Coexistence of the Water Shrew ( Neomys fodiens ) and the Common Shrew ( Sorex araneus ), Competing Species, in a Fluctuating Environment: Sociability and Space Use in a “Dry” Habitat

—Sustainable (no trend in the abundance of interacting species) coexistence of species can be maintained due to fluctuations in their abundance and distribution over habitats in a heterogeneous environment. The water shrew and common shrew, coexist in near-water areas and occasionally in “dry” habitats, where the water shrew periodically invades. Given the well-known overlapping of food niches of these species, one would expect the discovery of mechanisms that reduce competition; however, in “dry” habitats, we did not find such mechanisms. The use of space is characterized by a random overlapping of the home ranges of animals. In the preference test (a container with an animal versus an empty container), individual repeatability of sociability was found in tests with a conspecific stimulant, but was absent in tests with a stimulant of another species. The presence of the water shrew (as a stimulant) in the behavioral test did not increase the anxiety of common shrews, but only increased the thoroughness of exploration. No reaction of the water shrew to the common shrew was found in interspecific tests. In the absence of specific adaptations aimed at the spatial segregation of animals, the coexistence of the water shrew and the common shrew is quite well explained by “fluctuation-dependent” models of coexistence.


INTRODUCTION
Interspecific interactions are one of the leading ecological and evolutionary factors in increasing biological diversity and community formation (Day and Young, 2004;Grant, P.R., and Grant, B.R., 2006;Pfennig et al., 2007;Levine et al., 2017;Adler et al., 2018). Ecologically close species inhabiting a common territory inevitably compete for vital resources (Bigon et al., 1989;Adler et al., 2018). Interspecific competition can be weakened due to the ecological separation of species and is considered as one of the driving forces for the emergence of interspecific differences (Hutchinson, 1957;Bigon et al., 1989;Freeman and Herron, 2001;Day and Young, 2004;Grant, P.R., and Grant, B.R., 2006;Pfennig et al., 2007). At the same time, the redundancy of co-occurring species is well known, and Hutchinson's (Hutchinson, 1959) question about the nature of the excess species diversity still has no clear answer (Day and Young, 2004). Presumably, stable coexistence (no long-term trend in abundance) of ecologically close species can be maintained due to fluctuations in abundance and space occupied in a heterogeneous environment (Chesson, 1994(Chesson, , 2000. The "fluctuation-dependent" model includes relative nonlinearity and storage effect models (Chesson, 2000). The "relative nonlinearity" model estimates the reproduction rate of competing species (r) taking into account the variation of the limiting factor. A more pronounced nonlinearity of r makes the species competitively weaker, but the variation of the limiting factor has a stabilizing effect, preventing the elimination of the weaker species. The "storage effect" stabilizing coexistence arises as a combined result of three components: "differential response" to the environment, "covariance between environment and competition," and the "buffer effect." The "differentiated response" implies differences in the response of species to environmental fluctuations; "covariance" implies differences in the level of competition associated with fluctuations (for example, deterioration of the environment weakens competition, and its improvement increases it); the "buffer effect" compensates for the decrease in r of the invader species in an unfavorable state of the environment and, in small mammals, may be due to different biotopic preferences (Chesson, 1990(Chesson, , 2000. Given the morphological uniformity, sympatry, and syntopicity of many shrew species, their similar ECOLOGY environmental requirements, and well-known high metabolism, one would expect particularly intense competition for space in these mammals (Churchfield, 1990. Shrews (Soricidae) are predators of medium to small invertebrates (Churchfield, 1990). Evolutionary success in this taxonomic group is determined by the "unique chewing mechanism" provided by the morphological features of the cranial skeleton, which determines the possibility of simultaneously holding and eating prey (Zaitsev, 2005). The morphological features allow animals to use prey from different taxa of invertebrates, but limit them in other food sources. The food niches of shrews overlap by 40-90%, and the overlap of food niches is higher in species of similar size (Churchfield and Sheftel, 1994;Churchfield et al., 1999;Churchfield and Ryhclik, 2006). Competing insectivorous species mutually influence the food base (Dickman, 1991). According to the principle of competitive exclusion, such species must not form stable multispecies communities. However, 4-9 species often coexist in shrews (Soricinae) (e.g., Churchfield and Sheftel, 1994;Churchfield et al., 1997Churchfield et al., , 1999Churchfield and Rychlik, 2006).
