INTRODUCTION
The diet of many predators can change during their life history, and the mechanism of such a shift is often linked to the relative size of predator and prey (Sullivan, Reference Sullivan1988; Winemiller, Reference Winemiller1989; Hughes et al., Reference Hughes, Burrows and Rogers1992; Giaretta et al., Reference Giaretta, Araujo, Medeiros and Facure1998). In most cases, the larger predators have a broader diet than smaller individuals of the same species (Shine et al., Reference Shine, Harlow and Keogh1998; Olabarria et al., Reference Olabarria, Incera, Garrido, Rodil and Rossi2009). Predators can switch to a different kind of major prey (Hughes et al., Reference Hughes, Burrows and Rogers1992; Clements & Rawlings, Reference Clements and Rawlings2014) or narrow their diet during ontogeny (Aristov et al., Reference Aristov, Poloskin and Zhernakova2003; Moura et al., Reference Moura, Figueiredo, Farias, Serra-Pereira, Neves, De Fatima Borges and Gordo2008; Da Rosa et al., Reference Da Rosa, Canavero, Maneyro and Camargo2011). We expect that the change in the population structure of predators (e.g. due to successful recruitment) can lead to the diet shift. For instance, juveniles may broaden their diet by including smaller suboptimal species when the availability of the major prey of appropriate size is low.
Moonsnails are size-selective predators (Edwards & Huebner, Reference Edwards and Huebner1977; Kingsley-Smith et al., Reference Kingsley-Smith, Richardson and Seed2003; Quijon et al., Reference Quijon, Grassle and Rosario2007), widely known for their ability to drill the shells of their prey (mainly bivalve molluscs) (Carriker, Reference Carriker1981; Huelsken, Reference Huelsken2011 and others). The drilled shells can provide information about the rations and occurrence of certain predator species (Vignali & Galleni, Reference Vignali and Galleni1986; Kabat, Reference Kabat1990). There is evidence of voracious behaviour of moonsnails in the literature and they are believed to regulate the abundance of their prey (Wiltse, Reference Wiltse1980).
The Iceland moonsnail Amauropsis islandica is a naticid that can dwell on the upper subtidal and lower intertidal flats of the White Sea. The feeding mode and the long-term changes in the intertidal populations of this predator had been monitored since 2001 at the Kandalaksha Natural State Reserve (Aristov, Reference Aristov and Koryakin2013). The programme revealed that the adult Amauropsis moonsnails feed on common intertidal bivalves, such as Macoma balthica (L., 1758), Mya arenaria (L., 1758) and Mytilus edulis L., 1758. Occasionally, Littorina littorea (L., 1758) gastropods and other littorinids supplement the diet. The preferred prey of adult Amauropsis is Macoma balthica (Aristov & Granovitch, Reference Aristov and Granovitch2011).
In the last 4 years, the number of young Amauropsis snails in our surveys has increased (Aristov, Reference Aristov and Koryakin2013), and the number of small Macoma has dropped below the level that could support the juvenile predators. This made us assume that the small moonsnails consume other prey besides clams. The sizes of the other known potential victims generally lay outside the size range accessible to the juvenile predators, and the identity of the prey that can supplement the diet of juvenile Amauropsis was yet to be revealed. Among the other molluscs inhabiting the White Sea sandflats, only the mudsnail Hydrobia ulvae (Pennant, 1777) is small enough for juvenile moonsnails. In addition, it was very abundant at the site throughout the monitoring period (Aristov, Reference Aristov and Koryakin2013). We conducted a survey of the sandflat molluscan assemblages and caging experiments to assess the role of H. ulvae in the diet of A. islandica juveniles and to compare it with M. balthica as a prey.
MATERIALS AND METHODS
Study area
Both the survey of molluscan assemblages and caging experiment were held in July – early August 2014 at the sandflats of the Youzhnaya inlet (67°00′ N 32°34′ E, Ryashkov island, Severny archipelago, Kandalaksha State Nature Reserve, Figure 1).
