Predator-Prey Interactions between Shell-Boring Beetle Larvae and Rock-Dwelling Land Snails

Drilus beetle larvae (Coleoptera: Elateridae) are specialized predators of land snails. Here, we describe various aspects of the predator-prey interactions between multiple Drilus species attacking multiple Albinaria (Gastropoda: Clausiliidae) species in Greece. We observe that Drilus species may be facultative or obligate Albinaria-specialists. We map geographically varying predation rates in Crete, where on average 24% of empty shells carry fatal Drilus bore holes. We also provide first-hand observations and video-footage of prey entry and exit strategies of the Drilus larvae, and evaluate the potential mutual evolutionary impacts. We find limited evidence for an effect of shell features and snail behavioral traits on inter- and intra-specifically differing predation rates. We also find that Drilus predators adjust their predation behavior based on specific shell traits of the prey. In conclusion, we suggest that, with these baseline data, this interesting predator-prey system will be available for further, detailed more evolutionary ecology studies.


Introduction
Gastropods are among the most diverse groups of animals, and the most readily observable aspect of this diversity is in their shell form and ornamentation.Gastropod shell shape is generally considered to evolve under a strong direct influence of biotic and abiotic agents, and, at least in marine gastropods, predator-prey interactions are paramount among these [1][2].On land, however, the influence of predators on snail shell evolution is less clear [3].
One particularly diverse group of land snails is the clausiliid genus Albinaria, of which the .100species occur abundantly on limestone rocks throughout Greece and surrounding regions (fig.1A) [4], with most species occupying small, usually nonoverlapping ranges [5][6][7].The animals actively forage on microflora on the rocks during the wet months (roughly October to April), but estivate, usually with their apertures firmly sealed to the rock, and often in dense clusters, during the dry part of the year.Species differentiation is most apparent in various shell traits, such as radial ribbing, apertural folds and lamellae, and the structure of the clausilium (a door-like aperture closing apparatus and a Clausiliidae synapomorphy; figs.1B, C).
Although Albinaria snails are preyed upon by a wide range of predators, including rodents, molluscivorous snails, and carabid beetles [8], their chief enemies appear to be larvae of the elaterid beetle genus Drilus (fig.1D-F).These enter Albinaria snails by boring a hole through the shell wall (fig.1G), killing and eating the snail's soft body, and, after molting or pupating, leaving via a second bore-hole (fig.1H) [9][10]; but see below.Often, the entry and exit holes can be distinguished by their shape [8].In many locations, more than 50% of the empty shells carry such Drilus bore-holes [11].
Given their intimate relationship with Albinaria and their shells, we expect that Drilus might be an important selective agent in Albinaria shell evolution.Indications of non-random predation by Drilus already exist, since Mesher and Welter-Schultes [11] found that on the island of Dı ´a (off Crete), Drilus attack was high in the three native Albinaria species, but low in a fourth, possibly introduced, species.To assess such possibilities further, we have investigated in more detail the interaction between Albinaria and Drilus in various regions in Greece.

Biogeography and Diversity of Drilus Predators
We first investigated the geographic variation in predation rate, by mapping the proportions, relative to the total number of empty shells, of shells with Drilus bore-holes, which have a characteristic size and shape, and are easily distinguished from other types of shell damage-see fig.S1 and [8]-throughout Crete.We did this by measuring the proportion of shells with Drilus holes in each of 1,160 museum samples.(With ''sample'' we mean a number of empty shells from a single location, usually taken in a 50650 m area; see Data files S1 and S3.)These show a pattern of high, but regionally varying predation rates, with an average of 0.23960.168(fig.2).In the Peloponnese and Kephalonia, where our geographic coverage was much less complete, we found predation rates that appeared lower than in Crete, but still substantial: average = 0.08260.079(n = 18 samples, 7,450 shells in total; in view of the mixed character of the samples, we refrained from testing for significance in the Crete vs. Pelopon-nese+Kephallonia difference).Drilus predation therefore accounts for a large proportion of total adult mortality [12].
We then used a combination of larval and adult characters, and mtDNA sequencing, to determine the number of Drilus species in the region, as well as their prey specificity (see Data file S2).We found that at least nine species can be distinguished (for the taxonomic and phylogenetic details we refer to Ref. [13], and Kundrata et al. unpublished manuscript; we provide a preliminary map of the species' distributions in fig.S2).For the four species that occur in Crete, we did not assess prey specificity, but for the remaining five, occurring in the Peloponnese, Zakynthos, and Kephalonia, we were able to do this by investigating bore holes and exuviae in shells of the entire snail fauna in 39 locations.The results (Fig. 3A) show that Drilus includes obligate Albinariaspecialists (e.g., Drilus ''L'') as well as generalists that only rarely attack Albinaria (e.g., Drilus ''E'').Although these results may be somewhat confounded by differences in the snail species abundance distributions, the data from two locations, Paralı ´on A ´stros and Koutro ´ufa, where two Drilus species (D. ''D'' and D. ''E'') occur syntopically, confirm the differences in prey specificity between these two species (Fig. 3B).As in other members of the genus, our Drilus species tend to have small geographic ranges (e.g., Drilus ''D,'' ''L,'' and ''M'' may have range lengths of 50 km or less), which may be explained by the fact that Drilus females are wingless (fig.1E) [14].

