**4. Discussion**

The main aim of this investigation was to ascertain the extent to which rock-rubble groynes can support similar ecological communities to natural rocky shores, which several studies have identified as being under pressure [32]. Our findings sugges<sup>t</sup> that although rock-rubble groynes support fewer species of intertidal macro-organisms than natural rocky shores (when measured using small quadrats across groyne-sized extents), these differences are not consistent across habitat (i.e., rock or pool) and tidal level (i.e., high, mid, low). On the exposed rock surfaces, the relatively small difference in species richness between the groynes and the natural rocky shores may reflect no more than a species–area effect over the entire ecosystem, i.e., the total area of rock-surface habitat is relatively small on the groynes compared with the natural rocky shores when viewed at a beach level. On the rock surfaces, the abundance of macro-organisms is, if anything, higher for the groynes. In the pools, it is much lower. Species that only occur on the groynes are associated with the higher abundance of mussels and barnacles (e.g., the Greenleaf worm) or they utilise the sandy habitat that is not present on the rocky shores (e.g., brown shrimp). Alternatively, the species found only on or predominantly on the natural shores appear to be those most sensitive to desiccation (e.g. *Desmarestia aculeata)*. Much of the difference between structures appears to be due to the paucity of rock pools on the groynes, especially complex pools (Table 5).

Differences in the complexity of the intertidal zone, particularly in relation to water-retention capacity, may therefore explain some of the results. Natural rocky shores typically have higher complexity, with more microhabitats providing shelter from desiccation and other stresses [43]—for example groves, gullies and cracks, all of which can maintain moisture at low tide. In algae, higher rates of water loss speed up the decline of photosynthesis and respiration [57], and when desiccation past a critical water content occurs, irreversible damage results [58]. Retention of water is also vital for mollusc survival and maintenance [59]. Incorporating rock pool features on artificial structures has been shown to increase the diversity of intertidal taxa, sometimes by four- to nine-fold [24,60–62]. The structure of rock pools on groynes was relatively simple. The predominant formation of pools on groynes was where substrate had become saturated and water had built up against the rocks, with a few pools in the higher tidal level having been formed by chemical weathering. The simplistic structure of these pools meant only certain species (e.g., *Palaemon serratus* with anti-predatory mechanisms of immobility and cryptic appearance [63]) could survive there.


**Table 5.** Mean numbers of rock pools in the sampling sections (±1 standard error).

While clear delineations of communities exist across the tidal levels for rock and pool habitats combined (Figure 4), there was a di fference between generalist species (e.g., *P. serratus, Anemonia viridis*) at the higher end of axis 2 and specialist species (e.g., *Halichondria panicea, Adocia cinerea)* at the lower end of axis 2 in the DCA. These results sugges<sup>t</sup> that distinct communities of species that are sensitive to desiccation exist at all tidal levels, but currently the structure of rock pools on groynes only support generalist species. It should be noted that pool depth and perimeter were not measured within this study, following similar approaches used on groynes [21,23], and while precautions were made to minimise variations among pools (e.g., pools of similar size; grouping of quadrats), there is the possibility that di fferences between pools may be due to pool depth and complexity [64,65]. Thus, while our results identified lower biodiversity in rock pool habitats on groynes, there may be potential for artificial rock pool habitats to replicate those found on natural shores, and further research into which species benefit from rock pool presence, as well as which type of rock pools harbour the most species on groynes needs to be undertaken.

Furthermore, in times of low amplitude, a characteristic of the diurnal currents in the English Channel (in Dorset) causes the e ffects of the varying topography to become pronounced, causing shallow water harmonics which are responsible for a double-tide [66]. This means that there is a relatively long stand of high water, and for approximately 16 hours of the 24-h cycle, the water level is above the mid-water spring tide level [67]. Consequently, while Sandbanks may not be as complex as a natural shore, it retains water in other ways through this double-tide and the intertidal species are subject to reduced levels of desiccation, which could explain the higher biodiversity observed there (Figure 3, Supplementary Information 5). We felt it was important to include Sandbanks in our study due to the contrasting results found in previous research [21,23], but it should be noted that tidal regimes have a large impact on the ecological communities, and while pairing sites by geographic proximity removed some of this uncertainty, the tidal regime may be a compounding factor within the county variable implemented in the ANOVAs (Tables 1 and 2).