The study area. The dot marks the Youzhnaya inlet at the Ryashkov island (Kandalaksha Bay, the White Sea), where the field survey and caging experiment were carried out.
Youzhnaya inlet is the sheltered shallow bight at the south of Ryashkov island, where the tidal amplitude is about 1.8 m. A stream flows into the head of the bight and fresh water runs over the tidal flat during ebb. The tidal sediments consist of fine and coarse sand with a trace of clay. There is no distinct intertidal-subtidal boundary marked by a ridge of boulders; large boulders (up to 1.7 m high) are scattered across the low intertidal zone of the inlet. A dense population of the lugworm Arenicola marina (L., 1758) dwells within the lower zone, together with common intertidal species such as Macoma balthica, Hydrobia ulvae, Microspio theeli (Söderström, 1920) (Aristov, Reference Aristov and Koryakin2013).
Survey of natural molluscan assemblages
In July 2014, we collected Hydrobia ulvae from 20 cores sampled with a 4 × 10−3 m2 tube corer and sieved on a 0.5 mm mesh, Macoma balthica from 14 cores with an area of 3 × 10−2 m2 and Amauropsis islandica from 14 cores with an area of 2.5 × 10−1 m2 that were sieved on a 1 mm mesh. To ensure compatibility of the abundance estimates from different sieving methods, all Hydrobia snails smaller than 1 mm in width were excluded from the subsequent analyses. We counted live A. islandica, H. ulvae and M. balthica specimens and empty shells of the two latter species – intact and perforated. Only unarticulated bivalve shells were collected. We measured the width of Amauropsis and Hydrobia, and the length of Macoma shells with a microscope micrometer. The live molluscs (except Amauropsis) were wet weighted using spring balances with 0.001 g accuracy.
Caging experiments
To assess A. islandica feeding rates on both types of prey (H. ulvae and M. balthica) and possible species-selectivity in feeding, we set up a caging experiment. We encaged five Amauropsis snails with different prey: Hydrobia (HP, 4 replicates times 50 ind.), Macoma (MP, 4 replicates times 50 ind.), and Macoma together with Hydrobia (HMP, 8 replicates, 25 ind. of each species). As a control, we separately exposed Hydrobia (HC, 3 replicates times 50 ind.), and Macoma (MC, 3 replicates times 50 ind.). The cages were exposed for 16 days (19 – 3 August 2014). The number of molluscs in the cages was chosen so that the total density of prey in the cages was equal. The sizes of prey were selected so that the total weight of soft tissues, estimated using a previously established relationship (Poloskin, Reference Poloskin1995), was similar between the species within a cage. Before the experiment, on 16–19 July 2014, we collected 80 ind. of Amauropsis islandica 5–6 mm in height, 550 ind. of Hydrobia ulvae 1.7–2.5 mm in width and 550 ind. of Macoma balthica 3–5.5 mm in length.
The 22 cages, made of plastic boxes (10 × 10 × 10 cm, 1 L) and screened with 1 mm capron mesh, were filled with a 5 cm layer of sand. Predators and prey were added, and the cages were placed in one block at the intertidal sandflat in the Youzhnaya inlet near the chart datum. The cages belonging to different treatments were arranged checkerwise within the block.
After the exposure, the content of the cages was sieved through the 1-mm mesh. The molluscs and their shells – intact and perforated – were counted and measured using a microscope micrometer or vernier calipers. During the experiment, we found that small H. ulvae individuals crept into all cages from outside. Consequently, we used only the data on Hydrobia larger than 1.5 mm in width.
Statistical methods
We used parametric descriptive statistics: the estimates of all means are reported with standard errors. The pairwise comparisons between the means were made with Welch's t-test which allows for unequal within-group variances. We compared the proportions of drilled shells between species using χ2-test (Agresti & Franklin, Reference Agresti and Franklin2013).
