Temporal and spatial comparisons of limpet sizes and densities.

To examine the impacts of temporal changes in harvesting on the densities of limpets, the commonly harvested C. granatina and S. argenvillei and the rarely harvested S. granularis were compared between years (1970 vs. 2017) and sites. A two-way ANOVA (with years and sites as fixed factors) was applied for each species. For the same species, to evaluate spatial differences in harvesting among sites in 2017, two-way ANOVAs with fixed factors protection level and site were employed. For temporal comparisons of shell lengths of C. granatina and S. granularis, t-tests were applied.

For both temporal and spatial comparisons, the data were first tested for normality and homogeneity of variance using Shapiro-Wilk’s and Levene’s tests, respectively and, where necessary, data were square-root transformed to meet these assumptions. In cases where significant differences were found, Tukey pairwise tests were used as post-hoc tests.

Effects of Mytilus on zonation of Scutellastra granularis.

To evaluate the impacts of the invasion of the alien mussel M. galloprovincialis on the zonation patterns of S. granularis, two-way ANOVAs (with fixed factors years and zones) were applied to density and size data for this limpet. The data were again tested for normality and homogeneity of variance and square-root transformed to meet the model assumptions. If significant effects were found, Tukey multiple comparison post-hoc tests were used to explore their significance.

Temporal changes in community composition.

To visualise changes in community composition between years (1970 vs. 2017) and among the five zones, an unconstrained ordination based on Bayesian ordination analysis with two latent variables was fitted to the historical and current community data using the lvsplot in the R package BORAL [45].

To test for statistical differences in percentage cover (taxon-abundance) between years and among zones, we used a full factorial multivariate generalized linear model (manyglm, R package “mvabund” [46, 47]). Adherence to model assumptions was based on plotting the residuals against the fitted model [46]. To probe differences and to identify taxa that were important contributors to differences, univariate tests were additionally run for each taxon, using the p.uni = “adjusted” argument.

Spatial differences in community composition.

The Bayesian ordination and generalised linear model-based approaches used in the temporal analysis were not suitable for these analyses since the high number of species (dependent variables) prevented the models from converging. Multivariate percentage cover data for the four sites compared for spatial differences were thus visualised by an ordination using non-metric multidimensional scaling (MDS) performed in PRIMER (Plymouth Routines in Multivariate Ecological Research, Version 6.0).

To statistically investigate differences in community composition among sites of different protection levels, a two-way PERMANOVA with factor ‘site’ nested in ‘protection level’ was performed on the species-abundance data. The nested design was chosen to account for site-specific differences due to local geomorphology. Because the zones reflect different biotopes (i.e. different benthic communities), the analyses were done separately for each zone. Pairwise comparisons were done to test for differences in community composition among sites. Prior to the analysis, data were standardised (by sample) and square-root transformed. To further explore the differences among sites and zones and to establish the species or groups mainly responsible for any observed differences, SIMPER (similarity percentage) analyses were performed. A cumulative cut off of 90% was applied and only taxa contributing >10% to differences are presented.

Temporal comparisons.

Densities of the frequently harvested limpet C. granatina were similar at the two sites in 1970, but showed substantial and significant reductions at both sites by 2017. The interaction between year and site was also significant, because the decline at Wireless Point was much greater than at Wireless Island (Fig 2A; S1 Table). The density of the other commonly harvested limpet, S. argenvillei (present at wave-exposed Wireless Island only), also declined significantly between the years (t = 5.28, df = 9, p < 0.001; Fig 2C). In contrast, densities of the rarely harvested limpet S. granularis remained constant between years at both sites (Fig 2E; S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2.

