Functional Ecology 1997 11, 472–483 A biomechanical hypothesis explaining upstream movements by the freshwater snail Elimia A. D. HURYN*‡ and M. W. DENNY† *Aquatic Biology Program, Department of Biological Sciences, University of Alabama, Tuscaloosa, Alabama 35487, USA, Department of Zoology, PO Box 56, University of Otago, Dunedin, New Zealand and †Hopkins Marine Station, Department of Biological Sciences, Stanford University, Pacific Grove, CA 93950–3094, USA Summary 1. Many taxa of freshwater invertebrates show active upstream movements, particularly the snails. Hypotheses explaining this behaviour invoke the search for food or space, compensation for drift, avoidance of predation and hydrodynamic effects. The pervasiveness of upstream movements among remote lineages of snails (two subclasses, three orders, 10 families), however, suggests that snails may move upstream for mechanical rather than adaptive reasons. 2. It is proposed that upstream movements by snails are a function of torque on the snail’s foot generated by hydrodynamic drag on the shell. When subject to high broadside drag-forces on their shells, snails are able to reduce torque and stabilize orientation only by directing their anterior aspect upstream. 3. Movements of the freshwater pleurocerid snail Elimia were studied by following marked free-ranging individuals in six streams in Alabama, USA (four species, eight populations). 4. Populations showed either no net movement (two streams) or significant upstream movements ranging to a mean of ≈ 40 m over a 3-month period (four streams). Movement patterns were stream specific rather than species or population specific. Within populations showing significant upstream movements, snails with shell lengths ≤ 10 mm showed little net movement. Larger snails showed movements from 0 to > 200 m upstream. 5. A torque-constrained random walk model was used to perform a post hoc test of the hypothesis that upstream movements were a function of torque on the snail’s foot generated by hydrodynamic drag on the shell. The model predicted upstream and sizedependent movement patterns that approximated those observed for snails in the field. Key-words: Biomechanics, hydrodynamics, snails, streams, torque Functional Ecology (1997) 11, 472–483 Introduction Movement patterns of invertebrates in streams have attracted continuous interest for the better part of this century. Although most effort has been concentrated on downstream movements as drift (Waters 1972; Brittain & Eikeland 1988), active upstream movements have also been documented for many taxa (Söderström 1987; Bergey & Ward 1989), particularly the snails (Table 1). Hypotheses explaining upstream movements by snails have invoked the search for food or space (Paulini 1963), compensation for downstream drift (Carpenter 1928; Schneider & Frost 1986), avoidance of predation (Schneider & Lyons 1993) and con© 1997 British Ecological Society ‡Current address: Department of Applied Ecology and Environmental Sciences, University of Maine, 5722 Deering Hall, Orono, Maine 04469–5722, USA. straints imposed by body architecture and hydrodynamics (Haynes, Taylor & Varley 1985). Although numerous, none of these hypotheses has been rigorously examined and all remain conjectural. Upstream movements have been documented for a remarkably diverse assemblage of snails in temperate and tropical streams worldwide (Table 1). This assemblage, spanning two subclasses, three orders and 10 families is the result of repeated invasions of streams by marine ancestors over geological time (Hutchinson 1967; Graham 1985). From an evolutionary perspective, the character ‘upstream movement’ that defines the assemblage must logically be either a plesiomorphy derived from ancestral marine forms and not a specific adaptation to life in freshwater streams, or have been derived independently in different lineages. The pervasiveness of upstream movements among 472 473 Upstream movements by snails Table 1. Summary of snail families containing freshwater taxa (Davis 1982) and reports of upstream movements. Grades of evidence are: Indirect-a = movements inferred from appearance of animals in reaches where they previously where not observed, Indirect-b = movements of genetically distinct subpopulations; Direct-a = observation of individuals actively moving upstream, Direct-b = movements of marked individuals, Direct-c = movements of radionuclides, Direct-d = flume studies; ? = source of evidence not stated; – = not studied Subclass Superfamily Family Prosobranchia Neritinacea Nertinidae Neritina latissima N. granosa Hydrocenidae Viviparacea Ampullariidae Vivparidae Campeloma sp. C. decisum Valvatacea Valvatidae Cerithiacea Pleuroceridae Elimia catenaria E. clavaeformis E. proxima E. proxima E. semicarinata Pleurocera sp. P. acuta Melanopsidae Syrnolopsidae Thiaridae Melanoides tuberculata Thiara granifera Rissoacea Assimineidae Baicaliidae Bithyniidae Hydrobiidae Potamopyrgus antipodarum P. jenkinsi Cochliopina tryoniana Lepyriidae Pomatiopsidae Pyrguliidae Stenothyridae Buccinacea Buccinidae Volutacea Marginellidae © 1997 British Ecological Society, Functional Ecology, 11, 472–483 Pulmonata Acroloxidae Ancylidae Chilinidae Latiidae Lymnaeidae Lymnaea pereger Physidae Physa sp. Physella integra Planorbidae Biomphalaria glabrata B. glabrata Location Evidence Source Costa Rica Hawaii – Direct-a Direct-a – Schneider & Lyons 1993 Ford & Kinzie 1982 – – – – Illinois Michigan Indirect-a Direct-b,d Shelford 1913 Bovjberg 1952 – – – Georgia Tennesee North Carolina Virginia Kentucky Illinois Kentucky – – Direct-d Direct-b Direct-b Indirect-b Direct-b