The genus Agaronia includes dominant predators in the eulittoral zone of dissipative sandy beaches of the tropical Eastern Pacific, which show specific adaptations to this environment such as swash-surfing locomotion. We studied A. propatula in its natural habitat in El Salvador and Costa Rica, and performed field experiments to obtain insights into its ecology, behaviour and sensory physiology. Agaronia propatula is not attracted by carrion and preys mostly on the ubiquitous beach snail Olivella semistriata. This, however, reflects community composition rather than prey specialization; A. propatula is an investigative hunter and will, quite literally, attack everything that moves (with the notable exception of echinoids). Prey is identified at short range by tactile and, to a lesser degree, by chemosensation located in the propodium. We found no evidence for long-distance sensory capabilities; A. propatula rather seems to rely on the regular physical structure of its wave-dominated environment when it moves between its shallow subtidal resting zone and its upper intertidal hunting grounds where potential prey predictably congregates. On the other hand, behavioural patterns such as the rapid yet haphazard cruising of foraging individuals, or the complex prey capture sequence in which the prey is transferred to a metapodial pouch, are similar in A. propatula and Oliva. Thus, our results lead us to speculate that the development of behavioural features that proved adaptive in the intertidal environment was essential in the evolution of Agaronia from Oliva-like ancestors.
Sandy beaches are among the most hostile environments for animals and plants (McLachlan & Brown, 2006). They are shaped by physical factors, in particular the rhythm of the waves and tides, and are characterized by the permanent turnover of the sediment on or in which the organisms live (Raffaelli & Hawkins, 1999; Little, 2000). Sandy beach organisms show specific behavioural adaptations to their dynamic environment (Palmer, 1973; Chelazzi & Vanini, 1988; Brown, 1996). Various molluscs, for example, move swiftly by ‘surfing’ the swash waves, using parts of their bodies as underwater sails. However, the bulk of the available literature focuses on the small number of model species for which quantitative ecological and behavioural results have been produced; in the case of surfing molluscs, these are clams of the genus Donax and Bullia whelks (Brown, Stenton-Dozey & Trueman, 1989). Whether these results can be generalized and are applicable to other taxa remains unclear.
The macrofauna of dissipative sandy beaches of the Panamic faunal province (Pacific American coasts from Baja California to northern Peru) is often dominated numerically by one of two swash-surfing suspension feeders, the snails Olivella semistriata and O. columellaris (Caenogastropoda: Olivellidae) which may reach densities of thousands of individuals per beach metre (Olsson, 1923/1924; Schuster-Dieterichs, 1956; Aerts et al., 2004; own unpublished data). It would appear that the removal of suspended particles from the sea water by these abundant gastropods is a key element of the cycling of organic matter on these beaches, but no quantitative data are available. Similarly, little is known about the most important predators of the Olivella species, namely members of the genus Agaronia (Caenogastropoda: Olividae). Predation by Agaronia species on Olivella species can be easily observed on Panamic beaches and has been noted by several authors (Olsson, 1956; Seilacher, 1959; López, Montoya & López, 1988; Metz, 1995). However, no detailed information on behavioural and physiological aspects of the predator–prey relationships among the olivid gastropods that occupy key positions in the food webs of Panamic sandy beaches has been published.
Here we report field studies on Agaronia undertaken in El Salvador and Costa Rica with two goals in mind. First, we intended to characterize the predator's behaviour and sensory physiology with regard to its prey-capturing capabilities, in order to define this one piece of the food-web puzzle of Panamic sandy beaches. Second, we expected that a better understanding of the sensory ecology of Agaronia would contribute to the understanding of the adaptive diversification of olivid gastropods. In many aspects of its morphology and behaviour, Agaronia closely resembles Oliva, as for example in the peculiar manner in which the foot is used in prey capture and handling (Rupert & Peters, 2011). On the other hand, the genera differ in several important morphological, sensory and behavioural features. For instance, Agaronia lacks cephalic tentacles and eyes, whereas Oliva possesses both (Gray, 1839; Olsson, 1956). Agaronia readily performs ‘tail’ autotomy in response to attacks, although the ‘tail’—the hind part of the foot—plays an essential role in prey capture (Rupert & Peters, 2011); autotomy has not been recorded for Oliva so far. Oliva species generally are nocturnal, whereas the activities of (at least Panamic) Agaronia can be studied during the day. Panamic Agaronia species have successfully occupied the eulittoral of dissipative sandy beaches, showing specific behavioural adaptations to this demanding habitat such as circatidal migrations by swash-surfing (unpublished results), a behaviour not observed in Oliva.
The taxonomy of Agaronia appears plagued by problems similar to those analysed in detail in Oliva by Tursch & Greifeneder (2001). The number of Panamic Agaronia species and their distinguishing criteria differ from author to author (Keen, 1971; López et al., 1988; Sterba, 2004), making it difficult to establish species identities routinely in the field. At this stage, however, we see no evidence for any ecological or behavioural differentiation among the intertidal, swash-surfing, actively hunting Agaronia populations that inhabit dissipative sandy beaches of the Central American west coast. Being unable to resolve the taxonomic issues at this time, we will refer to our study object as Agaronia propatula (Conrad, 1849) for pragmatic reasons (compare Rupert & Peters, 2011).
