Correlational patterns of species diversity, swimming ability and ecological tolerance of non-marine ostracoda (Crustacea) with different reproductive modes in shallow water bodies of ağrı region (Turkey)

Abstract A total of 52 ostracod taxa (35 live spp. 17 subfossils) were collected from 70 shallow aquatic bodies with 11 different habitat types in Ağrı province (Turkey). Fabaeformiscandona acuminata is a new record for the Turkish ostracod fauna while 29 species were new for the province. Three types of habitats (stream, ditch and pond) contained the highest numbers of 24, 15, 11 species, respectively. Species were clustered into four main groups based on their distribution among habitats. Canonical Correspondence Analysis was able to explain 73.9% of the correlation between the 18 most frequently found species and five environmental variables. Water temperature and elevation were found to be the most effective factors on ostracods. Species with and without swimming setae tend to be found mostly in lentic and lotic habitats, respectively. Besides, species with swimming setae showed relatively higher ecological optimum and tolerance values for pH, electrical conductivity and water temperature than species without setae. The numbers of non-swimmer sexually reproducing species were twice the number of parthenogenetic species. Parthenogenetic species tend to have higher tolerance and optimum values for different environmental variables than sexual species. Results suggest a strong correlation among reproductive modes, swimming ability and ecological tolerances of the species.


Introduction
Understanding how species inhabit their habitat(s) is crucial for several reasons, such as estimating patterns of species distribution, analyzing suitability of habitats, calculating species diversity, and using the information in habitat management (Hall et al. 1997;Krausman 1999;Akdemir and K€ ulk€ oyl€ uo glu 2011;Uçak et al. 2014;Yavuzatmaca et al. 2015;K€ ulk€ oyl€ uo glu et al. 2016). Species may or may not be found together in the same types of habitat depending on spatial and temporal conditions. When their habitat requirements are different, species diversity may be increased by habitat diversity. Accordingly, one may consider some kind of mutual relationship between species and habitat diversity (but see details in McGuinnes 2000). This idea can also be applied for ostracods.
Ostracods, being small (ca. 0.5-5.5 mm) aquatic invertebrates, can be found in almost all kinds of aquatic habitats from deep sea (5000 m b.s.l.) (Benson 1972;Brandão and Yasuhara 2013) to high mountain waters and lakes (>4000 m a.s.l.) (Laprida et al. 2006;Mischke et al. 2007) as long as the habitat conditions are suitable (K€ ulk€ oyl€ uo glu and Sarı 2012; Uçak et al. 2014). Hence, variability of dispersion within this wide elevation range (ca. 10,000 m) in a variety of aquatic habitats may also be related to species broad ecological tolerances along with different adaptive values (K€ ulk€ oyl€ uo glu, , K€ ulk€ oyl€ uo glu and Sari 2012. They are one of the most abundant and diverse taxonomic groups in Crustacea and show different reproductive modes. Some species are only known as parthenogenetic (only females found in the population) while some others represent sexual (both males and females present) or mixed populations (McKenzie 1971;Butlin et al. 1998). According to Chaplin et al. (1994), parthenogenetic ostracod species can have better survival chances in freshwater habitats where environmental conditions are more or less stable. In contrast, species with sexual reproduction can do better in habitats where conditions are more changeable. Thereby, ostracods show different species-specific responses to those changes in environmental conditions and factors (Wise 1961;Delorme 1991;K€ ulk€ oyl€ uo glu 2003;K€ ulk€ oyl€ uo glu et al. 2003).
Ostracods having a pair of calcified valves, can walk, crawl, climb, swim and even jump in the water by using their extremities with a combined motion of their first (A1) and second antenna (A2). Detailed knowledge about the taxonomic and evolutionary importance of these characters has already been outlined in the literature (Martens 1987;Maddocks 2000;Meisch 2000;Karanovic 2005;Smith and Kamiya 2008). Almost all freshwater ostracods carry setae on A1 while the presence or absence of setae on A2 is one of the more distinguishing differences frequently used in taxonomic keys. The swimming ability (and eventual escape behavior) of ostracods not only helps them change their location in water but in turn also affect the availability and vulnerability of them to predators (e.g. fish). If presence or absence of a character (i.e., swimming setae) allows non-marine ostracods to survive in different habitats, the character or trait can be considered of "adaptive value". The character(s) can also help offspring succeed in new environments (Ghalambor et al. 2007) when its occurrence can be correlated to habitat type and/or conditions in which species tend to be found more often.