Mean (+1SE) densities (left) and shell lengths (right) in 1970 and 2017, for (A, B) Cymbula granatina, (C, D) Scutellastra argenvillei and (E, F) S. granularis. No historical length-data were available for S. argenvillei and it was absent from Wireless Point. Shell length data for were available for only one site (Wireless Island). Asterisks between years and different letters above error bars indicate significant differences (P < 0.05).

https://doi.org/10.1371/journal.pone.0294404.g002

Shell lengths of C. granatina and S. granularis differed significantly between the years at Wireless Island, with the commonly harvested C. granatina decreasing in size (t = 1.8738, df = 149, p = 0.031), and the rarely harvested S. granularis increasing (t = -4.769, df = 99, p < 0.001) (Fig 2B and 2F). No historical data existed for sizes of S. argenvillei, preventing a temporal comparison of sizes for that species (Fig 2D), and for C. granatina and S. granularis, historical size data were available only for Wireless Island.

Spatial comparisons.

Densities of C. granatina were significantly less at the harvested areas than in the no-take areas, with the density at Scarborough South being intermediate and not differing significantly from those at the other sites (S2 Table, Fig 3A). Sizes (i.e. shell lengths) of C. granatina also showed significant effects of protection level and sites (S3 Table; Fig 3B), being smallest at Wireless Point, largest at Scarborough South (the site that lies farthest inside the sanctuary area), and intermediate at Kommetjie and Scarborough North.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3.

Mean (+1SE) densities (left panels) and shell lengths (right panels) of the limpets (A-B) Cymbula granatina, (C-D) Scutellastra argenvillei and (E-F) Scutellastra granularis at the four study sites. Different letters indicate significant differences according to posthoc tests and refer to comparisons within each panel.

https://doi.org/10.1371/journal.pone.0294404.g003

Scutellastra argenvillei attained greater average densities inside the no-take area, with by far the highest densities found at Scarborough South, while few individuals occurred at the other sites (S2 Table; Fig 3C). The average shell length of this species was unambiguously larger inside the no-take Scarborough sites than in the harvested Wireless Point and Kommetjie sites (S3 Table; Fig 3D).

Densities of the rarely harvested S. granularis differed significantly among sites, but not between protection levels (S2 Table; Fig 3E). While density was low at Wireless Point, it did not differ statistically from that of Scarborough North, and density at Scarborough South was not statistically different from that at Kommetjie (Fig 3E). Its shell lengths were also not affected by protection levels (S3 Table), with Scarborough North and South being intermediate between the low values at Kommetjie and the high values at Wireless Point (S3 Table; Fig 3F). Mean density thus appeared to be negatively related to mean size, although it was not possible to test this statistically with a sample size of only four sites.

Temporal comparisons.

Differences in rocky-shore community composition at both sites existed between the years 1970 and 2017 (Fig 5), but the magnitude of the differences varied depending on the intertidal zone considered, as reflected in a significant Year × Zone interaction (S5 Table). At Wireless Island (Fig 5A), two clusters were distinguishable: 1970, characterised by greater abundances of the limpets S. barbara, C. oculus and C. granatina and encrusting algae; and 2017, when corticated and ephemeral algae, the alien mussel M. galloprovincialis and, to a lesser extent, S. granularis, were more prevalent. At Wireless Point (Fig 5B) the distinction between years was less clear-cut but still significant, with the 1970 cluster being driven by a greater abundance of encrusting algae, C. granatina and the ribbed mussel Aulacomya atra, and the 2017 cluster by more ephemeral algae, S. granularis and C. oculus (Fig 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5.

Unconstrained ordination with two latent variables based on taxon abundance in individual quadrats, showing differences between years and different zones (see key), at (A) Wireless Island and (B) Wireless Point. The position of species and functional group names in the ordination portrays indicator species/groups characterising each of the respective years.

https://doi.org/10.1371/journal.pone.0294404.g005

Exploring changes in the abundances of individual taxa over time (S6 Table) M. galloprovincialis was originally absent from Wireless Island, but had become well established by 2017, and ephemeral and corticated algae had increased significantly. At the more wave-sheltered and more heavily harvested Wireless Point, M. galloprovincialis failed to become established. Other effects were confounded by interactions between year and zone, but indicated that A. atra and C. oculus declined, whereas ephemeral algae increased. In the case of A. atra the interaction reflected its disappearance on the low shore and appearance on the high shore; C. oculus consistently declined but at variable rates among zones; ephemeral algae experienced increases in some but not all zones.