The coexistence of shrews may be facilitated by their difference in size: a difference in the size of preferred prey and a narrowing of the food niche in small species are expected (Kirkland, 1991;Churchfield and Sheftel, 1994;. However, almost complete coincidence of food niches and shared use of microhabitats were found in the Sorex araneus-Sorex minutus pair (more than a twofold difference in size) in the mountain forests of Slovenia (Klenovšek et al., 2013). Discussing the results, the authors of the latter publication intend to pay attention to the possibility of niche temporal division, which may hinder the implementation of competitive exclusion.
A number of shrew species are characterized by pronounced abundance dynamics. In particular, the common shrew demonstrates fluctuations in abundance throughout its range, which are both cyclic (Bol'shakov et al., 1996;Bobretsov, 2004;Sheftel, 1989;Tkadlec and Stenseth, 2001;Tast et al., 2005) or noncyclic (Henttonen et al., 1989;Churchfield et al., 1995;Tomášková et al., 2005). In areas with periodically changing climate humidity, the abundance and biotopic distribution of the water shrew can also change dramatically (Panov and Karpenko, 2004). Taking into account fluctuations in abundance and distribution in habitats in the water shrew-common shrew pair of species, we can talk about a "natural model" for studying "fluctuation-dependent" mechanisms of coexistence. The ranges of these species almost completely coincide (Mitchell-Jones et al., 1999;Harris and Yalden, 2004). The common shrew occurs in almost all habitats of the temperate and boreal zones, is able to use a wide range of food objects, and easily switches to the most abundant and available (both seasonally and locally) prey (Churchfield, 1990;Hanski, 1994). The water shrew is a semi-aquatic species with specific adaptations for swimming and diving (Greenwood et al., 2002;Champneys, 2012), while the common shrew is a terrestrial species (Churchfield, 1990;Shchipanov et al., 2019). However, despite differences in preferred habitats, both species coexist in wet areas: marshes, wet meadows, and along the banks of various watercourses (Churchfield, 1984a(Churchfield, , 1984bRychlik, 2000;Churchfield and Rychlik, 2006;Czabán et al., 2015). Both species are characterized as territorial (Krushinska and Rychlik, 1993;Rychlik, 1998) and could be expected to manifest the most intense interference competition leading to the displacement of one of the species. However, the common shrew cannot be considered as a territorial species in the full sense of the word. The animals do not specifically protect their territory, and the separation of individuals in space is determined by random interference (Shchipanov, 2021). The use of radioactive labeling revealed overlapping areas in water shrews as well (Lardet, 1988). The coexistence of the water shrew (Neomys fodiens) and the common shrew (Sorex araneus) in "dry" habitats has not been studied previously.
The present study (1) analyzes the long-term dynamics of these species on a model plot in a "dry" habitat, (2) characterizes the use of space and its sharing by both species, (3) studies sociability in single-species and interspecific preference tests, and (4) attempts to compare the obtained results with "fluctuation-dependent" models of coexistence, using known data on the ecology of the compared species.