MATERIAL AND METHODS
Study sites and field observations
Observations of the natural behaviour of undisturbed Agaronia propatula were made on 29 beaches during seven field trips (four trips to northwest Costa Rica with 12 different locations from the Nicaraguan border to the city of Puntarenas; three trips to El Salvador with 17 locations along the entire coast) from November 2009 to May 2011. Types of behaviour that were observed regularly such as swash-surfing, rhythmic burrowing and the stereotypical prey-capture sequence were photographed and filmed with digital cameras. Videos for publication were prepared with ImageJ v. 1.38x (http://rsbweb.nih.gov.ij) and QuickTime 7 Pro (http://www.apple.com/quicktime/). Systematic field experiments (described below) were conducted at two dissipative sandy beaches, each of about 5 km length, which had dense A. propatula populations as well as suitable infrastructures: Playa Grande in the Marine National Park ‘Las Baulas’ in northwestern Costa Rica (10°20′N 85°51′W) and the public beach of El Cuco in eastern El Salvador (13°10′N 88°06′W). At these beaches we obtained prey spectra by recording observed attacks and examining the contents of metapodial pouches.
Field experiments were conducted during the early rainy season in May/June 2010 (Costa Rica) and May 2011 (El Salvador) and focused on the role of mechanic and chemical signals in prey detection by A. propatula. For all tests, actively hunting specimens (i.e. animals that crawled over the sediment surface at low or intermediate tide) of intermediate size (shell length 28–40 mm, 2.2–6.9 g) were selected and the responses to stimuli applied at defined directions and distances were recorded. To ensure that each individual was used for one test only, animals that had been subjected to a test were collected and released after our work at a particular beach section had ended for the day. At least two tidal cycles were completed between consecutive working sessions at a given beach section.
To study the response to mechanical stimuli, natural prey and artificial baits that either moved or remained motionless were offered. Actively crawling Olivella semistriata of intermediate size (shell length 11–15 mm, 0.16–0.39 g) were picked up carefully and placed in the path of hunting A. propatula at 5–6 cm distance from the anterior edge of the hunters' propodium. These O. semistriata usually continued to crawl and, if they did, served as moving natural prey. Other O. semistriata were ‘immobilized’ by rolling them between finger tips, which caused them to retract and remain in their shells for up to 3 min. These specimens were also placed in the paths of hunting A. propatula and served as motionless natural prey. The responses to moving and motionless natural prey were compared to the responses to a variety of artificial baits (compare Fig. 3A below): plastic sticks (7–8 cm long, 0.70–0.80 g), wooden sticks (7–8 cm, 0.32–0.39 g), plastic-covered paperclips (bent to 7–8 cm length, 1.24 g), cotton swabs (4 cm, 0.16–0.19 g when wetted with sea water) and empty turritelliform shells (3.0–3.5 cm, 1.1–1.4 g) taken from the beach. To serve as a moving artificial bait, an item was held so that one of its tips touched the sediment 3 cm in front of a hunting A. propatula and was gently but rapidly vibrated. Motionless baits were items that were simply laid into the path of a hunter at about 6 cm distance. Each cotton swab was used only once to avoid carrying mucus or scent substances in the cotton material. The other items consisted of more inert materials; they were thoroughly washed with sea water after each test and reused. The proportion of tests in which an item evoked an attack response was recorded.
Responses to olfactory stimuli were tested using water that had been loaded with the scent of potential prey species. To obtain scented water, 20 O. semistriata, 40 Echinolittorina conspersa/dubiosa, 15 Donax sp. or 5 A. propatula were incubated in 40 ml sea water in a plastic tube (3 cm diameter) for at least 10 min. Similarly, one medium-sized (4–5 cm diameter) sand dollar (Mellita sp.) was kept in 10 ml sea water in a plastic Petri dish for the same period. Droplets of the scented fluids were placed either in front of a hunting A. propatula, or 5–8 cm above and 3 cm in front of the animal on the beach slope so that the scented water ran into the path of the hunter (compare Fig. 4 below). The number of cases in which this stimulus evoked an attack strike was recorded and compared to the attack rate evoked by the control (plain sea water).
To identify body regions in A. propatula that were responsive to mechanic and olfactory stimuli, a moving artificial bait (plastic stick as described above) or droplets of Olivella-scented sea water were applied at different positions next to the hunters' bodies as detailed in Figure 6 below.
Categorical data that could be represented in 2 × 2 contingency tables were compared using Fisher's exact probability (FEP) test (two-tailed). A helpful online calculator with explanations can be found on the website maintained by Richard Lowry at http://faculty.vassar.edu/lowry/fisher.html (accessed last on 7 January 2012). Mean values of continuous data were compared using Student's t-test (two-tailed).
General observations: functional morphology of foot
Agaronia propatula shows the typical features of the genus, lacking a defined head, tentacles, eyes and an operculum (Olsson, 1956). The anterior of the foot is shaped like a double-sided ploughshare. A pronounced groove separates the narrow propodium from the large metapodium (Fig. 1A). The latter has two lateral folds that can cover the anterior portion of the shell and the entrance of the mantle cavity. More posteriorly, the metapodium is flat and visually divided in most specimens by a distinctly pigmented transversal line, which marks an autotomy plane (Rupert & Peters, 2011). Irrespective of these morphological subdivisions, the foot represents two biomechanical elements, as becomes apparent when the snails perform searching movements while being held up in the air. In this situation, the anterior part of the foot, consisting of propodium and the anterior metapodium, forms a pointed ‘snout’ that probes the surroundings, while the posterior metapodium flattens and becomes concave ventrally (Fig. 1B; Supplementary material 1). The biomechanical antagonism of the two foot components in locomotion becomes apparent when hunting specimens accelerate by switching from the usual smooth gliding mode into a ‘galloping’ mode in which the posterior and anterior foot portions move forward alternately; we clocked ‘galloping’ snails such as the one seen in Supplementary material 1 at velocities in excess of 9 mm/s. Agaronia burrows by thrusting stepwise into the sediment at a shallow angle, indicating that it employs the common ‘penetration anchor’ vs ‘terminal anchor’ mechanism analysed by Trueman & Brown (1992; Supplementary material 1). As in other burrowing marine gastropods, the subdivision of the foot of Agaronia provides the antagonistic actuators required (posterior foot: penetration anchor; anterior foot: terminal anchor).
