Along‐ and across‐shore components of the spatial distribution of the clam Tivela mactroides (Born, 1778) (Bivalvia, Veneridae)

Beach clam fisheries often provide an alternate income for traditional fishermen, but this activity is highly vulnerable to urbanization of coastal areas. Tivela mactroides is a beach clam commonly harvested in the Caribbean (Venezuela) and on the south‐eastern Brazilian coast. Only localized records on its spatial structure in size and abundance are available to support management of this clam. The present study showed that the spatial distribution in abundance and size of T. mactroides in Caraguatatuba Bay, state of São Paulo, Brazil, has both along‐ and across‐shore (intertidal and subtidal) components. The nature of this distribution should be taken into account in the design of further population or monitoring studies concerning this species. Clam density varied along‐shore, with a tendency towards higher numbers of individuals in the stations near the northern side of river mouths of this bay. Lower densities were related to the occurrence of sea‐stars and sand‐dollars. The intertidal and subtidal across‐shore distribution of clam abundance and size varied between and within areas (southern and northern). The individuals from the southern area were concentrated at intermediate tidal levels (0.2–0.6 m in relation to mean low water), whereas in the northern area the clams were more homogeneously distributed. The uppermost limit of their distribution was 0.8 m (southern) and 0.4 m (northern), and an upwards tendency towards increased size was more evident in the northern area. In the subtidal, abundance and individual size tended to increase from the deepest (4.5 m) to the shallowest (0.5 m) levels only in the southern area. Individual clams in the northern area tended to be smaller and more abundant than in the southern area. The species had an ontogenic distribution: small‐sized individuals were recorded only in the subtidal. Possible causes for these patterns and for the differences between areas (human exploitation, organic pollution, beach morphodynamics, differential recruitment, and adult–larvae interactions) are discussed, and specific studies are suggested to evaluate formally the proposed hypotheses.


Introduction
Commercial, recreational or traditional fisheries of beach clams are common but vulnerable activities that may require special attention from managers because of their importance to local and traditional people . The lack of background population information for decision makers is one of the major problems that complicates the development and implementation of adequate conservation strategies for these resources.
The venerid clam Tivela mactroides (Born, 1778) is commonly harvested on beaches in the Caribbean (Venezuela) and on the south-eastern Brazilian coast . Its range extends from Venezuela to Brazil (states of Pará to Santa Catarina) and includes some Caribbean islands (Rios 1994). It is an important component of the macrofauna of sandy beaches in the state of São Paulo in south-eastern Brazil (Narchi 1972). Tivela mactroides inhabits sandy beaches exposed to wave action, and is especially abundant near river mouths . A latitudinal variation in its vertical distribution was suggested, but not formally evaluated, by McLachlan et al. (1996); populations at higher latitudes occupy the intertidal zone (Narchi 1972), and those in lower latitudes are predominantly subtidal (Etchevers 1976). Although ontogenic segregative distribution, in which the larvae tend to establish apart from the adults, commonly occurs in beach clams , it has never been described for T. mactroides.
Most beach clams, including T. mactroides, occur in intertidal and shallow subtidal habitats; however, population studies generally focus on only one of these habitats (see McLachlan et al. 1996 for a review). This bias is also common for other marine bivalves (Moore and Lopez 1969;Schaeffer-Novelli 1980;Arruda-Soares et al. 1982;Thompson 1982;Hanekom 1986;Monti et al. 1991;Castro-Ortiz and García-Domingues 1993;Clasing et al. 1994;de Montaudouin and Bachelet 1996) and is explainable by the difficulties in sampling in the surf zone and in standardizing collecting methods and effort, especially because of differences in the spatial scale of subtidal habitats in relation to intertidal ones.
