Our study covered a large proportion of the types of rocky intertidal habitats on Helgoland Island. Only the open west coast could not be surveyed, because it was off limits for research during our fieldwork. The west coast faces the North Sea directly, so it is subjected to a higher degree of wave exposure than our studied shores. Other surveys of rocky intertidal species richness have been conducted recently on Helgoland shores (Reichert et al. 2008; Valdivia and Molis 2009), but they did not sample the entire intertidal range. Our data covering the full intertidal range indicate that neither overall richness nor diversity followed the predicted unimodal trend across elevation. Support for ESM predictions was only partial, since richness did increase from high to middle elevations but remained similar between middle and low elevations. The predicted increase in diversity from high to middle elevations only occurred at Kringel, with the opposite trend occurring at Bunker and no significant change at Nord-Ost Hafen. From middle to low elevations, diversity remained similar at Kringel and Bunker and increased at Nord-Ost Hafen. Hereafter, we discuss our findings with the goal of improving the understanding and testing of the ESM.

An important issue to consider is the domain of application of the ESM. The unimodal curve for overall richness and diversity (Scrosati and Heaven 2007) and the curves for the underlying interspecific interactions (Bruno et al. 2003) apply to the full stress gradient across which a regional biota can occur. Thus, the presence of species of other biotas might render the ESM unable to make accurate predictions. For example, communities in transition areas between two contiguous biotas share species to some extent (Kent et al. 1997; Attrill and Rundle 2002; Gastner et al. 2009). In rocky intertidal habitats, high elevations are often only occupied by truly intertidal species, which decrease in richness toward the upper intertidal boundary without sharing the substrate with the terrestrial species (e.g., vascular plants) that occur above (Raffaelli and Hawkins 1996). Thus, ESM predictions (low richness and diversity, and abiotic stress determining dynamics) are often met at the high intertidal zone (Bertness et al. 1999, 2006; Russell et al. 2006; Scrosati and Heaven 2007; Petes et al. 2008). In contrast, because of its limited aerial exposure, the low intertidal zone includes some species that primarily belong to subtidal biotas. Thus, the overlap of intertidal and subtidal biotas at low elevations might not allow ESM predictions to apply there. It is worth noting that the few studies that have quantified overall benthic richness (primary producers plus consumers) across the full intertidal range have found that richness increases from high to low elevations. This trend has been found for southern South America (Ingólfsson 2005), the central E Pacific (Lubchenco et al. 1984), NE Pacific (Ingólfsson 2005; Konar et al. 2009), NW Atlantic (Scrosati and Heaven 2007), N Atlantic (Ingólfsson 2005), and NE Atlantic (this study). Whether this pattern is universal for rocky intertidal systems remains unknown, but it suggests that the ESM might not be a good predictive tool for transition zones between biotas.

We initially evaluated removing from analyses the species appearing to be mostly subtidal (those occurring only at low elevations; Table 1), although that was an estimate because the full vertical distribution (elevation and depth) is unknown for many of such species. In any case, a quick analysis following that routine did reveal a unimodal trend for “intertidal” richness for Nord-Ost Hafen (with a total of 33, 40, and 28 species at low, middle, and high elevations, respectively) and Bunker (30, 33, 29), but Kringel retained the original trend (32, 31, 25). However, even when the corrected richness decreased at low elevations, it is unclear whether that resulted from effects of the “subtidal” species that were in the quadrats but were ignored, or whether it truly resulted from the interaction between the environment and the “intertidal” species as predicted by the ESM (Bruno et al. 2003). The ideal test of ESM predictions using low elevations would experimentally exclude the predominantly subtidal species and then assess richness once the community readjusts. ESM predictions on richness have been tested through mensurative approaches in other aquatic systems, such as lakes (Locke 1996). Although only zooplankton species were assessed, that study found a weak unimodal richness trend across a stress gradient determined by pH changes. Overlap of different biotas seemed not to occur in those lakes, suggesting that ESM-predicted richness patterns might indeed occur when communities from the same regional species pool are considered.

Another issue to consider relates to the interspecific interactions behind the predicted patterns in richness and diversity (summarized in the Introduction). Is there evidence in our data that interspecific interactions changed in importance across elevation as predicted by the ESM (Bruno et al. 2003)? The success or failure to find such evidence may give clues as to what aspects may be relevant to understanding our results in relation to the ESM. For example, the red alga Mastocarpus stellatus was the dominant basal species at middle elevations at Nord-Ost Hafen (reaching 90% cover), suggesting that this species may be competitively dominant there. This would agree with the predictions made by the ESM for intermediate stress levels (Bruno et al. 2003). The morphologically complex thallus of this alga (Dudgeon and Johnson 1992) might attract a number of small species that would otherwise be absent, resulting in total richness peaking at middle elevations, as the ESM predicted (Fig. 1). Our richness data did not show such a peak, but species size may have influenced this outcome. Because of constraints in logistics and taxonomic expertise, we restricted observations to organisms larger than 1 mm. Thus, if very small species were in fact drawn to middle elevations by M. stellatus, our sampling approach might have missed them, potentially explaining the lack of a unimodal pattern for our richness data. This type of limitation is a universal problem affecting intertidal diversity studies (Kimbro and Grosholz 2006; Russell et al. 2006; Thompson et al. 2007; Konar et al. 2009; Valdivia and Molis 2009). In fact, we know of no community study, aquatic or terrestrial, that has not suffered from this drawback. Increased efforts to identify even the smallest species might thus improve the accuracy of trend determination in ESM research. The above discussion must not be taken, however, as an indication that ESM-predicted patterns in richness did occur in Helgoland but our surveys failed to detect them. For example, at Bunker, the combined cover of all morphologically complex algae (M. stellatus and Fucus spp.) actually peaked at high elevations. Then, were these algae promoting a high occurrence of very small species there, total richness would still not follow a unimodal pattern.