Temporal comparisons of densities of limpets.

The densities of commonly harvested limpet species declined substantially over the years. Previously, populations of C. granatina and S. argenvillei formed dense beds with densities of respectively 70-288/m² and 162-216/m², which equate well with those at other unharvested sites [17, 19, 55, 56]. Currently, the populations of C. granatina and S. argenvillei at Wireless Point and Wireless Island have declined by 56% and 97% respectively. These declines far exceed those reported elsewhere for experimental thinning of C. granatina and S. argenvillei, which were reduced respectively down to 33.4% and 25% of their original densities for individuals ≥60 mm and ≥75mm [17]: yet even at those less severe levels of depletion, ecosystem effects were detectable [55]. The depletion levels we recorded are more comparable to those reported for other limpet species considered to be heavily harvested, such as 50% for S. longicosta [57] and 75% for C. oculus [9] on the east coast of South Africa, decimation and the threat of global extinction of Scutellastra mexicana in the eastern Pacific [58–60] and the >90% reduction and local extinctions of Patella ferruginea in the Mediterranean [59] and Patella candei in the Canary Islands [61]. In contrast, the rarely harvested S. granularis displayed a (non-significant) average 25% increase in density over the years, supporting the idea that harvesting was the main cause of the declines of harvested species. Intensification of harvesting is attributable to a rise in human populations in the vicinity of Cape Town [2, 43] and near the Wireless study sites in particular, where the steadily growing low-income settlement of Masiphumelele was established in the 1980s.

Comparing the two harvested sites, the significantly greater densities of C. granatina at Wireless Island, compared to Wireless Point, probably reflects relatively low harvesting pressure at Wireless Island because it is detached from the mainland and less accessible. Inaccessibility of rocky shores is well known as a factor that reduces harvesting pressure on intertidal gastropods [4, 62–65]. In line with this, greater densities of Phorcus turbinatus, Patella ferruginea and Patella ulyssiponensis have been reported on inaccessible islands compared to accessible mainland sites in Italy [26, 66, 67], and for Haliotis spadicea and Scutellastra longicosta in South Africa [68, 69], the mussel M. galloprovincialis in Portugal [70, 71] and the alga Durvillaea antarctica in Chile [72]. However, the higher density of S. granularis at Wireless Island than at Wireless Point may be because it is indirectly benefitting from the greater exposure to wave action there, as the alien mussel M. galloprovincialis reaches greatest densities in areas of strong wave action [73], and provides secondary habitat that boosts the numbers of juveniles of S. granularis [15, 16, 18, 34, 35].

Temporal comparisons of shell lengths of limpets.

The decline in shell lengths of the commonly harvested C. granatina over time is consistent with effects of harvesting because large individuals are targeted as food [1, 22, 31, 74], a phenonemon that has been recorded even from prehistoric times [75, 76]. Declines in shell lengths of harvested species, e.g. Lottia gigantea [31], Patella caerulea (as P. candei crenata), Patella aspera [74] and Tegula funebralis [22], are common worldwide, especially in areas near residential developments that are frequently visited by shellfish gatherers [74]. However, other factors such as competition and resultant declines of food sources can also reduce average sizes of limpets [77]. Even gastropods that are not harvested, such as Nucella lapillus and Siphonaria lessonii, may decrease in size if food availability is reduced by competition [23, 78]. However, competition is unlikely to explain the reduction in size of C. granatina since its size reduced in parallel with a reduction in density. In its adult stage it becomes a ‘trapper’ that relies on trapping drift algae and is therefore relatively independent of local epilithic algal supplies [56]. Moreover, S. granularis, which is rarely harvested, increased in shell length over time in the same area. When harvesting was prevented in an MPA in Chile, the shell lengths of the two commonly harvested limpets Fissurella picta and L. gigantea increased despite their density rising [78]. These observations strongly indicate that harvesting, not competition, caused the declines in sizes of harvested species. These facts, together with the observation that sizes of S. granularis increased over time–likely as a result of reduced competition from harvested limpets–strengthen the argument that harvesting was the cause of decline in shell lengths of C. granatina.

