Discussion
The spider community showed large regional changes along the Baltic Sea seashore despite comparatively small salinity differences (5 vs. 7‰). Several spider species (Pardosa agrestis , P. agricola, Arctosa leopardus and Alopecosa cuneata ) were almost exclusively located on the higher salinity sites compared to lower salinity sites by the Baltic Sea shore and by inland lakeshores. At the same time, other taxa (Pardosa prativaga, P. amentata and Pirata spp.) had the opposite distribution pattern, and this pattern was seemingly not explained by either prey availability or actual spider diets. In fact, there were no detectable diet differences between spider species or between spiders captured on shores with different salinity levels. Instead, spider diets varied between shores with or without thick beds of stranded wrack, a gradient that did not affect spider community structure. Consequently, and because the species shift only occurred on coastal sites and not on corresponding inland sites, it seems that coastal spider communities are directly affected by the saline conditions.
High salinity has several negative impacts on spiders and other arthropods, by reducing both survival and reproduction (Pétillonet al. 2011; Puzin et al. 2011; Foucreau et al.2012). Even though none of the species found on the Baltic shorelines can be considered true halophilic and are usually not found on more marine seashores (Pétillon et al. 2008), it seems reasonable to assume that species vary in their sensitivity to saline conditions. However, please note that previous studies on wolf spiders tested the responses of individuals at a much higher salinity (>30‰) than in our gradient, and it is unclear to what extent that their conclusions could be extrapolated to our study. Irrespective of the mechanisms, our data in combination with previous studies suggest a gradient in salinity thresholds of the dominant wolf spider species on marine shorelines in northwestern Europe where P. prativagatypically dominates low salinity sites, P. agricola dominates intermediate salinity sites and P. purbeckensis dominates high salinity sites. The species abundance distributions of wolf spider communities are often highly skewed with one dominant species having more than 60% of all individuals and a tail of rare species. Even though low salinity sites are not always dominated by P. prativaga , two-thirds are dominated by this species and then more rarely by P. amentata , P. palustris and some other species (see also Meriste, Helm & Ivask 2016).
Whereas the restriction to low salinity sites can likely be explained by salt sensitivity, the corresponding absence of other species at the same low salinity sites seems more puzzling. First, it is evident that the absence from low salinity sites is not absolute as both Pardosa agrestis and P. agricola are frequently reported also from inland habitats in central Europe and more rarely inland also from northern Europe (GBIF.org). Moreover, studies on P. purbeckensis , perhaps the most halophilic species, suggest that fitness is not reduced on low salinity sites (Pétillon et al. 2011). It is possible that some other habitat characteristics restrict the occurrence in low salinity sites or that distributions are restricted by species interactions. Several wolf spider species are known for intraguild predation of other wolf spider species, at least in the laboratory, and dominance is mainly governed by size differences (Buddle, Walker & Rypstra 2003; Rypstra & Samu 2005; Rickers, Langel & Scheu 2006; Rypstra et al. 2007; Turney & Buddle 2019), but no study this far has evaluated the role of intraguild predation on the spatial distribution of wolf spiders.
Whatever the reason is for the difference in wolf spider community composition, the patterns are not likely explained by different dietary niches among spider species or by differences in prey availability. Both this, and previous studies using either molecular gut content analysis or other methods, indicate large overlaps in the diet of wolf spider species (Mellbrand & Hambäck 2010; Verschut et al. 2019). Diet differences observed in this study instead seem to depend on whether spiders were collected on sites with or without accumulated wrack, but these diet shifts did not coincide with shifts in the wolf spider community. By far the most abundant prey group in the wolf spider guts on sites with either wrack or no wrack were dipterans (typically taxa with smaller individuals) and to some extent homopterans. This general prey composition of wolf spiders is of course well-known from non-molecular studies (e.g., Nyffeler 1999), but the relative importance of small dipterans is perhaps larger in our study habitats. Some differences between molecular and non-molecular studies may occur because the former provide an improved representation of small prey items, which are easily overlooked in non-molecular studies due to more rapid consumption. In either case, wolf spiders are likely quite opportunistic predators where prey choice perhaps depend more on encounter probabilities and catchability of prey in their selected habitat than on prey qualities. This opportunistic behavior is perhaps also reflected in the different number of prey species, where the number is higher in southern sites, as expected, and in sites with no wrack. Similarly, diet consistency was also higher on wrack sites, and both patterns observed for wrack sites may reflect that wrack beds are dominated by a small set of detritivorous species. More surprising was the higher diet consistency of spiders on southern non-wrack sites compared with northern non-wrack sites, despite the lower total prey diversity observed for the spiders in the southern region.
Even though opportunism seems to be a dominant pattern, particularly dark-winged fungus gnats (Sciaridae) are underrepresented in wolf spider guts despite their comparatively high occurrence at these sites, similar to what was found previously (Verschut et al. 2019). The reason for spiders to avoid fungus gnats may be that they represent low quality food (as suggested by Toft & Wise 1999b; Toft & Wise 1999a). Diet differences between sites with or without accumulated wrack otherwise reflect availability, even though we refrained from testing the availability-use relationship due to the bias in SLAM traps. Many small flies often occurring on wrack beds, such as Drosophilidae, Ephydridae, Sepsidae and Sphaeroceridae are underrepresented in Malaise type traps on shore lines because these flies tend not to stick to the ground. In either case, these small detritivorous flies that likely developed in or close to the decomposing wrack made up more than 75% of all prey in spider guts when collected from sites with heavy wrack beds and the diet composition was surprisingly similar for spiders collected on northern and southern wrack beds. More unexpected was perhaps the low frequency of chironomids in the spider gut contents, particularly in the non-wrack sites. In a previous study (Verschut et al. 2019), not far from the sites included in this paper, chironomids dominated the spider gut contents and particularly late in the season. In this study, there were no seasonal differences and spiders on non-wrack sites instead consumed a range of terrestrial prey groups, such as Homoptera and various terrestrial Diptera (Chloropidae, Empididae, Dolichopodidae etc.), and it seems that spiders were less strongly connected to the nearby marine environment than previously assumed. In either case, this variability among studies indicate how dynamic food choice of spiders may be.
To summarize, our study indicates that quite a small difference in salinity caused the species composition of wolf spider communities to change almost completely. The mechanism underlying this community shift is less obvious, both why species disappear in the high salinity and in the low salinity ends, but we can conclude that prey availability or differences in the trophic niche between species is likely not involved.