Indirect-a Direct-b – – Kreiger & Burbank 1976 Burris et al. 1990 Crutchfield 1966 Dillon 1988 Mancini 1978 Shelford 1913 Houp 1970 – – Martinique Puerto Rico Indirect-b Direct-a Pointier et al. 1993 Chaniotis et al. 1980 – – – – – – – – – New Zealand Britain Costa Rica – SE Asia – SE Asia Indirect-b Direct-b,d Direct-a – ? – ? Wallace 1992 Haynes et al. 1985 Schneider & Lyons 1993 – Davis 1982 – Davis 1982 – – – – – – – – – – – – – – – – – Britain Indirect-a Hynes 1960 Michigan New Mexico Direct-c Indirect-a Ball, Wojtalik & Hooper 1963 Noel 1954 Brazil Puerto Rico Direct-b,d Direct-d Paulini 1963 Jobin & Ippen 1964 474 A. D. Huryn & M. W. Denny remote lineages of freshwater snails suggests that this behaviour may result from an architectural constraint imposed by the general gastropod morphology rather than being a specific adaptation to existence in freshwater streams; some taxa may move upstream for mechanical rather than adaptive reasons. In this paper, upstream movements by the freshwater snail Elimia (Pleuroceridae) are explained by invoking torque generated by the interaction of hydrodynamic drag with gastropod architecture. The utility of this mechanical hypothesis is then demonstrated by a simulation model showing how torque may provide a simple explanation for the widespread and well-documented occurrence of upstream movements by snails. STUDY SITES Six streams were selected for study, three each from two major physiographic provinces of Alabama (USA), Valley & Ridge and Piedmont. All study reaches are of moderate gradient (< 2%) and have channels characterized by long reaches of bedrock (> 50 m) punctuated by short riffles. Average width of channels at base flow ranged from 3 to 9 m. Average discharge based on semi-monthly measurements during 1990–91 ranged from 76 to 210 l s–1. Channels of study sites in the Valley & Ridge province (Hendrick Mill Branch – Blount County; Alligator Creek, Rocky Branch – Bibb County) are largely carbonate bedrock. Channels of study sites in the Piedmont province (Choccolocco Creek, Marys Creek, South Fork Terrapin Creek – Winston County) are primarily phyllite bedrock. The snail Elimia cahawbensis occurs at Alligator Creek, Hendrick Mill Branch and Rocky Branch. Elimia carinifera and E. variata occur only at Hendrick Mill Branch and Alligator Creek, respectively. Elimia fascinans occurs at Choccolocco Creek, Marys Creek and Terrapin Creek. Additional information about the life history and ecology of Elimia is contained in Huryn, Koebel & Benke (1994) and Huryn, Benke & Ward (1995). Fig. 1. Schematic diagram showing a broadside projection of Elimia cahawbensis where c.a. refers to the centre of projected area, c.m. refers to the perpendicular axis of the columellar muscles and ‘lever’ is the distance from centre of projected area to the perpendicular axis of the columellar muscles. The lever is the distance over which torque will be generated because of drag forces on the shell when perpendicular to flow. Methods and materials MOVEMENTS OF SNAILS Approximately 1000 snails (range = 747–1139) were collected from each population and individually marked using numbered microtags (Freilich 1989). Once marked, snails were released midway between the beginning and end of a 100-m study reach. All releases were made between 18 March and 8 April 1990. Following the release, surveys for marked snails were made at all sites at ≈ 3-month intervals for one year (June/July 1990; October 1990; February 1991; April 1991). Surveys began at the downstream end of the study reach (50 m below the release point) and continued upstream until there appeared to be little probability of finding additional marked snails. This procedure was repeated a minimum of three times on each date. When a marked snail was found, its number, position and shell length was recorded and it was released. Additional detail concerning methods is contained in Huryn et al. (1994, 1995). MECHANICAL FORCES AND SNAIL SHELLS Torque is generated when a force is exerted on a lever arm; it tends to cause an object to rotate. Since the columellar muscles of Elimia are offset from the centre of the shell’s lateral projected area and drag forces act near the centre of area, drag on the shell is likely to impose torque on the columellar muscles and foot (Fig. 1). It is proposed that such torque will cause a snail in flowing water to act much like a weather vane – that is, sufficient drag forces on the snail’s shell will cause it to rotate until its anterior end faces upstream. Once in such a position the frontal projected area is symmetrical about the axis of the collumellar muscles, the torque is eliminated and the orientation is stable. When exposed to appropriate flow velocities this imposed shell orientation will effectively steer snails in an upstream direction. If the posterior rather than the anterior aspect of the shell is projected into the current (Fig. 1), torque may similarly be eliminated. This position is unstable, however. Empty shells of the four species of Elimia were filled with wax (to simulate the presence of the snail’s body) and mounted on a circular plate (7-cm diameter). This plate was in turn held flush with the wall of a wind tunnel (24-cm2 cross section) by a drag transducer (Denny 1989, 1995). The snail shell was then subjected to a free-stream wind velocity of 28·8 m s–1, and the resulting force on the shell and plate was recorded. In a separate experiment, friction drag on the bare mounting plate was measured and subtracted from the overall force to yield the force on the shell. Each shell was tested in two orientations, one broadside to flow