MATERIALS AND METHODS
Area and Animals The material was collected in the vicinity of the village of Bakanovo in the Staritskii district of Tver oblast (56°18′ N; 34°54′ E). The main data were obtained at the capture-mark-recaptur grid ("dry" habitat) in 2014-2021. The site is located in an abandoned part of the village; borders a ravine into which a stream flows in spring and autumn and in wet years; and neighbors two small ponds (Fig. 1). The territory is overgrown with nettles and fireweed, and there are patches of cereal forbs. There is a plot of young birches mixed with alder, separate old lindens, and willow bushes. Traps (87 in total) on the grid were set as five lines with an interval of 10 m between the lines (four lines with 18 traps in each and one line with 15 traps) and 7.5 m between traps in the line. The observed area is 0.65 ha.
Animals were marked in accordance with the protocol developed for shrews (Shchipanov et al., 2000): an original trap was used, rolled oats dipped in unrefined sunflower oil were used as bait, and a check was made at least every 1.5 h. After checking, the traps were left opened, nonoperative, and available for free visit. In this study, we limited ourselves to one check per day. In this mode, the animal can move freely for more than 90% of the time of the day, and the frequency of capture of an individual reflects the frequency of its presence at the location of the trap. The animals were marked by toe clipping. In shrews, such marking does not affect the survival of animals (Shchipanov et al., 2005). For water shrews, numbers that required cutting off a maximum of two toes were used. The animals caught repeatedly during one trapping session (14 days) were considered resident. A total of 223 common shrews (1005 captures) and 29 water shrews (65 captures) were caught on the site in August 2014-2021. In trapping along streams, 38 common shrews and 21 water shrews were caught.
The behavior was studied in 2021. Animals were removed from the site for three hours in the course of inspection of the grid (until 9 a.m.), placed in pits with food in abundance (chopped chicken hearts and grasshoppers), and used in tests immediately after the end of the inspection. The tests were completed no later than 11:30 a.m., then the animals were returned to the place of capture. Individuals in repeated (and different-type) experiments were used on different days. A total of 20 underyearlings of common shrews and 15 water shrews (12 young water shrews weighing up to 16 g and three adult females weighing >17 g) participated in the tests. Of these, 17 common shrews participated in 36 single-species tests, and ten water shrews participated in 21 single-species tests. Eleven water shrews and six common shrews participated in 12 interspecific tests of water shrew-common shrew (stimulant), and 16 common shrews and eight water shrews participated in 22 experiments of common shrew-water shrew (stimulant). To assess the frequency of sociability in single-species tests, the data obtained in the first and second tests of an individual were used (nine common shrews and four water shrews; 18 and eight tests, respectively); in interspecific tests, the results of the first single-species and interspecific tests of an individual were used (14 common shrews and seven water shrews, 28 and 14 tests).

Wetness of the Area Estimates
The humidity in the "dry" habitat was characterized using the amount of precipitation falling in the warm season (May-August) for the nearest (20 km from the observation site) town of Staritsa (Pogoda i klimat). Summer drying of the temporary stream occurred in 2016, 2019, and 2020.

Sociability Assessment
The propensity of the animals for social contact (sociability) (Moy et al., 2004) was assessed in a modified preference test (Crawley, 2000). The scheme of the experiment and the evaluation of the validity of the results were described in detail earlier Demidova, 2020, 2022). The test was performed in a square open field (50 × 50 cm), with two wiremesh cylindrical containers (10 cm high and 8 cm in diameter), in one of which a stimulating animal was  placed. The animals could visually, olfactorily, and acoustically interact through the mesh wall. The cover of the container was solid and prevented contacts. The containers were placed in fixed positions (Fig. 2). The focal animal was manually released from a plastic cup in the center of the arena, and then the operator left the room. The test was recorded on a Sony Handycam video camera in MPEG-2 format. The test record was digitized using the EthoVision XT (Noldus) software package; from the moment the operator left the room, the net time of the digitized test was 5 min. The distance covered by the focal animal, staying time, and velocity in the zone were evaluated. The following virtual zones were distinguished ( Fig. 2): BZ, border zone, 3 cm strip along the walls of the arena; zones of the container with an animal (AZ) and empty container (FZ), 3 cm around the walls of the containers; covers of the container with an animal (AC) and empty container (FC); and the central zone (CZ), the space of the arena, with the exception of the area of the listed zones (Fig. 2). The choice of a 3 cm band was determined by the fact that the digital mark of an individual falls within 3 cm from the border of the object explored by an animal (Fig. 2).

