Coexistence of two newt species in a transition zone of range overlap

Theory suggests that spatial segregation of similar, co-occurring species may be driven by alternative innate life history and dispersal strategies, and that it operates through catastrophic events. An inventory of the evolutionary closely related small-bodied newts Lissotriton helveticus and L. vulgaris in the northwest of France demonstrated the species’ spatial partitioning, with L. vulgaris dominating in two pond-rich and historically disturbed coastal areas and L. helveticus prevailing inland where ponds are sparser. Population numbers were followed over several decades (1975–2021) in a pond within the narrow (ca. 2000 m wide) species transition zone. Early in the temporal survey (1986) a massive die-off was observed of two-third of the L. helveticus breeding population from a late frost event. Yet, the contribution of L. helveticus to the newt assemblage was more or less stable around 60%, even though the total population size fluctuated by an order of magnitude. Lissotriton vulgaris and a third species, Ichthyosaura alpestris, made up ca. 30% and 10% of the total till 1993, after which date their relative contributions reversed. These data suggest that a state shift may have occurred among the latter two species and that the assumed twospecies dynamics of Lissotriton underlying the study has been an oversimplification. The local decline of L. vulgaris is paralleled by the loss of well-vegetated ponds from the wider agricultural terrain that affects this species more than L. helveticus and I. alpestris.


Introduction
The competitive exclusion principle asserts that, when resources are in short supply, no two non-hybridizing species can exploit the environment in exactly the same way and coexist. Syntopic species will therefore differ along one or more niche dimensions (Hardin, 1960;Mueller, 2019). When it was shown that coexistence rather than exclusion of closely related species is the rule, this principle gradually changed into the 'competitive niche shift principle' (den Boer, 1986). Whether the strict or more lenient perspective is taken, to further test and document the principle, it would be advantageous to select a study system that covers a wide array of environmental conditions and to not be restrained by dimensions of time or space. The European small-bodied smooth and palmate newts (Lissotriton vulgaris and L. helveticus) provide an appropriate research system, because they are morphologically similar and closely related species with widely overlapping ranges. They breed in discrete aquatic habitats (mostly ponds) where they are usually abundant and frequently co-occur. In two pioneering papers, Griffiths (1986Griffiths ( , 1987 reported on the overlapping niche dimensions of adult newts during their aquatic breeding phase and concluded that there was no evidence for resource partitioning in the species' temporal, micro-habitat or feeding profiles, suggesting that population regulation occurred during another phase of life. In newts this may be the larval stage, or the terrestrial juvenile and adult stages. Resource partitioning in the aquatic larval stage remains to be studied and not much is known about the species' terrestrial habitat preferences, though L. helveticus may be prevailing at higher altitudes than L. vulgaris (Cooke & Ferguson, 1975;Feldmann, 1981), in forested areas, on heathland and sandy soils ( van Gelder, 1972;Geraeds, 2009) and in nutrient-poor and acidic environments, possibly related to geology (Cooke & Frazer, 1976;Denton, 1991).
I here investigate the long-standing issue of a differential habitat utilization of smooth and palmate newts along wide spatial and temporal axes. This work demonstrates the importance of long-term research on populations in order to understand the drivers of population trends. As newts seem to be at the mercy of catastrophic and density-independent factors (e.g., freezing and drying ponds), the observed patterns are highly stochastic. Yet, this study suggests that the species do possess different life histories and dispersal strategies and respond differently to environmental perturbations and it shows how intrinsic and environmental factors can differentially affect the abundance of coexisting species.

Materials and methods
Amphibian survey data for the western, coastal part of the département (department, dept.) Pas de Calais in the northwest of France were obtained over the 1974-2021 period (see e.g., Zuiderwijk, 1980;Arntzen et al., 2017). The data were recompiled to obtain the frequency (F) of Lissotriton helveticus, the palmate newt (F h = N h /(N h + N v ) and L. vulgaris, the smooth newt (F v = 1 -F h ) in ca. 400 ponds across the landscape. For 105 ponds (this excludes the 'focal' pond described below) sample sizes were over the N h + N v = 9 threshold that was considered the minimum to conduct a meaningful analysis. A contour map of F v was produced with MyStat software (Systat Inc., 2008), with the logarithm of (N h + N v ) as a weighing factor to give more emphasis to larger samples. An indication of pond density was obtained by calculating the average Euclidian distance of the pond under consideration to the other nearest one to ten ponds (Ponddist1 -Ponddist10). Hardness and salinity of pond water were determined with a Hach dr/3 Photometer as Ca+ in mg/l (58 observations) and Cl-in mg/l (28 observations).