and the second with the shell’s anterior end upstream. The coefficient of drag (Cd) was calcu- 475 Upstream movements by snails lated as a function of the force experienced by the shell (N) in either frontal or broadside projection by the following equation (Vogel 1994): Cd = (2F)/(DV2A), eqn 1 where F is drag force (N, measured by the drag transducer), D is the density of air (1·20 kg m–3 at 20·5 °C), V is the velocity of the air (m s–1) and A is the projected area (m2). Projected areas were obtained by digitizing scaled photographs of the shells in the anterior or broadside orientation. Given the Cd, the projected area of a shell, and current speed and density of water, drag forces can be estimated by solving for F in equation 1. Reynolds numbers (Re) of shells in the wind tunnel (104) were higher than the maximum estimated for free-ranging snails at the study sites (103). Measurements of Cd for marine snails show little variation as a function of Re (range = 103–105; Denny 1995), however, and it was assumed that Cds were constant over the range of Re values experienced by snails in the study streams. The centre of area for the broadside projection of shells for each species was located by hanging cutouts of tracings of the perimeter of photographs of each shell from a pin at three or more separate positions. Vertical lines drawn from each pin position intersected at the approximate centre of area. The distance between the centre of area for the broadside projections and the axis of the columellar muscles (approximated as the centre of the aperture) was used to estimate the lever arm over which the drag on the shell will act (Fig. 1). Torque on the columellar muscles of Elimia was estimated as the product of the drag on the shell (N) and length of the lever arm (m). It is assumed that torque on the collumellar muscles will influence direction of movement by rotating the snail on its foot. The mechanical effects of hydraulic forces on snail shells should be scaled to an index of an individual’s ability to resist torque, which may change with size. For convenience we calculated torque as a function of foot area (Denny, Daniel & Koehl 1985). Foot area, measured from snails relaxed with menthol, was regressed against shell length to provide predictive equations (P < 0·05, Table 2). SNAIL ORIENTATION AND TORQUE The relationship between water velocity and orientation of individual snails was measured in Hendrick Mill Branch (E. cahawbensis, E. carinifera) and Marys Creek (E. fascinans) using plastic glitter (silver) as tracer particles. Glitter was released into the stream and photographed with a 35-mm SLR camera at 1/15, 1/30 or 1/60 s as it passed over targeted snails. A ruler placed in the photographic field provided scale needed to estimate water velocity and shell dimensions. Water velocity was calculated from the scaled length of streaks caused by moving glitter and the camera’s shutter speed. No attempt was made to assign particles to a specific vertical position within the water column. With some experience, however, glitter could be released so that most particles passed within a few centimetres of targeted snails. Snails with the anterior aspect of their shells oriented directly into the current were assigned values of 0°. Deviation of the snail’s longitudinal axis from 0° was measured with a protractor. For each snail, the torque that would be experienced if the shell was broadside to the current was estimated. These data were used to construct plots of torque m–2 foot area vs snail orientation. TORQUE AS A DETERMINANT OF SNAIL MOVEMENT PATTERNS Measurements of water velocity were used to construct a two-dimensional matrix of cells that represented the diversity and spatial distribution of actual hydraulic conditions in each study reach. Measurements were made at 1-m intervals across transects placed perpendicular to the stream channel. The transects were located at 2-m intervals parallel to the study reach. Water velocity was measured ≈ 2 cm above the substrata with an electromagnetic flow meter (Flo-Mate 2000, Marsh-McBirney, Inc., Frederick, MD). Although measurements of velocity were made ≈ 2 cm above the substrata, these data are relevant to drag forces actually experienced by snails. The boundary layers in the study streams are probably tur- Table 2. Summary of parameters used to calculate torque/m2 foot area for four species of Elimia. Broadside and anterior refer to coefficients of drag (Cd) measured for single shells at each orientation. For comparative purposes, Cds estimated for spheres under similar hydraulic conditions (Denny 1993) are given after the diagonal. The broadside projection and foot area can be estimated for each species by the following equation (P< 0·05): area (m2) = cLb, where L is shell length (m). Lever coefficient is the proportion of total shell length that occurs between the main axis of the columellar muscles and the centre of the shell’s projected area © 1997 British Ecological Society, Functional Ecology, 11, 472–483 Species Broadside/sphere Cd Anterior/sphere Cd Lateral projection (m2) c b Foot area (m2) c b E. cahawbensis E. carinifera E. fascinans E. variata 0·51/0·45 0·43/0·45 0·46/0·44 0·38/0·44 0·50/0·43 0·47/0·43 0·42/0·43 0·35/0·44 0·2089 0·2748 0·3990 0·3636 0·8903 0·0162 0·2914 0·8549 1·9057 2·0509 2·0479 2·0022 2·3463 1·6338 2·0144 2·2290 Lever coefficient 0·24 0·22 0·27 0·23 476 A. D. Huryn & M. W. Denny © 1997 British Ecological Society, Functional Ecology, 11, 472–483 bulent as a result of flow separation caused by virtually infinite levels of bed roughness upstream (Hart, Clark & Jasentuliyana 