Structural and Physical Factors as Predictors of the Species Distribution and Diversity of Dragonflies and Damselflies (Odonata) in an Upland Storage Reservoir

albistylum and Calopteryx virgo were distinguished as the indicator species (IndVal analysis) for the upland river-reservoir hydrological system. Our results highlight some design features of reservoirs that may help to maintain the diversity of odonates, as well as many other groups of aquatic organisms, as the former are well-known indicators of general biodiversity.

hydrographic network, possibly even forming a substantial part of it (e.g. DONG et al. 2019), and therefore constitute an important habitat for wildlife. It is known that reservoirs may degrade rivers and adversely affect their biodiversity (KORNIJÓW 2009;BUCZYÑSKI 2015); but there is data indicating that, if appropriately designed and then properly managed, such reservoirs can also be beneficial (KORNIJÓW 2009;BUCZYÑSKI 2015;WU et al. 2019). This suggests that existing facilities may need to be modified. For this reason, analyses of biodiversity are very important in the case of reservoirs with different designs and methods of utilisation, functioning in different catchment areas and landscape types.
In the present paper, the dragonfly and damselfly assemblages of an upland water storage reservoir in East-Central Europe are analysed, along with stretches of the feeder rivers immediately above the reservoir. We focused on the structural features of the habitat and the immediate surroundings of the reservoir/rivers, as well as on certain physical factors, e.g. water movement, since these can easily be altered during the design of a reservoir or its possible modifications, in order to protect biodiversity. The object of the study -Wióry Storage Reservoir -was chosen because of its considerable habitat diversity. It is a water body that is typical in the uplands of Central and East-Central Europe. Dragonflies and damselflies (Odonata) were chosen as the research objects because they are good indicators of both the environmental quality and biodiversity (BULÁNKOVÁ 1997;GERLACH et al. 2013;KIETZKA 2019). Our aims were to determine: (1) the species richness and composition of the odonate fauna in the whole hydrological system, i.e. the reservoir itself and the stretches of the feeder rivers immediately above it; (2) the most significant habitat drivers of the species distribution; (3) indicator species characteristic of this type of hydrological system; and (4) the function of the storage reservoir in the conservation of the odonate species richness.

Study area
Wióry Storage Reservoir (50°57¢04.2¢¢N, 21°10¢27.2¢¢E) is situated in Southern Poland, 145 km south of Warsaw and 37 km north-east of Kielce. It is located in the centre of the Kielce Upland, at the foot of the highest part of the OEwiêtokrzyskie Mountains, which is an isolated, low-altitude mountain range. This area lies within the Tertiary tectonic uplift, with outcrops of Early Triassic to Late Cretaceous folded structures. Differences in resistance to weathering of the rocks have led to the formation of numerous hills, depres-sions and flat-bottomed river valleys (KONDRACKI 2011;SOLON et al. 2018).
The soil in the study area includes loess, brown soil and chernozems; there are also deposits of quartzite, while the lower-lying ground is overlain with sand and Quaternary, mostly red, clay (MARCINEK et al. 2011). The climate resembles a submontane one, being cooler and more humid when compared to nearby regions (KONDRACKI 2011). The average air temperatures during the period of 1990-2010 were as follows: annual: 6ºC; February: -3ºC; July: 17ºC. In addition, the annual average precipitation was in excess of 600 mm (INSTYTUT METEOROLOGII I GOSPO-DARKI WODNEJ 2020).
Wióry Reservoir is situated in the lower course of the OEwioelina, a right-bank tributary of the Kamienna River, while its southern part is fed by the Pokrzywianka. It was built in 1980-2005 and became operational in 2007 (G£ÓWNY URZ¥D STATYSTYCZNY 2020). The reservoir is almost completely surrounded by highly fertile arable land, with small areas of fallow land and extensively managed meadows. The only woodlands situated along the reservoir's 40 km long shoreline are small ones located near the dam and below the mouth of the Pokrzywianka, covering a distance of around 3 km (GEOPORTAL 2020; GMINA PAW£ÓW 2020).
The research was conducted at seven sites: four by the reservoir itself and three by the feeder rivers ( Fig. 1). Sites 3 and 4 were representative of riverine habitats beyond the maximum extent of backflow from the reservoir. The OEwioelina and Pokrzywianka are both small upland-submontane rivers, with alternating dammed and naturally-flowing stretches and with quite strong currents. Site 5 lay in the backflow area of the Pokrzywianka, where the current was very slow, and the emergent and floating vegetation in the littoral zone were typical of lentic waters. The other study sites were situated by the reservoir itself. Sites 1 and 7 were in small bays with diverse emergent, floating and submerged vegetation. The shallow littoral zone was wide at Site 7 but narrow at Site 1, which meant it was almost totally exposed in summer when the water level was low. Site 2 lay in a stretch of straight shoreline with well-developed vegetation of a similar structure as that in Site 7. Site 6 was also in a section of straight shore, where there was strong water movement and sparse, periodically occurring submerged plants.

