Resource partitioning confirmed by isotopic signatures allows small mammals to share seasonally flooded meadows

Abstract Meadows in river deltas are characterized by a high diversity and abundance of small mammals. However, neither their spatial arrangement nor differences in their use of microhabitat can necessarily explain the dense co‐occurrence of sympatric species. We investigated how several small mammal species share a seasonally flooded meadow of limited size, testing predictions (P1) that herbivore, granivore, insectivore, and omnivore species are separated in time (dominant in different years), (P2) that sympatric species undergo isotopic partitioning, and (P3) that there are intraspecific differences in diet. Stable carbon and nitrogen isotope signatures in the hair of seven synantropic shrew, vole, and mice species were used as a proxy for their diet. We found that the three most abundant species in eight of the nine years were from different diet groups. However, based on the number of species in the functional groups, the state of small mammal community was considered unfavored in five out of the nine investigation years. In years with the greatest dominance of Apodemus agrarius, the small mammal community was characterized by decreased diversity and Micromys minutus was either in low abundance or absent. In 2014 and 2016, years of low abundance or absence of M. oeconomus, M. agrestis, and M. glareolus were both recorded in high numbers. Differences in the isotopic signatures of the three most abundant small mammal species in the community were clearly expressed and core areas in the isotopic space were separated, showing their dependence on different dietary resources. Intraspecific dietary separation between young and adult animals was observed only in M. oeconomus. Thus, the high species diversity of small mammals and the formation of their community in this investigated flooded meadow are maintained by isotopic partitioning (segregation in dietary space) and by changes in their number over time (shifting dominance).

Small mammal communities are not randomly assembled, they follow so-called "assembly rule" (Fox & Kirkland, 1992). It says that "each species entering a community will tend to be drawn from a different group until each group is represented, and then the rule repeats" (Fox & Brown, 1993). According to Fox (1987), we should expect a single species from each of the different dietary groups to form the community in years with low small mammal diversity, with increasing resources thereafter allowing the addition of a second species from each group, then a third, ultimately resulting in a favorable community structure (see Data analysis). As these small mammal groups reflect dietary separation, isotopic partitioning also should be expected (Calandra et al., 2015;Hwang, Millar, & Longstaffe, 2007).
The aim of the study was to investigate the pattern of coexistence of several small mammal species in a seasonally flooded meadow, based on the working hypothesis that, in order to coexist in a small area, species should be separated not only in dietary space but also in time. We supposed that separation in dietary space would operate for a single year, while shifting dominance would operate over the longer periods, this additionally reflecting the differing resilience of various species to floods. We tested three predictions: P1-sympatric species of the same group (herbivores, granivores, insectivores, and omnivores) are separated by time, that is, dominate in different years, P2-sympatric species are separated in dietary space, thus differ in isotopic signatures, and P3-intraspecific differences between various demographic groups are present (assuming intraspecific competition for food). P3 is based on our previous and ongoing research (Balčiauskas, Skipitytė, Jasiulionis, Balčiauskienė, & Remeikis, 2018;Balčiauskas et al., 2016), where we found some intraspecific segregation in the isotopic space in yellow-necked mice (Apodemus flavicollis) and bank voles (Myodes glareolus) living in great cormorant colonies, an environment where foods are scarce, thus necessitating competition.

| STUDY SITE
We studied the small mammal community of a flooded meadow (55°19'26.23"N, 21°20'24.15"E) near Rusnė settlement (55°20'10''N; 21°18'54''E) in the Nemunas River Delta, situated in western Lithuania ( Figure 1). The delta is on the border of two major biogeographical regions in Europe (European Environment Agency, 2002), namely the boreal and continental, and thus, the small mammal community encompasses species from both.
The area of the site is quite small (7.05 ha, with a perimeter of 1,070 m) and is flooded every year (Balčiauskas et al., 2012b), with the duration of submergence dependent on flood height. Regardless of flood level, the trapping site is totally flooded for only a short time each spring. Spring floods normally start around 19 of March, and the average duration of flood is 16 days (Floods, 2018). In the study area, spring floods effectively eradicate the small mammal communities in the meadow, but the negative effects are short-term and high small mammal diversities are restored during the summer period from enclosing levees, serving as refugees during flood (Balčiauskas et al., 2012b).
The area consists of a polder system with artificially raised embankments to protect against high spring floods. The meadows are surrounded by ditches, overgrown by reeds and partially by shrubs   Notes. Diet preferences marked with superscripts: a-insectivores, b-granivores, c-herbivores, d-omnivores (according to Butet & Delettre, 2011;Churchfield & Rychlik, 2006;Zub et al., 2012;Pernetta, 1976). Shannon's H measures diversity of the small mammal community, Simpson's c the dominance. Trapping effort is expressed in trap nights pooled. Each year, we used 6-31 lines of 25 snap traps, each set 5 m apart, the number of lines depending on the number of trapping sessions (Table 1). We positioned the traps according to the perimeter of the site in all years, the trap lines being close to drainage ditches (2-10 m) and adjacent reed belts. In 2009, traps were additionally set on a diagonal transect ( Figure 1). Traps were set for three days, checked once a day, and baited with bread crust and sunflower oil.

