Introduction

The behavior of the inhabitants of intertidal zones is linked to the predictable tidal cycles, and the involvement of endogenous clocks has been demonstrated in several species (^{Naylor 2010}). Especially in mollusks, however, the distinction of endogenously controlled circatidal behaviors from direct responses to tide-dependent changes of external stimuli has proven difficult. For instance, the tide-related behavior of Hydrobia ulvae, an abundant gastropod of European estua-rine habitats, has been scrutinized for half a century with equivocal results (^{Newell 1962}, ^{Little and Nix 1976}, ^{Barnes 1986}, ^{Vieira et al. 2010}). Other examples are the large-scale tidal migrations of certain sandy beach mollusks, which appear highly adaptive in the intertidal environment. Nonetheless, it seems that mollusk tidal migration is "not synchronized by an intrinsic mechanism but results from behavioral responses to changing physical conditions" (^{McLachlan et al. 1979: 433}; see also ^{Ansell 1983}, ^{McLachlan and Hesp 1984}, ^{Ellers 1995a}, ^1995b).

Olivella columellaris (Sowerby 1825) and O. semistriata (Gray 1839) (Olivellidae, Caenogastropoda) are common sandy-beach gastropods of the tropical American west coast (^{Olsson 1956}) with unusual feeding habits. Both species use 2 pairs of lateral appendages, which are not found in other Olivellidae, to expose semispherical mucus sheets to flowing water (^{Troost et al. 2012}). The mucus, together with captured plankton and detritus, is eaten in short intervals (^{Seilacher 1959}). So far, this mode of suspension feeding has been noted only in the backwash zone of sandy beaches. Since this zone moves with the tide, efficient feeding requires tidal migrations, which both species achieve by using their expanded feet as underwater sails. It seemed plausible to interpret underwater sailing locomotion and suspension feeding as functionally linked elements of a specific adaptation to the conditions in the sandy beach intertidal zone (^{Schuster 1952}, ^{Schuster-Dieterichs 1956}, Seilacher 1959, ^{Friedrich 1969: 270}), leading to the claim (^{Vanagt et al. 2008}) that an endogenous circatidal clock is crucial for explaining the observed migration patterns (all these authors confused O. semistriata and O. columellaris; see Troost et al. 2012).

Extreme densities of O. columellaris are characteristic of northern Peruvian beaches (^{Olsson 1924}; compare ^{Ramírez et al. 2003}). Here we report observations from this region which suggest that the migration and suspension-feeding behavior of this species are more versatile than previously thought.

Materials and methods

In August 2013, we studied O. columellaris during field trips along the coast of the Department of Piura, Peru, and conducted systematic observations north of the village Colan (4°59'S, 81°4'W). To determine local densities, short (20 cm) rigid plastic pipes of 28, 14, or 10 cm diameter were pushed into the sediment to 8 cm depth, the pipe contents were collected and sieved (5 mm mesh), the O. columellaris retained in the sieve were counted, and the density was calculated as individuals per square meter (ind m^-2). Swash and backwash, flow in channels, suspension feeding, and sailing events (migration by snails employing their foot as an underwater sail) were documented on videos taken with digital cameras (DSC-H20, Sony, Tokyo, Japan) and analyzed using QuickTime Pro v7 (http://www.apple.com/quicktime).

Results

Olivella columellaris is most easily observed when it is feeding in the backwash. The animals remain burrowed while the swash passes over them upwards on the beach slope. After the flow has reversed and the velocity of the backwash has decreased to such an extent that larger sediment particles are not transported anymore, the snails unfold their lateral appendages to expose the mucus sheets to the flow (Fig. 1). Evidently, O. columellaris cannot perform suspension-feeding at high flow velocities (as reported for O. semistriata by ^{Seilacher 1959}). Consequently, the backwash periods during which suspension feeding is possible rarely exceed 20 s. At our study site at Colán, the density of O. columellaris in the backwash ranged from 1,000 to 20,000 ind m^-2. In contrast, fewer than 10 ind m^-2 were found above the highest reach of the waves at any stage of the tide, indicative of the efficiency of the animals' migration in the backwash zone.

Figure 1 Olivella columellaris in its natural habitat near Colan, Peru, suspension-feeding in the backwash; flow direction is bottom to top. In these animals that are burrowed in the sand, only the propodium (most anterior portion of the foot) is visible. The propodial appendages suspend transparent mucus sheets, which the flow expands into a semispherical shape. Horizontal field width is 5 cm. 