Characteristics and Indices Used
The abundance of animals (N) at the grid was estimated as the number of resident animals that were present on it. Following Cantoni (1993), abundance was estimated on a line along streams as the number of individuals caught per 100 m of line. The abundance variation (nonlinearity) was estimated using the index S I = SDlogN (Stenseth and Framstad, 1980;Henttonen et al., 1985).
To characterize the combination of residents in space quantitatively, we estimated the proportion of traps visited jointly (H j ) with conspecifics or, depending on the task, jointly with individuals of an alien species: where T x is the number of traps visited only by the home range owner and T j is the number of traps covisited with neighbors. The combination of spatial activity A j was estimated as the relative frequency of visits to the traps of the site by the owner, own captures, (C h ), and other individuals of its own or another species, alien captures C a : Sociability was assessed using two groups of indices: the preference (I) and mobility (D) indices Demidova, 2020, 2022). The preference indices characterize the preference for exploring a "social" object, a container with an animal (stimulant). The time and distance preference indices I t and I d were calculated as where t is the time of presence and d is the distance covered in the zones of the container with the stimulant AZ and empty container FZ. The indices vary from -0.5 to +0.5; a positive value is the predominant activity in the stimulant zone.
The mobility index characterizes the relative density of the survey of the zone analyzed (Z): where d Z is the distance covered in the analyzed area, d T is the total distance covered in the test, S Z is the area of the analyzed zone, and S a is the total area of the arena. The index is 1 if the density of traces in the zone corresponds to a random walk around the area.

Data Analysis and Statistics
The normality of the distribution was tested in the Shapiro-Wilk test. If necessary, the data were normalized using the logarithm or arcsine transformation (in the case of I t and I d , 0.5 was added to the index value). The variation in normally distributed data is shown as ±SD in the text and figures; otherwise, explanations are given. Correlations of normally distributed data and non-normally distributed data were estimated by Pearson's coefficient (r) and Spearman's rank coefficient (R S ), respectively. Synchronicity in the abundance dynamics was assessed using Spearman's rank correlation coefficient for logarithmic abundance rates, as previously proposed for voles (Saitoh et al., 1998). Parametric or nonparametric tests (indicated in the text) were used when comparing samples, depending on the type of data.
To estimate the possibility of a random coincidence of traps visited in the space by N individuals, each of which, on average, visits k (rounded to the nearest integer) traps, the expected number of jointly visited traps was determined in a computer experiment. In doing this, the model generates random numbers in Nk trials. The results of 1000 iterations performed in the Excel 2016 environment using the built-in random number generator were used to determine the mathematical expectation and 95% confidence interval (CI), 1.96SD (for more information, see the study by Shchipanov, 2021).
To identify sustainable individual behavior the adjusted repeatability R (Nakagawa and Schielzeth, 2010): within-group variance among individuals, which was divided by the total variance, was assayed using the rpt function in the rptR package (Stoffel et al., 2017). In each calculation, the individual animal number (ID) was a random and the experiment number was a fixed factor. To determine the 95% CI, 1000 iterations were carried out. The statistical significance of R was assessed using permutations and the maximum likelihood test (LRT) in the rptR package.