A few previous studies (e.g. McGregor and Kesling 1969;Kamiya 1988) revealed that ostracods' soft body parts as much as carapace shape were important for their reproductive and copulatory behavior in relation to their adaptation to different (micro)habitats. Indeed, swimming setae can greatly help ostracods' movement ability. For example, Higuti et al. (2007) demonstrated that the number of swimming ostracods increased more during wet than dry seasons in the Parana River floodplain (Brazil). Similarly, Szlauer-Łukaszewska (2008) reported that species without swimming setae were more frequently found on the river bottom than in flooded basins nearby in Odra River (Poland). Most recently, Akdemir et al. (2016) found that non-swimmer ostracod species collected from a variety of aquatic bodies of Gaziantep province in Turkey preferred lotic habitats more than lentic habitats (but see Discussion).
Although these studies showed the importance of the extremities for ostracod success in different habitats and/or conditions, they did not focus on possible relationships among ostracods' reproductive modes, species diversity, ecological tolerance and swimming abilities. On the other hand, species habitat selection can be affected directly or indirectly by its reproductive success and survival chances as ultimate reasons (Hilden 1965;Litvaitis et al. 1996).
Thus, the aims of this study are to (1) complete the first extensive study on ostracods in A grı province, (2) reveal regional (gamma) species diversity in the area, (3) estimate and compare ecological tolerance and optimum levels between parthenogenetic and sexually reproducing species, and (4) seek possible association for habitat preferences of the species with or without swimming setae on A2.

Site description
Our study area includes the province (aka city) of A grı (39 05 0 -40 07 0 N, 42 20 0 -44 30 0 E) located at 1640 m a.s.l., covering ca. 11,376 km 2 of surface area in east of Turkey (A grı 2017). Because of high mountain ranges consisting of an Alpine-Himalayan volcanic system from east to west, the area is known for its harsh climatic conditions where winters are long and severe while summers are hot. B€ uy€ uk A grı Da gı (B€ uy€ uk A grı Mountain) is located in the area and is the highest peak in Turkey and Europe with about 5137 m of elevation. The average annual precipitation of the area is about 529 mm where the average air temperature is about 6.1 C.

Sampling and laboratory analyses
We randomly collected 115 samples from 11 different aquatic habitat types (lake, pond, pool, wetland, river, stream, waterfall, spring, ditch, trough, thermal water) in A grı Province during 13-17 May 2014 ( Figure 1). Twelve variables (pH, dissolved oxygen (DO, mg/l), per cent oxygen saturation (% sat.), water and air temperature (T C), electrical conductivity (EC, mS/cm), total dissolved solids (TDS, mg/l), salinity (ppt), atmospheric pressure (mmHg), moisture (%), wind speed (m/s), elevation (m)) were recorded in situ (Table 1). A Testo 410-2 model anemometer was used to record air temperature, wind speed and air moisture while basic geographical data (elevation) were recorded with a geographical positioning system (GARMIN Etrex Vista H GPS). The other variables were measured by YSI Multi Professional Plus.
Each sample collected with a hand net (200 lm in mesh size) from not deeper than 100 cm of depth was fixed in 70% ethanol in plastic containers (250 ml) in situ after water quality measurements were taken. Samples were kept in a cooler during the field work and brought back to Limnology Laboratory of Department of Biology where each sample was washed and filtered through standard sized sieves (0.5, 1.0, 1.5 and 2.0 mm mesh size) under tap water and kept in 70% ethanol. Individual specimens were separated from the bulk material under a stereomicroscope (Olympus ACH 1X) but only adults were mounted in lactophenol solution. Carapaces and valves were kept in micropalentological slides. During taxonomic description done under a light microscope (Olympus BX-51), both soft body parts and carapace structures were used following the taxonomic key of Meisch (2000) and Karanovic (2012). Species dissected with males are assumed to have sexual reproduction while species without males are considered as parthenogenetics. Slides and other materials were deposited in the Limnology Laboratory of Abant _ Izzet Baysal University Bolu, Turkey and are available upon request.

