemed to be higher in N. h. sicula than in N. h. helvetica, with the greatest variation in central and southern Italy.

With respect to pattern, the eponymous side bars were present in most barred grass snakes of both subspecies (for the five groups of pure N. h. helvetica, in 89.1–94.0%; for the three groups of N. h. sicula, in 90.1–96.8%; see also Fig. 1c). The individuals without side bars included melanistic snakes and representatives of the picturata phenotype, as well as a few other individuals without side bars, including some rare uniformly grey-colored snakes (see for examples, Figs 2, 3, 6). However, N. h. sicula shows high percentages of individuals with large side bars in all three geographic groupings (65.8–89.2%, with the lowest value in northern Italy), whereas large side bars occur only in 38.0–58.2% of individuals in the five groupings representing pure N. h. helvetica (Fig. 1c). The two subsamples from southern France and southern Switzerland containing representatives of both subspecies and admixed individuals had intermediate values (62.4% and 71.1%).

Many individuals of the nominotypical subspecies have plain grey backs, and the side bars are the only dorsal pattern elements in 48.0–68.0% of the five groupings of pure N. h. helvetica (melanistic individuals disregarded). The percentages for snakes without additional dark body pattern for southern France and southern Switzerland were 19.6% and 31.1%. If a dark dorsal pattern was present, the individual elements were mostly small in pure N. h. helvetica, with percentages ranging between 25.5% and 44.0% (melanistic individuals again disregarded), and in the two intermediate subsamples from southern France and southern Switzerland, the values were 49.7% and 35.6%, respectively (melanistic individuals disregarded). In contrast, in the three groups of N. h. sicula, most individuals had an additional dark dorsal pattern. In northern, central and southern Italy, only 6.3%, 0.6% and 11.0% of the barred grass snakes had no additional dark dorsal pattern, and among the ones with a dark dorsal pattern, 29.4%, 52.6% and 49.2% of snakes had large dark dorsal elements (picturata phenotype and melanistic snakes not considered; compare Fig. 1d).

Some barred grass snakes with additional dark body spots were individuals with a tessellated pattern (Figs 1e, 4). Such snakes were rare in the five subsamples of pure N. h. helvetica, with percentages ranging from 0% to 2.8%. In southern France and southern Switzerland higher percentages were observed (9.5% and 24.4%). Tessellated individuals were also more common in N. h. sicula, with the highest percentage (21.5%) in the north of Italy, while in the center and south 10.6% and 8.4% occurred. A similar, but inverse, pattern was observed for the frequency of a body pattern closely resembling that of the asp viper, Vipera aspis (Linnaeus, 1758) (Figs 1f, 5). In the distribution range of N. h. helvetica, this asp pattern was rare, with percentages ranging between 1.0% (Germany) and 2.0% (Benelux and Great Britain). However, in northern Switzerland 14.5% N. h. helvetica had an asp-like body pattern. Natrix helvetica sicula showed this pattern much more frequently. The highest percentage occurred in the south of Italy with more than half of the snakes having this pattern (54.2%). The values for the subsamples from central and northern Italy were 38.0% and 16.6%. The percentages for southern France (28.6%) and southern Switzerland (13.3%) resembled N. h. sicula.

Two rare traits were never found in N. h. helvetica and occurred only in N. h. sicula, and these rare traits were most frequent in the southern part of the distribution range. This was the case for the presence of a red- or orange-colored snout (Figs 6a, 7a, c, 8a). Only six individuals out of 2173 barred grass snakes from northern Italy showed this coloration (0.3%), compared to eight out of 317 snakes from central Italy (2.5%) and 44 out of 266 snakes from southern Italy (16.5%; File S2b). Another unique trait of N. h. sicula concerns the occipital pattern. The normal character state in N. helvetica consists of a bipartite pattern, on each side with a light lunar mark (crescent) that often fades with age and a posterior elongated dark spot. However, in southern Italy, these dark elements were often differently shaped and rather short and dorsally merged, forming in 27.8% of the studied snakes a closed ring (Fig. 8b). Rarely, more than one ring could even be present, i.e., the foremost side bars also formed closed rings (Figs 6b, 8c), and this occurred only in central and southern Italy (0.9% and 14.1%, respectively). Except for southern Italy, a neck ring was only observed in two snakes from northern Italy (0.2%). This difference was related to the general shape of the dark spots on the occiput. There was more than twice the percentage of moderately or short occipital dark spots in southern Italy than in the other two Italian groupings (39.9% in southern Italy versus 14.5% and 16.3% in northern and central Italy), i.e., in southern Italy the occipital dark spots tended to be generally shorter than in central or northern Italy (Fig. 8d). In N. h. sicula, in particular in snakes from southern Italy, the light occipital lunar markings fade earlier than in the nominotypical subspecies (Fig. 8e). This is reflected by a high percentage (53.7%) of snakes without light crescents in southern Italy; the other two groups of N. h. sicula (central Italy: 26.4%, northern Italy: 21.5%) showed intermediate values compared to those of the nominotypical subspecies (1.0–18.0%) or the subsamples from southern France (15.3%) and southern Switzerland (6.8%). Moreover, in the north of the distribution range of N. h. sicula, the light crescents tend to be dorsally wider separated than in N. h. helvetica (Fig. 8f). Accordingly, barred grass snakes from the northern Italian subsample had the lowest percentages of individuals with dorsally closed light lunar markings (8.2%/8.6%, second value disregards melanistic snakes or representatives of the picturata phenotype). The values for central Italy (10.9%/11.8%) and southern Italy (10.7%/16.4%) were higher, with the second percentage for southern Italy resembling the values calculated across all 369 N. h. helvetica (or 362, if melanistic snakes were disregarded) for which these traits were available (16.5% and 16.9%, respectively).