1996). Turbulent boundary layers are characterized by time-averaged velocity gradients. Energetic eddies, however, will regularly reach almost to the substratum and bring with them velocities that are near those of the mainstream. Hart et al. (1996), for example, measured fluctuations of water velocity as great as 80 cm s–1 over intervals of < 0·1 s within 2 mm of the benthos in a turbulent stream. Except for individuals that are small enough to hide in the viscous sublayer (in the order of 0·1 mm), snails are likely be subject to intermittent velocities approximately equal to the mainstream. These intermittent velocities will place the largest torque on the snails, and are therefore the ones that are most critical in determining hydraulic-dependent behaviour (Denny 1994). Based on this reasoning, we assume that measurements of the mainstream flow ≈ 2 cm above the bed provide realistic information about the flow environment closer to the substratum. Water velocity matrices constructed for each study stream formed the basis for a computer simulation of how torque might act to steer movements of snails in streams. The simulation consisted of placing a virtual snail of specified length (5, 10, 15 or 20 mm) into the water velocity matrix at a cell equivalent to the actual release position in each stream (‘0,0 m’) and allowing it to step from cell to cell in a random direction unless the snail entered a cell where torque m–2 foot area exceeded an empirical threshold derived from plots of torque m–2 foot area vs snail orientation (see ‘snail orientation and torque’, above). In cells where torque exceeded the empirical threshold, the next step was to the adjacent cell upstream. Each snail was allowed to take 720 steps (= 720 ‘m’). This value was selected on the basis of how far a specimen of Elimia might conservatively be expected to move over a 3-month summer period (Beetle 1970; Burris, Bamford & Stewart 1990; A. D Huryn, personal observation). To allow lateral steps and longitudinal steps to be of equal distance, water velocities were assigned to each 1-m2 cell following the assumption that flow conditions were constant between each 2-m transect. After each series of steps was completed, the final position of the snail was recorded. This procedure was repeated 1000 times for each of the four length classes. In some cases virtual snails rapidly moved off the original matrix, and therefore matrices were extended by sequentially repeating randomly selected transects of water velocity measurements in an upstream or downstream direction. Results of the model were in the form of frequency distributions of probable final positions for different length classes and species of snails for each study site. The success of the model was assessed by comparing actual distributions of marked snails with simulated distributions. Since we were originally interested in examining only the effect of torque on movement patterns, snails of all sizes were assumed to move at equal velocities. However, the model was later modified to incorporate size-dependent differences in movement speed as indicated for Elimia by Krieger & Burbank (1976). It is emphasized that the objective of the model was heuristic rather than predictive per se, and is offered to show how torque may influence movements of snails in streams. Results MOVEMENTS OF MARKED SNAILS Numbers of snails recaptured during the initial survey ranged from 4 to 13% of the total individuals marked among the eight populations. However, the proportion of snails captured more than once over all successive surveys was also low, ranging from 3 to 19% of the total recaptured (mean = 9%). The low incidence of repeated recaptures of unique snails indicates that marked snails were often passed over rather than having moved exceptional distances or losing tags. After 3 months of free-ranging, significant netupstream movement was observed for every population except E. cahawbensis at Rocky Branch and E. fascinans at Terrapin Creek (one sample t-test, P < 0·05, Figs 2 and 3). Maximum upstream movements of individual snails ranged from 58 m for E. fascinans at Choccolocco Creek (Fig. 3) to 200 m for E. carinifera at Hendrick Mill Branch (Fig. 2). Maximum downstream movements ranged from 3 m at Choccolocco Creek to 47 m for E. fascinans at Marys Creek. However, few snails were recovered at positions substantially downstream of the release point (Figs 2 and 3). Furthermore, no snails were ever found during additional intensive field work that was conducted several hundred metres downstream from the release point in each stream throughout the duration of the study. Although this work was not directly related to the snail project, field personnel were instructed to report marked snails. Average position at recapture for E. cahawbensis in Alligator Creek and Hendrick Mill Branch, and for E. carinifera in Hendrick Mill Branch was significantly further upstream than for other populations (ANOVA, Tukey’s HSD, P < 0·05). These differences are apparently not species specific. For example, after 3 months, populations of E. cahawbensis showed average positions of 1, 28 and 39 m at Rocky Branch, Hendrick Mill Branch and Alligator Creek, respectively (Figs 2 and 3). This range brackets the variation observed among all eight populations. Snails with lengths ≤ 10 mm showed little movement from the point of release compared to larger snails (Figs 2 and 3). Mean position at recapture was not significantly influenced by date at Alligator Creek, Hendrick Mill Branch and Terrapin Creek (ANOVA, P > 0·05). However, this factor was significant at Rocky