Statistical analyses
Canonical Correspondence Analysis (CCA), a multivariate ordination technique, was used to understand most effective environmental factors on ostracod species. During which, rare species were removed and the data was long-transformed and tested with Monte Carlo Permutation test (499 permutation) (ter Braak 1987;Birks et al. 1990;McGarigal et al. 2000). We also ran Detrended Correspondence Analysis (DCA) (software package CANOCO for Windows 4.5) to confirm the suitability of our data for CCA. We found length of gradient in DCA as 3.2, suggesting that our data was good for CCA.  Atmp ¼ atmospheric pressure (mmHg); DO ¼ dissolved oxygen (mg/l); %DO ¼ per cent oxygen saturation (% sat.); EC ¼ electrical conductivity (mS/cm); Elev ¼ elevation (m); Moist ¼ moisture (%); Ta ¼ air temperature ( C); TDS ¼ total dissolved solids (mg/l); Sal ¼ salinity (ppt); Tw ¼ water temperature ( C); Wind ¼ wind speed (m/s).
C2 software program with a transfer function of weighted averaging regression (Juggins 2003) was used to estimate tolerance and optimum values of 18 ostracod species (found at least 3 or more times in different sites) for five environmental variables. As stated by ter Braak and Barendregt (1986), optimum estimates are proportional to the frequency of species in the habitats sampled. Since species occurrence frequencies can vary depending on geographical, biological and/or environmental conditions of habitats, levels of tolerance and optimum values may show variations among the species. Accordingly, knowledge of species tolerance and optimum estimates can be used to evaluate present and future habitat conditions. We used the Coleman heterogeneity test for species occurrence patterns in 11 different aquatic habitats. This method allows us to compare the levels of heterogeneity of habitats (Coleman 1981). If the expected curve shows a sharp increase or deviation from the origin, heterogeneity of habitats is greater than our expected values by random sampling only (Coleman et al. 1982;Colwell and Coddington 1994).
Also, alpha (Shannon-Wiener Diversity index) and beta species diversity values were calculated in Species Diversity and Richness version 4 program (Seaby and Henderson 2006). The range of S-W index values were considered between 1.5 (low) and 3.5 (maximum) (Magurran 1988).
The analysis of Similarity Percentage (SIMPER) was applied to determine the similarities and percentage contribution of individual ostracod species and habitats. The analysis based on Bray-Curtis measure of similarity makes it easy to point the species with the most important observed patterns of similarities amid others. Besides, Two Way Indicator Species Analysis (TWINSPAN) was used to show clustering relationships of the species among habitats. At each division, the effective habitat(s) where species tend to be found more are shown. The last two analyses were provided in the Community Analyses Package (CAP 4.1.3.) program (Seaby and Henderson 2007). Additionally, t-test with unequal variances (at 0.05 critical levels) in Microsoft Excel 2010 program was used to compare the mean numbers of parthenogenetic and sexual species with or without swimming setae on the second antenna.
The most common three species were found in 8 (Candona neglecta, Ilyocypris bradyi) and 7 (Heterocypris incongruens) of 11 different aquatic types. SIMPER results outlined the fact that these three species alone were responsible for about 70% of the total similarities (Table 3)    and 1.95, respectively) when Whittaker beta diversity was 3.326. Three types of habitats (stream, ditch and pond) consisted of the highest number (24, 15, and 11) of species, respectively. Coleman curve (Figure 2) was slightly found above but closer to the observed ostracod species acquisition curve pointing some samples heterogeneity possibly due to random samplings applied during the study. The ratio of swimmers (20 spp.) and nonswimmers (15 spp.) was found to be different for each group. For example, within the groups of swimmers, 65% and 35% of the species were parthenogenetic (13 spp.) or sexually reproducing (7 spp.), respectively. Within the groups of nonswimmers, numbers of species with sexual reproductive mode was twice the parthenogenetics where 67% and 33% of the species were sexual (10 spp.) or parthenogenetic (5 spp.), respectively. The ratio of reproductive mode (17 sexual and 18 parthenogenetic spp.) was not significantly different among the species. However, dominancy of each group with or without swimming ability was found to be different; for instance, 41% and 59% of the sexual species were nonswimmers or swimmers while the ratio was significantly different for the parthenogenetic species favoring nonswimmers (72%) over swimmers (28%). Non-swimmer species were abundantly found in shallow and changeable habitats such as trough, spring, pool and wetland while swimmer species were most common in deeper and relatively stable habitats such as lake, stream, pond, ditch and thermal waters (Table 2). Based on species occurrences amid different habitat types (Figure 3), TWINSPAN dendrogram separated species in four main clustering groups. Besides, swimmer species exhibited higher optimum and tolerance values for pH, EC, Tw, and DO (for optimum value) than non-swimmer ostracods. Regardless of their swimming ability, in general, parthenogenetic species illustrated higher tolerance and optimum values for these variables herein used than sexual species (Table 5). CCA diagram displayed 73.9% of the correlation between 18 most frequently occurring species and five environmental variables with low variance (8.9%) where water temperature (p ¼ 0.042, F ¼ 1.743) and elevation (p ¼ 0.002, F ¼ 3.721) were the most effective factors on ostracods (Figure 4).