The two elements of the foot are essential for the prey capture mechanism of A. propatula, which resembles that described previously for Oliva species (Marcus & Marcus, 1959; Olsson & Corvo, 1968; Taylor & Glover, 2000; Kantor & Tursch, 2001; Supplementary material 1). However, in contrast to most Oliva species, A. propatula hunts on beach sediments that are wet but exposed to air between incoming waves. Prey capture of Agaronia is therefore easily observed in the wild. Once a potential prey item is located below the sand surface, the snail extends the anterior foot rapidly to grasp the prey. If the detected prey is located above the surface, the anterior foot is raised during its rapid forward movement so that the prey is grasped from above. Such above-surface strikes make a surprisingly violent impression on the unsuspecting observer; they are particularly fast and powerful if they are carried out in the forward direction, because in this case they are supported by a simultaneous backward movement of the posterior metapodium, which balances the momentum of the striking anterior foot. In contrast, strikes to the left or right are visibly slower. In most cases in which the initial strike is unsuccessful, the snail immediately launches a second and sometimes even a third strike. These, however, usually are directed some 10–30° to the left or right of the direction of the initial attack (Supplementary material 2).
After a successful strike, the anterior foot with the captured prey bends ventrally and backward. At the same time, the posterior foot turns concave ventrally and bends ventrally and forward; like a dredger bucket, it takes up sediment and water but also encloses the oncoming anterior foot, which holds the prey item. This movement of the posterior metapodium is so powerful that it usually causes the animal to ‘somersault’ forward. The anterior foot is then pulled forward out of the pouch formed by the posterior metapodium, leaving the prey within the pouch. Next, the pouch contracts which, under favourable conditions, becomes visible as a forceful expulsion of water and sediment from the pouch (this can be seen in the movie by Rupert & Peters, 2011: Supplementary material). Finally, the snail burrows rapidly into the sediment, dragging the prey within the metapodial pouch to a depth of several centimetres, where it can be safely consumed. Burrowing always follows a successful attack; we never observed A. propatula collecting multiple prey items in the pouch during a continuing raid as reported for several Oliva species (Taylor & Glover, 2000; Kantor & Tursch, 2001). Since burrowing with, as well as without, prey occurs at similar speed in a stepping fashion, the most posterior, pouch-forming portion of the foot does not seem to be an essential functional element of the penetration anchor.
Similarly, the expansion of the posterior foot is not functionally essential for swash-surfing. Normally, surfing animals expand the complete foot and then are dragged along the sand surface by the water flow, lying on their right or left side (Fig. 1C). However, at 2 to 1h before low tide, large numbers of A. propatula utilize the backwash of the last waves that reach across the intertidal plane to surf back to the lower beach zone that will remain submerged. We found that up to one-third of these swash-surfing individuals carry prey in their pouches, even when little or no hunting activity had been observable at the same location since the preceding high tide. This is circumstantial evidence suggesting that burrowed A. propatula keep captured prey in their metapodial pouches over prolonged periods rather than swallowing it. Thus it appears that A. propatula consumes prey while holding it in the pouch, as Oliva species do (Marcus & Marcus, 1959; Kantor & Tursch, 2001).
Field observations: prey species and their detection
At Playa Grande (Costa Rica), the prey most frequently attacked and subdued by A. propatula (88% of all observed attacks and examined pouches) was the ubiquitous Olivella semistriata (Table 1). Equal proportions of the remaining attacks were directed at burrowing clams (Donax sp.), unidentified small (<1 cm) crustaceans and smaller individuals of its own species. The prey spectrum at El Cuco (El Salvador) was more diverse (Table 1, Fig. 2) which seemed to reflect the composition of the beach community. Common taxa of Panamic beaches such as burrowing clams (Donax sp.) and mole crabs (Emerita sp.) can be found at Playa Grande, but one has to search for them. In contrast, these animals literally are ‘all over the place’ at El Cuco and are collected effortlessly by the local population for food. Olivella semistriata was still the most frequently attacked prey species at El Cuco, but its proportion was reduced to some 60%, with bivalves contributing almost one quarter of all prey items (Table 1). While gastropods subdued by an A. propatula were always small relative to their predator (Fig. 2C, D), the less mobile bivalve prey items frequently reached the size of the Agaronia that had captured them (Fig. 2E, F). Interestingly, all mature Emerita sp. that we found under attack by an A. propatula were very much larger than their attacker (Fig. 2G, H) and had a soft and elastic carapace, indicating that they just had moulted. In all observed cases, the Agaronia were struggling, without immediate success, to burrow into the sand with the oversized prey. Whether such attacks on large mole crabs ultimately lead to the death of the prey remains unclear.