The inventory revealed a marked spatial separation of Lissotriton helveticus and L. vulgaris (see results below). To investigate the species' long-term population dynamics under conditions of syntopy, a study pond was selected located at the transition of L. vulgaris to L. helveticus dominated areas. Additional selection criteria were permission to access, small pond size and substantial newt populations, to provide sample sizes with sufficient statistical power to analyse rigorously. The selected pond with coordinates 50.820 N and 1.604 E is located in rough pasture between the villages Ambleteuse and Audresselles adjacent to the Selles brook. The surrounding area was initially communal land and classified as a nature reserve since 2012 (Réserve naturelle régionale du pré communal d' Ambleteuse). Aerial photography available at https:// remonterletemps.ign.fr reveals that the study pond was created in between July 14, 1971 and January 1, 1972, possibly for the purpose of sand extraction (supplementary fig. S1). Dimensions were ca. 15 × 20 m. The pond was detected as a site for amphibian reproduction in 1975 and followed up to the present day. A further newt species surveyed at the pond alongside L. vulgaris and L. helveticus was the alpine newt, Ichthyosaura alpestris, whereas the northern crested newt, Triturus cristatus, was seen just once.
Observed pond depth in spring varied from 90 to zero cm (i.e., dry). Pond half-life in the area has been estimated at 20 years on average (Curado et al., 2011) so that the chance for a randomly selected pond to survive the study period of 47 years was about one in five. Natural succession reduced pond size and depth, and it was further diminished by the dumping of debris from the beach and construction waste in 1993-1994. Encroachment by shrubs and trees fully shaded the pond within a local spinney (diameter 40 m) by 2003 (supplementary fig. S2). In spring 2021 pond size was down to ca. 9 × 14 m with a depth of 35 cm. In 1975In -1981In , 1984 and 2021 adult newts were sampled in spring or early summer to determine species composition. In spring 1987 the pond was dry and in 1983, 1986, 1988-1995 and 2012 population size estimates were made through a multiple capture procedure of marked and unmarked individuals, by either using funnel traps (in 2012) or dip nets (the other years). Batch marking was applied by the clipping of a single finger. Population size estimates (N̂) were made in the breeding period for the entire newt population of three species, i.e., L. helveticus, L. vulgaris and I. alpestris, on the -substantiated -assumption of equal catchability of species (Arntzen, 2002;Arntzen & Zuiderwijk, 2020). The pooling of species was deemed necessary because it is tedious to estimate small population sizes with an acceptable level of confidence. Immigration and emigration were assumed to be absent, amounting to a 'closed' population. Recaptured individuals were taken into account with the 'weighted mean' method (N̂W M ; Begon, 1979), or excluded from the calculations for the 'removal' method (N̂R, Pollock & Otto, 1983, as implemented in the Capture program by Rexstad & Burnham (1991), accessible at https://www.mbrpwrc.usgs.gov/software/capture.html). The results on logN̂W M and logN̂R correlate (r = 0.84, df = 12, P <0.01), with the tendency that logN̂W M would exceed logN̂R by a factor 1.05 on average (range 0.997-1.180) (for details see the Results section). I here utilize the more conservative N R estimates that come with a 95% confidence interval (ci95). population dynamics of syntopic newts | 10.1163/18759866-bja10028 Capture efficiency is defined as the number of different individuals observed in an annual sampling period divided by N R .
Females of both Lissotriton species have similar appearance (Veith & Dorr, 1985;Arntzen et al., 1998), with one of the differences being that L. helveticus females appear a little heavier than L. vulgaris females. Because different volumes may indicate different fecundities (see Verrell, 1986), I counted the number of yolked oocytes in females of both species in early spring, i.e., before the onset of oviposition. These females were found dead in the focal pond directly after a freezing incident just prior to the annual breeding period (see below).

Species distribution
A total of 17,537 adult Lissotriton captures was made (average sample size per pond 167.0, range 10-2162) with more L. helveticus (74%) than L. vulgaris (26%) (table 1). Lissotriton vulgaris was the more numerous species in and around the coastal zones 'Dunes de la Slack' and the so-called 'bomb crater area' south of Cap Griz Nez, whereas L. helveticus was more numerous inland ( fig. 1). The contour plot of species frequencies describes the transition in species composition from 0.4 < F v < 0.6 over a distance of ca. 2000 m. At all localities with large sample sizes (N h + N v > 200 in 20 ponds) both species were found. Lissotriton vulgaris was rare (F v < 0.05) in six of these localities whereas L. helveticus was never rare.
Correlations of F v and chemical parameters were marginally insignificant for hardness of the pond water (Spearman's correlation coefficient, r s = -0.249, N = 58, P < 0.10) and for salinity (r s = -0.336, N = 28, P < 0.10). The data are presented in supplementary table S1.
Species composition F v was significantly negatively correlated with the average distance to other ponds for Ponddist3-Ponddist10 but not for Ponddist1 and Ponddist2, indicating that L. vulgaris is more numerous when ponds are close together (supplementary table S2).