| Small mammal trapping
The total trapping effort was 7,651 trap nights, and 1,359 individuals of 11 species were captured (Table 1). Presented in the Supporting information

| Stable isotope analysis
To test predictions P2 and P3, hair samples were collected in 2015 from 81 individuals of the seven small mammal species for stable isotope analysis (Table 2). We clipped off a tuft of hair from between the shoulders of each specimen and stored it dry in separate bags.
Scissored samples were weighed with a microbalance and packed in tin capsules, and stable isotope analysis was then carried out.
Carbon and nitrogen stable isotope ratios were measured using an elemental analyzer (EA) (Flash EA1112) coupled to an isotope ratio mass spectrometer (IRMS) (Thermo Delta V Advantage) via a ConFlo III interface (EA-IRMS).
Carbon and nitrogen isotope data are reported as δ X values (where X represents the heavier isotope 13 C or 15 N) or differences from given standards, expressed in parts per thousand (‰), and are calculated according to the formula: where R sample = 13 C/ 12 C or 15 N/ 14 N of the sample, R standard = 13 C/ 12 C or 15 N/ 14 N of the standard.

| Data analysis
The diversity of the small mammal community was expressed using the Shannon-Wiener diversity index, H, on the base of log 2 (Krebs, 1999), while dominance was expressed using the Simpson's index c (Golet et al., 2013;Krebs, 1999;Zhang et al., 2007). Diversity of the community was compared to other habitats and territories of different size in Lithuania, data from Balčiauskas and Juškaitis (1997).

We checked if there was a correlation between diversity and
A. agrarius dominance (this a generally uncommon species in the country, but strongly dominant during most of the investigation). Dominance was calculated as a percentage of the total number of trapped individuals. Pearson's r was used as dominance values were distributed normally.
Prediction P1 was tested according to the rule of equal representation of functional groups (insectivores, granivores, herbivores, and omnivores) in a small mammal community (Fox, 1987).
Accordingly, if the difference between the numbers of species trapped in these four groups in any year is >1, the state of the community is considered unfavorable. The distribution of favored and unfavored states of the small mammal communities in the Rusnė flooded meadow is presented in Supporting information Table S2.
The influences of species, as well as intraspecific differences (between males and females, and between the three age groups), on the carbon and nitrogen stable isotope values were tested with parametric ANOVA, using Wilk's lambda test for significance. Differences between groups were evaluated with post hoc Tukey test.
3.5.0 (https://cran.r-project.org/bin/windows/base/rdevel.html) for the five most numerous small mammal species, having five or more individuals investigated for δ 15 N and δ 13 C. Positions of seven small mammal species, including those with sample size n < 5, in the isotopic biplot were shown using SigmaPlot ver. 12.5. All other calculations were performed using Statistica for Windows ver. 6.

| Diversity of small mammals in the flooded meadow
Eleven  (Table 1).
An increasing dominance of A. agrarius was negatively related to the diversity of the small mammal community (r = −0.74, n = 9, p = 0.02). In the years of the strongest dominance of A. agrarius, the small mammal community consisted of 4-5 species, with a low abundance or absence of M. minutus, a species belonging to the same granivorous group (Table 1).

| Temporal changes
Throughout the investigation, the composition of the small mammal community followed the expected numbers of species in functional groups (differences from the expected numbers were not significant). However, in five out of the nine years, the state of the small mammal community was unfavored.  (Table 1).
Thus, based on the frequent deviations from the species assembly rule, P1 prediction was not fully confirmed, but a change in small mammal numbers over time (shifting dominance) was clearly demonstrated.

| Interspecific differences in dietary space
The range of stable isotope values, though overlapping, showed a separation of several species (Figure 3, Supporting information  (Figure 3a). The difference between average δ 13 C values in granivores and herbivores was 5.6%, while between granivores and omnivores it was 5.9% and between omnivores and insectivores 4.3%. However, the difference between herbivores and omnivores was just 0.3% (Supporting information Table S4).
The dietary niches of the most abundant species (core ellipses in the isotopic space) were separated and did not intersect (Figure 3b)

| Intraspecific differences in dietary space
Differences in the stable isotopes in the hair of male and female small mammals were not significant in general for δ 15 N (F 1,10 = 1.36, p = 0.27) or δ 13 C (F 1,10 = 1.51, p = 0.31), nor in some separate species (Supporting information Figure S1). No significant differences between stable isotope values were found between age groups in A. agrarius, M. glareolus, and M. agrestis (Figure 4a-c). Thus, prediction P3 for most of the analyzed species was not confirmed. In M. oeconomus, differences in the stable isotopes in the hair of young, subadult, and adult small mammals were significant for δ 13 C (F 2,31 = 3.34, p = 0.048) and near-significant for δ 15 N (F 2,31 = 3.21, p = 0.054) (Figure 4d). However, the difference expressed in percentage was not large: juveniles of M. oeconomus were characterized by 1.4% lower δ 13 C than adult animals and 10.5% higher δ 15 N.