|Prey species||Prey taxon frequency|
|Playa Grande (%)||El Cuco (%)|
|Olivella semistriata (Gastropoda)||88||61|
|Agaronia propatula (Gastropoda)||3||2|
|Donax sp. (Bivalvia)||4||18|
|Tivela sp. (Bivalvia)||—||4|
|Pitar sp. (Bivalvia)||—||2|
|Unidentified small (<1 cm) crustaceans||4||4|
|Emerita sp. (immature, <2 cm; Crustacea)||—||2|
|Emerita sp. (mature, >5 cm, with soft exoskeleton; Crustacea)||—||7|
|Prey species||Prey taxon frequency|
|Playa Grande (%)||El Cuco (%)|
|Olivella semistriata (Gastropoda)||88||61|
|Agaronia propatula (Gastropoda)||3||2|
|Donax sp. (Bivalvia)||4||18|
|Tivela sp. (Bivalvia)||—||4|
|Pitar sp. (Bivalvia)||—||2|
|Unidentified small (<1 cm) crustaceans||4||4|
|Emerita sp. (immature, <2 cm; Crustacea)||—||2|
|Emerita sp. (mature, >5 cm, with soft exoskeleton; Crustacea)||—||7|
Prey frequencies include directly observed attacks and prey removed from metapodial pouches. Pouches never contained more than one item.
Among the echinoderms on Pacific sandy beaches of Central America, lunulate sand dollars of the genus Mellita are common (Harold & Telford, 1990). We found the animals regularly in the lower eulittoral of Playa Grande and El Cuco. Since we had never observed Agaronia attacking Mellita sp., we tested their interaction by placing a sand dollar in the path of a crawling snail so that they moved towards each other. Upon contact, the sand dollar continued on its way unimpressed, whereas the snail stopped briefly and then moved to the left or right, apparently to evade the sand dollar (five tests with different animals gave the same results). Thus, A. propatula seems to avoid echinoids. Similarly, carrion appears of little interest to A. propatula, as we never observed snails approaching, let alone feeding on, the dead jellyfish, crabs, fish, sea turtles, birds or fishery and kitchen wastes that we found washed-up on the beaches. This is an obvious contrast to the scavenging habits of South African Bullia whelks, the standard textbook example of swash-surfing snails (Brown, 1982).
Agaronia propatula preferred wet sand occasionally inundated by incoming waves and numerous animals crawled about the sediment of the lower and mid-eulittoral at low and intermediate tide. Evidently, these crawling specimens were actively hunting, as they almost always attacked moving prey that crossed their paths. In contrast, individuals that were resting motionless on the surface—this was most frequently observed on relatively dry sand when the tide was lowest—or that had started to burrow into the sediment seemed to be in a ‘peaceful’ mode, as we never observed them attacking approaching prey; neither could we provoke these animals to tackle artificial baits or respond to the scent of prey (compare field experiments below).
Interestingly, A. propatula typically attempted to burrow into the sand when they had landed at the end of a swash-ride, even if the spot where a surfing snail came to rest was located within a dense accumulation of O. semistriata. As we did not observe a single instance of an animal starting to hunt immediately after swash-surfing, it appears that A. propatula do not initiate swash rides to pursue specific prey items detected over long distances, as has been reported for Bullia digitalis (Odendaal et al., 1992). This conclusion is supported by the fact that long-shore distributions of hunting A. propatula and O. semistriata did not correlate. While generally abundant, the local densities of O. semistriata varied along the beaches, with high-density spots that moved or dispersed and reassembled elsewhere with each tidal cycle. Agaronia often formed similar hotspots but, contrary to our initial expectation, dense accumulations of O. semistriata were of no guiding value when we searched for locations with high densities of A. propatula to perform the field experiments described below.
Crawling, motivated hunters never probed the sediment surface with their siphon, and generally appeared unaware of prey at distances >3 cm. Potential prey animals were usually attacked when they were moving within about 2 cm in front of the predator. In contrast, motionless O. semistriata were frequently ignored, in some cases even if the hunting Agaronia literally bumped into them (Supplementary material 3). Crawling O. semistriata leave characteristic tracks in the sand. On a few occasions we observed A. propatula striking at recently made Olivella tracks (Supplementary material 3) or following these tracks over significant distances (up to 40 cm) finally to make a catch. In most cases, however, tracks that an Olivella had left—sometimes only seconds before a hunting Agaronia crossed them—evoked no response whatsoever from the predator (Supplementary material 3). These observations suggested that prey detection occurred over short distances only and depended mostly on tactile cues, possibly supported by olfactory stimuli under certain conditions. We decided to clarify the situation by controlled field experiments.
Field experiments: tactile stimuli
When active O. semistriata were picked up from the sand and placed in front of a hunting A. propatula so that their paths crossed shortly afterwards, the crawling Olivella was always attacked when it approached the front of the Agaronia to <2 cm (Fig. 3B). However, when the prey was first ‘deactivated’ by rolling it between finger tips for a few seconds, which caused it to retreat into its shell, and then placed into the path of a hunting Agaronia, the attack rate decreased to about 30% (Fig. 3B). We repeated this experiment with five types of nonliving bait: sticks of plastic and wood, paper clips, cotton swaps and empty gastropod shells (Fig. 3A). When any of these baits was gently vibrated while it touched the sand roughly 2 cm in front of the anterior edge of the foot of a hunting A. propatula, it was almost always attacked (Fig. 3B). Placing the same baits motionless about 6 cm before an Agaronia reduced the attack rate to some 10%, although the hunters always made contact with the bait item shortly after we had placed it in their paths (Fig. 3B). Evidently, motion transduced by the water-saturated sediment as a tactile stimulus is an essential signal that hunting A. propatula employ to distinguish live potential prey from random dead objects that they encounter in their environment.