Behavior of Drilus preying on Albinaria
Unexpectedly, we found that Drilus may successfully attack Albinaria without leaving any bore-holes in the shell.We opened 645 dead Albinaria shells from the Peloponnese, and out of 169 shells that had been preyed by Drilus (judged by the presence of exuviae or a live larva), 60 contained exuviae but showed no trace of a bore-hole.This means not only that predation rates calculated from bore-hole frequencies are underestimates, but also that Drilus employs more than one attack strategy.
To understand better the predatory behavior of Drilus, we obtained three live pseudopupae (an immobile resting stage) from field-collected Albinaria shells, as well as live Albinaria, and used these for observations in the laboratory.Of these three, one remained in pseudopupal state.The other two (both Drilus ''L'') molted into active larvae.One (obtained from an A. edmundi) entered an Albinaria edmundi via the aperture and killed it.It remained inside for 28 days, bored a hole in the shell wall to exit, but then died while emerging.However, the third one (obtained from an A. discolor) stayed alive and active for almost two years, and during that time consumed eight adult prey individuals.In all eight predation events (which took place under dark conditions in a box with several loose Albinaria individuals, i.e., not adhering to any substrate), the larva entered the snail via the aperture (after having inspected several potential prey), never boring a hole in the shell wall nor into the clausilium (Video S2).After entering a snail, it apparently attacked and ate (part of) the snail immediately, because it moved fragments of dried, undigested snail tissue outside of the aperture within three days.In total, it would remain in a shell for 22-32 days, except for one very lengthy stay inside a prey shell that lasted from September 19 th until May 1 st , and possibly indicated hibernation.Each time the larva exited from an empty prey shell, it left behind an exuvia, meaning that the number of larval stages may be much larger than the three to four that had been suspected previously [15][16].In all cases except one, the larva did bore a hole from within the prey shell to exit.Boring was done with the jaws and a copious amount of (possibly acidic) saliva and, based on the one occasion when it was observed from start to finish (Video S1; fig.4) took seven hours.
These observations confirm that a Drilus may enter an Albinaria prey either by apertural-entry or by shell-boring (see also Video S2 for these behaviors in Cretan species).Similarly, it may also exit the (empty) prey shell either via the aperture or via a new borehole.As in other snail-predator interactions [16][17][18], each of these entry and exit strategies have their advantages and disadvantages.For example, apertural entry as well as exit may be hindered by the clausilium and/or by apertural folds that project into the lumen of the whorl.On the other hand, it may be less costly than boring in terms of time, energy, and saliva (if the saliva indeed helps dissolve the shell).Entering and perhaps also exiting the shell by boring may be hindered by the presence of dense and/or tall radial ribs on the shell surface, but is more costly.Finally, it should be noted that shell features that impede predator entry will primarily impose a selection pressure on the prey (by ensuring its survival), whereas those that impede predator exit will primarily select for predator traits (by causing the predator to die within the prey shell).