Field sampling
Sampling was carried out once a month at each site by 4-5 observers, from May to September 2020. Observations of the adult dragonflies and damselflies were conducted between 10:00 and 16:00 hrs on sunny days, with little if any wind. Species were identified with the naked eye or by using 10x binoculars. In the case of any doubt, an individual was caught with an entomological net for a closer examination. Once the species was identified, it was released back into the environment. The sampling sites encompassed shore sections 100 m long or longer, but the number of individuals were always calculated per 100 m. The species recorded were assigned to one of six categories: 1 -1 individual per 100 m; 2 -2-10; 3 -11-20; 4 -21-50; 5 -51-100; and 6 ->100 individuals per 100 m. Metamorphosis, tenerals, reproductive behaviour and feeding patterns were also recorded.
The following structural and physical habitat parameters were measured or defined at each study site: water movement (current or standing water) (code for the further analyses: WM), percentage of woodland within a 100 m radius of the site (For), cover of submerged and floating plants (SFP), cover of emergent vegetation (RB -reed beds), width of the shallow littoral zone (SLZ), degree of siltation (Silt), distance from the dam to the site (Dist), and the degree of shade (Shad). Each factor was measured or coded on a scale from 0 (absence / undeveloped) to 4 (complete cover / strongly developed). The scores were assigned on the basis of field observations and by the analysis of aerial photographs (GEOPORTAL 2020).

Biocoenotic and statistical analyses
All of the species were classified as 'autochthonous' if tenerals or intensive reproductive behaviour were observed; as 'probably autochthonous' if the reproductive behaviour was occasional (1-3 encounters) or numerous individuals were observed in a suitable habitat; and as 'recorded' in all other cases.
Non-metric multidimensional scaling (NMDS) with a minimum spanning tree was used to discover the faunistic similarities between the sites. Calculations were based on the Bray-Curtis index. The similarity percentage method (SIMPER) was employed to identify the species responsible for assemblage discrimination between the riverine and reservoir sites. In addition, by applying an indicator species analysis IndVal (DUFRENE & LEGENDRE 1997), we were able to determine and graphically visualise the indicator species in both groups of sites. In the IndVal analysis, the statistical significances (p<0.05) were estimated by 9999 random reassignments (permutations) of sites across the groups. The analyses were performed in the PAST 4.09 program (HAMMER et al. 2001).
A partial canonical correspondence analysis (pCCA) was used to display the effects of environmental variables on the species compositions of the dragonflies and damselflies. To eliminate the influence of time, we used a 'month' as a covariate. A forward selection of explanatory variables with 999 test permutations was undertaken in order to filter the significant (p<0.05) parameters primarily responsible for species variation. Significance tests of the first canonical axis and of all the axes were also performed. Since a collinearity was detected when fitting the variables, the variable 'distance from the dam (Dist)' was excluded from the analysis. On this basis, a pCCA biplot was generated showing the preferences of particular species for particular plants and substrata (marked with different colours). The analysis was performed in Canoco 5.0 (TER BRAAK & ŠMI-LAUER 2012).
Basic statistical analyses (tests) were done with the Statistica 13.0 software. GUILFORD's (1973) interpretation of the magnitude of significant correlations was adopted.