Branch, Choccolocco Creek and Marys Creek (P < 0·05); apparently because of a small but consistent down- 477 Upstream movements by snails © 1997 British Ecological Society, Functional Ecology, 11, 472–483 Fig. 2. Left. Plots of the position and size observed for four populations of free-ranging marked snails released in Hendrick Mill Branch and Alligator Creek. Large points indicate positions ≈ 3 months following release. Small points indicate positions of recaptured snails ≈ 6 months to 1 year following release. Panels are arranged in order of decreasing net upstream movement. The arrow along the y-axis indicates the length at which ash-free dry mass is ≈ 2·2 mg. This biomass threshold has been shown to distinguish different levels of expected food satiation by Elimia (see text). Right. Plots of the frequency distributions of different positions attained by snails of different lengths (5–20 mm) predicted by a torque-constrained random walk model. Each plot represents 1000 iterations of the model. 478 A. D. Huryn & M. W. Denny Fig. 3. Left. Plots of the position and size observed for four populations of free-ranging marked snails released in Choccolocco Creek, Marys Creek, Terrapin Creek and Rocky Branch (see caption for Fig. 2 for additional explanation). Right. Plots of the frequency distributions of different positions attained by snails of different lengths (5–20 mm) predicted by a torque-constrained random walk model. Each plot represents 1000 iterations of the model. © 1997 British Ecological Society, Functional Ecology, 11, 472–483 stream movement of average position during winter. For example, the average position observed at Choccolocco Creek ranged from 12 m upstream of the release point in October to 5 m in February. A similar pattern was observed at Rocky Branch (3–0 m) and Marys Creek (10–7 m). Snails were apparently capable of the maximum movements observed during the 1-year study within 3 months following release (Figs 2 and 3). Compared with movement activity during the initial 3-month period of observation 479 Upstream movements by snails (spring–summer), movement activity of Elimia may have been limited by declining water temperature (e.g. Kreiger & Burbank 1976). Seasonal variation in movement activity may explain the apparently lower rates of movement following spring–summer (Figs 2 and 3). DRAG Coefficients of drag were measured for a shell from each of the four snail species (Table 2). Measurements of total drag forces on each shell were repeated at least three times. Cds calculated for each shell represent the average of these measurements; standard errors were always < 15% of the mean. Cds for broadside projections were similar among species. For example, the standard error among shells from all species was ≈ 7% of the mean (range = 0·38–0·51, mean = 0·45, standard error = 0·03; Table 2). Cds for anterior projections were similar to those calculated for broadside projections (range = 0·35–0·50, mean = 0·44; Table 2). For either projection, Cds of snails were similar to those of spheres with the same Re (0·43–0·44; Table 2). © 1997 British Ecological Society, Functional Ecology, 11, 472–483 SNAIL ORIENTATION AND TORQUE Orientation and torque m–2 foot area was measured for E. cahawbensis (n = 55), E. carinifera (n = 58) and E. fascinans (n = 64) in Hendrick Mill Creek and Marys Creek. It was assumed that snails oriented into the current with the long axis of their shells within 45° of a vector parallel to the direction of flow were moving upstream; snails aligned within 45–135° of the vector were moving in a direction perpendicular to flow; snails aligned 135–180° were moving downstream. A plot of shell orientation by larger snails (> 10 mm length) vs torque m–2 foot area (Fig. 4) shows that E. cahawbensis and E. carinifera in Hendrick Mill Branch orient in an upstream direction under flow conditions that would generate a broadside torque ≥ 0·15 N m–2 foot area. Elimia fascinans in Marys Creek orient in an upstream direction under conditions that would generate a broadside torque ≥ 0·25 N m–2 foot area. Threshold torque was not measured for E. variata so a conservative torque of 0·25 N m–2 foot area was used for modelling purposes. With few exceptions, snails ≤ 10 mm in length Fig. 4. Plot of torque m–2 foot area (N m–2) for broadside shells vs shell orientation for two length classes (≤ 10 mm, lower panel; > 10 mm, upper panel) of Elimia cahawbensis and E. carinifera in Hendrick Mill Branch and E. fascinans in Marys Creek. Snails are assumed to be moving upstream when oriented against and within 45° of the direction of flow. 480 A. D. Huryn & M. W. Denny were not observed under flow conditions that generated torques > 0·15 N m–2 foot area. When shells were exposed to torques below the apparent threshold, orientation of E. cahawbensis and E. fascinans was not significantly different from a random distribution (P > 0·05, χ2 goodness of fit for circular frequency distributions; Zar 1984). However, the orientation of E. carinifera departed significantly from a random distribution because of a bias toward individuals oriented in neutral or upstream directions (P < 0·05, Fig. 4). FLOW CONDITIONS AMONG STREAMS Point measurements of water velocity among study reaches varied from 0 to ≈ 0·7 m s–1 (Fig. 5). In general, Valley & Ridge streams (Alligator Creek, Hendrick Mill Branch, Rocky Branch) had a higher range of velocities than Piedmont Streams (Choccolocco Creek, Marys Creek, Terrapin Creek; Fig. 5). Water velocities covering ≈ 20% of the stream bed were capable of generating torques exceeding the apparent threshold for 20-mm long specimens