Discussion and conclusion
Hartmann (1964) was the first scientist to report one species (Heterocypris rotundata) from two samples of swamp water at Toprak Kale region of A grı. Most recently, Rausuli et al. (2014) reported 10 species (Cypris pubera, Eucypris virens, E. lilljeborgi, Trajancypris serrata, T. clavata, Heterocypris incongruens, H. rotundata, H. barbara, Potamocypris villosa, P. unicaudata) from five (three temporary water bodies and two lakes) sites in A grı. Three of their species (E. lilljeborgi, H. barbara, P. unicaudata) were new reports for the area. During the present study, we did not find these four species previously reported by these authors. In sum, at the moment, there are 39 ostracod species known from A grı which showed the highest known alpha diversity value (2.14) reported in Turkey. Actually, comparing the number of species and considering those undescribed taxa (not listed here) suggests that species diversity of the area is much higher than what we know at the present. Three species (C. neglecta, I. bradyi, H. incongruens) were highly responsible for the results of finding high values of occurrences. Indeed, these species or so called "cosmoecious species" (K€ ulk€ oyl€ uo glu 2013) not only have broad geographical distribution but also have wider tolerance ranges than many (if not all) other freshwater ostracods in the world. During the present study, they contributed almost about 70% of the total similarities among the sites. These results correspond with those of previous studies in Turkey (Sarı and K€ ulk€ oyl€ uo glu 2010;Akdemir and K€ ulk€ oyl€ uo glu 2014;Yavuzatmaca et al. 2015Yavuzatmaca et al. , 2017K€ ulk€ oyl€ uo glu et al. 2017a, 2017b where these species along with a couple of others increased the similarity values among the habitats studied. Although biotic factors (i.e., competition, predation) can affect diversity and richness, this part is out of the scope of the present study and not discussed here. However, frequent co-occurrence patterns and dominancy of these species can be affected by a variety of abiotic factors such as habitat type, water quality and seasonality as well. For example, when Rossetti et al. (2004) did not find any of these three species mentioned above out of 19 species in 16 wetland sites of Northern Italy, Iglikowska and Namiotko (2012) reported the highest number of species from lake sites and the lowest from peat-bogs in Lapland (Poland) although these species were not dominant. In Germany, Scharf and Viehberg (2014) collected ostracods from the historical town moat of Bremen where I. bradyi and C. neglecta were subdominant but H. incongruens was absent. In contrast, none of the species found during the present study was reported from polygon ponds in the tundra of the Indigirka Lowland, north-east Siberia (Schneider et al. 2016). Notwithstanding the presence of differences among the habitats, previous works (e.g. K€ ulk€ oyl€ uo glu, , K€ ulk€ oyl€ uo glu and Sari 2012, K€ ulk€ oyl€ uo glu, , K€ ulk€ oyl€ uo glu, Yavuzatmaca,   Table 2. Uçak et al. 2014;Yavuzatmaca et al. 2015Yavuzatmaca et al. , 2017 clearly pointed out and suggested that natural habitats (e.g. ponds, streams, springs, and creeks) with relatively high alpha values appear to contain higher ostracod diversity and abundance than artificial habitats. Indeed, during the present study, we found supportive evidence that species tend to be more abundant and diverse in three types of aquatic habitats (streams, ditches and ponds) than others. The number of species with (20 spp.) or without (15 spp.) swimming setae was different. It seems that occurrence of the setae can be related to habitat types and some of those effective factors. For example, we found that shallow or temporary aquatic  habitats with unpredictable conditions tend to have more non-swimmer species (or these kind of species prefer such habitats) than deeper and relatively stable habitats where swimmers are common. Although there were no data on the ratio provided by the author, Hoff (1942) stated that species with or without setae were common in stagnant and flowing waters of Illinois, USA. Also, comparing the species characteristics in the moats of Bremen and Greifswald (Germany), Scharf and Viehberg (2014) stated that two swimming species (Cypria ophtalmica, Cypridopsis vidua) were not found in Greifswald because of relatively faster water flow velocity. These two species prefer stagnant waters. The authors also provided the ratio between nonswimmer and swimmer species in Bremen moats and Greifswald as 9:12 and 5:2, respectively. During the present study, the ratio between nonswimmers to swimmers was 15:20 (or 3:4) similar to Bremen. All these results indicate that species habitat preferences appear to be correlated to their swimming abilities.