Although certain studies have addressed specific aspects of the biology of T. mactroides, there is only limited and fragmented information on the spatial distribution, in size and abundance, of this species on both along-and across-shore scales. The present study aimed to describe the spatial distribution of T. mactroides in Caraguatatuba Bay, considering its entire living area, i.e. the intertidal and subtidal zones, by addressing the following questions: 2. Is the distribution of T. mactroides (abundance and shell size) homogeneous at different intertidal levels (intertidal across-shore distribution)? 3. Is the distribution of T. mactroides (abundance and shell size) dependent on depth (subtidal across-shore distribution)? 4. Are the physical characters similar between and within beach areas at specific tidal levels? 5. Is the distribution of T. mactroides (abundance and shell size) similar between and within beach areas with different environmental conditions in relation to specific tidal levels?

Study area
Caraguatatuba Bay is located on the northern coast of the state of São Paulo, Brazil, from Arpoar Point (23u43925.30S, 45u24907.10E) to Camaroeiro Point (23u37941.10S, 45u24902.40E) ( Figure 1). It is bordered by a large sandy beach (about 16 km long), with moderate wave energy because of the shadowing effect of São Sebastião Island. Three important rivers (Juqueriquerê, Lagoa and Santo Antô nio) flow into different parts of Caraguatatuba Bay, constantly importing terrigenous sediment. Beach characteristics vary markedly in a north-south orientation. The southernmost part has a wide intertidal ultradissipative terrace (800 m). In the central part, the beach is still gently sloping but has a narrower intertidal zone. The northern part has a more heterogeneous slope with a low tide dissipative terrace subjacent to a steeply sloped reflective high intertidal and berm. In the bay, T. mactroides is harvested exclusively in the intertidal zone and mainly in the southern area. All samples were taken in autumn 2002, from 2 to 5 April in the subtidal, and from 13 to 16 April in the intertidal. This concentrated sampling effort controlled for any temporal effect. A particular methodology was employed for each question, as follows.
Along-shore distribution-abundance patterns (Question 1) Fourteen sampling stations were established in Caraguatatuba Bay, 1000 m apart from each other, from Arpoar to Camaroeiro points (see Figure 1). Three 30-m long dredgings (speed 2 knots) were conducted perpendicular to the water line in the shallow subtidal zone (about 1.5 m deep) at each station, at neap high tide periods. This precaution was taken to exclude interference from human exploitation (observed in the intertidal zone) on clam density, and to avoid the higher spring tide variation that would prevent the boat from approaching the shallowest areas. A rectangular dredge (80625 cm) with 3-mm internal mesh size was used to collect the sediment, which was sieved in a 3-mm mesh. The individuals sampled in each transect were counted and returned alive to the sea. The mean number of individuals per transect (fourth-root transformed) was then compared among stations (fixed factor, 14 levels) through a one-way model I ANOVA, under the null hypothesis that the mean number of individuals was similar among the defined beach stations. The data were transformed after the heterogeneity of variances was verified through the Cochran test (Underwood 1997). A Student-Newman-Keuls test (SNK test) was employed a posteriori to identify a valid alternative hypothesis.
Intertidal across-shore distribution-abundance and size (Question 2) Two structurally different intertidal areas (southern and northern, Figure 1) were selected to describe the vertical (across-shore) distribution of T. mactroides. These environments had characteristic slopes (see below), the flattest, widest and more homogeneous being located in the southern part of the bay, and the steepest, narrowest and more heterogeneous in the northern. One 50-m wide sampling sector perpendicular to the water line was then selected in each area, based on burrow marks in the sediment. Because the objective here was not to compare the two areas but rather to address the vertical distribution of the clams, the sectors were neither sorted nor replicated, but were selected taking into account areas with high densities of clams.
Because of differences between areas in slope and intertidal width, different numbers of intervals were established to describe the zonation of this species: 12 in the southern and 11 in the northern. In each sector, 8-m intervals were used in the flattest parts of the intertidal and 4 m in the steepest part, thus allowing a relatively homogeneous vertical screening of this population. Six random samples (quadrats of 0.5060.50 m, sampled to a depth of 10 cm) were taken during spring low tides in each interval, totalling 72 and 66 samples in the southern and northern areas, respectively. The samples were washed in the field using a 3-mm mesh sieve. The individuals were counted and returned alive to the sea. The data (number of individuals) for each interval were then averaged (n56) and plotted in relation to intertidal distance from water line.