Results
A total of 25 species of dragonflies and damselflies were recorded, 22 of which occurred in the basin of the storage reservoir. All but one of these 22 species were autochthonous or probably autochthonous. Only two species were recorded at all the sites -Ischnura elegans and Sympetrum sanguineum. The other species were found mainly by the rivers or the reservoir ( Table 1).
The arrangement of the study sites on the NMDS plot ( Fig. 2) showed that the riverine Sites 3 and 4 exhibited the greatest faunistic similarity (85%) and occupied the opposite end of Coordinate Axis 1 to the reservoir sites. The fauna of the riverine Site 5 displayed the strongest similarity to the fauna of the reservoir sites (especially Site 2 -67%), while Coordinate Axis 1 indicated the gradient associated with the water movement/flow (from lotic to lentic habitats). In turn, Coordinate Axis 2 differentiated the sites in terms of the aquatic vegetation and shoreline type, i.e. small bays with the lushest vegetation (Sites 1 and 7 -negative axis values) vs. sites with a straight shoreline and less abundant vegetation (Sites 2 and 6 -positive axis values).
The evident diversity of the overall species richness and a strong differentiation of their numbers was apparent ( Fig. 3), but these results were statistically insignificant (ANOVA p=0.28 and Kruskal-Wallis test p=0.20, respectively). Nonetheless, the values of both these factors were strongly correlated: Spearman's rank correlation R S =0.82 (p<0.05). Within the storage reservoir and inflow section, Site 7 stood out in both respects, as the other sites were nearly as species-rich in dragonflies and damselflies but differed in their numbers. A comparison of particular sites and their habitat conditions indicated that the key factors were probably the quantitative richness of plants (favouring high densities of odonates) and water movement in the littoral zone (limiting the number of odonates).
The differences between the species richness and the abundance of odonates at the study sites, analysed with respect to sub-orders, turned out to be statistically highly non-significant: number of dragonfly species -Kruskal-Wallis test p=0.45; abundance of dragonflies -Kruskal-Wallis test p=0.52; number of damselfly species -Kruskal-Wallis test p=0.45; and abundance of damselflies -Kruskal-Wallis test p=0.35.
The following ecological groupings were recorded: rheobionts (3 species), rheophiles (2), limnophiles (1) and eurytopes (19). Their presence at the study sites is illustrated in Fig. 4. Rheobionts were present almost everywhere, but their large populations, as well as reproduction activities and development stages were re- Table 1 Dragonflies recorded at the study sites: -autochthonous species, -probably autochthonous species, -recorded species, FR -feeder river, RI -river (inflow section), SR -storage reservoir. Maximum density: 1 -1 ind./100 m, 2 -2-10, 3 -11-20, 4 -21-50, 5 -51-100, 6 ->100  corded only along the rivers. By contrast, rheophiles were numerous and autochthonous in the reservoir, with their success probably owing to water movement in the shallow littoral zone at a given site, so long as it was not too strong or where the movement was moderated by littoral plants. This fact is best illustrated by the small number of rheophiles at Site 6, where the water movement was distinctive: the larvae were probably not able to survive on the mineral bottoms and littoral plants were practically absent. The only limnophile (Anax parthenope) was recorded solely at the sites with well-developed vegetation, with a spatial structure resembling the phytolittoral of a eutrophic lake (with zones consisting of amphiphytes, helophytes, nymphaeids and elodeids). Eurytopes were both species-rich and abundant at the majority of sites. Their species richness was low only in at the river sites.
A similarity percentage analysis (SIMPER) showed that just five odonate species were responsible for the differences between the riverine and reservoir sites, contributing 72% to the total dissimilarity. The overall average dissimilarity of the entire river-reservoir system was 87.4%. However, those species represented different ecological groups: the eurytope Ischnura elegans with the highest percentage (24.4%), followed by the eurytope Sympetrum sanguineum (18.2%), the rheophile Platycnemis pennipes (15.8%), the rheobionts Calopteryx splendens (8.2%) and the thermophilous eurytope Erythromma viridulum (5.6%). It is worth mentioning that, except for Sympetrum sanguineum, all of these species were strongly associated with the presence of plants. An indicator species analysis (IndVal - Fig. 5) -as an alternative to the SIMPER -showed that Ischnura elegans (67%), Orthetrum albistylum (33%), O. cancellatum (31%) and Sympetrum sanguineum (25%) had the highest in- dicator values for the reservoir sites. The key species for the riverine sites were Calopteryx virgo (47%), C. splendens (26%) and Platycnemis pennipes (21%). Ischnura elegans, Orthetrum albistylum and Calopteryx virgo were determined to be the best candidates as indicators for this upland river-reservoir hydrological system; this result was also confirmed by the statistical significances (p<0.005).
A partial CCA for all the sites (river -inflow section -storage reservoir) explained 34.2% of the variability in the dragonfly occurrence. The permutation test results for the first axis and all the axes were significant: pseudo-F=4.02, p=0.001; pseudo-F=1.8, p=0.002, respectively. Three of the predictors were statistically significant: submerged and floating plants (SFP), which explained 13% of the variability, and which in turn made a 38% contribution to the whole of the variation (p=0.001); width of the shallow littoral zone (6.3%, 18.5%, p=0.005); and water movement (4.7%, 13.6%, p=0.05) (Fig. 6). The first axis was defined by SFP (r=-0.83) and the second by SLZ (r=-0.6). Species including Calopteryx splendens, C. virgo, Gomphus vulgatissimus, Chalcolestes viridis and Aeshna cyanea exhibited a positive coordinate on Axis 1 and were closely associated with the fastest current/water movements. The same applied to the number of rheobionts. Erythromma viridulum, Anax imperator, A. parthenope, Orthetrum albistylum and O. cancellatum were strongly associated with the presence of submerged and floating plants. Meanwhile, Enallagma cyathigerum, Coenagrion puella and Somatochlora metallica were associated with the widest shallow littoral zone, whereas Aeshna affinis and A. mixta had intermediate values for this factor, in contrast to the libellulid species (e.g. Sympetrum spp. and pioneering Libellula depressa).
No statistically significant Spearman rank correlations (R S ) were found between the variables and the species richness and abundance of the two odonate sub-orders. The only exceptions were moderate correlations between the cover of emergent vegetation on the one hand, and the anisopteran species richness (R S =0.46) and abundance (R S =0.48) on the other. In the pCCA (Fig. 6), the anisopteran and zygopteran species were similarly arranged around the vectors of statistically significant factors. Only the cover of submerged and floating plants was associated with distinctly more anisopteran species, which tallied with the Spearman rank correlations, as described above.