Species numbers in the focal pond -early observations
Dip-netting the focal pond in 1975 yielded 31 L. helveticus and ten L. vulgaris (F v = 0.24) which is in line with the pond's location in the species transition zone ( fig. 1). In early March 1986 the combination of a low water table in combination with a 'false spring' was fatal for a large cohort of newts, presumably through freezing or suffocation in ca. 20 cm of ice, water and mud. On March 12 and 13, the ice had largely melted and 505 L. helveticus (280 males, 225 females), eight L. vulgaris (four males, four females) and one I. alpestris male were observed dead. It is unclear if the casualties had been hibernating in the pond or that they were early immigrants. Found alive were three males and five females of L. helveticus. The population size estimates later in the season suggested that the incident affected the breeding population of L. helveticus (505 dead out of N = 721, 70%) more strongly than the other species (L. vulgaris -eight dead out of N = 274, 3%; I. alpestris one dead out of N = 27,4%) and further suggests that the latter two species arrived at the pond later in the season, after the catastrophic event had occurred. Also found dead were 94 presumably aquatically hibernating Rana temporaria juveniles, along with one adult male and five juveniles alive. All corpses were without signs of decomposition. They were collected by hand, preserved in ethanol for later storage at the Naturalis Biodiversity Center, with collection numbers zma.rena.7819-7823. The average number of yolked oocytes (Ñoo) was counted in 12 L. helveticus (Ñ oo = 242.4,  Those years are alternated by 1986 after the mass mortality event and 1989, 1991 and 1993, when relatively low L. helveticus frequencies were observed.

Discussion
Amphibians are under pressure around the globe from a wide variety of causes of which habitat deterioration is the prime one (Green et al., 2020). In particular pond loss, that terminates the population and weakens the metapopulation structure through the increase of dispersal distances, has been implied to negatively affect population survival in theoretical (Halley et al., 1996;Rustigian et al., 2003;Swanack et al., 2009) and empirical studies (Pope et al., 2000;Joly et al., 2001;Arntzen et al., 2017). It has, however, also been suggested that differential dispersal profiles may rapidly evolve, even over historical times, such as in an invading species (Rollins et al., 2015;Hudson et al., 2016) or in stable versus unstable breeding habitat configurations in landscapes subjected to anthropogenic disturbances (Cote et al., 2017;Cayuela et al., 2020a), to the benefit of species survival. The inventory of two closely related and morphologically similar species of small-bodied newts, L. helveticus and L. vulgaris, showed their preponderance in pond-sparse and pond-dense areas, respectively. The Lissotriton species in Pas de Calais therewith appear to present a fine working example of the principle of competitive exclusion. While Year 10.1163/18759866-bja10028 | arntzen it might be argued that resources are not in short supply and that L. helveticus and L. vulgaris are not in competition, a simple thought experiment suggests otherwise. I predict that if L. vulgaris was removed from the dunes L. helveticus would colonize the vacated ponds. This notion is supported by the wide presence of L. helveticus in the coastal dunes of the Vendée, a department located outside but adjacent to the L. vulgaris range (Doody, 1991; Groupe herpétologique des Pays de la Loire, 2021). From an evolutionary perspective, a mechanism for limiting competitive exclusion is to adopt alternative life history and dispersal strategies, which are then reinforced through natural selection. This evolutionary process reduces competitive interactions and increases opportunities for colonization in one species and nutrient acquisition in the other, and often operates though catastrophic events (Roxburgh et al., 2004). In view of interspecific competition, we are thus posed with the question if, as theory suggests, L. helveticus and L. vulgaris differ in life history parameters such as dispersal strategies and their response to environmental change.