Abundance
The common shrew predominated in terms of abundance: in the "dry" habitat on the grid for the entire observation period, the number of water shrews was 10.6% of the total number of both species caught; in "wet" habitats, it was 35.6%. According to the geometric mean abundance (for all the years of capture), the predominance of common shrews is even more noticeable: on the grid, 1.3 water shrews versus 30.1 common shrews; along watercourses, 1.4 water shrews versus 3.2 common shrews. The dynamics of the abundance of these species on the grid are independent (Rs = 0.47, p > 0.2). The common shrew did not show significant fluctuations (Fig. 3): S I on the grid was 0.09, i.e., much less than 0.5, the critical value for fluctuating populations (Henttonen et al., 1985). Fluctuations in the abundance of water shrews in the "dry" habitat were strongly pronounced, S I = 0.79. In near-water areas, the abundance of water shrews almost did not change during the years of surveys (Fig. 3): We could correctly assess the effect of territory wetness only for a "dry" habitat, i.e., the grid. On the grid, water shrews were found to have a significant negative correlation between the abundance and total precipitation at the end of summer, in July and August, but only a tendency (p < 0.1) to an increase in abundance in the case of lower total precipitation over the summer and drier August (Table 1). The summer amount of precipitation of the current year is not related to the presence of running water in the stream, which remained there in dry years and apparently depended on the moisture accumulated in the soil. If we mark years with flowing water as 1 and dry years as 0, we find that the abundance of water shrews on the site positively correlates with the presence of water in the stream (R S = 0.80, p < 0.02).

Use of Space
Sufficiently complete data for analysis were obtained only in the "dry" habitat. Resident common shrews visited, on average, 72.6 ± 13.2% of traps (out of all 87 on the grid); in 2021, they visited 77%. From  30 to 61% (49% in 2021) of the traps in the home range of the common shrew were also visited by neighboring common shrews. The share of traps visited jointly (H j ) correlated (R S = 0.91, p < 0.01) with the share of traps (out of all 87) visited by resident common shrews (Fig. 4).
It was previously shown that the number of covisited traps in the common shrew in 2014-2020 corresponded to the expectation for a random match of trap numbers (Shchipanov, 2021). This situation did not change in 2021 either. In total, common shrews visited 68 traps, of which 30 traps were visited by two or more individuals, i.e., H j = 0.49. The average number of traps in the home range of an individual was 4.5. The total number of traps in the ranges of all 25 resident animals was Nk = 113. With this number of tests, an average of 32 trap numbers match in the model, the 95% CI boundaries are 26-38 matching numbers. Thus, the actual number of traps visited by common shrews jointly with their neighbors (30) does not go beyond the confidence interval expected for a random coincidence. The space visiting is proportional to the share of jointly visited traps. The actual frequency of visits to traps by site owners jointly with "aliens" (A j = 0.54 (74 : 88)) does not differ (χ 2 = 1.58, df = 1, p > 0.20) from the expected one (0.51 (82 : 80)).
In 2021, at a high density on the site, we managed to estimate the probability of a random coincidence of the visited traps for water shrews as well. In total, water shrews visited 26 traps, of which six traps were visited by two or more individuals, i.e., H j = 0.23. The average number of traps in the home range of the water shrew is four, and the total number of traps in the ranges of all eight resident animals is Nk = 32. An average of four trap numbers randomly coincide in the model, and the 95% CI ranges from 1.1 to 7.5. Thus, for water shrews, the number of traps visited jointly with their neighbors (6) is also within the confidence interval expected for a random coincidence. The frequency of visits to traps by home range owners jointly with "aliens" A j = 0.18 (8 : 36) does not differ (χ 2 = 0.52, df = 1, p > 0.47) from the expected value, 0.23 (10 : 34) for a ratio that is proportional to the share of traps visited jointly.
The joint presence of the common shrew and water shrew in the traps corresponded to a random coincidence of the numbers of the visited traps. Out of 68 traps visited by common shrews, 25 were also visited by water shrews. Random trap numbers were generated for each of the species in separate models. Coincident numbers were determined in the water shrewcommon shrew model pairs based on the results of 1000 iterations. As a result, the average trap numbers randomly matching in both species was 26, and the 95% CI ranged from 20 to 32. Thus, the actual number of traps visited jointly by both species (25) does not fall outside the confidence interval expected for a random match.