Agaronia propatula that had ‘caught’ an artificial bait performed the usual behavioural sequence. The ‘prey’ was transferred to the metapodial pouch and the successful hunter attempted to burrow into the sand (Supplementary material 4). This behaviour did not require further movement of the artificial prey, indicating that while potential prey was identified by its motility before the launch of an attack, the subsequent handling of the captured prey did not depend on additional prey-specific signals. Because the artificial baits were much larger than the standard prey O. semistriata (Fig. 3A), they could not be enclosed completely in the metapodial pouches, as similarly observed with mature Emerita sp. (Fig. 2G, H). Nevertheless, the hunters kept their catch in close grip and could be lifted from the sediment by the bait (Supplementary material 4). However, while A. propatula that had captured O. semistriata in their pouches always burrowed into the sediment successfully, hunters that attempted to burrow while clinging to an artificial bait item frequently released the bait before burrowing was complete (Fig. 3C). Those burrowing attempts that were finished (as defined by the complete disappearance of the animal, not the bait) tended to take longer with artificial bait than with O. semistriata (Fig. 3C). These results suggested that A. propatula released artificial bait items because it had difficulties handling very large prey, rather than because it could distinguish between real prey and artificial objects once an object classified as prey had been captured. We concluded that A. propatula identified moving objects as prey even if these objects consisted of materials and had shapes of which the predators had no prior experience.
Field experiments: olfactory stimuli
Motionless O. semistriata were attacked more frequently by A. propatula than any of the five motionless artificial baits, although the statistical significances of the differences varied (0.058 > P > 0.005, FEP test; Fig. 3B). This indicated that A. propatula had a limited capability of identifying natural prey and distinguishing it from dead objects in situations in which the primary criterion for prey identification—motion—was absent. To test whether this capability was based on olfactory stimuli, we prepared ‘Olivella water’ by incubating O. semistriata in a small volume of sea water. When two to five droplets of this water (which we presumed to contain Olivella-specific scent substances) were placed carefully about 2 cm before a hunting A. propatula, strikes were triggered in some 80% of the trials (Fig. 4A). Control tests with sea water did not usually result in attacks (Fig. 4A; see also Fig. 5). Obviously, hunters performing ‘Olivella water’-induced strikes literally clutched at thin air (Supplementary material 5). We concluded that A. propatula can identify its main prey, O. semistriata, by a water-soluble substance(s) emitted by the prey.
Although the above results (Fig. 4A) appeared unambiguous, we decided to exclude the slightest possibility that strikes that seemed to be provoked by scent substances in ‘Olivella water’ were actually triggered by vibrations that occurred when the droplets were placed on the sediment. To this end, we selected hunting A. propatula that moved more or less perpendicular to the slope on relatively steep parts of the intertidal zone. Series of droplets of ‘Olivella water’ were placed 5–8 cm uphill of and about 3 cm in front of the crawling A. propatula, so that the flow of the fluid into the path of the hunting snail could be followed as sediment particles were moved downhill. When the hunting Agaronia reached the freely flowing ‘Olivella water’, strikes were launched in three quarters of all trials (Fig. 4B). In contrast, attacks occurred at a negligible rate in control experiments with sea water (Fig. 4B). Thus, our conclusion that A. propatula is capable of sensing O. semistriata olfactorily was corroborated.
In the experiment shown in Fig. 4B, strikes always occurred in the forward direction, i.e. in the direction in which the snail moved. After their vain attacks, the hunting snails never turned uphill but continued on their original path. This implied that the direction of flow of the scented water as such was not perceived as meaningful information by A. propatula.
Echinolittorina conspersa and E. dubiosa are small Panamic periwinkles restricted to the rocky supralittoral fringe (Williams & Reid, 2004), and therefore are unlikely to meet A. propatula in its habitat. However, Echinolittorina-scented water evoked the same responses as Olivella-scented water (Fig. 5; in the field, we did not distinguish between the two similar Echinolittorina species which formed mixed populations on the rocks adjacent to our study beaches). Evidently, the capability of A. propatula to identify potential prey olfactorily is not restricted to the species it encounters regularly. The predator also responded to the scent of Donax sp. as well as to that of its own species (Fig. 5). The lower attack rate in response to these stimuli as compared to the responses to Olivella and Echinolittorina scents (Fig. 5) cannot be interpreted as evidence for a lower sensitivity, because the actual concentrations of the relevant scent compound(s) were not controlled in our tests. Water carrying the scent of sand dollars (Mellita sp.) failed to induce attacks but caused Agaronia to turn away from the stimulus in some 20% of the trials (Fig. 5). This reaction, which we never observed following exposure to scents of other potential prey species, resembled the response to live Mellita specimens described above. While the reason for this avoidance of echinoids remains unclear, we can conclude that prey identification in A. propatula is supported by an olfactory sense which is versatile enough to allow for the exploitation of unusual prey such as an unlucky Echinolittorina flushed from its rock onto an adjacent sandy beach.
Field experiments: localization of tactile and olfactory receptors
We had frequently observed that close encounters and even collisions between A. propatula and O. semistriata provoked no reaction from the predator, especially when the potential prey approached the predator from the side or back (Supplementary material 3). Moreover, the leading edge of the propodium of crawling A. propatula often performed rapid ruffling movements that are suggestive of a continuous probing of the environment (Supplementary material 5). These observations suggested that A. propatula surveys only a limited, forward-facing sector of its immediate surroundings by means of sensory organs in the propodium. To test this hypothesis, we compared the reactions to tactile (vibrating plastic stick) and olfactory (‘Olivella water’) stimuli that were applied at four different positions around the hunting predator. Both types of stimulus provoked attacks at high rates when applied in front of the snail and on its left or right side at the level of the anterior end of the shell (Fig. 6). When applied behind the crawling animal or laterally at the level of the shell apex, attack rates were negligible (Fig. 6). We concluded that the tactile and olfactory receptors that enabled prey identification in A. propatula were restricted to the most anterior part of the foot, probably the propodium.