Spatial differences in densities of limpets.

Protection played a major role in determining the densities of the commonly harvested C. granatina and S. argenvillei. The higher densities of C. granatina in the Scarborough no-take areas compared to the harvested Wireless Point and Kommetjie sites indicates it is benefiting from the greater protection in no-take areas, as has been demonstrated for limpet populations elsewhere [8, 9, 22, 79]. The case for S. argenvillei is less clear-cut, as its abundances were greater at only one of the protected sites. The four- to 15-fold differences in densities of C. granatina indicate that the no-take areas in the reserve succeed in maintaining the exceptionally high natural densities that this species achieves in other unharvested parts of the west coast [56]; and this is probably true for S. argenvillei as well. Similarly, higher densities have been reported for Cymbula oculus [9, 57], Perna perna [80] and S. longicosta [57] in MPAs on the South-east coast of South Africa, and for Patella ferruginea in Sinis-Isola MPA, Italy [67], following exclusion of harvesting from rocky shores.

The lack of protection effects for S. granularis was expected as this species was not collected by harvesters [39]. The low density of S. granularis at Wireless Point compared to Kommetjie and Scarborough South was, however, unexpected. Density of this species is enhanced in areas with a high cover of M. galloprovincialis, due to improved recruitment and survival of juveniles of the limpet in mussel beds [15, 16, 18, 34, 35]. Mytilus galloprovincialis was almost absent from Wireless Point. Differences in the densities of S. granularis may therefore be a secondary effect of the relative abundance of M. galloprovincialis, and unrelated to differences in harvesting. The most likely explanation for the scarcity of M. galloprovincialis at Wireless Point is that this site lies in the lee of Wireless Island and therefore experiences less wave action, and wave action supplies particulate food for the mussel [73]. However, the reasons for the high abundance of S. granularis at Kommetjie, where M. galloprovincialis was also scarce, remain unclear.

Spatial differences in shell lengths of limpets.

Because of the common practice for harvesters to select large individuals, mean sizes of targeted species are frequently reduced on harvested shores because of the truncation of larger sizes [9, 10, 30, 31, 33, 69, 81]. The presence of larger C. granatina and S. argenvillei at the protected Scarborough sites than at the harvested sites of Kommetjie and Wireless Point accords with size-selective harvesting [39]. The argument that harvesting is the root cause of this pattern in these species is supported by the fact that the rarely harvested S. granularis did not experience a reduction of size in harvested areas: indeed, its sizes at one of the harvested sites, Wireless Point, were 12–19% greater than at either of the protected sites. The differences in shell lengths of C. granatina and S. argenvillei translate to 8–12% declines, falling within the range of declines reported for other molluscan populations under intense harvesting pressure: e.g. 8.6–10% for Anadara scapha, Gafrarium tumidum and Modiolus auriculatus in New Caledonia [10, 11], 20–30% for C. oculus in South Africa [9] and 28% for Lottia gigantea in California [30].

Temporal changes in the zonation of Scutellastra granularis.