Based on the topographic slope, the tidal levels corresponding to the heights 20.2, 0.0, 0.2, 0.4, 0.6, and 0.8 m in relation to MLW (mean low water, based on the forecasts from tide tables, DHN 2002) were standardized in both sectors. These tidal levels coincided with some of the intervals cited above, and allowed direct comparisons of the vertical patterns of abundance and size of this species between areas. This methodology also allowed comparisons based on tidal heights instead of distances from MLW, which may be highly variable depending on the beach slope. To reduce sampling effort, only the individuals collected at these levels were measured (total shell length, mm).
The mean number of individuals (fourth-root transformed only in the southern area) and the mean shell length were compared among tidal levels (fixed factor, southern-five levels: 20.2, 0.0, 0.2, 0.4, and 0.6 m, northern-four levels: 20.2, 0.0, 0.2, and 0.4 m; only tidal levels where T. mactroides occurred were considered in these analyses) using a one-way model I ANOVA for each area independently, under the null hypotheses that (1) mean number of individuals and (2) mean shell size are similar among the above-specified tidal levels. The SNK test was employed a posteriori to identify valid alternative hypotheses.
Subtidal across-shore distribution-abundance and size (Question 3) Four depth intervals [0-1 m (mean depth class value 0.5 m), 2-3 m (2.5 m), 4-5 m (4.5 m), and 6-7 m (6.5 m)] were established in both areas (southern and northern) in relation to MLW. Five stations were then sorted in each area at each depth interval, totalling 20 sampling stations (replicates) in each area. One sample was taken at each station at neap high tide periods, dredging the sediment (at 2 knots) for a distance of 50 m parallel to the water line using a rectangular dredge (80625 cm with 3-mm internal mesh size). The individuals collected were counted, measured (shell length, mm), and then returned alive to the sea. The number of individuals (fourth-root transformed) and the shell size were compared between areas (fixed factor, two levels) and among depth intervals (fixed factor, three levels: 0.5, 2.5, and 4.5 m; no individuals were recorded in the 6.5 m interval) through a two-way model I ANOVA under the null hypotheses that the depth gradient in (1) mean number of individuals and (2) mean shell size is the same between areas. If the interaction term (area6depth) was significant, the depth differences in abundance and size were tested with the a posteriori SNK test for each area independently.

Comparisons between the intertidal areas-physical variables (Question 4)
The two intertidal areas (southern and northern) were then compared in relation to physical variables (beach profile and sediment). For this purpose, four sub-areas were sorted in each study area and the beach profile was measured (n53) in each one. The profiles were described by measuring the elevation (cm) in 4 m intervals, from the vegetation (southern) or berm (northern) to the 0.0 m tidal level. For each transect, three variables were recorded: mean elevation (cm), standard deviation of elevation (cm), and total height of the beach (cm). The mean elevation was obtained by averaging all the elevation values recorded. The standard deviation (SD) of elevation was calculated based on these values to represent the degree of heterogeneity in beach slope, i.e. if the profile is more (smaller values of SD) or less (larger values of SD) continuous. Total beach height was the vertical distance between the highest part of the beach (vegetation or berm) and MLW. These three variables were compared between areas (fixed factor, two levels) and sub-areas (random factor, four levels) through a nested two-way model III ANOVA under the null hypothesis that they do not differ between nor within (among sub-areas) areas. The a posteriori SNK test was employed to verify the degree of profile homogeneity within each area independently.
The calculated profiles enabled the identification of the position of 0.0 and 0.4 m tidal levels. These levels were chosen due to the evident variation in number of individuals and in shell size within this interval (see below). Sediment samples (n53) were taken at these two tidal levels. The granulometric analyses were employed through dry screening of the sediment (Suguio 1973) using the Wentworth (1922) scale. The values of mean grain diameter and sorting coefficient were calculated according to the Folk and Ward (1957) method. The mean grain size (phi units) and sorting coefficient (phi units) of the sediment were then compared between areas, sub-areas nested in areas, and tidal levels (fixed factor, two levels) using a nested three-way model III ANOVA under the null hypothesis that the intertidal variation in these variables is independent of sub-areas (interaction term: sub-area (area)6tidal level) or of areas (interaction term: area6tidal level). The a posteriori SNK test was employed to verify specific differences among sub-areas and between tidal levels and areas if interaction terms were significant.