Unexpectedly, we found suspension-feeding O. columellaris also under permanent water cover in lower beach zones. Here the snails expanded their mucus sheets when the incoming water flow stopped, and kept them exposed until the reversed seaward flow had picked up enough speed to transport sediment grains of significant size. This behavior has not been described before, presumably because its observation requires clear visibility in the lower swash zone, a rare condition. We observed it only with wave amplitudes below 20 cm around noon, when the sun stood high providing excellent illumination.

Olivella columellaris readily changes its migration behavior where the timing of the tidal progression is modified, as we repeatedly observed around artificial structures such as groins or jetties. We decided to document a representative case of such behavioral modification quantitatively. At our study site, several coast-parallel lines of boulders had been placed just below the high-water line to prevent erosion. At ebb tide, water from pools that had formed above these boulders streamed seaward, carving narrow channels of 1 to 4 cm depth into the tidal plain (Fig. 2a). Generally, the snails remained in these channels longer than on the adjacent plain. Suspension feeding correlated with seaward flow, and because the latter was extended in the channel compared to the plain, snails in the channel could feed longer (example shown in Fig. 2b, c). The animals on the plain responded to the comparatively poor feeding conditions at their location, and utilized the backwash to sail seaward while the snails in the channel did not (Fig. 2c). To evaluate the consequences of this behavior, we determined the density of the snails at the points defined in Figure 2b at the end of the observational period documented in Figure 2c. At levels corresponding to the upper channel, there were no snails left on the plain (P1, P2), while several hundred individuals per square meter were found in the channel (C1, C2; Fig. 2d). At the level at which the time-courses (Fig. 2c) had been determined, about 100 ind m^-2 were still present on the plain (P3) but in the channel the density was over 10-fold higher (C3; Fig. 2d). Below the channel's lower end, densities were similar in the channel and plain (C4, C5, P4, P5; Fig. 2d). Similar observations were made in numerous flow channels, and periods of continuous feeding within channels in excess of 10 min were recorded repeatedly (Fig. 2e).

Figure 2 Opportunistic suspension feeding of Olivella columellaris. (a) Flow channels forming at retreating tide below a row of boulders; the horizontal width of the photograph is 19 m along its bottom edge. (b) Sketch (drawn to scale) of a flow channel in which the behavior of the snails was analyzed; the locations of 10 collection points for density determination are indicated. (c) Time-courses of water currents (lw, landward; sw, seaward) and feeding activity in the channel and on the plain at the positions marked C3 and P3 in (b). The asterisk indicates a sailing event: numerous snails emerged on the plain and sailed seaward with the backwash. (d) Densities of O. columellaris at the points defined in (b) at the end of the time-course in (c). (e) Time-course as in (c), showing almost continuous feeding activities in a channel over more than 20 min. Large asterisks: sailing events involving >20 snails; small asterisks: sailing events involving <20 snails. 

Discussion

The existence and significance of endogenous tidal rhythms in some intertidal organisms cannot be doubted (^{Naylor 2010}). It does not follow, however, that the behavior of intertidal creatures generally is controlled by endogenous circatidal clocks as implied, for example, by ^{de la Iglesia and Johnson (2013)}. The assumption that an animal might benefit from or even require such an endogenous clock is plausible if the animal's life style is tightly related to the progression of the tides. This seemed to be the case in O. columellaris, which had appeared to depend on its ability to migrate with the backwash zone for its nutrition. Our observation of feeding animals in lower beach zones suggests that O. columellaris is capable of suspension feeding at any depth, if only the sediment surface experiences water flows of appropriate velocities. Thus, the linkage between feeding and tidal migrations in this species is not as strict as previously assumed. Moreover, animals in flow channels performed suspension feeding opportunistically as long as water currents permitted, and did not leave their locations unless feeding conditions deteriorated. Consequently, their tidal migrations were delayed, or even canceled. We have made analogous observations in numerous similar cases along the coast of north Peru. Our findings support the notion that tidal migrations in mollusks result from direct behavioral responses to changing conditions (^{McLachlan et al. 1979}). If an endogenous circatidal clock should exist in O. columellaris (^{Vanagt et al. 2008}), one would have to conclude that its influence on the behavior of the snails is easily overruled by external stimuli.