Our study is the first to report the effect of the alien M. galloprovincialis on the zonation patterns of S. granularis. Invasion by this mussel has shifted the centre of recruitment of S. granularis from the infratidal and bottom zones to the mid and high shore. Before the arrival of M. galloprovincialis, densities of S. granularis decreased upshore and its average size increased upshore–to the extent that Branch [77] classed it as one of a group of ‘migratory’ species that settle low on the shore and then shift progressively upshore–presumably escaping intense low-shore competition as it ages and growth diminishes the threat of desiccation [82]. The arrival of M. galloprovincialis increased densities of this limpet at Wireless Island, but concentrated its recruits in the mussel beds, with a downward movement of adults as they shifted from the mussel beds to vacant space below the beds. These findings agree with results of others who have reported lower densities of S. granularis where M. galloprovincialis is absent or scarce along the west coast, and reductions in the abundance of S. granularis in the infratidal zone after arrival of M. galloprovincialis [15, 16, 18, 34, 35]. Facilitation of gastropods by mussels is not unique to the sites where we worked. For example, in Chile the limpet Siphonaria lessoni and the whelk Nacella magellanica attain higher densities in beds of the mussel Perumytilus purpuratus than on bare rock [83].

Prior to the arrival of M. galloprovincialis, the zonation of S. granularis reflected a pattern typical of many ‘upshore’ species, such as Lottia digitalis, Collisella subrugosa, Helcion concolor or Siphonaria guamensis, for which recruitment occurs low on the shore and individuals migrate upshore as they become larger [1, 77, 84–86]. However, the invasion by M. galloprovincialis has altered the size zonation patterns of S. granularis: recruits and juveniles are now found predominantly in the mid to upper zones and adults in the infratidal zone. Mytilus galloprovincialis has ameliorated the previously harsh conditions on the high shore, which usually limit settlement (or survival) of recruits to the low shore, so that M. galloprovincialis promotes survival of recruits and juveniles of S. granularis [16, 34, 35]. Similar interactions have been observed on the south coast of South Africa, between S. granularis juveniles and barnacles, where high-shore barnacles moderate physical conditions, increasing recruitment of S. granularis there, but decreasing its size [87]. The presence of large-sized S. granularis in the infratidal zone indicates a downward migration of individuals as they grow into the zone where there are few individuals of M. galloprovincialis. Previous studies have shown that adult S. granularis inhabit patches of bare rock that are devoid of M. galloprovincialis and small ones predominate in mussel beds [18]. The downwards movement may reduce competition for space in the high shore where primary space suitable for adult S. granularis is limited, and algal food supplies reduced. Experimental removal of M. galloprovincialis results in an increase in the density of adult S. granularis on the primary space that becomes available [16]. Tanaka et al. [86] reported a similar positive correlation between the amount of bare rock and the sizes of the limpet Collisella subrugosa on the high shores of Brazil. This demonstrates the need for adequate primary space by adult limpets. Limitations on space availability on the mid- to upper zones due to monopolisation by M. galloprovincialis have forced adult limpets to reverse their previous upshore migratory behaviour [77].

Temporal changes in community composition.

Substantial changes in community composition took place between 1970 and 2017, for which there are several explanations. One obvious agent of change is the arrival and establishment of the alien Mediterranean mussel M. galloprovincialis, which drives shifts in rocky-shore community composition in South Africa in a number of ways [34, 35]. Due to its high tolerance of desiccation, high growth rate and fecundity, M. galloprovincialis monopolises mid to high zones and creates a complex habitat that reduces physical stress and supports a rich interstitial fauna [18, 19, 34, 35, 88]. The secondary habitat formed by this mussel increases the density of S. granularis in the top and high zones and while reducing its densities in the infratidal and low zones where they were previously most abundant. Following the arrival of M. galloprovincialis, increases in the abundance of S. granularis juveniles in mussel beds and diminishment of adults on the shrinking primary space have previously been reported on the west coast of South Africa [15, 16, 18, 34, 35].