The higher algal abundance at Sandbanks may also be explained by the rock surface heterogeneity. Most seaweeds attach to the rock through a combination of etching, gripping or glue, and carbonate rocks tend to support such attachment more than other rock types [68,69], and the physical and chemical substrate properties are known to a ffect organism adhesion and persistence [70]. The Sandbanks groynes are made of limestone, while the groynes at Sheringham and Sidmouth are made of syenite and granite, respectively (Table SI1). However, there was no significant di fference in algal richness or abundance between structures in Devon, although there was a significantly higher abundance of sessile animals on the Sidmouth groynes. The relatively low algal abundance at Sheringham and Sidmouth may be due to the simpler structure of rock material. We paired sample sites based on proximity rather than rock type in order to address questions pertaining to the spread of geographic distributions and compare nearby biological assemblages. Therefore, we are unable to separate possible effects of structure type and rock type; however, our results (Tables 1 and 2, Figures 2 and 3) do sugges<sup>t</sup> that features associated with the structure type may have a stronger influence than rock type alone. While geographic proximity was a key feature of our analysis, the distance between structures in Devon was approximately 100 km. Stokes Beach is publicly-accessible and not characterised by steep cliff faces (as is typical in the region), and represented the closest suitable, safe and accessible rocky shore, rather than necessarily the closest in Euclidean space. Future research should aim to investigate the role of type in the biological assemblages of groynes, as well the conservation value of ensuring that rock type of artificial structures mimics nearby rocky shores (as a form of ecological engineering).

The higher abundance of sessile animals on rock-rubble groynes has important conservation implications. The dog whelk *Nucella lapillus* was recently included in the OSPAR List of threatened and/or declining species and habitats. We found that this species was more prevalent and abundant on the groynes where there was a higher abundance of sessile animals (e.g., mussels and barnacles), which make up a substantial part of *N. lapillus'* diet [71]. The simple structure of groynes provides refuge for many generalist species, which in turn provide trophic support to the wider ecosystem, resulting in clear communities of dog whelks and barnacles (Figure 4). We also found *Pomatoschistus minutus* on both natural shores and groynes. While the total abundance was higher on natural shores (Table 3), the presence of this species on groynes suggests that they have the potential to mimic natural shores and rock pools, ye<sup>t</sup> currently do not support equivalent abundances. Interestingly, the DCA identified *P. minutus* within communities of Sea Oak *Halidrys siliquosa* and *Corallina o*ffi*cinalis* (Figure 4)*. H. siliquosa* supports a range of invertebrate epifauna and epifloral (including red algae such as *C. o*ffi*cinalis*), suggesting that the trophic interactions among these species are vital for the presence of rare species. Moreover, the role of ecosystem engineers [72] in generating microhabitats that provide refuge by reducing the impact of abiotic stresses in coastal ecosystems is important for intertidal communities. The incorporation of biotic interactions in statistical models is a key research frontier in spatial ecology [73,74], and our results (Figure 4) sugges<sup>t</sup> such interactions are key to determining representative communities on artificial structures. However, more research is needed to link the abiotic and biotic factors responsible for intertidal communities on artificial structures and highlight key ecosystem engineering designs that could be introduced to increase biodiversity during construction.