of E. carinifera in Hendrick Mill Branch (e.g. 0·2 m s–1, Fig. 5). At the other extreme, the maximum velocity in Terrapin Creek was well below that required to produce threshold torques for 20-mm long specimens of E. fascinans (Fig. 5). TORQUE-CONSTRAINED RANDOM WALK MODEL Results of the torque-constrained random walk model predicted that stream and snail size should differentially influence movement patterns of snails (Figs 2 and 3). The largest net-upstream movement was predicted for E. carinifera and E. cahawbensis in Hendrick Mill Branch and E. cahawbensis in Alligator Creek (Fig. 2). The Hendrick Mill Branch model predicted complete lack of downstream movements because of super-threshold torques imposed on all size classes by rapid water velocities (+ 0·7 m s–1) through a bedrock chute below the release point. Consequently, all snails were forced to proceed upstream. The smallest net upstream movements were predicted for E. fascinans in Terrapin Creek, where no net movements were predicted at all (Fig. 3). Size-dependent movement patterns were clearly indicated among model results for E. carinifera (Hendrick Mill Branch), E. variata (Alligator Creek) and E. fascinans (Choccolocco Creek) (Figs 2 and 3). Frequency distributions predicted for length classes > 10 mm were strongly skewed in an upstream direction, whereas frequency distributions for snails ≤ 10 mm were, with the exception of E. carinifera (see comments concerning Hendrick Mill Branch above), symmetrically distributed about the release point. Size-dependent movements were either not apparent (E. cahawbensis, all sites; E. fascinans, Terrapin Creek) or only vaguely apparent (E. fascinans, Marys Creek) among the remaining populations (Figs 2 and 3). Since size-dependent differences in snail behaviour were not incorporated into the original model, size-dependent movement Fig. 5. Proportion of stream bed covered by different water velocities for six streams. Black bars indicate water velocities predicted to produce torques resulting in upstream movements for 20-mm long snails. The approximate maximum length of Elimia is ≈ 20 mm. 481 Upstream movements by snails patterns are because of differences in the torque and the ratio of torque to foot area experienced among length classes. COMPARISON OF MODEL RESULTS WITH FIELD OBSERVATIONS Average positions for snails predicted by the torqueconstrained random walk model were significantly correlated with field observations (Fig. 6). Discussion The torque-constrained random walk model was successful in approximating patterns of upstream movement observed among populations of Elimia in the field (Fig. 6). The model also showed how differences in channel form may underlie much of the variability observed among snail populations. For example, the lack of movement of snails to reaches below the release point in Hendrick Mill Branch, as both predicted by the model and observed in the field, is apparently the result of a high-velocity chute ≈ 1–2 m below the release point. Although minor in spatial extent, model results indicated that this channel features greatly influenced movements of snails because it imposed super-threshold torques on all length classes. Consequently, random walks by all snails were forced to progress upstream from the release point, which resulted in substantial net movements compared with other sites. Other effects of channel form among streams were indicated by both field observations and model results for different populations of E. fascinans. For example, lack of upstream movement and the symmetrical distribution of individuals about the release point for E. fascinans in Terrapin Creek is attributable to lack of water velocities required to exceed torque thresholds (Figs 3 and 5). In contrast to Terrapin Creek, moderate levels of upstream movement by © 1997 British Ecological Society, Functional Ecology, 11, 472–483 Fig. 6. Correlation of observed positions of snails following 3 months of free-ranging movements with model predictions. The diagonal line indicates the trajectory of points expected if observed positions are equal to predicted positions. larger length classes of E. fascinans in Choccolocco Creek is attributed to patches of water velocity that exceed torque thresholds (Figs 3 and 5). The torque-constrained random walk model predicted that snails will accumulate in reaches with currents below threshold velocities, but pass rapidly through reaches with currents above threshold velocities. This is reflected in the complex longitudinal distributions of individuals observed for length classes that progressed rapidly upstream (Figs 2 and 3). Although not tested in the present study, Gore (1983) showed that in the Buffalo River (Arkansas) large individuals of Elimia potosiensis were significantly more abundant in regions of low current velocity compared with small individuals. Since the present study indicates that large snails may at times be subject to larger effective torques than smaller snails, the distribution reported by Gore (1983) is consistent with our model. Although comparison of model results generally approximated field observations (Figs 2, 3 and 6), size-dependence predicted by the model was usually not as marked as that observed in the field. In particular, the extremely clumped distribution of small snails (e.g. < 10 mm; Figs 2 and 3) was clearly not predicted by the model. The striking contrast between movement patterns of large vs small snails may be attributed to a number of factors. First, small snails may simply avoid flow conditions that generate threshold torques as suggested by Fig. 4. Second, small