This study was carried out on intertidal cohesive platforms at Copacabana (38° 14′ S, 57° 46′ W), 5 km northeast of Miramar, Buenos Aires Province, Argentina (Figure 1A). These platforms are located at the base of an 8–12 m high cohesive cliff (known as Barranca Parodi in the stratigraphic literature; Figure 1B) developed in the Chapadmalal Formation (3.2–4.5 million years BP), which is primarily composed of reddish-brown, sandy siltstones originating from reworked loess deposits and sparse indurated caliche levels and nodules (Zárate and Fasano, 1989; Isla et al., 2015; Rico et al., 2020). The shore in this area faces southeast, which is the direction of the strongest swells (Fiore et al., 2009), and can be classified as ‘‘exposed’’ according to the MarLin wave exposure categories (i.e., an open coast facing away from prevailing winds but with a long fetch, and where strong winds are frequent)¹. Current velocities on the surface of the study platforms can be as high as 13 m s^–1 (Gutiérrez et al., 2018). The tidal regime is semidiurnal and microtidal (0.80 m mean amplitude; Servicio de Hidrografía Naval, Argentina)². Current coastal retreat rates in this area average 0.8 m yr^–1 (Isla et al., 2018).

As is common for cohesive and soft rock platforms (e.g., Ledesma-Vázquez and Johnson, 1994; Gómez-Pujol et al., 2014; Moses, 2014), the studied platforms are cut by interconnected rills that run mostly perpendicular to the shoreline and are, thus, fragmented into elongated benches of varying size from a few square decimeters to a few square meters (see Figure 1C). The relatively small mussel, Brachidontes rodriguezii (up to 55 mm in length, most individuals <30 mm length), forms dense and primarily single-layered beds (200–800 ind. dm^–2; Arribas et al., 2015; Gutiérrez et al., 2015) along the top and along the upper sides of the benches located within the mid intertidal zone (Figure 1D). The height of these mid intertidal benches varies between 50 and 80 cm, though their top surfaces are relatively level with each other (i.e., height differences are mostly associated with variations in rill depth). Their lower sides and the bottoms of rills are periodically buried and abraded by sand and gravel (both sourced from nearby beaches; see Lamarchina et al., 2021) and, thus, lack biotic cover (Figure 1D). Abrasion and undercutting of the benches is likely a significant mechanism of overall platform erosion independently of the biotic platform-lowering effects investigated here.

Sixteen benches (0.4–0.8 m^–2 top surface area) were selected on June 29, 2018, and mussels were removed from half while the remaining benches were left as mussel-covered controls. The mussel layer was pulled by hand or with the aid of a putty knife when strongly attached, but taking care not to scrape the bench surface. Treatments were assigned systematically to ensure adequate interspersion. Platform height, substrate hardness, flow/abrasion rates, near-surface temperatures, substrate moisture, and subsurface salt concentrations were subsequently measured over the course of the experiment (see section “Measurements of Response Variables”). The experiment was ended on November 21, 2018, following the incipient recruitment of barnacles, Balanus glandula, to the mussel-removal benches. While the size (<1 mm) and surface area covered by these barnacles (<5%) were small so as not to interfere with our measurements and sampling around this date, we predicted that barnacle coverage would become significant in the following weeks due to continuing recruitment and individual growth. This was confirmed in late December when the barnacle cover reached 20–30%.

Repeated measures ANOVA (Neter et al., 1990) was used to evaluate the effect of Treatment (i.e., mussel removal; fixed factor) and Time (i.e., sampling date; random repeated measures variable) on platform height loss. Beta regression (i.e., a technique to model continuous random variables that assume values in the 0–1 interval; see Cribari-Neto and Zeileis, 2010) was used to evaluate the effect of Treatment on flow/abrasion rates (proportional weight losses from the plaster discs) and substrate hardness (Shore A units ranging between 0 and 1), as well as the effect of Treatment and Depth (fixed variable) on substrate moisture (proportional weight loss after oven drying) and salt content (salt weight/sample weight). All analyses were run in R (R Core Team, 2020) using the “lme4” (Bates et al., 2015) and “betareg” (Cribari-Neto and Zeileis, 2010) packages for the repeated measures ANOVA and beta regression, respectively.

Mussel cover efficiently buffered high temperatures during air exposure; near-surface temperatures on the mussel-covered benches during a period of exposure on a hot spring day (11:10 AM to 2:05 PM) were lower than on the cleared bench surfaces (Figure 2). This is in line with previous studies showing the dampening effects of algae or sessile invertebrate covers on the temperature of intertidal consolidated substrates during daytime air exposure (Coombes et al., 2013, 2017; Baxter, 2015; Gowell et al., 2015; Scrosati and Ellrich, 2018). The heat-attenuating effects of mussels were striking, with reductions ranging between 7 and 12°C (Figure 2). This is attributed to substrate shading and, possibly, to increased evaporative cooling due to the water retained in the interstitial spaces of the mussel beds as well as the mussel bodies themselves (see Helmuth, 1998). Evaporative cooling mechanisms have also been proposed to explain lower substrate temperatures in the presence of seaweeds and barnacles (Gowell et al., 2015; Coombes et al., 2017).