Comparisons between the intertidal areas-abundance and size (Question 5)
The abundance and size of T. mactroides were also compared between areas. The individuals were collected using six samples (quadrats of 0.5060.50 m) at each tidal level (0.0 and 0.4 m) and sub-area (the same as described above), totalling 48 sampling units in each area. The sediment was washed in a 3-mm mesh sieve. The individuals were then counted, measured (shell length, mm), and returned alive to the sea. The abundance (individuals/quadrat) and size of T. mactroides were then compared between areas, subareas nested in areas, and tidal levels using a nested three-way model III ANOVA. Specifically, this design was developed to test the null hypothesis that the intertidal patterns of distribution in clam size and abundance did not depend on the area or sub-area being sampled (interaction terms: area6tidal level and sub-area (area)6tidal level, respectively). The a posteriori SNK test was also employed to verify specific differences among sub-areas and between tidal levels and areas if the interaction terms were significant.

Along-shore distribution-abundance patterns (Question 1)
A total of 5177 individuals of Tivela mactroides was obtained in the three 30-m long dredgings at each of the 14 sampling stations in the shallow subtidal of Caraguatatuba Bay. These clams showed an irregular along-shore distribution in this habitat (fourth-root transformation; ANOVA: F (13, 28) 522.13, P,0.001; Figure 2). In general, the southern part (stations 2-7) of the bay had higher and more constant numbers of individuals than did the northern part (stations 8-14). Fewer clams were found at stations 1, 10, 11, and 13. At station 1, the sediment was very fine and the beach profile was characteristic of a tidal flat. Stations 8-13 yielded sea-stars (Astropecten marginatus Gray, 1840, 5-30 individuals per dredging), which generally had juvenile clams in their stomachs (visible externally by protuberances). A large number of sand-dollars (Mellita quinquiesperforata (Leske, 1778), 30-50 individuals per dredging) was also recorded at station 13. At station 14, where the number of clams was relatively high, there were many hermit crabs (Isocheles sawayai Forest and Saint Laurent, 1967, 20-30 individuals per dredging).
Intertidal across-shore distribution-abundance and size (Question 2) For this purpose, 2532 individuals were sampled in Caraguatatuba Bay along intertidal levels in one 50-m wide perpendicular sector in each of the two areas studied: southern (72 quadrats, 1377 individuals) and northern (66 quadrats, 1155 individuals). In the southern area, individuals were mainly found in the strip between tidal levels 0.2 and 0.6 m, whereas in the northern area, the clams were more homogeneously distributed across the intertidal area ( Figure 3). The higher across-shore variation in abundance in the southern area was demonstrated by contrasting the abundance of T. mactroides in the five previously defined tidal levels ( Figure 4; Table I). The size of the individuals also varied in relation to the tidal levels in both areas (Table I), with a tendency towards increased size from the lowest to the highest level ( Figure 4). This tendency was more evident in the northern area.
Subtidal across-shore distribution-abundance and size (Question 3) A total of 2317 individuals was sampled at the five depth intervals in both southern (1572 individuals) and northern (745 individuals) areas. The number of individuals in the subtidal showed higher variability among depths in the southern than in the northern area ( Figure 5), and the two-way ANOVA showed that the way clam density varied among depths differed between areas (significant interaction term: area6depth, Table II). In the southern area, abundance increased from the deepest (4.5 m) to the shallowest (0.5 m) subtidal depth intervals, whereas no variation was evident in the northern area ( Figure 5). No individuals were sampled in the 6-7 m depth interval, in either area.