The biomass of the two types of prey consumed in each cage of the experiment was reconstructed using previously fitted equations. For Hydrobia we used the equation W = 2.90 L1.53, where W is wet weight of Hydrobia in mg, and L is the width of the last whorl in mm. The regression was obtained for 83 H. ulvae from the same location. Despite its poor support (R 2 = 0.46), the average differences between the observed and predicted biomass of the live Hydrobia did not exceed 7 ± 2%. We deduced the biomass of Macoma using the equation W = 0.16 L2.96, where W is the wet weight in mg, and L is the shell length in mm (Maximovich et al., Reference Maximovich, Gerassimova and Kunina1993). The average differences between the observed and predicted biomass of the live Macoma were 8 ± 3%.
We estimated the total per capita prey consumption as the biomass of each type of prey consumed in a cage per number of predators, which survived until the end of exposure. We compared the per capita prey consumption of each prey species between single-prey and mixed-prey treatments (HP vs. HMP treatments for H. ulvae and MP vs. HMP treatments for M. balthica) using the analysis of covariance (Quinn & Keough, Reference Quinn and Keough2002). In each of the two models, we controlled for the predator size using an average shell length of Amauropsis in a cage as a covariate. The residuals of both models were checked for normality using quantile plots, and for the variance heterogeneity using Breusch–Pagan test (Breusch & Pagan, Reference Breusch and Pagan1979). The assumptions were met for the model of H. ulvae consumption. The variance heterogeneity for the model of M. balthica consumption was successfully corrected by log10-transformation of the response variable. As both models contained only two treatments, the differences between the levels were evaluated using the coefficients of the fitted models that were t-tested for significance. The differences between the treatments in the model of the per capita consumption of M. balthica, where the predictor was log10-transformed, were transformed back to the original scale.
All calculations were performed using R 3.1.0 (R Core Team, 2014) in RStudio (RStudio, 2014). We used car and lmtest packages for diagnostics of regression assumptions (Zeileis & Hothorn, Reference Zeileis and Hothorn2002; Fox & Weisberg, Reference Fox and Weisberg2011) and ggplot2 to produce the figures (Wickham, Reference Wickham2009; Chang, Reference Chang2012).
RESULTS
Natural molluscan assemblages
The total density of Amauropsis islandica at the study site around the caging area was 21.1 ± 5.6 ind. m−2. About 14% (2.9 ± 1.5 ind. m−2) were juvenile snails with shell height less than 6 mm. The average density and biomass of Macoma balthica was lower than in the case of Hydrobia ulvae (Figure 2). Despite this, Amauropsis snails fed mostly on Macoma, as indicated by the higher perforated/intact ratio than for Hydrobia ulvae (4.69 vs. 1.26, χ2-test, P < 0.01). Still, Hydrobia supplemented the diet of the predator.
Abundance of the two prey species Hydrobya ulvae and Macoma balthica and their empty shells in the natural assemblage: (A) average biomass of live mollusks; (B, C) average density of live molluscs and their empty shells (intact and perforated) for Hydrobia and Macoma respectively. Error bars are ±1 SE.
The size ranges of the perforated shells of the two prey species mostly overlapped (Figure 3). The minimal height of the drilled Hydrobia shells (range 1–2.4 mm) was smaller than the length of the smallest drilled Macoma shells (range 1.5–12.8 mm). The prey sizes later selected for experiments were potentially accessible for moonsnails, as indicated by the size structure of the drilled shells.
The size distribution of drilled shells of Macoma balthica and Hydrobia ulvae in the natural assemblage.
Caging experiments
In the caging experiment, perforated prey shells were found only in the cages with Amauropsis. The distribution of drilled and undrilled shells revealed that the moonsnails were able to consume M. balthica and H. ulvae of the sizes offered (Figure 4). The range of the drilled shells did not change between treatments, indicating that we effectively controlled for the predator size selectivity.
The size distribution of perforated shells of Macoma balthica and Hydrobia ulvae in the caging experiment. (A) Hydrobia as a single prey in HP treatment; (B) Macoma as a single prey in MP treatment; (C, D) Hydrobia and Macoma together in HMP treatment.
Almost all mortality of H. ulvae and M. balthica in the cages with Amauropsis was caused by predation. A total of 8 mud snails (2%) and 2 clams (<1%) died due to other reasons. Such a low level of natural mortality during the experiment indicated good caging conditions.