Potential Evolutionary Impact of Predator-Prey Interaction on the Prey
We may predict that more heavily-ribbed Albinaria species [19] are better protected against boring (and therefore show a lower proportion of bored shells and/or more bore-hole failures) and that Albinaria species with a more completely obstructed aperture [20] are better protected against apertural-entry (and therefore show more bored shells).As a preliminary test of these predictions, we scored numbers of bored and intact shells in pairs of syntopic Albinaria species, exposed to the same Drilus species, but differing in either radial rib height or apertural obstructions.Our results (Table S1) do not provide consistent support for the predictions.At Argı ´nia, the strongly ribbed A. adrianae has, as predicted, a lower proportion of bored shells than the smooth A. contaminata, but at Po ´ros, no difference was found.Bore-hole failures do not differ among the four populations.At one of the three sites where a species with a partly-obstructed aperture (an N-type clausilium) cooccurs with a species with a completely obstructed (G-type) aperture, the latter has fewer bored shells, while at the other two locations, no difference was found.Obviously, these results are inconclusive since they are based on small numbers of populations and should be repeated with a larger-scale study, including more species and more locations.
We also obtained some data that suggest that predation by Drilus may be non-random within a single Albinaria population and hence may cause natural selection on variable traits.In a population of A. krueperi, preyed upon by Drilus ''G'', we found that predation (derived from exuviae in the shells) was predominantly towards the smaller individuals; however, such a response was not found in a population of A. adriani, preyed upon by Drilus ''D'' and ''E'' (Table S2).In addition, in a population of A. caerulea from the Cyclades, we found that predation risk (derived from bore-hole data) was greater in snails estivating in isolation than in snails estivating in clusters (fig.3C), which may be one way in which such clustered estivation (a conspicuous behavioral pattern in most Albinaria species) evolved.

Potential Evolutionary Impact of Predator-Prey Interaction on the Predator
In contrast to the evolutionary impact that the predator may have on the prey, the reverse may also be the case.Albinaria snails are, at least during the dry months of the year, often firmly sealed against the substrate, whereas most of the other species that we recorded as Drilus prey tend to estivate in the soil.We may therefore expect that Drilus species that are more specialized in feeding on Albinaria, have a greater tendency to bore holes for entering, and, probably more importantly, for exiting the shell (an inability to do the latter would mean death for the Drilus larva).Indeed, we find a positive correlation between Albinaria-specificity and hole-boring tendency among the five Drilus species from the Peloponnese and Kephalonia (fig.3A).
Finally, another indication that the prey shell morphology affects the predator was obtained when we compared the positions of entrance holes in shells of species with an N-type clausilium (A.argynnis, A. discolor, and A. solicola; all from the Eastern Peloponnese) with those in shells of species with a G-type clausilium (A.adriani, A. edmundi, and A. campylauchen, all roughly sympatric with the previous three).We found that in the N-type group, more shells had a bore-hole in the outer shell wall between the aperture and the clausilium than in the G-type group (fig.3D).Since the N-type species have a clausilium that does not completely close off the aperture, whereas the G-type clausilium does, these data may indicate that predator populations faced with G-type prey have evolved a shell entry strategy by which the hole is bored at a Maps were drawn using inverse kriging distance calculation in R v.2.15.2 [23], with packages gstat [24], maptools (R v.0.8-27), rgdal (R v.0.8-12) and rgeos (R v.0.3-2).doi:10.1371/journal.pone.0100366.g002position beyond the obstruction.However, the same difference is seen in Monemvasia, where the N-type A. discolor lives syntopically with the G-type A. campylauchen.This suggests that, rather than an evolved behavioral difference between predator populations, the predator may also be able to detect the clausilium type before boring a hole and that the difference in strategy is due to a behavioral response, rather than an evolved, fixed behavior (fig.3D).

Conclusion
Given the complexity of this predator-prey system, with multiple predator species, multiple prey species, varying prey specificity, and complex small-scale biogeographic patterns, we will not hazard a conclusion on the impact of these interactions on shell evolution in Albinaria.However, we think we have shown that detailed study of the natural history of these ecological interactions reveal a potential for such evolutionary responses.Elements from our study may provide starting points for further work, specifically targeted at understanding the possible predator-prey arms races or patterns of escalation in this system, an approach that has proved successful in a similar terrestrial system consisting of tropical micro-snails and shell-boring slugs [21][22].