To test the assumption of the influence of trap occupancy on the frequency of their visits by individuals of different species, we estimated the frequency of falling into traps visited by conspecifics and "alien" species. As a result, no dependence was found between the frequency of visits to the trap by one of the species and the frequency of visits to the trap by another species (R s = 0.096, p > 0.6), and the number of visits to traps by conspecifics and alien species (the median of the number of common shrew catches in the traps visited by both conspecifics and water shrews is equal to two in both cases) do not differ (Mann-Whitney U = 487.5, p > 0.37).

SS SN
A significant positive correlation between the number of traps occupied by common shrews and log of the abundance of water shrews in the period 2014-2021 (r = 0.78, p < 0.023) shows that the suitability of the site space for resident individuals of both the common shrew and the water shrew changes synchronously.

Behavior
The total activity in the test, i.e., the average speed of movement and the total distance covered in the test, are significantly (ANOVA, ID, random, p < 0.05) higher in the common shrew (Fig. 5). In common shrews, activity in the open field in the presence of the water shrew stimulant significantly (t = 3.65, df = 32, p < 0.001) decreased: the total distance in single-species tests was 2960 ± 1124 cm, and the total distance in tests with the water shrew was 1671 ± 903 cm. In water shrews, changes of the total distance in tests with the common shrew stimulant (Fig. 5) was insignificant (p > 0.18).
The distribution of spatial activity in different zones of the arena (D-index) in the water shrew and common shrew is the same (Fig. 6). In all types of tests, the visiting density in the central part of the arena (D CZ ) is significantly less than one (Table 2). Mobility in single-species tests in the central part of the arena (D CZ ) in the common shrew is significantly (t = -2.34, df = 56, p < 0.023) less than in the water shrew, which indicates the greater caution of the animals of this species. In interspecific experiments, the common shrew significantly decreased movements in the border zone (t = 2.41, df = 58, p < 0.019) and increased mobility in the zone of an empty container, FZ (t = -2.3, df = 58, p < 0.023); other changes in interspecific tests of both species are not significant (Fig. 6).
The density of visiting container covers did not differ from one in both species, which allows us to speak of a random visit to this zone. However, it can be noted that the density of visiting the cover of an empty container that is a nonsocial explored object in intraspecies tests was higher in common shrews than in water shrews at the trend level (Mann-Whitney U = 272.5, p < 0.08), indicating a greater thoroughness of exploration.
The density of visits to container zones was in all cases significantly higher than the density of movements in the border zone (Table 3), which indicates the interest of animals in the exploration of these  Table 2), and medians (Me) are shown for non-normally distributed data. Significant differences with single-species tests (p < 0.05) are shown as parentheses.  (Table 4). In interspecific tests, this difference disappears in the common shrew, since the speed of movement in the empty container exploration zone (FZ) decreases highly significantly (t = 3.96, df = 31, p < 0.001). No noticeable changes were observed in water shrews in interspecific tests (Table 4).
The preference in distance (I d ) ranged from -0.48 to 0.28 in common shrews (on average, it was 0.01 in single-species tests and -0.06 in tests with the water shrew stimulant); in water shrews, it ranged from -0.44 to 0.43 (on average, it was -0.01 in single-species tests and 0.01 in tests with the common shrew stimulant). In all samples, I d did not differ from zero. Given the significant differences in the velocity in the zones of the object, the preference in time (I t ) is more indicative. This index in the common shrew in singlespecies tests varied from -0.49 to 0.38, averaged 0.11 for the sample, and was significantly (signs test) greater than zero (n = 38, %I t > 0 = 68, p < 0.034). In water shrews, I t (from -0.41 to 0.49) averaged 0.09 and did not differ significantly from zero. In interspecific tests, I t in common shrews significantly (t = -2.58, df = 57, p < 0.013) decreased or became slightly negative (-0.04), not differing from zero. No significant changes were found in water shrews; I t averaged 0.06.