The feeding rates of Amauropsis islandica juveniles in the cages with Hydrobia ulvae only (1.4 ± 0.2 mg day−1 ind.−1) were significantly lower than in the cages with Macoma balthica (6.4 ± 0.5 mg day−1 ind.−1, t-test, t 3.92 = 10.36, P < 0.01; Figure 5).
Average feeding rates of Amauropsis islandica in the caging experiment in the treatments where Hydrobia ulvae (HP) or Macoma balthica (MP) were offered alone, or together (HMP). Error bars are ±1 SE.
We compared the total per capita consumption of H. ulvae and M. balthica between the treatments where they were offered separately and together. The average shell length of Amauropsis in the cage that was used as a covariate affected the prey consumption in neither of the two models. Since initially all the moonsnails were of the same size, the absence of the predator size effect means that either the predator growth did not depend on prey consumption, or the 16-day exposure was too short to register the effect on shell growth.
The consumption of Hydrobia significantly differed between the treatments where Hydrobia was alone and where it was mixed with Macoma (F 1, 9 = 16.13, P < 0.01). As predicted by the model, each predator in the HP treatment consumed during the exposure period on average 15.0 ± 3.7 mg of Hydrobia more than in the HMP (t-test of the linear model coefficient, t 9 = 4.02, P < 0.01). The proportion of Hydrobia consumed did not exceed 30% in both treatments. Amauropsis juveniles consumed 0–29% of Hydrobia in the mixed treatment (12.3 ± 3.5%) and 12–25% when Hydrobia was offered alone (16.5 ± 2.9%).
The log10-transformed total per capita consumption of Macoma also significantly differed between treatments where Macoma was alone and where it was accompanied by Hydrobia (F 1, 9 = 16.21, P < 0.01). The predicted consumption of Macoma in the treatment where it was alone was 1.6 ± 0.2 times higher (back-transformed scale) than when it was mixed with Hydrobia as an alternative prey (t-test of the linear model coefficient, t 9 = 4.03, P < 0.01) (Figure 5). In some cages of the mixed treatment the predators consumed almost all available Macoma, presumably because the total starting amount of Macoma in the mixed treatment was too low. By the end of the experiment, from 50 to 100% (81 ± 7%) of Macoma had been eaten. At the same time, in the treatment where Macoma was offered as a single prey, only 56–71% (65 ± 3%) of clams had been drilled by the end of the exposure period.
DISCUSSION
Our results confirm that juvenile naticids can feed on prey uncommon for adults. Under natural conditions Amauropsis islandica moonsnails were able to prey upon Hydrobia ulvae mudsnails, although they mostly consumed their usual prey – Macoma balthica clams. The size distribution of shells suggested that Hydrobia were consumed by smaller moonsnails, and the caging experiment confirmed that the juvenile moonsnails were able to feed on the mudsnails.
The juveniles of predatory gastropods can sometimes consume different items compared with adults. For example, young muricids can attack spirorbid polychaetes (Palmer, Reference Palmer1988), some naticids can prey upon gastropods (Berry, Reference Berry1982), while the juveniles of other moonsnails are able to perforate the tests of Foraminifera (Saidova & Beklemishev, Reference Saidova and Beklemishev1953). During ontogeny the diet of predators either narrows (Moura et al., Reference Moura, Figueiredo, Farias, Serra-Pereira, Neves, De Fatima Borges and Gordo2008) or, more often, broadens (for review see Wilson, Reference Wilson1975). In any case, smaller prey items can be displaced by larger ones as the size-selective predators grow and new types of prey become accessible (Da Rosa et al., Reference Da Rosa, Canavero, Maneyro and Camargo2011). The possible reason for such displacement is an unprofitable gain/spent energy ratio. For example, the diet of the growing muricids typically changes from the less energetically favourable barnacles to bivalves (Hughes et al., Reference Hughes, Burrows and Rogers1992). Similar size-related diet shifts happen in other consumers (for review see Werner & Gilliam, Reference Werner and Gilliam1984).
