**3. Results**

#### *3.1. Species Richness and Abundance*

We recorded 81 species in total: 27 algae, 6 lichens, 11 sessile animals and 37 mobile animals (Supplementary Information 4). We found 48 species on the groynes and 71 species on the natural rocky shores. Species richness was higher on the natural rocky shores than on the groynes; however, this varied by species group and habitat type (Figure 2a,b). On rock surfaces, species richness on natural shores was significantly higher than that for groynes (Figure 2a, Table 1); however, when species groups were disaggregated, the di fference was not significant for algae and there were significantly more sessile animal species on groynes. In the rock pools, species richness was substantially higher on natural rocky shores than groynes for all species groups except sessile animals (no significant di fference) and lichens (none recorded in pools) (Figure 2b, Table 1). Indeed, the rock pool habitat represents a key di fference between the groynes and the natural rocky shores. Of the 81 species we recorded, 66 were present in rock pools, including 21 only recorded in rock pools. Of these 21, twelve were only in natural rock pools (5 algae, 7 mobile animals), three only in pools on groynes (all were mobile animals) and six in both (5 mobile animals, 1 sessile animal). Species richness varied significantly between counties and across tidal levels, and these factors interacted with structure type to varying degrees to a ffect the magnitude of species richness di fference between groynes and natural rocky shores (Figure 3, Table 1).

**Figure 2.** Comparison of species richness on (**a**) rock surfaces and (**b**) rock pools, and of species' abundance on (**c**) rock surface and (**d**) rock pools, between rock-rubble groynes and natural rocky shores. Error bars represent 1 standard error of the mean. Abundance for algae, lichen and sessile animals is shown as percentage cover, and for mobile animals is shown as the total count.

**Table 1.** Results of the analysis of variance (ANOVA) testing the e ffects of County, Type, and Level, on average species richness per quadrat, averaged over a set of up to three quadrats. 'Type' is the rocky structure type (natural vs. artificial). 'Level' is the tidal level (low, medium, high). 'County' refers to the three stretches of coastline (Devon, Dorset, Norfolk). \*\* *p* < 0.01, \* *p* < 0.05.


**Figure 3.** Comparison of species richness on (**a**) rock surfaces and (**b**) rock pools, among tidal levels and counties. Error bars represent 1 standard error of the mean.

Findings were di fferent when we considered abundance (Figure 2c,d, Table 2). We found no significant di fferences in abundance between structure types for algae or mobile animals in the rock habitat, although in pools the abundances were significantly higher on natural shores. Again, we found a significantly higher abundance of sessile animals on the rock habitat of groynes than natural shores, although this pattern was not significant for pool habitats. Abundance also depended strongly on the tidal level, county, and habitat studied (Supplementary Information 5). For example, in Devon, algal abundances on natural shores increased towards low tide on the rock habitat while it decreased towards low tide in pool habitats. Di fferences also existed across structures, with algal abundances in Norfolk decreasing towards low tide on groynes in pools but increasing on natural structures in pools (Supplementary Information 5). Abundance di fferences between structures were notable for mobile species, particularly in rock pools (F = 56.97; Table 2). The abundance of mobile animals was generally greater on natural rocky shores, with this di fference prominent in Dorset. However, we observed nearly the opposite pattern in Devon, with abundances of mobile animals not significantly di fferent across structures in pools and on rocks at most tidal levels, and abundance higher at low tide for groynes than natural shores.

**Table 2.** Results of the analysis of variance (ANOVA) testing the e ffects of County, Type, and Level, on average species abundance per quadrat, averaged over a set of up to three quadrats. 'Type' is the rocky structure type (natural vs. artificial). 'Level' is the tidal level (low, medium, high). 'County' refers to the three stretches of coastline (Devon, Dorset, Norfolk). \*\* *p* < 0.01, \* *p* < 0.05.