A second obvious contributor to the observed changes in community composition over time is the intensification of harvesting, which was non-existent or minimal at the surveyed sites in 1970 and has since become a common practice at the two Wireless sites (and at Kommetjie) [39]. A reduction in targeted limpet species and increases in various macroalgal groups that proliferate under reduced grazing pressure are common effects of harvesting [7, 21, 33, 89–96]. The types of algae that came to predominate differed among sites, with ephemeral species increasing at Wireless Point and corticated species at Wireless Island. This may reflect differences in the abundances of limpets, thus influencing the types and amounts of algae that develop [20, 92, 94, 97–102]. Cymbula granatina used to be common at both sites but was depleted to a greater extent at the more easily accessible Wireless Point, where only few individuals remain on the low shore, and ephemeral algae have taken over the intertidal zone. The drastic decline in numbers of C. granatina is unlikely due to the arrival of M. galloprovincialis, as this mussel largely inhabits mid to high zones and prevails on exposed shores [19], whereas this limpet mainly occupies the low shore on sheltered to semi-exposed shores where it can trap drift algae [56]. Thus, the decline in C. granatina can more likely be ascribed to increases in the human population adjacent to the rocky shore and associated intensification of harvesting [2, 39].

Reductions in the densities of limpets affect not only algae, but predators that eat limpets. It has even been suggested that a contributory factor to the extinction of the Canarian Black Oystercatcher Haematopus meadewaldoi was over-exploitation of limpets [103]. Conversely, arrival of M. galloprovincialis in South Africa has boosted food supplies for the African Black Oystercatcher H. moquini and contributed to its rebounding population [104, 105].

Spatial differences in community composition.

Differences in the community composition between rocky-shore sites experiencing high harvesting pressure and those protected from harvesting have been reported many times in South Africa and worldwide [7, 11, 33, 95, 106–108]. Experimental removal of limpets has also often demonstrated this effect [96, 100], leading to increases in macroalgae. Increased macroalgal cover may in turn inhibit the settlement and recruitment of species that require primary substrate or a cover of crustose algae to settle, while increasing settlement of those species that live within algal beds. This has been observed in MPAs along the south and east coasts of South Africa, where protection elevates the abundance of sessile fauna e.g. barnacles and mussels, whereas algal mats are more abundant at harvested sites [7]. Similar to these studies, the harvested sites we surveyed (Wireless Point and Kommetjie) were dominated by extensive mats of Ulva spp., Pachymenia orbitosa and Mazzaella capensis, linked to the to the depletion of limpets such as C. granatina (and, at Wireless Island, S. argenvillei), creating opportunities for growth of opportunistic algae like Ulva spp. [7, 33, 106, 107]. The dominance of a single genus (Ulva) in the community of Wireless Point while at the other harvested site, Kommetjie, P. orbitosa and Sarcothalia stiriata were abundant suggests different degrees of disturbance between these sites. Dominance by opportunistic algae such as Ulva spp. reflects a shift between herbivores and algae, with shellfish-gathering increasing algal mats, while protection against harvesting strengthens the limpet/algal interaction, as reflected in the rarity of opportunistic algae at the Scarborough sites. Increased algal growth may have knock-on consequences for the functioning of the ecosystem, including reduced feeding and altered settlement of mussels, and smothering of other organisms through overgrowth [7, 33, 106, 107].

The differences in community composition between sites that experience high versus low harvesting pressure are usually due to commonly harvested species being abundant in protected sites, whereas at harvested sites algae and algal-associated species become distinguishing taxa [7, 8, 33, 106]. For example, Deepananda and Macusi [97] reported dominance of opportunistic algae such as Gracilaria cassa, Valoniopsis pachynema and Padina boergesenii in the community composition of harvested sites in Sri Lanka. Hence, the abundance of Ulva spp., an opportunistic species, at the harvested sites we examined was not surprising, and is a common response in areas where grazers have been depleted [20, 90, 92, 94, 98–100].