Interestingly, all strikes observed in these experiments (Fig. 6) were directed towards the location at which the stimulus had been applied, implying that A. propatula is capable of the exact localization of nearby sources of sediment vibrations as well as attractive smells.
The genera Agaronia and Oliva have always been considered to be closely related (e.g. Gray, 1839; Olsson, 1956; Tursch & Greifeneder, 2001: 67). Therefore it is no surprise that our investigations into the behavioural and sensory ecology of A. propatula revealed close similarities to some of the behavioural patterns found in Oliva. For example, the anterior and posterior parts of the foot can act as biomechanically independent units or even antagonists, as becomes evident when prey is seized by the anterior foot and then transferred to the posterior foot for storage in the metapodial pouch. This complex action proceeds in a practically identical manner in both taxa (Marcus & Marcus, 1959; Olsson & Corvo, 1968; Taylor & Glover, 2000; Kantor & Tursch, 2001; Rupert & Peters, 2011; Supplementary material 1). It should be noted that a similar behavioural sequence has evolved in naticids (Gonor, 1965; Edwards, 1969). Similarly, Oliva as well as A. propatula seem incapable of accurate long-distance localization of their prey, which needs to be in the immediate vicinity of the leading edge of the propodium for proper identification (Marcus & Marcus, 1959; Zeigler & Porreca, 1969: 11; Kantor & Tursch, 2001; Fig. 6). While the habitat of at least three Oliva species extends from deeper water into the eulittoral (O. undulata, Olsson, 1956: 164; O. tigridella, Taylor & Glover, 2000; O. oliva, Tursch & Greifeneder, 2001: 194; Yu.I. Kantor, personal communication), most members of the genus dwell in subtidal sediments (Van Osselaer et al., 1994). On the other hand, A. propatula performs swash-surfing, a highly specialized behaviour of intertidal sandy beach molluscs, which has not been reported from Oliva so far. Thus, A. propatula appears like an Oliva that has adapted to the conditions of the sandy beach intertidal more completely than the other Oliva species found in this habitat. This idea suggests an evolutionary scenario in which a clade represented by modern Agaronia separated from the Oliva main line(s) through the development of behavioural features that enabled a more efficient exploitation of the harsh intertidal environment. It will be interesting to determine whether the ecology of Agaronia species in other faunal provinces supports this hypothesis. This comparative approach should be expanded further by studies on swash-surfing, South American Olivancillaria (Olividae) species, which apparently have also conquered the sandy eulittoral (Caetano, Veloso & Cardoso, 2003).
Comparing food spectra of A. propatula and Oliva species, one may conclude that the former differs from the latter in its preference for one prey species. On our study beaches, O. semistriata was by far the prey most frequently taken by A. propatula, in agreement with remarks in the literature (Seilacher, 1959; López et al., 1988; Metz, 1995). However, these observations do not necessarily imply a specialization or preference of the predator for this particular prey. Olivella semistriata is the numerically dominant macrofaunal element on many northern Panamic sandy beaches and accumulates in the backwash zone, where it can apply its specialized suspension-feeding mechanism (Seilacher, 1959; this author mistook O. semistriata for O. columellaris, but his observations, interpretations and excellent drawings are highly instructive nonetheless). Consequently, the dominance of O. semistriata in the prey spectrum of A. propatula may simply reflect the relative abundance of potential prey species in the community in which Agaronia hunts (for general discussion of related problems in ecology, see Underwood, Chapman & Crowe, 2004). Several findings support the idea that A. propatula is an opportunistic predator. First, species other than O. semistriata contribute significantly to the prey spectrum at Playa de Cuco with its rich species composition (Table 1). Second, the scent of Echinolittorina conspersa and E. dubiosa, upper rocky-shore species whose habitat does not overlap with that of A. propatula, triggers the same attack responses as that of O. semistriata (Fig. 5). Third, various artificial objects are attacked if they displayed a key character that distinguishes most living objects from dead ones—autonomous movement (Fig. 3A, B). Since Agaronia always attempts to reach a feeding position underneath the sediment surface once it has ‘caught’ such an artificial object (Fig. 3C), it seems that the familiarity of size, shape, smell or taste is no requirement for prey identification. This ‘investigative hunting’ style of A. propatula provided us with an opportunity to teach a cheeky know-all tourist on Playa Grande some respect for the key predator on the beach, as Agaronia can be provoked to strike at human toes set before them. In this context, the avoidance of Mellita sp. is intriguing. These echinoids must possess a very powerful deterrent indeed to keep a predator at bay which unhesitatingly pounces on unknown objects and prey specimens several times its size. We may speculate that Mellita is toxic for A. propatula, but this hypothesis requires experimental corroboration.