Additional Methods Details
Sampling of live specimens was done at the locations with the following coordinates: 37u09.6949N, 22u48.7459E and 37u20.1329N, 22u45.7699E. No specific permissions were required for the accomplished field sampling activities by the University of Patras.The field studies were conducted exclusively in public and not protected land areas.The field studies did not involve any endangered or protected species.Sampling was Figure 3. A, five Drilus species from the Peloponnese have different specificities for Albinaria as prey, and concomitant propensities to bore holes in the shell (calculated as the number of bore holes divided by the total number of prey).B, difference (P,0.05,Fisher's exact test) in prey specificity between Drilus ''D'' (more Albinaria-specific) and ''E'' (less Albinaria specific) in two localities where both species occur syntopically (these data are not included in fig.3A).C, differences in Drilus predation rate (dark portion of the bar) between solitarily (''sol'') and groupwise (''agg'') estivating snails of A. caerulea in four 5 m 2 plots in the islands of Paros, Naxos, and Thira (voucher numbers RMNH.MOL.84354-84363,RMNH.MOL.85192, and RMNH.MOL.85193).An aggregate was defined as a cluster of .20 snails, with distances of ,2 cm separating them.A snail was considered solitary if it was .20 cm distance from a conspecific (significance tested with Chi-square test).D, positions of entrance holes in shells of species with an N-type clausilium compared with those in shells of species with a G-type clausilium, shown for the entire eastern Peloponnese as well as for the site Monemvasia, where both clausilium types occur microsympatrically.

Figure 1 .
Figure 1.Greek Albinaria snails and their Drilus predators.A, Albinaria hippolyti from Crete (photo: V. Wiese).B and C, the clausilium, shown in the shell aperture after removal of the left lateral shell wall (B shows a less-obstructing, N-type clausilium, C shows a more obstructing, G-type clausilium).D and E, a male and a female, respectively, of a yet undescribed Drilus species from Crete (scale: 2 mm).F, a full-grown larva of Drilus ''L'' from the Peloponnese (same scale as D and E).G, an estivating A. discolor from the Peloponnese, with a Drilus exit bore hole.H, a Drilus ''L'' exiting from its prey, an A. menelaus from the Peloponnese.doi:10.1371/journal.pone.0100366.g001

Figure 2 .
Figure 2. Map of Crete and surrounding islands, showing contours of regionally varying Drilus predation rates (given as percentages attacked shells per sample) in Albinaria, derived from bore-hole frequencies in 1,160 museum samples from Naturalis Biodiversity Center, Leiden, The Netherlands, Haus der Natur, Cismar, Germany, and Natural History Museum, Budapest, Hungary.Maps were drawn using inverse kriging distance calculation in R v.2.15.2[23], with packages gstat[24], maptools (R v.0.8-27), rgdal (R v.0.8-12) and rgeos (R v.0.3-2).doi:10.1371/journal.pone.0100366.g002 Figure3.A, five Drilus species from the Peloponnese have different specificities for Albinaria as prey, and concomitant propensities to bore holes in the shell (calculated as the number of bore holes divided by the total number of prey).B, difference (P,0.05,Fisher's exact test) in prey specificity between Drilus ''D'' (more Albinaria-specific) and ''E'' (less Albinaria specific) in two localities where both species occur syntopically (these data are not included in fig.3A).C, differences in Drilus predation rate (dark portion of the bar) between solitarily (''sol'') and groupwise (''agg'') estivating snails of A. caerulea in four 5 m 2 plots in the islands of Paros, Naxos, and Thira (voucher numbers RMNH.MOL.84354-84363,RMNH.MOL.85192, and RMNH.MOL.85193).An aggregate was defined as a cluster of .20 snails, with distances of ,2 cm separating them.A snail was considered solitary if it was .20 cm distance from a conspecific (significance tested with Chi-square test).D, positions of entrance holes in shells of species with an N-type clausilium compared with those in shells of species with a G-type clausilium, shown for the entire eastern Peloponnese as well as for the site Monemvasia, where both clausilium types occur microsympatrically.(P-values are derived from Fisher's exact test.)doi:10.1371/journal.pone.0100366.g003

Figure 4 .
Figure 4. Still from Video S1, showing a Drilus ''L'' larva (in the lab) creating an exit bore-hole from within a prey Albinaria meleaus, followed by the lengthy procedure by which the larva emerges from the shell.doi:10.1371/journal.pone.0100366.g004