Spatial patterns in the population structure of the whelk Stramonita haemastoma (Linnaeus, 1766) (Gastropoda: Muricidae) in the Canarian Archipelago (eastern Atlantic)

The goal of this study was to determine the consistency of the vertical distribution patterns of the predatory whelk, Stramonita haemastoma, in the rocky intertidal zone of the canarian archipelago (eastern atlantic) across a hierarchy of five orders of magnitude of horizontal spatial variability (from tens of m to hundreds of km). in general, this species showed a consistent vertical zonation pattern across islands, with the majority of the specimens (74.65%) found in the mid intertidal zone. This result most likely reflects the whelk’s preference for a habitat with a large amount of potential prey but minimal stress induced by swells and desiccation. The mean abundance (0 to 1.73 ± 0.40 ind m-2, mean ± Se) and size structure (7 to 45 mm in shell length) of S. haemastoma in the canarian archipelago was considerably lower compared to those from continental areas in the atlantic ocean, which suggests that there are some differences between these populations. natural mechanisms (e.g. isolation or poor recruitment events) and human perturbations (e.g. exploitation of intertidal resources) might influence the spatial distribution patterns.

spatial and temporal scales (benedetti-cecchi et al., 2000;moreno, 2001;Thompson et al., 2002;Fraschetti et al., 2005).The analysis of this variability is a key requisite for understanding the importance of different processes, and for constructing (and generalizing) predictive models of the ecological patterns of intertidal assemblages (Levin, 1992;underwood and chapman, 1996;benedetti-cecchi, 2001).
Predatory prosobranch molluscs, such as the globally-distributed whelk Stramonita haemastoma (Linnaeus 1766), have been widely recognized as key organisms that determine the structure and organization of intertidal communities (connell, 1961;menge, 1978;edwards et al., 1982;Liu et al., 1991;Rilov et al., 2001).Spatio-temporal variability in the population structure of this species has been attributed to differences in abiotic processes, such as habitat structure and the degree of wave exposure (e.g.Richardson and brown, 1990;Rilov et al., 2001Rilov et al., , 2004;;brown et al., 2004).Similarly, limited prey availability, in terms of abundance and biomass, has been suggested to be a possible mechanism behind the low abundances and growth rates of these intertidal carnivores on some coasts (mckillup and butler, 1983;burrows and Hughes, 1990;mckillup and mckillup, 1997;ota and Tokeshi, 2000).
The goal of this study was to test the consistency of the vertical distribution patterns of S. haemastoma on the rocky coasts of the canary archipelago (eastern atlantic) by identifying patterns of spatial variability over scales from tens of m to hundreds of km. in addition, bibliographic data was used to determine whether there are differences in the population structure of this species (abundance and size structure) between insular systems and continental areas of the atlantic ocean.

Study area and sampling
The study was carried out on the coasts of the seven islands that constitute the canarian archipelago (eastern atlantic) and in the group of small islets that constitutes the chinijo archipelago, during march 2003 (Fig. 1).These intertidal systems are rocky platforms of volcanic origin (mainly basalts), with a width between 30 and 130 m. differences in age between the eastern ('old' islands) and the western islands ('young' islands) (carracedo, 2001) have led to dissimilarities in the mean length of these platforms (eastern islands = 64.66 ± 6.32 m, western islands = 45.16 ± 8.51 m; mean ± Se). in general, three zones can be recognized in the rocky intertidal zone of the canary islands according to classical zonation schemes (Stephenson and Stephenson, 1949;Lawson and norton, 1971): the high, mid, and low intertidal zones.The high zone is characterized by a cyanobacterial film of blue-green bacteria, the mid zone is dominated by the cirriped barnacle Chthamalus stellatus, and the lower zone is covered by a conspicuous canopy of turf-forming algae.
Three localities 1 to 50 km apart were randomly chosen at each island.in addition, 3 randomly-selected sites, tens of m apart, were selected within each intertidal zone at each locality.We haphazardly placed ten 50 x 50 cm (0.25 m 2 ) quadrats several meters apart on each of the 3 sites within each intertidal zone and locality.We counted and measured (shell length to the nearest mm -SL) all the S. haemastoma within each quadrat.as a result, we adopted a hierarchical design that spans five orders of magnitude of horizontal spatial variability, and incorporates islands (10 4 -10 5 m apart), localities (10 3 -10 4 m apart), sites (10 1 -10 2 m apart) and quadrats (10 0 -10 1 m apart).