General remarks on the dragonfly fauna
The species richness of dragonflies and damselflies varies greatly according to the geographic region (BERNARD et al. 2009;BOUDOT & KALKMAN 2015;KALKMAN et al. 2018), which is why the odonates from the Wióry Storage Reservoir are best compared with the data from Poland as a whole, as well as the physiographical macroregion in which the reservoir is situated. A total of 25 species were recorded in the study area (34% of the 74 species hitherto recorded in Poland) (BERNARD et al. 2009;BUCZYÑSKI et al. 2019), with 22 of them in the storage reservoir (30%). This comprised 40% and 35%, respectively, of the 63 species already recorded in the Kielce Upland (BERNARD et al. 2009;GRZÊDZICKA 2010;STAOEKOWIAK 2014;GWARDJAN et al. 2015;STAOEKOWIAK 2015;2017;STAOEKOWIAK & SOWA 2018; GWARDJAN 2020; BUCZYÑSKI & BIE-LAK-BIELECKI 2021). These numbers are significant.
The species richness of dragonflies and damselflies at the Wióry Reservoir site was significant in the context of the data gathered at similar sites elsewhere in Poland. The upland areas of the country form quite a compact latitudinal belt, and since they are of a very similar latitude and altitude (KONDRACKI 2011), they are comparable. From 9 to 37 species have been recorded (av. = 19.5±8.6) in those Polish upland reservoirs for which extensive data is available (BUCZYÑSKI 2001;BUCZYÑSKI & TOÑCZYK 2004;BUCZYÑSKI 2008;CUBER 2008;TOÑCZYK & STANKIEWICZ 2008;¯URAWLEW 2013;BUCZYÑSKI 2015;BUCZYÑSKI et al. 2020). Thus, the number of species recorded at the Wióry Reservoir was well above the average for comparable reservoirs.
Data from some other reservoirs indicates how important the spatial structure of a habitat may be for odonate diversity. For example, the shore of the Zemborzyce Reservoir (near Lublin, SE Poland) is partly concreted and not greatly diversified; only a small part has a shallow bay with a natural shore and shallow littoral, where the vegetation and structure are similar to that of eutrophic lakes. 28 dragonfly and damselfly species were recorded at this particular place (82% of the total), but no more than 12 species (35%) were found elsewhere in that reservoir. Moreover, the density of imagines was 5-11 times greater in the shallow bay (BUCZYÑSKI 2015). The key factors in relation to the total odonate richness in another small water body, located in Pietermaritzburg (RSA), were determined to be the areas of shallow water, and the gain and loss of various odonate species with the advancing succession of vegetation and changes in its spatial composition (SUH & SAMWAYS 2005).