The crawling speed of O. semistriata, as well as that of A. propatula, increases with body size (unpublished observations). This fact, together with the observation that gastropods found in metapodial pouches were always smaller than the predator, whereas subdued bivalves often reached the predator's size (Fig. 2C–F), suggests that prey mobility, but not prey size, is a limiting factor for hunting success in A. propatula. This interpretation is further supported by the low proportion of highly mobile crustaceans in the prey spectrum and the relatively small body sizes of the few crustaceans that actually were found in metapodial pouches (Table 1). Adult mole crabs seemed to be an exception (Fig. 2G, H), but all the individuals attacked had not yet hardened their exoskeleton following moulting. These mole crabs were unable to move and burrow efficiently, due to the transiently nonfunctional state of their exoskeleton. Given their agile nature, it seems inconceivable that a biomechanically fully functional, adult mole crab could fail to escape an approaching Agaronia.
On wet sand, A. propatula monitors a narrow band of about 2 cm width around the edge of its anterior foot, within which potential prey is detected reliably by tactile and olfactory receptors that appear to be localized in the propodium (Fig. 6). Due to its rich innervation, the leading edge of the propodium has been suggested to be the “most important sense organ” in Oliva (Marcus & Marcus, 1959). Chemosensory sensitivity of the propodial edge also has been demonstrated for Bullia whelks (Hodgson & Brown, 1985), which resemble Agaronia in their swash-surfing habits and the fact that they forage on wet sand. Within the narrow surveillance zone around the propodium, A. propatula determine the position of the source of a stimulus quite precisely: in experiments in which the stimulus was applied within 1–2 cm of the edge of the propodium, attacks were generally directed towards the source (Fig. 6). Such precision seems to result from the snail's ability to interpret the point on its propodium where the stimulus is perceived first as a marker of the direction towards the source of the stimulus. This interpretation is supported by experiments in which snails crawled towards a stream of Olivella-scented water that flowed perpendicular to the snail's path (Fig. 4B). Strikes carried out when the snail moved into the stream of scented water were always directed forward in the direction of the snail's movement, presumably because it was the most anterior portion of the propodium that came into contact with the stimulus first. The flow of the scented water as such did not provide useful information for hunting Agaronia, as the animals never turned towards the source of the stream in these experiments. Evidently, the snail's ability to integrate spatially the information acquired by the propodial chemoreceptors is limited, despite the precision in localizing stimuli originating from within the surveillance zone of the propodium.
Although hunting A. propatula screen only a small forward-facing area of not more than 5–10 cm2, they crawl rapidly on more or less random trajectories with speeds approaching 1 cm/s, thus increasing the chance to contact potential prey. Similarly, Oliva species “often pass by a morsel two or three times before finding it” even if it is only centimetres away (Zeigler & Porreca, 1969: 11; compare Kantor & Tursch, 2001), probably because food identification requires direct contact with the receptors in the propodium (Marcus & Marcus, 1959: 119). Interestingly, Oliva species make up for their lack of efficient long-distance food localization by speed of movement, which often enables them to reach food through haphazard paths before slower competitors with superior navigation capabilities arrive on straight trajectories (Kantor & Tursch, 2001). In contrast to A. propatula, most Oliva species forage under water and take up water through their siphons, apparently detecting the presence—as opposed to the location—of food using chemoreceptors on the osphradium (Olsson & Corvo, 1968; Zeigler & Porreca, 1969; Kantor & Tursch, 2001). Osphradium-based sensory capabilities play no detectable role in A. propatula hunting on tidal planes exposed to air, but this does not necessarily exclude the possibility that the perception of water-borne chemical stimuli is relevant for the approach to potential prey while an Agaronia is submerged. Such long-distance perception could be hypothesized to operate when Agaronia perform swash-surfing to reach the beach zone where feeding O. semistriata concentrate. However, our field observations provided no support for this idea. Agaronia propatula showed no tendency to home in on particularly dense accumulations of its prey—as they should if they were attracted chemically to the prey over long distances. In addition, A. propatula reliably started to burrow into the sand at the end of each swash ride, which would not be expected if swash rides were undertaken to reach food items precisely located over long distances.
The lack of evidence for long-distance chemoperception of prey in A. propatula may be surprising, as it well known that gastropods employ olfactory signals in a multitude of physiological and ecological contexts (Croll, 1983). Various marine gastropods use chemical signals to detect the presence of potential predators (Harvey, Garneau & Himmelman, 1987; Marko & Palmer, 1991; Rochette et al., 1998; Jacobsen & Stabell, 2004; Aschaffenburg, 2008). Marine gastropods may also find their food thanks to their chemoreceptory abilities, a fact that has long been exploited in whelk fisheries (McQuinn, Gendron & Himmelman, 1988). Moreover, Buccinum undatum (Rochette, Morisette & Himmelman, 1995; Rochette, Dill, Himmelman, 1997) as well as Littorina scutulata (Keppel & Scrosati, 2004) have been shown to perform risk assessment if olfactory cues signal the presence of both food and predators. Nonetheless it is unlikely that A. propatula utilizes chemoreception of water-borne substances to detect its prey over significant distances on the beach, not only because we have not found supporting evidence, but also because of the fundamental physical problems involved in long-distance perception in this unique environment. If a foraging gastropod is to detect and approach food over significant distances with the help of its olfactory capacities, water must flow continuously from the food to the gastropod (Weissburg, 2000; Weissburg et al., 2002; Webster & Weissburg, 2009). This ‘informative flow’ may be turbulent (Ferner & Weissburg, 2005) but it must not cease or change direction, or else the forager loses its bearings (Lapointe & Sainte-Marie, 1992). Obviously, this requirement is not met on dissipative sandy beaches where the water is permanently mixed and changes its direction of flow continuously. Consequently, A. propatula and other intertidal gastropods may well ‘smell’ that food is present somewhere on the beach, but it is hard to see how they could locate their food over significant distances through the detection of water-borne chemicals. This conclusion seems to disagree with the report that the intertidal scavenger Bullia digitalis locates dead mussels by chemoreception followed by directed swash-surfing towards the food (Odendaal et al., 1992). However, Odendaal et al. (1992) studied Bullia approaching carrion floating in the swash. As these authors explained, a cloud of scent substances may remain stable relative to the position of its emitter if the emitter moves in unison with the water body in which it floats. This situation is distinct from the case of stationary emitters such as congregations of O. semistriata at the water line. It also is worth noting that in the experiments performed by Odendaal et al. (1992), B. digitalis detected food only when it was closer than 2 m, and that the bait was approached successfully in <30% of the trials even when the initial distance between snail and bait was <0.5 m.