Statistical analysis
We tested the significance of the differences in the mean abundances between the three intertidal zones across the set of spatial scales by means of a mixed, partially nested, anoVa model (underwood, 1997).The model incorporated the following factors: (1) 'island' (fixed factor with eight levels corresponding to the seven islands plus the chinijo archipelago), (2) 'zone' (fixed factor with three levels corresponding to the high, mid and low intertidal zones, and orthogonal to the previous factor), (3) 'Locality' (random factor nested within 'island' with three levels) and (4) 'Site' (random factor nested within 'Locality', 'island', and 'zone', with three levels).due to the high heterogeneity of variances, which persisted after data transformation (cochran's test, P<0.01), we considered a significance level of α = 0.01 instead of α = 0.05 to avoid an increase in the Type i error; anoVa is robust to heterogeneity of variances, particularly for large balanced studies (underwood, 1997).When the anoVa detected significant differences for 'island' and/or 'zone', either as an interaction or as a main effect, further analyses were performed by using the SnK a-posteriori multiple comparison tests to resolve differences.To estimate the contribution of each horizontal spatial scale (island, locality, site and quadrat) to the total variability of each intertidal zone, we calculated the corresponding variance components, and subsequently the percentage of contribution of each spatial scale.We tested for differences in mean sizes (SL) among islands and intertidal zones using 95% confidence intervals (underwood, 1997;martínez-González et al., 2001).
abundance patterns of S. haemastoma in each intertidal zone varied among localities (zone x Locality (island), P<0.0001, Table 1). in contrast, the zoning pattern was consistent among islands (island x zone, P>0.3, Table 1), although some inconsistencies were observed in some islands (e.g.Tenerife and Gran canaria showed greater abundances in the lower intertidal zone, Fig. 2).Generally, the greater abundances were observed in the mid intertidal zone (SnK, P<0.01; Table 2), where 159 individuals (74.65%) were found (Fig. 2).differences among quadrats at the lowest spatial scale accumulated most of the total variance in our survey (>72%, Table 3), which evidences the important role played by small-scale spatial variability in the distribution patterns of this species.We also detected high variability from site to site within islands, zones and localities (Sites (island x zone x Locality), P<0.01, Table 1). in fact, this factor explained 11.14%, 11.13% and 6.35% of the total variability for the low, mid and high intertidal zones respectively (Table 3).
The mean size for the overall study was 21.88 ± 0.40 mm SL (mean ± Se) and ranged between 7 and 45 mm SL.We obtained a unimodal and symmetric size-frequency distribution, with the majority of individuals (>80%) in the intermediate size ranges (16 to 30 mm SL, Fig. 3).Table 4 shows the mean sizes per island, as well as the mean sizes in each intertidal zone, for the entire survey.in general, the mean sizes were similar among islands, though some minor differences were observed (Table 4).
at oceanic scales, S. haemastoma can be found with densities from a few ind m -2 (this study) to     ler, 1985), Gulf of mexico (brown and Richardson, 1987), azores (Spence et al., 1990), mediterranean (Rilov et al., 2001), canary islands (this study).densities up to hundreds, and sometimes >1000 ind m -2 (e.g.Venezuela, Fig. 4).Similarly, its size range can vary considerably, from 40 to 45 mm SL to >80 mm SL (Fig. 4). it stands out that low values of both density and size were generally observed in insular systems (i.e. this study and the azores), in contrast to continental areas, where mean values were considerably larger (Fig. 4).