The greater abundances of C. granatina and S. argenvillei at the protected sites we surveyed (Scarborough North and South) indicates direct benefits of protection through no-take areas in this MPA. Similarly, the greater contributions of M. galloprovincialis to community composition at the protected Scarborough sites showed that this alien species is (ironically) also benefitting from the protection from harvesting. The greater contributions of the crustose coralline alga Chamberlainium spp. at the two protected sites reflect a strong association between Chamberlainium spp. and Scutellastra cochlear [109, 110], as average densities of this limpet were also greater in the protected sites, even although it is not harvested. It is also a reflection of a more general positive association between encrusting coralline algae and limpets, which has been well established in North America [111, 112] as well as on the west coast of South Africa [110, 113].

The predominant occurrence of Siphonaria spp. at harvested sites likely reflects a reduction in competition from patellid limpets, as several experimental tests have shown that siphonariids are inferior competitors to patellid limpets [114] due to their less efficient radulae and removal of food [115, 116].

Another prominent driver of decadal changes in community composition on rocky shores is climate change. Our analysis does not probe this, but geographic shifts of distribution and changes in abundance on the coast of South Africa have been securely linked to climate change for several species, including the kelp Ecklonia maxima [117], the rock lobster Jasus lalandii [118], and the brown mussel Perna perna [119]. These shifts have been attributed to changes in the intensity of south-easterly winds, leading to a cooling of nearshore waters on the south-east coast, and to increases in low-oxygen events on the west coast [43, 119].

Comparable dramatic cascading temporal and spatial effects have been demonstrated in Tasmania, where an expansion of the range of the urchin Centrostephanus rodgersii has been attributed to a combination climate change and overfishing of predatory fish [120, 121]. With the appearance of this urchin in Tasmania, and the subsequent increase in its abundance, kelp beds have been transformed into urchin barrens, with consequent diminishment of energy flow and biodiversity. Such urchin effects are widespread, as is apparent from a global analysis of urchin dynamics [122]. Our data similarly reveal considerable temporal and spatial changes in community composition and functioning that are related, in this case, to both harvesting of a key grazer and the effects of a dominant invasive alien, Mytilus galloprovincialis, with cascading effects up and down the food web, associated with alteration of habitat structure and ensuing effects, as we outline below.

Correlations between limpet densities and algal cover.

The negative correlation that emerged between limpets and ephemeral algae is not new: many previous studies have shown that removal of limpets increases ephemeral algae [94, 98, 99]. In our study, the highest abundances of these algae were at Wireless Point, where there were relatively few limpets compared to Kommetjie and Scarborough.

The parabolic relationship between corticated algae and limpet densities is more difficult to explain, although it is possible to speculate that disturbance (grazing) is a cause of low abundance at high limpet densities, and competition with opportunistic algae displaces them at low limpet densities.

The positive correlation between encrusting algae (dominated by Chamberlainium spp.) and limpets, particularly Scutellastra cochlear [123] is partially fuelled by the release of nutrients by S. cochlear, which are taken up by the surrounding algal ‘gardens’ this limpet occupies [124]. Encrusting algae constitute 80% of the diet of this limpet, and grazing intensity on the thalli promotes their growth rate [110, 124]. In addition, limpet grazing promotes encrusting corallines by preventing competitive overgrowth by foliar algae [110]. Steneck [112] has argued for a mutual dependency between the encrusting alga Clathromorphum circumscriptum and the limpet Acmaea testudinalis. Maneveldt and Keats [109] noted an absence of encrusting algae in the lower zones of west-coast areas where ever S. cochlear is absent. A similar positive relationship exists between encrusting algae and Patella aspera in Italy [125]. More recently, a mutualistic relationship between the limpet Patella ulyssiponensis and crustose corallines has been established in northwest Europe [41], raising the likelihood that the phenomenon is more widespread than is generally appreciated. This strengthens the argument that encrusting algae benefit from coexistence with grazers, particularly S. cochlear in this case.

Encrusting coralline algae have been shown to be preferential settlement sites for a range of invertebrates, including abalone Haliotis spp. [126] and the commercially important H. midae [127, 128]. Reductions in the abundance of grazers, with resultant effects on encrusting corallines, are therefore of more than academic interest.