The size distribution patterns for this species differed across depth intervals depending on the area sampled (significant interaction term: area6depth, Table II). The size of the individuals tended to increase from the deepest to the shallowest levels only in the southern area ( Figure 5). No clear size tendency among depths was apparent in the subtidal of the northern area. The abundance and size of the individuals in the shallowest subtidal depth (0.5 m) tended to be higher than at the same level in the northern area ( Figure 5). For the other depth intervals, the abundance and size of T. mactroides was roughly similar.

Comparisons between the intertidal areas-physical variables (Question 4)
The intertidal areas (southern and northern) and levels (0.0 and 0.4 m) were compared in relation to physical variables (beach slope and sediment constitution). Marked differences in the beach slope were shown between and within (sub-areas) the study areas (southern and northern, Figure 6), emphasized by the mean elevation, standard deviation of   Table I. The letters (southern) and the numbers (northern) represent the results of the a posteriori SNK test. The horizontal bar indicates non-significant differences in the SNK test. elevation, and total beach height ( Figure 7; Table III). The slope profile was wider, flatter, and more homogeneous in the southern area. The profile in the northern area was steeper, narrower, and more heterogeneous, showing higher elevations and standard deviation of elevation. This heterogeneity was evidenced by a low-tide terrace subjacent to a steeply sloping high intertidal. The northern area also showed higher variation among sub-areas in mean elevation and standard deviation of elevation (non-significant values in the southern area). The opposite situation was observed for the total height of the beach, with higher variability in the southern area and non-significant differences in the northern area ( Figure 7).
Comparison of the mean grain size and sorting coefficient of the sediment revealed a significant interaction between tidal levels and both areas and sub-areas (Table III). Data interpretation was only possible for each sub-area independently (Figure 8), but in general, the sand was coarser and more poorly sorted at the 0.4 m tidal level in comparison to the 0.0 m tidal level (for exceptions see Figure 8), and more marked variations in grain size and sorting coefficient between tidal levels were reported in the southern than in the northern area (note that the steeply sloping coarse sand of the northern area was not included in this analysis; this part of the intertidal was above the 0.4 m tidal level).

Comparisons between the intertidal areas-abundance and size (Question 5)
As with the physical variables, the intertidal areas (southern and northern) and levels (0.0 and 0.4 m) were also compared in relation to abundance and size of T. mactroides. The number of individuals, compared between the two tidal levels (0.0 and 0.4 m) of the southern and northern areas, did not show differences for any of the spatial scales tested (area, sub-area, and tidal level, Table IV). However, higher abundances tended to occur in the northern area than in the southern one and, when differences occurred, at the upper (0.4 m) tidal level than the lower one (0.0 m) (Figure 9). These differences between tidal levels were evident only when clam density was high (.20 individuals per quadrat).
The size of the individuals varied significantly between tidal levels only in the northern area (sub-areas analysed independently; Figure 9), with the larger individuals occurring at Table I. One-way model I ANOVA for the mean number of individuals per quadrat (fourth-root transformed only in the southern area, n56) and mean shell length (mm, n is variable) among tidal levels (fixed factor, southernfive levels: 20.2, 0.0, 0.2, 0.4, and 0.6 m, northern-four levels: 20.2, 0.0, 0.2, and 0.4 m).  Figure 5. Mean number of individuals (fourth-root transformed; ¡SE; n55) and mean shell length (mm; ¡SE; n is variable) in the two areas (southern and northern) and at different subtidal depths in Caraguatatuba Bay. The comparisons (two-way ANOVA) between areas and among depths are presented in Table II. The letters (southern) and the numbers (northern) represent the results of the a posteriori SNK test, and the horizontal bar indicates non-significant differences in the SNK test.

Source of variation
the upper tidal level. Despite the marked variation in shell size among sub-areas, individual clams tended to be larger in the southern than in the northern area ( Figure 9). These results showed a non-consistent zonation pattern in clam size and abundance among sub-areas (see Table IV, significant interaction factor: sub-area (area)6tidal level), although the over-all tendency of increase in number of individuals and shell size was equivalent in the two areas (see Table IV, non-significant interaction factor: area6tidal level). Table II. Two-way model I ANOVA for the mean number of individuals per transect (fourth-root transformed, n55) and mean shell length (mm, n is variable) between areas (fixed factor, two levels: southern and northern) and among subtidal depths (fixed factor, three levels: 0.5, 2.5, and 4.5 m).