| D ISCUSS I ON
We analyzed how several small mammal species, representing insectivores, granivores, herbivores, and omnivores, share a seasonally flooded meadow of limited size. With the re-occupation of the habitat after the spring flood, spatial arrangement may "pack" species of small mammals tightly, high floods giving chances to uncommon species to establish (Balčiauskas et al., 2012b). We identified separa-

| Small mammal diversity in the flooded areas
In general, higher species diversities are characteristic of larger areas (Balčiauskas & Juškaitis, 1997), but similar patterns are also found in seasonally flooded sites. For example, in the floodplains of the Sava River, 23 small mammal species were registered (Crnobrnja-Isailović et al., 2015), and in a much bigger area of the flooded Narewka River valley in Poland, the diversity was higher, with H = 2.46 and 11 species registered (Zub et al., 2012). The small mammal diversity in Rusnė did not differ from the bigger floodplains of the Vltava (H = 2.18, 8 species) and Danube (H = 2.21, 9 species) rivers (Bohdal, Navratil, & Sedlaček, 2016;Miklós, Žiak, & Hulejová, 2015).
Our diversity index (H = 2.13) was greater than that found in 95 out of 125 small mammal trapping sites across Lithuania, regardless of the size of these territories, which were in most cases significantly larger. Only in eight territories was the number of registered small F I G U R E 4 Intraspecific differences in the stable isotope values in the hair of young, subadult, and adult small mammals: a-Apodemus agrarius, b-Myodes glareolus, c-Microtus agrestis, d-Microtus oeconomus, black circles-adult, gray circles-subadult, white circles-young animals. Differences between young and adult animals in M. oeconomus were significant for δ 13 C and had a trend for δ 15 N (p = 0.054) mammal species larger than in the flooded meadow at Rusnė (re-calculated from Balčiauskas & Juškaitis, 1997).
Of note is a new small mammal species for the Baltic countries.

| Temporal changes in the dominant species
In the flooded meadow at Rusnė, high numbers of species sympatrically shared an area of limited size. Three species dominated during the nine years of investigation: the herbivorous M. oeconomus during two years, the granivorous M. minutus in one year and the granivorous A. agrarius in six years. An increase in M. oeconomus numbers was observed every fourth year (see Table 1), while A. agrarius dominated in the community for the last four years of the study period (2013)(2014)(2015)(2016).
It has to be noted that floods are extreme environmental phenomena, not only causing small mammal mortality, but also changing the dominant species and the resulting organization of the entire community (Thibault & Brown, 2008). Seasonal floods do not only have negative or even catastrophic effects on small mammal communities (Andersen, Wilson, Miller, & Falck, 2000), but can also influence the diversity of such communities positively (Golet et al., 2013). Generally, an increase in diversity after disturbance is observed, though such a relationship is not always linear and straightforward (Mackey & Currie, 2001). We previously found that flood height was a key factor influencing diversity and dominance in the small mammal community in the Rusnė flooded meadows. After low-level floods, A. agrarius was the dominant species, while high-level floods increased the chances for other species to dominate the meadow (Balčiauskas et al., 2012b). This corresponds to the situation described by Brown et al. (2001), where environmental perturbations can fully reorganize ecosystems, exceeding the ecological tolerances of dominant or keystone species; though changes may be buffered due to the compensatory dynamics of complementary species. In the investigated area, the 2010 flood in particular was very high (Balčiauskas et al., 2012b), and it was in this year that the two most abundant species were granivores A. agrarius and M. minutus. A. agrarius was shown to be the best colonizer of previously flooded areas within agricultural land (Zhang et al., 2007).
Thus, our recorded dominance of A. agrarius in the flooded meadow in six out of the nine years is not typical for Lithuania. There is no previous record of such dominance during earlier decades in various investigated habitats in the country (Balčiauskas, 2005;Balčiauskas et al., 2017;Balčiauskas & Juškaitis, 1997;Šinkūnas & Balčiauskas, 2006).
Several previous investigations have also confirmed favored states of small mammal communities in Lithuania (i.e., Balčiauskas & Juškaitis, 1997;Balčiauskas, 2005;Šinkūnas & Balčiauskas, 2006;Balčiauskas et al., 2017). However, in our flooded meadow, the community of small mammals was in an unfavored state (sensu Fox, 1987) in five of the nine years, and in one year, the two most numerous species, namely A. agrarius and M. minutus, were both granivores. According to Tulis et al. (2016), the negative interaction of A. agrarius occurs mostly with A. flavicollis, M. glareolus, S. araneus, and M. minutus. Hence, it is unusual to observe a high number of A. agrarius and M. minutus simultaneously.
One possible explanation at this locality lies in the abundance of a prevailing lush herbaceous vegetation (Wijnhoven et al., 2005) and of reed seeds at the flooded sites (Marques et al., 2015).
Alternatively, the changing structure of the small mammal community as it reoccupied the vacant area after a particularly high flood may also explain this unusual co-occurrence.