Amauropsis islandica is a size-selective predator like other naticids (Edwards & Huebner, Reference Edwards and Huebner1977). When Macoma is scarce, Amauropsis can benefit from the ability to feed on abundant, but apparently less desirable, Hydrobia snails. Prey switching during ontogeny has been reported recently (Clements & Rawlings, Reference Clements and Rawlings2014) for another naticid species, Lunatia heros (Say, 1822). The ability to diversify food sources is also important for other size-selective predators, especially when primary prey supply is unstable.
The conclusion about the non-random diet of Amauropsis juveniles was based on the proportion of drilled to intact prey shells – a common tool used to describe naticid feeding (Vignali & Galleni, Reference Vignali and Galleni1986; Artemieva et al., Reference Artemieva, Grishankov, Nikolaeva, Fokin, Shunatova and Yakovis2004). However, the method may give biased estimates because of the different dissolution rates of damaged and intact shells, or because of the species-specific rate of shell fragmentation, dissolution and removal by waves. The former was unlikely to affect the results, because there is evidence that disintegration rates of perforated and intact shells are similar (e.g. Dudley & Vermeij, Reference Dudley and Vermeij1978). Even if the damaged shells dissolved faster, we could only have underestimated the importance of Hydrobia. The latter source of bias could hardly alter the proportion of perforated shells among empty shells and thus was unlikely to affect the conclusions.
Unfortunately, we could not adequately assess the prey selection by Amauropsis juveniles, because of the high consumption of Macoma in the mixed-prey treatment. We believe that it indicated the ‘floor effect’ – the situation when almost all measurements come from the lower end of the range (Ruxton & Colegrave, Reference Ruxton and Colegrave2006). The potential level of Macoma consumption may be underestimated, because too few clams were offered initially. In contrast, no ‘floor effect’ was observed for Hydrobia. The consumption of Hydrobia could even be overestimated. The density of live Macoma in the mixed-prey treatment dropped by the end of experiment, so the starving moonsnails, who failed to find the clam, could become more inclined to feed on the second-best prey.
Although Amauropsis juveniles are able to consume H. ulvae, small Macoma remains the preferred prey of young and adult predators. It is possible that the density of the adult moonsnails may be indirectly regulated by predation of juveniles on Macoma, as predicted for other predator-prey systems (Osman & Whitlatch, Reference Osman and Whitlatch2004; De Roos et al., Reference De Roos, Schellekens, Van Kooten and Persson2008). On the other hand, the ability to feed on Hydrobia can help the young predators to survive the periods of primary prey shortage, until new clams migrate from the other biotopes (Hiddink, Reference Hiddink2003) or other usual prey species become accessible for growing predators.
We have shown that Amauropsis juveniles can consume different prey than adults. Other naticids may also be able to switch diet in a similar fashion. At least one of the other four White Sea naticids (Golikov, Reference Golikov, Starobogatov and Naumov1987) – Cryptonatica affinis (Gmelin, 1791) – also occurs in the intertidal, where it feeds on mussels Mytilus edulis (Alyakrinskaya, Reference Alyakrinskaya2002). It is possible that the juvenile Cryptonatica also may broaden their diet and include other small prey when the mussels of appropriate size are scarce. The similar ontogenetic diet shift may be predicted for any other size-selective generalist predator in the habitat where very few prey species are available or abundant.
ACKNOWLEDGEMENTS
We acknowledge Kandalaksha Natural Reserve administration for granting access to protected areas and providing land-base facilities. We wish to thank all students of the Laboratory of Marine Benthic Ecology and Hydrobiology for their selfless assistance in field sampling and maintaining of our experiment. Our special thanks go to Alexander Gornykh and Dr Vadim Khaitov for their help in data processing, as well as to Dr Alexey Sukhotin and Dr Eugene Yakovis who helped us to improve our ideas and methods. We are grateful to Yuta Tamberg for patient language revision. We also thank two anonymous referees for valuable comments on our work.