Another prediction of the ESM on interspecific interactions is that consumers should play larger roles as stress decreases (Bruno et al. 2003). Is there evidence in our data suggesting that consumer pressure may have been particularly strong at low elevations? Percent cover data (as well as other abundance measures, such as density and biomass) may not be accurate indicators of consumer pressure, but offer information for a basic analysis. On Helgoland rocky shores, periwinkles (Littorina spp.) and green crabs (Carcinus maenas) are important consumers (Janke 1990; Eschweiler et al. 2009). The most abundant periwinkles, Littorina littorea and L. obtusata, feed readily on the green algae Ulva spp. (Janke 1990) and on the brown algae Fucus spp. (Watson and Norton 1981), respectively, while the mussel Mytilus edulis is a preferred food item for C. maenas (Ropes 1968). From middle to low elevations on our studied shores, the abundance of these consumers generally remained similar or decreased (Table 1). In the only case in which it clearly increased (L. littorea at Kringel), the abundance of its preferred food (Ulva spp.) also increased toward low elevations. These observations suggest that consumer pressure may not be particularly strong at the low intertidal zone relative to higher elevations. Field experiments should be done to confirm this notion, which may also contribute to explaining why overall richness and diversity did not decrease from middle to low elevations, as the ESM had predicted.

An alternative possible explanation for ESM-predicted patterns in richness and diversity not occurring in Helgoland relates to the extent of the intertidal range. For example, the change from stress being the main factor structuring communities, to species coexistence and higher richness of basal species, to competitive exclusions among basal species, to predation disrupting such exclusions occurs from high elevations down the intertidal zone on the Pacific coast of North America (Paine 1974; Kim 1997; Robles et al. 2009), where tidal ranges reach 4–5 m. On shores with smaller tidal ranges (e.g., ~2 m in Helgoland) and a similar slope, the size and/or mobility of organisms might complicate the detection, or even the occurrence, of horizontal zones representing the ecological stages described above and, consequently, changes in total richness (which have not been quantified for western North America). We suggest that ESM-predicted patterns might become increasingly evident with increasing intertidal range.

A possible solution to working on shores with a limited intertidal range might be to increase the spatial resolution of sampling. For example, on each Helgoland shore, we sampled the full intertidal range between chart datum and the upper intertidal boundary. However, we combined data for quadrats from the lower, middle, and upper thirds of the range to describe average richness and diversity for the low, middle, and high intertidal zones. This is a common approach in intertidal ecology, allowing researchers to identify broad patterns (Menge and Branch 2001; Konar et al. 2009). As discussed above, our tests did not find a unimodal richness/diversity pattern across elevation. Thus, if full ESM-predicted patterns did occur on a shore with a limited intertidal range, sampling at a higher resolution might reveal them. For example, an intertidal range of 2 m could be divided in 8 elevation zones of 25 cm in vertical extent each, which could be sampled with 25 cm × 25 cm quadrats. This approach would require sampling almost 3 times more quadrats than a classic 3-zone study to maintain the same sampling effort at each zone. Nonetheless, future ESM tests might profit from following this approach.

A final consideration is due regarding the trends followed by richness and diversity in Helgoland. Both traits showed the same pattern across elevation at Kringel, but differences existed at Nord-Ost Hafen and Bunker. A similar situation occurred across the shores, as richness was lower at Kringel than on the other 2 shores, but diversity was similar in Kringel and Bunker. Logically, these differences were explained by evenness trends (Krebs 1999). Richness and diversity have been considered virtually as synonyms in many biodiversity studies (Whittaker et al. 2001, Wilsey et al. 2005, Stachowicz et al. 2007), but both traits may differ in spatial (or temporal) pattern mediated by evenness (Badano et al. 2002; Kimbro and Grosholz 2006; Scrosati and Heaven 2007; Wilsey and Stirling 2007). Since both richness and diversity help to understand community functioning, predicting how they vary with stress is thus important. Therefore, future ESM studies should quantify richness, diversity, and evenness simultaneously to facilitate ecological synthesis.