The repeatability (R) of preference indices was found only in intraspecies tests; i.e., both species were found to have sustainable individual differences in sociability with conspecifics and "social indifference" in relation to the "alien" species (Table 5).
The preference indices (I d and I t ) are highly significantly (p < 0.0001) correlated with each other: in the common shrew, r = 0.89, and in the water shrew, r = 0.95; however, in the common shrew, the preference in time, index I t positively correlates with the movement density in the stimulant zone (AZ) and on the cover of the container with the stimulant (AC), and the preference in distance, index I d is negatively correlated with movements in the zone (FZ) and on the cover (FC) of the empty container (Table 6). Thus, in common shrews, I t mainly reflects the social component of behavior and I d reflects the exploratory activity. In water shrews, both indices are significantly correlated with container zones: positively with the zone with a stimulant and negatively with the empty one. Movements on the cover of the container with a stimulant correlate with both preference indices, while movements on the cover of the empty container are in no way correlated with indices (activity on the covers of containers indicates whether the presence of a stimulant in the object is an obstacle to exploring the object). Mobility in the central and boundary zones of the arena is not correlated with preference indices in these species (Table 6)

DISCUSSION
Fluctuations in the abundance of the water shrew in constantly wet, near water habitats are usually not pronounced (Churchfield, 1984a;Michelat and Giraudoux, 2006). At the same time, the abundance of water shrews is associated with fluctuations in the water level: in the Kis-Balaton wetlands, the abundance and colonisation of habitats significantly increase during periods of greater flooding of the territory (Czabán et al., 2015). In the Baraba lowland, where the dry and wet phases of the climatic cycle alternate, the abundance of the water shrew increases many times during the years of the highest humidity, periodically reaching peak values, and then the water shrew occupies over 40% of the total catch of shrews (Panov and Karpenko, 2004). During the periods of increasing numbers, water shrews move into adjacent, drier habitats (Panov and Karpenko, 2004). Note that the abundance of water shrews in Baraba was associated not with the current abundance of precipitation, but with the water level in the swamps, which increased several years after heavy precipitation. In the case under consideration, the migration of the water shrew to "dry" habitats correlated with drier weather; however, it was observed only in those years when the stream that was closest to the site did not dry up. Apparently, the appearance of water shrews on the marking site was determined by a combination of successful reproduction in humid habitats and a shortage of favorable microhabitats near water that arose during the dry period. The level of the population density found by us in the "dry" habitat is comparable to the density in wet habitats: according to our data, it ranges from zero to 12.3 (at the time of the peak) individuals per ha, on average, 2.7 ind./ha; in water-cress beds, it is 3-5 ind./ha (Churchfield, 1984a). The abundance index along the stream, which is 1.4 individuals per 100 m (according to our data), is close to the abundance index along the canals in Switzerland: 1.8 per 100 m (Cantoni, 1993). The abundance of the common shrew on the site that is 30-40 individuals per ha corresponds to the known density levels of this species in herbaceous habitats (Shchipanov et al., 2019). The common shrew inhabits an area with a generally dense network of rivers and streams, and the periodic coexistence with the water shrew in "dry" habitats we observed is not unique. The diet of the water shrew living near water is dominated by aquatic animals (DuPasquier and Cantoni, 1992), but, in drier habitats and during the drier season (from mid-summer), the water shrew mainly uses terrestrial invertebrate species (Haberl, 2002;Churchfield, 1984b). Common shrews are sensitive to the food supply of the territory (Lukyanova et al., 2021). In wet habitats, competition between water shrews and common shrews can be reduced due to preferring different microhabitats (Churchfield and Rychlik, 2006). However, in a "dry" habitat, we found a positive correlation between the number of traps visited by resident common shrews and the number of resident water shrews. This suggests that favorable conditions for residents of both species are similar in this area.