#### *3.2. Species- and Community-Level Analysis*

Of the 21 species found only in rock pools, most were mobile organisms. Of these mobile species, *Idotea granulosa, Gibbula cineraria* and *Lacuna vincta* could potentially survive outside of pools; however, all are sensitive to rapid desiccation or feed on algal species that require water. The other mobile species were either shrimp or fish, which could not survive on open rock. The five species of algae that were only found in rock pools all require sublittoral habitats, except *Gelidium pusillum*, and were all found only on natural shores. The species found only in pools on groynes were generalist species, which were widely distributed and associated with sandy habitats, such as the brown shrimp *Crangon crangon.* Species only found on rock surfaces on groynes were the Greenleaf worm *Eulalia viridis* and the Dahlia anemone *Urticina felina*, with each only identified at one site: Sidmouth and Sandbanks, respectively. We identified only one Dahlia anemone at Sandbanks, while we found Greenleaf worms (in high abundances) in every low tide quadrat of one groyne at Sidmouth. Both species are habitat-specific, with the anemone requiring crevices and water, and the Greenleaf worm requiring mussel beds. Our dataset contains three species that are of conservation concern (Table 3), with the dog whelk *Nucella lapillus* occurring in higher abundances on the groynes and the other two at slightly higher abundances on the natural rocky shores.

DCA (Figure 4, Table SI4) revealed two main gradients in the di fferentiation of intertidal species composition. The first axis (DCA1) appears to be influenced by type and county and contains the most spread [eigenvalue 0.8114], while the second axis (DCA2) appears to be influenced by tidal level, with less spread [eigenvalue 0.5751]. Distinct communities were observable on the ordination plot. For example, communities found high on axis 1 are dominated by samples from natural shores at high tide and consist of predominantly lichens. Lichens favour areas of high stress and disturbance and are only found on bedrock or boulders [52], so were expected to be found on rocks in the high shore; however, their absence on groynes is notable and is perhaps related to the slow growth rates observed in marine lichens [53]. Species found low on axis 1 are those that occur on groynes in high abundance (e.g., Northern Acorn Barnacle *Semibalanus balanoides, Ulva intestinalis*) across all sites. Species found at either end of axis 2 include sponges and algae sensitive to desiccation at the lower end and fish and anemones at the higher end. These results sugges<sup>t</sup> that distinct communities of species that are sensitive to desiccation exist at all tidal levels. Overall, this analysis suggested only a slight di fferentiation between communities on groynes versus natural rocky shores, part of which reflects the absence of lichens on the relatively new surfaces of the groynes.

**Table 3.** Total abundance of the three species of conservation concern. For *Padina pavonica*, the units of abundance are percentage cover averaged across a set of up to three quadrats; for *Nucella lapillus* and *Pomatoschistus minutus* they are the total number of individuals per quadrat averaged over a set of up to three quadrats (see Methods). Conservation conventions and legislation: UK Biodiversity Action Plan (UKBAP), the Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR), and Bern Convention is the Bern Convention on the Conservation of European Wildlife and Natural Habitats.


**Figure 4.** Ordination plot of detrended correspondence analysis (DCA) results on all species for rock and pool habitats combined. Labels represent species (see Supplementary Information 4 for a full list). Significant (α < 0.01) environmental variables are represented as factor centroids with standard error ellipses.

#### *3.3. Species Ranges*

We observed only two species beyond the ranges recorded for them in the sources we used: the lined top shell *Phorcus lineatus* at Sandbanks (only just beyond its recorded range [54]) and the small periwinkle *Melarhaphe neritoides* at Sheringham (closest recorded presence near Ipswich, nearly 140 km away [55]). Our dataset contained only one species not native to Great Britain: *Austrominius modestus*, identified at Stokes, Sidmouth and West Runton (Table 4). Gibson et al. [50] identified this species as present at all three sites, while Avant [56] recorded this species as only present at Sidmouth and Stokes. When present on groynes, *A. modestus* was at higher abundances than on the paired natural rocky shores (Table 4).

**Table 4.** Mean abundance of *Austrominius modestus* on open rock at different shore levels. The unit is the percentage of cover.