*Austrominius modestus* was the only species identified in our study that is not native to Great Britain. As such, we cannot sensibly comment on the proposition that artificial structures support a higher richness of non-indigenous taxa, beyond stating that this species was located on both structure types in Devon, but only on the natural shore in Norfolk—contradicting the proposition. We did find higher abundances on the groynes in Devon than on the nearby rocky shores (Table 4), which supports other studies that have found higher densities of *A. modestus* on artificial structures in the UK [75]. However, the indigenous barnacle species were also more abundant on groynes than natural rocky shores, so the non-indigenous species may not differ from the natives in this respect. Possible explanations for the increase in barnacle abundance on groynes could be a function of larval supply, availability of space (i.e., lack of microflora), rock type and post-settlement mortality [76,77], meaning future studies should aim to quantify the possible risk associated with such factors in quantifying non-indigenous taxa spread via artificial structures.

Furthermore, we only observed two native British species beyond their previously-recorded ranges: the lined top shell *Phorcus lineatus* at Sandbanks and the small periwinkle *Melarhaphe neritoides* at Sheringham. According to Crothers [78], *P. lineatus* has an Eastern limit around Kimmeridge bay (Figure 1), due to a combination of unsuitable winter temperature and, importantly, lack of suitable habitat to the East. According to Pizzolla [55], *M. neritoides* is absent between Ipswich (approx. 140 km Southeast of Sheringham) and Scarborough (approx. 400km North of Sheringham). Both species are considered highly-mobile [78,79], and their establishment on the groynes suggests that these artificial

structures are providing alternative habitat for species that are limited by a lack of habitat rather than poor dispersal capacity. In sum, while rock-rubble groynes may provide non-indigenous species (whether defined nationally or internationally) with suitable habitat (along with indigenous species), we found no evidence that they enhance the dispersal of these species; more likely is that habitat availability is one of the most important factors limiting intertidal organism distributions [80].

We opted to delineate tidal heights based on the spring tidal levels rather than organism distribution limits for the aforementioned reasons. However, the double high tide at Sandbanks complicates this categorization. A continuous variable measuring distance from a specified point (e.g., the high spring tide mark) may have allowed for more flexibility in investigating the influence of tidal level on the ecological indicators; however, a simple distance metric would not account for variation in the distance and speed of tidal ranges. Discussion on the optimal method of measuring the seaward extent and tidal levels across di fferent study areas remains a priority for studies investigating intertidal ecology. Similarly, we did not measure vertical height. While vertical height influences desiccation rates of species, the range of vertical elevation across tidal levels was quite consistent across study sites, and as such, we are confident that any possible artefacts introduced through the chosen methodology have not biased our results. Similarly, we controlled for aspect through surveying the 'sheltered' side of the structure; while aspect in the sense of compass direction can also influence photosynthetic rates of algae [6,81,82], the predominant aspects of all groynes were East- and West-facing, minimising the influence of this in our study.

Finally, it is important to note that the age of an artificial structure will influence the ecological communities found. All groynes surveyed were of a similar age (Table SI1), so therefore, we are confident that cross-site error was minimised. However, this is an important consideration when discussing the ecosystem services provided by groynes, and particularly comparing with natural rocky shores, and when comparing groynes across di fferent studies. Our results sugges<sup>t</sup> di fferent patterns to those identified by Pinn et al. [21] and Firth et al. [23]. The methodological di fferences among all three studies (e.g., deconstruction of the ecological indicators, direct statistical testing, unpaired sites) may be su fficient to explain contradictory results; however, it is important to consider the dynamic nature of ecosystems, and the unlikeliness of a balanced equilibrium. More developed groynes (e.g., older) may explain the higher species richness identified in pool habitats on artificial structures by Firth et al. [23], suggesting that these structures may develop complexity with time and support more diverse ecosystems. Similarly, older groynes o ffer greater opportunity for invasion than younger groynes simply by being around for longer. Therefore, investigating spatiotemporal trends in biodiversity on artificial structures in urban environments should now be a primary avenue of research, to better evaluate their potential as a conservation opportunity, with the data from this study available in Supplementary Information 2 to encourage such studies.