snails may experience hydraulic forces different from those experienced by large snails because they experience lower Re values or because they are more strongly influenced by boundary layer phenomena (Vogel 1994). Third, small snails may not move at the same rate as large snails. For example, Krieger & Burbank (1976) showed that at 20 °C, specimens of E. catenaria < 10 mm in length moved 4–12 times slower than specimens > 10 mm in length. If snails with lengths < 10 mm are allowed to travel 10 times slower than snails > 10 mm, model results that closely mimic observed patterns are obtained (cf. Figs 2, 3 and 7). Finally, since foraging-related movements by Elimia are apparently size dependent, different levels of food limitation among sites may be involved. Hill, Weber & Stewart (1992) showed that under conditions of low food availability, small Elimia (< 2·2-mg ash-free dry mass (AFDM)) tended to be less food limited than large Elimia (> 2·2-mg AFDM). Food-limited snails and other grazing invertebrates have been shown to move at higher rates and more randomly than when food-satiated (Calow 1974; Hart & Resh 1980). Growth rates of snails among the Alabama study sites were strongly density dependent, apparently owing to variation in food availability (Huryn et al. 1995). The lowest growth rates were observed at Alligator Creek and Hendrick Mill Branch (‘high snail-biomass’, e.g. 3–5-g m AFDM m–2; Huryn et al. 1995). The length threshold that distin- Fig. 7. Plots of the frequency distributions of positions attained by E. cahawbensis of different lengths (5–20 mm) in three streams as predicted by a torque-constrained random walk model (to be compared with Figs 2 and 3). Each plot was derived from 1000 iterations of the model. Individuals 5 and 10 mm in length were moved at a rate 10 times slower than larger individuals (e.g. 72 1-m steps vs 720 1-m steps over a 3-month period). guishes the different movement patterns for large and small snails at these sites compares favourably with the biomass threshold of 2·2-mg AFDM/snail used by Hill et al. (1992) (Fig. 2), which suggests that food limitation may well be a factor underlying differences in movement patterns. As would be predicted, the distinction between movement patterns of large and small snails is less obvious at the remaining ‘low snail-biomass’ sites (e.g. < 2-g m AFDM m–2, Huryn et al. 1995) (Figs 2 and 3). Size, food limitation and hydrodynamics may all interact to influence movement patterns of snails. For example, DeNicola & McIntire (1991) showed that under conditions of high periphyton availability, the pleurocerid snail Juga foraged mainly on substrata that were sheltered from high flow velocities. However, when periphyton biomass was reduced to low levels, snails moved onto exposed substrata where they would more likely be subject to hydrodynamic drag and consequently, torque. The results of DeNicola & McIntire (1991), Hill et al. (1992) and the torque-constrained random walk model suggest that snails of different sizes should show different movement patterns under different levels of food limitation. Although factors contributing to upstream movements by freshwater snails are undoubtedly complex and multivariate, we suggest that the torque-constrained random walk model provides a testable null hypothesis that may be used to examine other explanations for this general phenomenon. Acknowledgements © 1997 British Ecological Society, Functional Ecology, 11, 472–483 We thank Mr Charles Williams, Mr M. C. Allgood, Jr and the Cahaba Hunting Club for allowing access to Alligator Creek, Hendrick Mill Branch and Rocky Branch, respectively. J. E. Freilich provided advice on the preparation of micro-labels and F. G. Thompson provided names for the snails. Field and laboratory assistance was provided by J. W. Koebel, A. C. Benke, J. W. Converse, B. Crossen, V. M. Butz Huryn, K. Miller, K. Petris and K. Suberkropp. An early draft of this paper was greatly improved by advice from S. D. Gaines, B. Statzner and an anonymous reviewer. This research was supported by grants from the National Science Foundation (BSR 88–18810 to A. K. Ward, G. M. Ward, A. C. Benke, R. J. Donahoe and J. M. Harlin), and a grant in aid from the University of Otago (New Zealand) to A.D.H. This paper is contribution 240 to the Aquatic Biology Program, University of Alabama. References Ball, R.C., Wojtalik, T.A. & Hooper, F.F. (1963) Upstream dispersion of radiophosphorous in a Michigan trout stream. Papers of the Michigan Academy of Science, Arts, and Letters 48, 57–64. Beetle, D.E. (1970) A photographic record of snail activity. Sterkiana 39, 1–7. Bergey, E.A. & Ward, J.V. (1989) Upstream–downstream movements of aquatic invertebrates in a Rocky Mountain stream. Hydrobiologia 185, 71–82. Bovbjerg, R.V. (1952) Ecological aspects of dispersal of the snail Campeloma decisum. Ecology 33, 169–176. Brittain, J.E. & Eikeland, T.J. (1988) Invertebrate drift – a review. Hydrobiologia 166, 77–93. Burris, J.A., Bamford, M.S. & Stewart, A.J. (1990) Behavioral responses of marked snails as indicators of water quality. Environmental Toxicology and Chemistry 9, 69–76. Calow, P. (1974) Some observations on the locomotor strategies and their metabolic effects in two species of freshwater gastropods, Ancylus fluviatilis Müll. & Planorbis contortus Linn. Oecologia 16, 149–161. 