As O.A. Zhigalskii notes (2007), the requirements of the species to the environment "indicate the possibility of the existence of interspecific competition rather than serve as direct evidence for it." In the "dry" habitat, we did not find spatial separation of resident common shrews and water shrews: the joint occurrence in space corresponded to the probability of a random coincidence of home ranges. No specific behavioral responses that could reduce the likelihood of direct contact between the animals were found. Both the water shrew and the common shrew are aggressive in both intraspecific and interspecific interactions, have a similar set of behavioral patterns, and, as in other shrews, the larger species dominates in contacts (Kalinin et al., 1998;Rychlik and Zwolak, 2006). The behavior of water shrews in interspecific interactions has been repeatedly studied in open-air cages and in contact tests. In particular, in the interactions of the common shrew and Mediterranean water shrew (Neomys anomalus), it was found that the number of aggressive contacts between resident animals was much less than in nonresident individuals of these species, and, taking into account the decrease in the number of conflicts, which was found (in this experiment) within 1-3 days, rapid mutual learning was assumed (Krushinska et al., 1994). In experiments, in the N. fodiens-N. anomalus pair, the smaller Mediterranean water shrew, which is the subordinate in interspecific contacts, avoids the larger common shrew, being guided, among other things, by acoustic and olfactory signals (Krushinska and Rychlik, 1993). Cohabitation with the water shrew could be expected to have a similar effect on the behavior of the common shrew. Contrary to expectations, we did not reveal any reactions aimed at avoiding or crowding out the "alien" species in the S. araneus-N. fodiens pair. Both species showed persistent individual differences (repetition of sociability) in single-species tests and indifference to the "social object" in interspecific tests. In interspecific tests, both species showed equal interest in exploring the container with an animal and the empty container. A reaction to the presence of the "alien" species was observed only in the common shrew: the general movement activity decreased significantly, but at the same time the attendance of the boundary zone of the open field and the speed of movement near the containers decreased (the thoroughness of exploration increased). Thus, the water shrew was perceived by common shrews not as a threatening subject, but rather as a disturbing object stimulating general exploration activity. This assumption is consistent with the observations in which common shrews dominated when being kept together with water shrews, and the number of intraspecific conflicts exceeded the number of interspecific ones (Köhler, 1985).
Thus, in dry habitats, we did not find spatial separation or any behavioral mechanisms capable of reducing the likelihood of direct contacts between the water shrew and the common shrew. At the same time, the considered situation is in good agreement with the models of "fluctuation-dependent" coexistence. Both species have a stable abundance trend . Both species have nonlinear population dynamics, but in "dry" habitats the species that is dominant in direct contacts, the water shrew (Rychlik and Zwolak, 2006), has a higher coefficient of nonlinearity (S I ), which, according to the "relative nonlinearity" model, reduces the pressure on the competitor (Chesson, 2000). At the place of invasion, the conditions that are favorable for the resident existence of both species coincide, and we see a positive covariance between the environment and competition. However, while the abundance of water shrews is associated with humidity of the territory, common shrews are not affected by this factor. The humidity of the territory can be considered as a "limiting factor," which is observed by us to vary in the water shrew. In this case, the "buffer effect" manifests itself as a division of preferred habitats under unfavorable environmental conditions: during low-water periods, the water shrew stays in near-water sites limited in area. Thus, considering the coexistence of the common shrew and water shrew, we can talk about the implementation of the "storage effect" as a result of three components: "differentiated response" to environmental changes, "covariance between the environment and competition," and the "buffer effect," as it is assumed in the models (Chesson, 1990(Chesson, , 2000.
It seems to us that the interactions in this pair of species can be a good natural model for testing the premises of the "fluctuation-dependent" coexistence that follow from the theoretical models and deserve deeper study.

FUNDING
This study was carried out within the framework of a State Assignment, project no. АААА-А18-118042490060-1.