Along-shore distribution
Except for a few stations, Tivela mactroides was roughly homogeneously distributed along Caraguatatuba Bay. A relatively high number of individuals was recorded at almost all the sampling stations, especially on the northern sides of the mouths of important rivers in this area, in accordance with previous reports by McLachlan et al. (1996) for other populations of this species. According to these authors, the water in river mouths has a high organicmatter content, thus contributing to the suspension-feeding habit of T. mactroides. Tivela mactroides dominated the benthic community of Caraguatatuba Bay, reaching densities of 388 individuals per m 2 (over-all mean 77 individuals per m 2 ) in the intertidal. Although dredging did not allow density estimates, we suppose that the density of smallsized individuals in the subtidal may reach considerably higher values. Other studies Figure 7. Mean elevation (cm; ¡SE; n54), standard deviation of elevation (cm; ¡SE; n54), and total height of the beach (cm; ¡SE; n54) in the southern and northern areas in Caraguatatuba Bay. Data are presented for each area and sub-area. The results of a nested two-way ANOVA are presented in Table III. The letters represent the results of the a posteriori SNK test, and the horizontal bars indicate non-significant differences in the SNK test. Table III. Results of the nested two-way model III ANOVA (sub-area nested in area, four levels) for the mean elevation (cm), mean standard deviation of elevation (cm, fourth-root transformed), and mean total beach height (cm) between areas (fixed factor, two levels: southern and northern; n53); results of the nested three-way ANOVA (sub-area nested in area, four levels) for the mean grain size (phi) and mean sorting coefficient (phi, n56) of the sediment between areas (fixed factor, two levels: southern and northern) and tidal levels (fixed factor, two levels: 0.0 and 0.4 m).  Prieto 1983) densities in areas where T. mactroides was also harvested. High densities are common in populations of this species, and are attributed to its suspension-feeding habit  Table III. The horizontal bars indicate non-significant differences in the SNK test. (Narchi 1972) and the wave action that constantly suspends the terrigenous sediments deposited on the bottom of the areas where this species occurs .

Source of variation
Other organisms co-occurred with Tivela mactroides in some parts of the subtidal region, and may at some level influence its abundance. The sea-star Astropecten marginatus was recorded (5-30 individuals per dredge) only between stations 8 and 12. Many individuals of this species had young (small) individuals of T. mactroides in their stomachs, which could be seen by external protuberances in the aboral surface of the sea-stars (personal observation). This finding may indicate that the predatory activity of A. marginatus on T. mactroides may be responsible for decreasing clam densities, at least at stations 10 and 11. The sand-dollar Mellita quinquiesperforata (30-50 individuals per dredge) was recorded only at station 13, and may be competing for space with T. mactroides. The presence of the sanddollar may prevent water circulation through the siphons of the clam and, as a consequence, cause clam mortality. Contrariwise, the density of T. mactroides was very high at the only station where the hermit crab Isocheles sawayai was recorded. This may suggest that this hermit crab species did not have any negative effect on the clam, although there are reports that the con-generic I. wurdemanni Stimpson, 1858 preys upon T. mactroides (Prieto 1987). Similarly, the hermit crab Clibanarius vitattus (Bosc, 1802) was considered by Prieto (1983) as the most important predator of this clam on a Venezuelan beach. Despite the indirect results found here, specific studies are necessary to demonstrate the importance of these co-occurring organisms (sea-stars, sand-dollars, hermit crabs, and even fish) in regulating the abundance of T. mactroides.