Lissotriton helveticus and L. vulgaris show a wide area of range overlap in the west of Europe covering large parts of Scotland, Wales, England, the Netherlands, Belgium, Luxemburg, France and Germany (Sillero et al., 2014). Also at a regional scale, the species mostly show overlap and are regularly found in syntopy (e.g., Zuiderwijk, 1980;Blab & Blab, 1981;Durkin & Cooke, 1984), where they share resources in their aquatic breeding habitat in space, time and food (Griffiths, 1986(Griffiths, , 1987. However, disparities exist in terrestrial habitat as well as aquatic habitat preferences ( van Gelder, 1972;Cooke & Ferguson, 1975;Cooke & Frazer, 1976;Feldmann, 1981;Geraeds, 2009;Denton, 1991, see also Introduction). In Pas de Calais both species are common and show a marked degree of spatial separation, with L. vulgaris prevailing inside, and L. helveticus outside the dune and bomb crater areas. The species separation may recently have sharpened, as seen from an analysis of population dynamics of syntopic newts | 10.1163/18759866-bja10028 species-habitat relationships over four decades that showed that L. vulgaris became increasingly associated with the dune environment and with well-vegetated ponds such as the bomb craters, whereas L. helveticus remained associated with forestation, arable land use and less vegetated ponds (Arntzen et al., 2017). The available knowledge on the species' terrestrial habitat preferences does not readily explain this spatial partitioning. For example, the extend of forestation does not seem to play a role, because locally nothing much changed, as shown by comparing the distribution of woodland on the 18th century maps by the Cassini family (accessible at https://www.geoportail/gouv.fr/donnees/ carte-de-cassini) with those of the present day ( fig. 1; see also Vallauri et al., 2012). Features that both L. vulgaris strongholds have in common are a coastal geographical position and a high density of more or less small and often temporary ponds (see also Curado et al., 2011;Arntzen et al., 2017). A marked spatial separation between the species has also been found in England, possibly related to geology, with soft-water ponds on millstone grit where L. helveticus is more common, abruptly transitioning to hard-water ponds on limestone where L. vulgaris prevails (Denton, 1991). However, with just presence/ absence data available, the sharpness of the species transition is difficult to assess and the question if 'pond density' may play a role in this differential distribution was not tackled. In Pas de Calais hardness of the water was not significantly associated with species composition but given the strength of the association (0.05 < P < 0.10) this parameter is a candidate for further investigations, along with salinity that shows a comparable association (supplementary table S1).
Among 91 amphibian species for which maximum observed salinity concentrations were documented (Hopkins & Brodie, 2015), L. helveticus takes seventh place and L. vulgaris takes 13th place. The maximum values are ca. two orders of magnitude higher than what was observed for 28 Lissotriton localities in Pas de Calais (supplementary table S1), which would indicate that the local environment poses no restrictions to either species in terms of salinity. An Atlantic climate affecting both species differently is unlikely at the spatial scale concerned so that, in the absence of other clues, pond density is the prime environmental parameter to be considered. It would be instructive to survey other dune areas in Pas de Calais, such as the Dunes d' Amont near Wissant and the Dunes de Hardelot near Hardelot Plage. Another case meriting further investigation is a string of ca. 50 L. vulgaris occurrences in the Loire valley, at the very southwestern edge of the species range, amidst an abundance of L. helveticus records (Groupe herpétologique des Pays de la Loire, 2021).
The Lissotriton species studied are morphologically similar, yet appear to be characterized by different life history parameter values. Lissotiton vulgaris thrives and, as seen from their relative numbers, replaces L. helveticus in Pas de Calais in two pond-rich areas that are situated in unstable landscapes. The study area was heavily shelled during the Second World War, with no bomb craters visible on aerial photographs of the wider Cap Griz Nez plateau on April 20, 1939 whereas there were thousands on June 12, 1949 (supplementary fig. S3). While most agricultural terrain has since been restored, 12 hectares of pasture remain untreated, with 10-20 craters that are deep enough to keep water up to the summer, at least occasionally. As for the Dunes de la Slack, aerial photography from 1929 onwards reveals a turbulent local history, going from largely barren, due to sand mining and Second World War military activities, to the gradual increase of vegetation. Water bodies were created for sand extraction, duck efficacy was high because of large annual total population sizes and reasonably high capture efficiencies (table 3) and although this could not have been foreseen at the start, the pond fortunately persisted till the present day and access remained possible. On the other hand, in some years the research period was fairly long (table 3) so that the assumption of a closed population was challenged. On the long-term, L. vulgaris numbers in the study pond have been decreasing and this appears to be associated with local habitat change. The pond has become smaller, more shaded and less vegetated (supplementary figs S1 and S2). This fits the aquatic habitat requirements of L. helveticus (Blab & Blab, 1981;de Fonseca & Jocqué, 1982) and that of the third newt species, I. alpestris (Feldmann, 1981;Roček et al., 2003;Thiesmeier & Schulte, 2010;van Overstraeten & de Fonseca, 1982) that has taken over the second place. The transition from L. vulgaris at 30% (from 1975 to 1993) to I. alpestris at 30% (observations in 2003, 2012 and 2021) may represent a 'tipping point' among alternate stable states (Scheffer et al., 1993, Dakos et al., 2019 that are supported by large, open and smaller, shaded pond conditions, respectively. It may though be noted that this concerns a single observation along with a post-hoc explanation. The pond character change and local decline of L. vulgaris are paralleled by the area-wide loss of well-vegetated ponds from agricultural land that affects this species more than L. helveticus and I. alpestris (Curado et al., 2011, Arntzen et al., 2017). Yet, the survey data indicate that L. vulgaris may be lingering around at low numbers, providing the species the opportunity to take advantage of changing environmental conditions when these are to arise, such as with the inadvertent pond creation from military activities in the Second World War.