The present investigation is another example for the value of Citizen Science repositories for certain scientific questions (O’Neill et al. 2023; Mason et al. 2025; for grass snakes see also the previous studies by Fritz and Ihlow 2022 and Fritz et al. 2023). However, it has to be kept in mind that photographic records alone allow only confirming the presence of a taxon and assessing a subset of morphology. Meristic data, measurements and invisible traits (e.g., ventral coloration) typically cannot be extracted, which is why photographic data cannot replace specimens in scientific collections that offer a full set of traits, including genetic data using museomics (Fong et al. 2023; Kapun et al. 2025).

The present dataset of photos of approximately 3800 barred grass snakes allows for the first time answering the question of whether the recently genetically delineated subspecies Natrix helvetica helvetica and N. h. sicula (see Schultze et al. 2020) can be told apart using coloration and pattern traits. That said, we acknowledge that our results are to some extent compromised by ontogenetic variation. Age-dependent variation is known in barred grass snakes, particularly with respect to the fading of the light lunar markings but also regarding the occurrence of very dark body coloration types (Thorpe 1979; Fritz et al. 2023; Di Nicola et al. 2025). The large sample sizes should minimize any such bias. Keeping these limitations and our purely descriptive approach in mind, we can conclude that coloration and pattern of representatives of the southern subspecies N. h. sicula are much more variable than in N. h. helvetica. In N. h. sicula, some conspicuous and unique phenotypes are mainly or completely confined to central and particularly southern Italy. However, there are also many individuals that could originate from any region of mainland Italy or Sicily according to their coloration and pattern. In the nominotypical subspecies, many individuals have a plain back coloration, and an extensive dark dorsal pattern is rare, facilitating for most individuals an easy morphological determination. Tessellated individuals, which are common in N. h. sicula, are extremely rare in N. h. helvetica. Except for northern Switzerland, this is also true for the asp pattern, which is largely restricted to N. h. sicula. In general, the majority of representatives of the nominotypical subspecies have plain backs and show, if present at all, only small dark back spots in addition to the side bars. The intermediate frequencies of many traits in those regions where the two subspecies are known to meet and hybridize (southern Switzerland, border region of southeastern France and Italy; Schultze et al. 2020; Schild et al. 2024) are in line with our conclusions.

It is noteworthy that Thorpe (1984) identified a similar morphological differentiation pattern using 41 traits (coloration, pattern, dentition, scalation, internal morphology, body proportions) for numerical phenetics. Although Thorpe’s (1979, 1984) concept of “Natrix natrix helvetica” included the currently recognized and genetically deeply divergent species N. astreptophora (Iberian Peninsula and adjacent southwestern France, Maghreb) and N. helvetica (Speybroeck et al. 2020; Schöneberg et al. 2023), he found northern populations of what is now N. h. helvetica morphologically little differentiated, whereas his samples from Sicily and southern mainland Italy (now N. h. sicula) were “among the most phenetically divergent populations” (Thorpe 1984).

With respect to coloration variants, we recorded the picturata phenotype in all three Italian subgroups. However, the only three representatives of this phenotype in our southern subgroup (Fig. 8a) are from close to the Salerno area identified by Bruni et al. (2022) as the southern distribution limit of the picturata morph. Thus, our findings are in line with their conclusion that this phenotype does not occur in southernmost mainland Italy and Sicily but in the more northern parts of the distribution range of N. h. sicula. In contrast, none of the photos of N. h. helvetica studied for this investigation showed the picturata phenotype. Yet, Fernández Guiberteau et al. (2015) and Meyer (2020) published photos of such snakes from the Val d’Aran (Catalonia) and the canton of Bern (Switzerland), respectively, providing evidence that the picturata phenotype occurs sporadically in the nominotypical subspecies.

Both representatives of the picturata phenotype and of melanistic or very dark snakes are very unevenly distributed across both the ranges of N. h. helvetica and N. h. sicula, with an obvious concentration of melanistic and very dark forms in the alpine region and southern Italy (Fig. 8a; File S2b). It is beyond the scope of the present investigation to explore this in depth, but it is noteworthy that Goldenberg et al. (2024) suggested that the occurrence of melanistic grass snakes in Italy correlates with regions in the south that have strong UV radiation. In contrast, the distribution of the picturata phenotype was inferred to correspond to regions in the north that have lower UV radiation. Also, the southern Italian records do neither match the hypothesis that higher elevations and lower temperatures favor melanism in grass snakes (see Bruni et al. 2022; Bury et al. 2022; Neumann et al. 2024; Martínez del Mármol et al. 2025) nor that darker coloration types are more advantageous in the north (Fritz and Ihlow 2022).