Across-shore distribution Narchi (1972Narchi ( , 1974 reported T. mactroides to be intertidal, occurring in the swash zone, in Ubatuba Bay in south-eastern Brazil. Etchevers (1976) and Prieto (1983) recorded it only on shallow subtidal flats in the Caribbean. McLachlan et al. (1996) suggested that these differences represent a latitudinal change in the distribution of this species. The present study showed that T. mactroides occurs from the upper intertidal zone until about 5 m in the subtidal. Narchi (1972Narchi ( , 1974 probably sampled only in the intertidal or used a sampling Table IV. Nested three-way model III ANOVA (sub-area nested in area, four levels) for the mean number of individuals per quadrat (fourth-root transformed, n56) and mean shell length (mm, n is variable) between areas (fixed factor, two levels: southern and northern) and tidal levels (fixed factor, two levels: 0.0 and 0.4 m). method that was not adequate to sample the smaller individuals in the subtidal. On the other hand, the microtidal regime in the Caribbean probably does not allow the exposure of the shallow subtidal flats where T. mactroides occurs. Apart from tides, which have a maximum range of 2 m in south-eastern Brazil, the areas studied by Narchi (1972Narchi ( , 1974,  Table IV. The horizontal bars indicate non-significant differences in the SNK test. Etchevers (1976) and Prieto (1983) are physically similar to the northern area of Caraguatatuba Bay. They have a steep, coarse-sand, upper intertidal-supralittoral fringe and a subjacent low tide, or shallow subtidal, fine-sand terrace and are typical of morphodynamically tide-dominated environments (Short 1996). The clams in the southern area preferentially occupied the strip encompassed between the 0.2 and 0.4 m intertidal levels, reaching a distribution limit at 0.8 m. In contrast, the clams in the northern area showed a homogeneous distribution in the intertidal zone and did not occur above the 0.4 m level. Vertical variations in the distribution of T. mactroides were also reported by Etchevers (1976). This author observed that T. mactroides occupied the deepest areas (up to 3 m) in the subtidal zone, where it co-occurred with two other bivalve species of the genus Donax. Donax hanleyanus Philippi, 1847 is a common species associated with intertidal coarse sediments of Brazilian reflective beaches (Cardoso and Veloso 2003), and was recorded in sympatry with T. mactroides by Narchi (1972Narchi ( , 1974, and in the present study in the northern area, where it occupied only the upper intertidal. Tivela mactroides never occupied this strip even when D. hanleyanus was absent. Therefore, these differences in shore-level distribution of T. mactroides are more likely attributable to morphodynamic along-shore changes rather than a competitive interaction with D. hanleyanus. The southern area is flatter, more homogeneous, and has finer sand than the northern area, thus allowing the water to reach higher levels and to drain more slowly. These differences in the intertidal distribution may also be a consequence of the larger size of the individuals in the southern than those in the northern area, i.e. the larger the individuals, the higher their tolerance to desiccation and the farther landwards their distribution.

Source of variation
In fact, individuals showed a tendency to increase in size along a gradient from the subtidal to the intertidal boundaries of their distribution. This peculiar distribution pattern has been observed in other high-density populations of beach clams around the world , but not for T. mactroides. Individuals seem to recruit in depths of about 2-5 m, ending their life cycle in the intertidal zone, thus leading to an ontogenic differential across-shore distribution. The vertical migratory process may be possible because of the passive drift of individuals generated by the environmental dynamics, such as waves and tides (see Etchevers 1976;Prieto 1983).
This distribution pattern, in which shell size tends to increase in an upshore direction, is similar to that of type 1 described by Vermeij (1972) for intertidal rocky shore species. According to this author, there are more pre-reproductive individuals in the zone of minimal mortality (subtidal) within the vertical range of the species because their survival is necessary to the maintenance of the population as a whole. The intertidal levels may be critical for T. mactroides, because of high mortality caused by natural disturbance, such as extended rainy periods and dry seasons, and extended low tide periods associated with high temperatures and sea storms. These stochastic events may be responsible for mass mortality of clams Defeo 2003; personal observation), and consequently responsible for regulating their densities and for causing periodic renewal of individuals in the population.