Environmental drivers and species distribution patterns
Based on the faunistic similarity (NMDS) or pCCA, the dragonfly and damselfly distribution patterns at the study sites revealed the significance of the predictors of structural and physical habitat characteristics. The structural predictors were interconnected: a shallow littoral zone was beneficial for the development of a dense cover of submerged and floating vegetation. The crucial function of such plants has been highlighted in the results of many papers (e.g. REMSBURG & TURNER 2009;BUCZYÑSKI 2015;NAGY et al. 2019;PERRON et al. 2021). The pCCA showed not only which elements of the habitat structure are important, but also that structural factors may predominate over those associated with the physical and chemical properties of the water or with the landscape. Similar results, especially regarding the key role of the littoral vegetation cover, were obtained by PETROVIÈOVÁ et al. (2021), who analysed the occurrence of dragonflies and damselflies in other kinds of artificial lentic waters. HALL et al. (2015) found structural and landscape parameters to be the most crucial factors for the presence of adult odonates; moreover, the percentage of species variation explained in their RDA was very similar to the results that we obtained. Such conclusions have also been drawn from studies focused on the occurrence of caddisflies (Trichoptera) in storage reservoirs (e.g. BUCZYÑSKA et al. 2016;BUCZYÑSKA 2019). Our inferences could thus relate to all kinds of aquatic invertebrates, as regards maintaining the high biodiversity of their habitats.
It was expected that the third key driver of odonate distribution would relate to the physical factors, especially water movement, in the hydrological system studied here. Since it has a considerable influence on both biotic components, such as the composition of the vegetation, and abiotic factors like the pH and oxygen content of the water, water movement is regarded as the most important factor in structuring the assemblages of aquatic environments (DEL-CLARO & GUILLERMO 2019). Moreover, while many insect taxa are amphibiotic, it is the water movement that primarily determines the ecological distribution of species, especially in relation to morphological and behavioural adaptations (DEL-CLARO & GUILLERMO 2019). Although rheobionts and rheophiles were relatively few in number compared with eurytopes within the whole river-reservoir system, they played a crucial part in the faunistic diversity and the odonate assemblages. Moreover, this group was significant in terms of distinguishing the indicators of the habitat differentiation (lentic vs. lotic waters). The SIMPER and IndVal analyses showed that dragonflies are not only good indicators of the habitat quality or general habitat richness (SAHLÉN & EKESTUBBE 2001;MARTIN & MAYNOU 2016), but are also well suited to a biotope characterisation (CLARK & SAMWAYS 1996).

The importance of storage reservoirs for odonate conservation
The data presented in this paper, as well as that published in the literature (MITRA 2000;SUH & SAMWAYS 2005;ANDREW et al. 2013;BARBOSA et al. 2019;BUCZYÑSKI et al. 2020), demonstrates that storage reservoirs may be important for the protection of dragonfly and damselfly diversity. The comparison made in the second paragraph of this section shows that in Poland, for example, an average of ca. 26% of the national odonate fauna can be recorded as present in a single upland reservoir. Their significance may be greater still on the scale of larger regions. BUCZYÑSKI (2015), for example, stated that more than 50 species were found in the storage reservoirs in East-Central Poland, i.e. 72% of the 69 species hitherto recorded in this region (BERNARD et al. 2009;BUCZYÑSKI et al. 2019). Such reservoirs may be valuable secondary habitats for stenotopic species, especially those preferring mesotrophic and eutrophic lakes (KOSKEN-NIEMI 1994;PETZOLD 2002;SHARMA & JOSHI 2007;CUBER 2008;FULAN et al. 2010;BUCZYÑSKI 2015). These inferences may be of great significance for biodiversity protection as a whole, as some studies have shown that odonates are useful bioindicators of general biodiversity (SAHLÉN & EKESTUBBE 2001;KIETZKA 2019).
Besides their economic function, storage reservoirs can play a highly beneficial role in nature conservation, especially if specific measures are taken to protect particular species and their assemblages, and to sustain the richest possible odonate fauna. In such cases, the spatial diversity of the reservoirs is crucialespecially the presence of more or less extensive shallow littoral zones and a rich, but not too dense, spatially-diverse vegetation (SUH & SAMWAYS 2005; data in this paper). Hence, new reservoirs should be designed in such a way that the water they contain should also permanently cover the shallower areas. The colonisation of the reservoirs by plants ought to proceed spontaneously and the subsequent succession should be monitored, since the odonate biodiversity and many other species benefit from the differentiation of successional stages, with the maximum biodiversity being achieved in the middle stages (BERNARD et al. 2002;SUH & SAMWAYS 2005;WILDERMUTH 2005;BUCZYÑSKI 2015). This may also make a reservoir attractive for angling and other forms of recreation. In this regard, the existing reservoirs should be appropriately modified. Storage reservoirs are among the most widespread anthropogenic waters (BIEMANS et al. 2011). Since they already exist and new ones are constantly being built, they can be utilised for the protection of odonates and other aquatic organisms, which should enhance the habitat resources and stabilise the populations of many species. Furthermore, with appropriate planning, such measures need not be costly.