A recent genetic investigation showed that N. h. sicula contains five parapatric and in part deeply divergent mitochondrial lineages that are paraphyletic with respect to those of N. h. helvetica and the subspecies endemic to Corsica and Sardinia (Schultze et al. 2020). The oldest mitochondrial divergence refers to the lineage from Sicily and Calabria and was dated to 6.8 million years ago, while the mitochondrial lineage of N. h. helvetica diverged only 320,000 years ago from its Italian sister lineage (Kindler et al. 2018a). However, using microsatellite loci, only N. h. helvetica and the subspecies from Corsica and Sardinia were distinct, whereas the barred grass snakes from mainland Italy and Sicily represented a single cluster. This pattern suggests several glacial refuges in Italy, particularly in the south, where the distinct mitochondrial lineages of the philopatric females persisted. However, on the nuclear genomic level, male-mediated gene flow seems to have erased any differentiation (Schultze et al. 2020). The geographically localized rare phenotypes of N. h. sicula in the south of the peninsula and Sicily, as well as the widely distributed phenotypes, fit this pattern. On the other hand, N. h. helvetica was inferred to have expanded its range in the Holocene from a single refuge in southern France (Kindler et al. 2018b), resulting in less morphological and genetic variation. This pattern conforms well to Hewitt’s (2000) paradigm of “southern richness and northern purity”. We hypothesize that the current admixture between N. h. helvetica and N. h. sicula in southern France and Switzerland occurred only during the Holocene range expansions of both taxa.

The dorsal pattern of N. helvetica resembling V. aspis (Figs 1f, 5, 9), largely restricted to N. h. sicula, merits special consideration. Using machine learning to compare 10 photos of N. helvetica with 84 photos of V. aspis of unknown geographic provenance, de Solan et al. (2020) suggested that this pattern represents a case of Batesian mimicry. This is generally supported by the field experiments conducted by Wüster et al. (2004) using plasticine snake models. The latter authors found that the presence of a zigzag band resembling that of adders, Vipera berus (Linnaeus, 1758), significantly reduced the number of attacks of avian predators on the models. Also, the zigzag band of certain Natrix maura (Linnaeus, 1758) was inferred to mimic another viper species, Vipera latastei Boscá, 1878 (de Solan et al. 2020), and the rather inconspicuous zigzag pattern of single individuals of N. natrix from Sweden and Bosnia and Herzegovina was interpreted as viper mimicry, too (Andrén 2024; Zimić et al. 2025). Thus, it seems very likely that the asp pattern in N. helvetica indeed mimics V. aspis.

However, the fact that the asp mimicry is largely restricted to N. h. sicula is unexpected. In most of France, the nominotypical subspecies of the barred grass snake and V. aspis are broadly sympatric, and there the asp pattern is extremely rare (Fig. 9). Most of this region was recolonized only during the Holocene by both species (see Kindler et al. 2018b for N. h. helvetica and Ursenbacher et al. 2006 for V. aspis). If the glacial refuge of the egg-laying N. h. helvetica was in allopatry from the live-bearing and perhaps less temperature-sensitive V. aspis, there may not have been enough time for the mimicry system to develop. In contrast, we can assume that the glacial refuges for N. h. sicula and V. aspis were located in Italy in close proximity or sympatry, because endemic lineages and subspecies of both taxa occur there, some presumably or evidently of pre-Pleistocene age (Ursenbacher et al. 2006; Schultze et al. 2020). Consequently, the mimicry system had significantly more time to evolve.

Regardless of these considerations, we can conclude that the two subspecies examined in the present study can be diagnosed as follows:

Natrix helvetica helvetica (Lacepède, 1789)

Most adults are darker than those of N. h. sicula and have more or less well-developed side bars and a plain back without an extensive additional dark pattern. If an additional dark back pattern is present, it typically consists of small dark speckles or spots. Tessellated individuals and snakes with an asp-like dorsal pattern are very rare. Melanistic individuals occur but the picturata phenotype is a rare exception. A red snout and anterior dark body rings are unknown. The posterior occipital dark spot is often longer than in N. h. sicula.

Natrix helvetica sicula (Cuvier, 1829)

Most adult individuals are lighter than in N. h. helvetica and have an extensive dark back pattern in addition to the side bars; back patterns resembling asp vipers and tessellated individuals are common. Melanistic individuals and the picturata phenotype occur; however, in the very south of the distribution range, the picturata phenotype is unknown. Individuals with a red or orange snout occur, particularly in the center and south of Italy. In southern Italy, individuals with anterior dark body rings occur. The posterior occipital dark spot is often shorter than in N. h. helvetica.