Another hypothesis for the absence of small individuals in the vicinity of the adults, i.e. in the intertidal zone, may be involuntary cannibalism (passive filtering) of larvae by suspension-feeding conspecific individuals (Woodin 1976). Fitch (1965), studying populations of the pismo clam T. stultorum (Mawe, 1823) in California, verified the absence of recruits in a higher adult-density environment and hypothesized that the conspecific adults remove the larvae from the water column before their settlement.
The sustainable use of this species may be guaranteed by the resettlement of individuals in the capture zone (intertidal and shallow subtidal up to 1.0 m depth), through migration of smaller individuals from the subtidal. Nothing is yet known about resettlement rates in this population, but it may be under the control of subtidal predators (sea-stars, fish, blue crabs, and hermit crabs), larval cannibalism, or bottom disturbance by shrimp trawling, which is a common fishing practice in Caraguatatuba Bay. In contrast, the ontogenic segregative distribution in this population may reduce the risk of incidental mortality from harvesting of other species of clams .

Comparisons between areas
Despite the great environmental variability along the 16 km extent of Caraguatatuba Bay, T. mactroides was able to occupy the whole area. Based on the universal consensus for sandy beach ecology that environmental characteristics, such as sediment grain size and beach slope, are the principal factors structuring the macrofauna (McLachlan 1983), spatial distribution of T. mactroides was compared between two particular environment types in Caraguatatuba Bay. The southern area had a gentle continuous beach slope, while the northern area had a steeper slope in the high intertidal associated with a low intertidal terrace. Sediment grain size and sorting coefficient were roughly similar for the two areas at the tidal levels considered (0.0 and 0.4 m), and thus did not explain the differences in clam density and size between these areas. Etchevers (1976) and Prieto (1983) hypothesized a transport of larvae and newly settled individuals by drift currents to explain differential along-shore distribution by size. This hypothesis cannot be applied to the situation in Caraguatatuba Bay, because it does not explain the absence of adults from the areas occupied by the juveniles in the across-shore gradient. Moreover, drift currents are a limited mode of transport for T. mactroides, which frequently remain buried instead of continuously migrating with the tides as observed for Donax serra (Rö ding, 1798) (Donn 1987). Passive migration in T. mactroides seems to be limited to the across-shore scale, and may continue during its entire life.
The higher density of T. mactroides in the northern area may be also caused by differential human exploitation of the clams between the two areas. In the northern area, fishermen collected small individuals (20-25 mm), while in the southern area the individuals were on average larger than 30 mm, a size generally selected by the fishermen in this region. Additional reasons for the smaller size of the clams in the northern area may be previous over-exploitation or mass mortality in this area, which could have provided space for new, small-sized migrating individuals. These arguments may suggest an important role of human exploitation in the abundance and size structure of this T. mactroides population, which merits further investigation.

Final considerations
This study showed that the structure of the population of T. mactroides in Caraguatatuba Bay varies over both along-and across-shore scales, and suggested that abundance and size of the individuals may be regulated by different factors. A complex and dynamic environment, with its intrinsic variability in biotic and abiotic factors, may be acting on populations of this species. Consequently, the spatial variability recorded here should be taken into account in the design of future population and monitoring studies on T. mactroides. Any study on the ecology of these clams should focus on both the intertidal and subtidal environments in order to record the population as a whole. The along-shore distribution also needs to be taken into account at different scales and hierarchically replicated (quadrats, transects, sub-areas, areas; see Underwood 1991). In addition, as the along-and across-shore pattern in density and size of the individuals was suggested to be influenced by pulses of recruitment, as also recorded for T. stultorum (Fitch 1950), the importance of long-term temporal replication is reinforced.
Finally, to understand the factors that influence the spatial and temporal structure of this population, as well as the ontogenic segregation, a series of experimental studies is needed to investigate larval preferences and tolerances, density-dependent recruitment, and the effect of predation. A long-term monitoring study of the environmental factors and of the population growth, maximum sustainable yield, and reproductive cycle would certainly furnish a solid basis for the establishment of future policies for the sustainable use of this resource.