| Isotopic partitioning
We expected that the tight packing of sympatric species and their segregation in dietary space would be reflected by stable isotope values. Testing two predictions, we found that isotopic partitioning may have helped maintain a high diversity of small mammals in the seasonally flooded meadow. Species were segregated in dietary space (confirming prediction P2), as was shown by analysis of stable isotopes from their hair. We interpret nearly full separation of the central ellipses as separation in dietary space. The only overlap in central ellipses, being less than 2%, was that between A. agrarius (dominant species in most years) and M. oeconomus.

Diversity (but not abundance) of resources in a limited area
presumably should also be limited, putting constraint on the differences in δ 15 N and, even more, in δ 13 C values. While differences in δ 15 N between insectivores, herbivores, and granivores were nearly 30%, differences in δ 13 C were a mere 5%. Thus, we have to interpret dietary separation with caution, possibly because of territorial limitation.
The widest trophic niche among the small mammals was occupied by A. agrarius, as the variance of stable isotope values in their hair was highest (see Figure 3a and Supporting information Table S3) and the core area largest (Figure 3b). A wider trophic niche supports stability in a species (Bearhop, Adams, Waldron, Fuller, & MacLeod, 2004;Wood et al., 2017), enabling the domination of A. agrarius in the area of investigation. We did not find intraspecific differences of stable isotope values in most of the investigated species (prediction P3 not confirmed), with some trend in M. oeconomus age groups only.
Isotopic partitioning of small mammal species is characteristic in other cases of limited space, such as under snow cover (Calandra et al., 2015;Merwe & Hellgren, 2016). The segregation of the isotopic niche spaces of small mammals, minimizing interspecific competition, allows sympatric species to coexist (Baltensperger et al., 2015), especially in grasslands, where small mammals are more plastic in their dietary preferences (Symes et al., 2013).
However, we found no other studies for comparison with regard to isotopic partitioning in small mammal species in a small area equivalent to the Rusnė flooded meadow. Although intraspecific dietary separation was found in A. flavicollis and M. glareolus living in the territory of a great cormorant colony, we interpret this as competition for scarce food resources and as adults feeding in the best habitats (Balčiauskas et al., 2016 and references therein). We suppose that abundant and diverse food in the cyclic habitat of the flooded meadow allows most species to avoid intraspecific competition. As for M. oeconomus, it is a relatively new species in Lithuania, arriving only about half a century ago (Balčiauskas, Balčiauskienė, & Baltrūnaitė, 2010), and thus, it may have a different strategy of habitat use.

| CON CLUS I ON S AND S I G NIFI C AN CE
We found that the small mammal community in the restricted area of flooded meadow maintained a high species diversity despite a cyclic stressor (flood) due to isotopic partitioning (segregation in In the future, with respect to climate change and the resultant expected increases in extreme flood events in northern Europe (Reader et al., 2014) and the arrival of new species due to changes in distribution ranges and consequent changes to communities and food webs (Baltensperger et al., 2015), knowledge of the formation of small mammal communities may help in the prognosis of ecosystem changes and predicting at-risk species.

ACK N OWLED G M ENTS
We thank Dr. Marian Litvaitis, Dept. of Natural Resources and the Environment, Univ. of New Hampshire, and five anonymous reviewers for their comments on several manuscript versions, and, of course, the patience of the Editors. We also thank Jos Stratford for polishing the language of the manuscript and Dr. Gintautas Vaitonis for graphic works.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S ' CO NTR I B UTI O N S
LB1 formulated the research idea, did statistical analysis, and drafted the manuscript; LB1, LB2, and MJ trapped small mammals; LB2 identified species, performed literature overview, and revised all manuscript versions; RS and MJ performed stable isotope analysis. All authors contributed critically to the drafts and gave final approval for publication.

DATA ACCE SS I B I LIT Y
Data from this study (stable isotope raw data matrix) available from the Dryad Digital Repository: https://doi.org/10.5061/ dryad.2rc8s7m.