483 Upstream movements by snails © 1997 British Ecological Society, Functional Ecology, 11, 472–483 Carpenter, K.E. (1928) Life in Inland Waters with Especial Reference to Animals. Sidgwick & Jackson, London. Chaniotis, B.N., Butler, J.M., Ferguson, F.F. & Jobin, W.R. (1980) Bionomics of Tarebia granifera (Gastropoda: Thiaridae) in Puerto Rico, an Asiatic vector of Paragonimiasis westermani. Caribbean Journal of Science 16, 81–90. Crutchfield, P.J. (1966) Positive rheotaxis in Goniobasis proxima. Nautilus 79, 80–86. Davis, G.M. (1982) Historical and ecological factors in the evolution, adaptive radiation, and biogeography of freshwater mollusks. American Zoologist 22, 375–395. DeNicola, D.M & McIntire, C.D. (1991) Effects of hydraulic refuge and irradiance on grazer–periphyton interactions in laboratory streams. Journal of the North American Benthological Society 10, 251–262. Denny, M.W. (1989) A limpet shell shape that reduces drag: Laboratory demonstration of a hydrodynamic mechanism and an exploration of its effectiveness in nature. Canadian Journal of Zoology 67, 2098–2106. Denny, M.W. (1993) Air and Water: The Biology and Physics of Life’s Media. Princeton University Press, Princeton, NJ. Denny, M.W. (1994) Extreme drag forces and the survival of wind- and water-swept organisms. Journal of Experimental Biology 194, 97–115. Denny, M.W. (1995) Predicting physical disturbance: mechanistic approaches to the study of survivorship on waveswept shores. Ecological Monographs 65, 371–418. Denny, M.W., Daniel, T.L. & Koehl, M.A.R. (1985) Mechanical limits to size in wave-swept organisms. Ecological Monographs 55, 69–102. Dillon, R.T., Jr (1988) Evolution from transplants between genetically distinct populations of freshwater snails. Genetica 76, 111–119. Ford, J.I. & Kinzie III, R.A. (1982) Life crawls upstream. Natural History 91, 61–66. Freilich, J.E. (1989) A method for tagging individual benthic macroinvertebrates. Journal of the North American Benthological Society 8, 351–354. Gore, J.A. (1983) Considerations of size related flow preferences among invertebrates used in instream flow studies. Developments in Ecology and Environmental Quality, Vol. 2 (ed. H. I. Shuval), pp. 389–398. Balaban Int. Publ., Jerusalem. Graham, A. (1985) Evolution within the Gastropoda: Prosobranchia. The Mollusca, Vol. 10 (eds E. R. Truman & M. R. Clarke), pp. 151–187. Academic Press, Orlando, Florida. Hart, D.D. & Resh, V.H. (1980) Movement patterns and foraging ecology of a stream caddisfly larva. Canadian Journal of Zoology 58, 1174–1185. Hart, D.D., Clark, B.D. & Jasentuliyana, A. (1996) Finescale field measurement of benthic flow environments inhabited by stream invertebrates. Limnology and Oceanography 41, 297–308. Haynes, A., Taylor, B.J.R. & Varley, M.E. (1985) The influence of the mobility of Potamopyrgus jenkinsi (Smith, A.E.) (Prosobranchia: Hydrobiidae) on its spread. Archives für Hydrobiologie 103, 497–508. Hill, W.R., Weber, S.C. & Stewart, A.J. (1992) Food limitation of two lotic grazers: quantity, quality, and sizespecificity. Journal of the North American Benthological Society 11, 420–432. Houp, K.H. (1970) Population dynamics of Pleurocera acuta in a central Kentucky limestone stream. American Midland Naturalist 83, 81–88. Huryn, A.D., Koebel, J.W. & Benke, A.C. (1994) Life history and longevity of the pleurocerid snail Elimia: A comparative study of eight populations. Journal of the North American Benthological Society 13, 540–556. Huryn, A.D., Benke, A.C. & Ward, G.M. (1995) Direct and indirect effects of geology on the distribution, biomass, and production of the freshwater snail Elimia. Journal of the North American Benthological Society 14, 519–534. Hutchinson, G.E. (1967) A Treatise on Limnology. Volume II: Introduction to Lake Biology and the Limnoplankton. John Wiley & Sons, New York. Hynes, H.B.N. (1960) The Biology of Polluted Waters. Liverpool. Jobin, W.R. & Ippen, A.T. (1964) Ecological design of irrigation canals for snail control. Science 145, 1324–1326. Kreiger, K.A. & Burbank, W.D. (1976) Distribution and dispersal mechanisms of Oxytrema (= Goniobasis) suturalis Haldeman (Gastropoda: Pleuroceridae) in the Yellow River, Georgia, U.S.A. American Midland Naturalist 95, 49–63. Mancini, E.R. (1978) The biology of Goniobasis semicarinata (Say) (Gastropoda: Pleuroceridae) in the Mosquito Creek drainage system. PhD thesis, University of Louisville, Louisville, Kentucky. Noel, M.S. (1954) Animal ecology of a New Mexico springbrook. Hydrobiologia 6, 120–135. Paulini, E. (1963) Field observations on the upstream migration of Australorbis glabratus. Bulletin of the World Health Organization 29, 838–841. Pointier, J.P., Thaler, L., Pernot, A.F. & Delay, B. (1993) Invasion of the Martinique island by the parthenogenetic snail Melanoides tuberculata and the succession of morphs. Acta Oecologica 14, 33–42. Schneider, D.W. & Frost, T.M. (1986) Massive upstream migrations by a tropical freshwater neritid snail. Hydrobiologia 137, 153–157. Schneider, D.W. & Lyons, J. (1993) Dynamics of upstream migration in two species of tropical freshwater snails. Journal of the North American Benthological Society 12, 3–16. Shelford, V.E. (1913) Animal Communities in Temperate North America. University of Chicago Press, Chicago. Söderström, O. (1987) Upstream movements of invertebrates in running water – a review. Archives für Hydrobiologie 111, 197–208. Vogel, S. (1994) Life in Moving Fluids, 2nd edn. Princeton University Press, Princeton, NJ. Wallace, C. (1992) Parthenogenesis, sex and chromosomes in Potamopyrgus. Journal of Molluscan Studies 58, 93–107. Waters, T.F. (1972) The drift of stream insects. Annual Review of Entomology 17, 253–272. Zar, J.H. (1984) Biostatistical Analysis (2nd edn). PracticeHall Inc., Englewood Cliffs, New Jersey. Received 22 January 1996; revised 27 November 1996; accepted 4 December 1996
© Copyright 2025 Paperzz