The ratio of swimmers and nonswimmers was critically important within each group. For example, within the groups of swimmers, 65% and 35% of the species were parthenogenetic (13 spp.) or sexual (7 spp.). In contrast, numbers of nonswimmer species with sexual reproductive mode was twice that of the parthenogenetics where 67% and 33% of the species were sexual (10 spp.) or parthenogenetic (5 spp.), respectively. Regardless, the swimming ability, numbers of species with different reproductive mode (17 sexual and 18 parthenogenetic spp.) was not significantly different between them. However, within each group, dominancy of species with or without swimming ability was found different; for instance, 41% and 59% of the sexual species were nonswimmers or swimmers, respectively. Contrary to this, the ratio was significantly different for the parthenogenetic species favoring nonswimmers (72%) over swimmers (28%). These results suggest that nonswimmer ostracods with sexual reproduction are better in unpredictable conditions since they are bottoming dependent inhabiting in or around the sediment with relatively low water velocity. This is opposite for swimmers with parthenogenetic reproduction in calm or stagnant waters. Our results are in accordance with previous works. For example, De Deckker (1983) mentioned that parthenogenetic ostracod species were mostly encountered from ephemeral environments while sexual species were found from stagnant habitats such as permanent or temporary lakes. K€ ulk€ oyl€ uo glu et al. (2012a) reporting more nonswimmer parthenogenetic species than swimmer sexual species in Ordu (Turkey), argued that sexual ostracods with swimming setae may increase their chance to find food, mate and energy that they demand in stable conditions. Contrary to our expectation, opposite results were also found by Iglikowska and Namiotko (2012) who reported the largest proportion of asexual ostracods (ca. 81.2%) from temporary waters while relatively stable habitats (e.g. peat-bogs and shallow lakes) displayed lower ratios (ca.11.3%). The authors pointed out that lakes and peat-bogs with generally more stable conditions favor a more sexual reproduction mode than parthenogenetic. What they stated was actually different the results we discussed above (see discussion in De Deckker 2002;K€ ulk€ oyl€ uo glu et al. 2012a). One of the possible explanations for this conflict may be based on seasonal differences in their sampling time from different habitat types in Lapland and Poland. It is well known that most of the non-marine ostracods portray differences in their seasonal occurrences (e.g. Hoff 1943;Ferguson 1944;Meisch 2000). Also, habitat types and aquatic conditions can show regional differences; therefore, such conditions may possibly affect species occurrences. However, these authors did not focus on species with or without swimming setae. Having setae may provide free movement in stagnant conditions by means of reducing competitive interactions and occupying different niches in the same habitat (McLay 1978a(McLay , 1978b but does not significantly help species in flowing or unpredictable waters unless species inhabit littoral zones with no or less flow. Our CCA diagram revealed that elevation and habitat types (in less extent) were the factors closely related to non-swimmer species while other factors (EC, Tw, habitat type and DO) appeared to be correlated to parthenogenetic swimmer species who show higher tolerance and optimum values for these variables. Similar results were outlined in Ordu (Turkey) by K€ ulk€ oyl€ uo glu et al. (2012a). Overall, these results may actually help to explain worldwide geographical distribution of many parthenogenetic species in a variety of habitats since they show wide ranges of tolerances. Indeed, TWINSPAN dendrogram clustered species into four different groups based on their occurrence among habitats but failed to distinguish their reproductive modes. It is probable that such failure may be due to habitat heterogeneity (McGuinnes 2000) as suggested by Coleman curve and environmental conditions (Kolasa et al. 1998) that play critical role in structuring species distribution.
Finally, from the perspective of evolutionary ecology, the fundamental goal is to understand the processes effective on contemporary patterns of diversity and species richness (Davis et al. 2014). In relation to this view, our results simply imply possible relationships among ostracod species diversity, their habitat preferences along with different swimming ability and reproductive modes. However, this does not apply whether such character(s) is/are environmentally induced variation or not. Besides, we cannot generalize this relationship for other organisms because habitat preferences and/or selection is known as a complex multi-scale process in many animals (Jedlikowski and Brambilla 2017), including ostracods. Nevertheless, our results open a new window for possible future studies.
Nurhayat Dalkıran is a hydrobiologist working on phytoplankton, provided discussion and interpretation on the results of the analyses.