Our interpretation agrees with the conclusion drawn by Kitching & Pearson (1981) from field observations in the naticid Polinices incei, a predatory gastropod that hunts bivalves on sandy beaches of eastern Australia. “It is plausible … that the adoption by some intertidal species of a non-olfactory method of prey detection is likely to be adaptive given the special environmental conditions which obtain on certain shores.” At intermediate and low tide, the sandy beach eulittoral “is washed frequently by incoming waves. For a predator and prey operating just below the sand surface, this will have the effect of washing out any chemical emissions, obviating the possibility of prey detection by olfaction, especially for a slow moving predator like the snail. Mechanoreception … presents an alternative which will be unaffected by wave action” (Kitching & Pearson, 1981: 320). In this context it is of interest that Navanax inermis, an efficient opisthobranch predator of the intertidal and shallow subtidal of the North American west coast appears to rely entirely on contact as opposed to distance chemoperception to locate acceptable prey (Paine, 1963).
While wave action certainly does not interfere with mechanical prey detection in those phases in which the sandy intertidal is exposed to air (Kitching & Pearson 1981), intertidal organisms may actually utilize wave action as a beneficial abiotic factor if they exploit the physical regularity that it establishes. This was recognized by Brown (1971) in a review of the lifestyle of swash-surfing Bullia species. Upon chemical detection of the presence of carrion, Bullia tend to emerge from the sand and surf towards the water line. There, they begin “crawling around almost at random”. In doing so, they never venture further up into drier beach zones or move back into the water, thus remaining within “a relatively narrow strip of beach—precisely that area in which washed-up animal remains are most likely to come to rest” (Brown, 1971: 303). It must be stressed that in this interpretation, as in the case of hunting Oliva (Kantor & Tursch, 2001), the smell of food betrays its presence, not its exact location. However, Bullia whelks can ‘predict’, as it were, the approximate location of the food item due to the regular physical properties of their habitat. Surfing in A. propatula appears to serve a similar purpose through a similar mechanism. Hungry individuals position themselves in the beach zone in which smooth backwash flows regularly occur, which is where the suspension-feeding O. semistriata congregates due to its own feeding requirements. Even without long-distance perception, Agaronia can ‘predict’ where prey accumulates at highest densities, and follow the prey during the tidal cycle. This requires migrations over a few hundred metres each day on typical dissipative beaches and swash-surfing is an energy-efficient mode of locomotion that enables A. propatula to remain in beach zones of maximum prey density. On the other hand, our finding that numerous specimens surfing downwards at retreating tide had prey in their pouches indicates that nonhunting individuals prefer lower beach zones, possibly to avoid predation by birds, terrestrial animals and larger members of their own species. Swash-surfing locomotion therefore not only facilitates the approach to prey, but enables A. propatula to utilize the different parts of dissipative sandy beaches with their dynamic yet predictable zonation.
Evidently, long-distance senses are not required for A. propatula to make sense of its environment; these snails can move over long distances between beach zones of high prey density and relatively safer locations, relying on the regular physical structure of their habitat alone. In terms of its sensory ecology, A. propatula seems to live in a mostly tactile world. This appears sufficient and appropriate as A. propatula's world is structured by factors that can be sensed and monitored by mechanoreceptors. Temporal structure is provided by fast and slow cycles of fluid-mechanic forces (waves and tides, respectively), while spatial structure is established by the direction of water flow (defining the physical and ecological gradients along the beach slope) and the sediment surface (separating the fluid phase in which to move from the semisolid phase in which to rest). Swash-surfing gastropods generally possess reduced sets of sensory organs, as indicated by the lack of eyes not only in Agaronia but also in Olivella and Bullia. It appears that their adaptation to the life on sandy beaches included the loss of sensory capabilities that became dispensable due to the physical regularity of this habitat.
This paper summarizes results from an IPFW Honors Project conducted by A.Z.C., which was supported by a grant from the June Enoch Scholarship Fund (we thank Prof. Shree Dhawale for enabling this support), from a research project financed by an IPFW Undergraduate Summer Research Grant to S.D.R. and from fieldwork by volunteer undergraduate research assistants (M.G., J.C.R. and A.S.S.). We thank Dr Pilar Santidrian Tomillo, the staff at the Goldring Marine Biology Station at Playa Grande, and the Leatherback Trust (www.leatherback.org) for logistic support and Dr David G. Reid for identifying the Echinolittorina species. Fieldwork in the Parque Nacional Marino ‘Las Baulas’ was carried out under the permit ACT-OR-D-015 from the Sistema Nacional de Áreas de Conservación, Ministerio del Ambiente y Energia, Costa Rica, to W.S.P. This paper is dedicated to Prof. Dieter Stefan Peters on occasion of his 80th birthday.