diScuSSion
This study shows that the predatory whelk, S. haemastoma, usually has a regular vertical distribution pattern in the rocky intertidal zone of the canary islands, with higher densities in the mid compared to the low intertidal zones, and an almost complete absence of individuals in the high intertidal zone.This probably reflects the whelk's preference for a habitat with a large amount of potential prey but minimal stress induced by wave exposure and desiccation.Stramonita haemastoma, like many predatory whelks, inhabits environments rich in potential food (moran, 1985;Rilov et al., 2001Rilov et al., , 2004)). it has a broad diet (butler, 1985;Rilov et al., 2002), that includes barnacles, limpets, top-shell snails and periwinkles on the canary rocky coasts (R. Ramírez pers.obs.), which are especially abundant in the mid intertidal compared with the low intertidal zone (navarro et al., 2005;Ramírez et al., 2005).Food availability, however, is not always an adequate predictor for whelk densities in littoral zones (Rilov et al., 2001), since other factors, particularly physical conditions (e.g.wave exposure or desiccation), may also affect its distribution patterns (moran, 1985;menge et al., 1994;Rilov et al., 2005).a way of minimizing both the risk of wave-induced dislodgment and desiccation stress is to hide inside shelters (menge, 1978;moran, 1985;Richardson and brown, 1990;Rilov et al., 2005).displacements caused by waves are probably common in the canary islands, since the coasts of the archipelago are exposed directly to oceanic swells from all directions, although waves with a northern component dominate in frequency.This fact, coupled with a considerable tidal amplitude (up to 3 m), results in almost constant daily exposure to strong wave action and desiccation.Therefore, the presence of shelters (e.g.crevices and holes), which are more common in the mid intertidal zone (Ramírez et al., 2008), seems to have a particular influence on the vertical distribution pattern of S. haemastoma in the canary islands.
our study has confirmed that variability at small spatial scales (1-10 m) is a general feature of benthic populations inhabiting the rocky intertidal zone (reviewed by Fraschetti et al., 2005).The abundance of S. haemastoma was extremely variable at small-scales, mainly between quadrats several meters apart.This patchy pattern is consistent with observations of this species from the mediterranean (Rilov et al., 2001) and with other intertidal molluscs from the study area (navarro et al., 2005;Ramírez et al., 2005).This pattern is possibly related to the predatory behaviour of S. haemastoma, which can feed in groups (brown and alexander, 1994).although group feeding does not seem to lower handling time, or increase per capita feeding rates, it could facilitate prey location, or could be beneficial for feeding on large preys (brown and alexander, 1994).Since preys are not large or very numerous in the archipelago (see below), detecting the scent of injured preys, or conspecific feeding, are potential factors that influence the distribution pattern of S. haemastoma (sensu brown and alexander, 1994).
The mean abundance and size structure of S. haemastoma in the canarian archipelago was considerably lower compared to continental areas in the atlantic ocean.For example, abundances were up to three orders of magnitude higher on the coasts of Venezuela (butler, 1985;Fig. 4).Similarly, the population structure was dominated by small-sized individuals, consistent with observations from the nearby azores archipelago (Spence et al., 1990;Fig. 4).These results reflect important differences between populations inhabiting oceanic archipelagos in the eastern atlantic and continental areas.organisms living on islands are constrained by the isolation and fragmentation of their populations (mcarthur and Wilson, 1967), while several paleoclimatic and human disturbances (reviewed by brown and Lomolino, 2000) can also limit patterns of dispersal, growth and survival of these populations (alexander and Roughgarden, 1996). in addition, poor recruitment events have been highlighted as a potential mechanism that explains the low abundances of intertidal macro-invertebrates found in these oceanic archipelagos (Hawkins et al., 2000).it is worth noting, moreover, that certain potential preys of S. haematoma are absent in the study area, or are naturally more abundant in continental areas, compared to insular systems in the eastern atlantic (nuñez, 1995;Hawkins et al., 2000).alternatively, human-induced perturbations may have accentuated the low abundances and sizes of S. haemastoma in the canary islands and adjacent archipelagos (azores and madeira).First, intertidal resources have been extensively exploited since prehistoric times (Hawkins et al., 2000;moro and Herrera, 2000;batista, 2001;cabrera, 2001), particularly species that constitute a potential prey for S. haemastoma, such as limpets and top-shell snails, and some species have been driven near to extinction (e.g.Patella candei), and reduced to small populations (nuñez et al., 2003(nuñez et al., , navarro et al., 2005(nuñez et al., , Ramírez et al., 2005;;Ramírez et al., 2008).The greater abundances of S. haemastoma in the lower, compared to the mid intertidal zone at Gran canaria and Tenerife (the two most populated islands), could be the result of the extremely low densities of potential preys that are routinely caught for human consumption, such as limpets (navarro et al., 2005).it is possible that S. haemastoma has therefore been 'forced' to move down-shore, facing a higher risk of wave-induced dislodgement, but probably finding larger amounts of suitable preys.Second, the socalled 'imposex' effect has been recently detected for S. haematoma from this area (R. Ramírez pers.obs.).This phenomenon affects the recruitment of many species, especially molluscs, and causes a notable reduction in their physiological performance, and even local extinctions (bryan et al, 1986;evans et al, 1996;Rilov et al., 2000).although sterilization of S. haemastoma females is not widely extended in the archipelago, this is occurring in areas that are greatly affected by TbT, such as industrial ports and their proximities (R. Ramírez pers.obs.). in summary, our results indicate that the distribution, abundance and size structure patterns of S. haemastoma in the canary islands most likely respond to several factors reported to influence this species in other geographic areas (e.g.food, availability of shelter, etc.) in conjunction with the specific characteristics of oceanic systems.Human-induced perturbations (e.g.harvesting, pollution, etc.) may also affect the populations of this whelk.more specific studies are necessary to disentangle the relative role of each of these processes.

acKnoWLedGemenTS
We are grateful to all the volunteers that helped in collecting the data, especially Rodolfo Jaraiz, Sandra Prieto, Patricia de la Fuente and adrian abelardo.

Fig. 1 .
Fig. 1. − map of the sampling localities at each island throughout the canarian archipelago (eastern atlantic).
Table2.− Summary of results of pair-wise comparisons of the abundance of Stramonita haemastoma between the three intertidal zones at the 24 localities.The numbers and percentages indicate the number of localities with a significant result (P<0.05intertidal;mi: mid intertidal; Hi: High intertidal.

Table 1 .
− Results of the anoVa testing for differences between islands (fixed), intertidal zones (fixed and orthogonal), localities (random and nested within islands) and sites (random and nested within islands, zones and localities) in the mean abundances of Stramonita haemastoma across the canarian archipelago.

Table 3 .
− analysis of the variance components for each intertidal zone.The contribution of each factor (ω²) is presented as a percentage (%).
SS: sums of squares; df: degrees of freedom; mS: mean squares; vc: variance components; ω²: contribution of effects.

Table 4 .
− mean sizes (mm SL ± Se) of Stramonita haemastoma at each island and in each intertidal zone (data pooled across islands).Significant differences in sizes, detected by 95% confidence intervals, are indicated as a footnote.
Significant differences : G>L, P, H Fig. 4. − Ranges of mean abundances and sizes for Stramonita haemastoma in different coastal areas in atlantic ocean waters (northern hemisphere).data sources: Florida and Venezuela (but-