As well as general reductions in temperature, mussel cover was found to reduce short-term, seconds-to-minute fluctuations in temperatures during air exposure, indicated by the more jagged temperature curves for the mussel-removed surfaces in Figure 2. By modulating these rapid temperature fluctuations (attributable to passing cloud and wind gusts) as well as the amplitude of daily thermal cycles (associated with tidal regimes and day–night cycles, e.g., Coombes, 2011; Pappalardo and D’Olivo, 2019), mussels are expected to reduce the frequency, magnitude, and rate of substrate expansion/contraction. While not likely sufficient to initiate breakdown via thermal “shock” effects (see Hall and Thorn, 2014), the buffering of such repeated, low-magnitude thermal changes is thought to limit the potential for thermal “fatigue” in a range of environments including on rocky coasts (Coombes and Naylor, 2012; Coombes et al., 2013; Hall and Thorn, 2014; Molaro et al., 2020). From the data in Figure 2, it can also be inferred that sudden temperature decreases occur when air-exposed substrates heated by the sun are rapidly cooled by the incoming tidal waters (the typical water temperature in this area and time of the year is 16°C) and that mussel layers can also modulate these changes. It remains to be evaluated whether buffering of such higher-magnitude and rapid thermal changes by mussels and other forms of biota alters the potential for thermal shock-related breakdown on coastal platforms (Coombes, 2011, 2014).

The dampening effects of mussel cover on high near-substrate temperatures were associated with concurrent decreases in substrate desiccation and salt crystallization relative to cleared areas. The moisture content of the mussel-covered substrates was significantly higher than that of the cleared substrates after a few hours of exposure to hot air temperatures at low tide, and this difference was consistent across the three 5-mm depth ranges of the substrate samples (Table 1 and Figure 3). This indicates that mussel covers slow substrate desiccation rates and reduce net substrate desiccation, at least during hot periods of tidal exposure. This has implications for the efficacy of wetting and drying as a weathering mechanism across the intertidal and splash zones in association with the repeated expansion and contraction and microtopographic fluctuation of the substrate (e.g., Kanyaya and Trenhaile, 2005; Trenhaile, 2006; Porter and Trenhaile, 2007). Indeed, the modulation of wetting and drying cycles (and associated expansion and contraction) by biota, including microorganisms (e.g., Mayaud et al., 2014), may have significant weathering implications for a range of substrates, especially clay-bearing types including mudstones (e.g., Stephenson and Kirk, 2000; Baxter, 2015) and the sandy siltstones of our study platform composed of 10–20% coarse-to-medium silts and 60–70% fine silt and clay (Rico et al., 2020).

The subsurface salt content of the platform material further indicates the moisture-modulating role of mussels. Specifically, we found higher concentrations of both chloride and sulfate ions in the upper 5-mm layer of the substrate from the mussel-removal benches relative to mussel-covered controls, while there was no difference in the deeper layers (5–15 mm) (Table 2 and Figure 4). This indicates that salt crystallization is reduced in the uppermost layer of the platform when covered by mussels—probably in association with their temperature-buffering and moisture-retaining functions. While such effects have been proposed based on laboratory simulations (see Gowell et al., 2015), this provides some of the first direct field evidence that biotic cover can reduce the efficacy of salt crystallization in the surface layers of intertidal platforms (also see Baxter, 2015 and Coombes et al., 2017). Salt weathering is thought to be important on rocky shores during hot weather, when desiccation during daytime tidal exposure causes the crystallization of dissolved salts in confined rock pores and associated destructive forces (Mottershead, 1989, 2013; Stephenson and Kirk, 2000; Porter and Trenhaile, 2007; Coombes, 2014). As well as reduced amounts of salt overall, lower crystallization frequencies and the less frequent hydration and thermal expansion/contraction of salt crystals within the upper layers of mussel-covered substrates due to the observed moisture/temperature-buffering effects may limit weathering relative to exposed substrates (e.g., Gowell et al., 2015; Coombes et al., 2017).

Based on the significantly greater weight losses of the plaster discs placed on mussel-covered benches relative to those without mussels (χ² = 12.81, df = 1, p < 0.001; Figure 5), our results indicate that hydrodynamic and abrasive forces are lower within mussel beds. This conforms to the findings of Carrington et al. (2008), who observed that water flow in the interstitial spaces of mussel beds is low relative to above mussel beds and in mussel-free estuarine bottoms. This provides some initial support for mussel beds reducing the mechanical impacts of sediment-laden waves and tidal flows on platform rock—especially very soft rock as studied here—both by physically armoring the surface and moderating near-surface water flows. Similar findings were reported by Baxter (2015) for mussel-colonized and artificially cleared patches on an intertidal mudstone platform in Wales, United Kingdom. Further work is now needed to determine whether such effects translate into altered rates of erosion and direct loss of material from the surfaces of colonized intertidal rock.

Mussel removal led to significant decreases in platform height (Table 3). Height losses in the mussel-removal treatments averaged 2 mm after 6 months, while the height of mussel-covered benches remained relatively constant over the same period, with minor variations attributable to measurement error as well as possible swelling and shrinking in response to wetting-drying effects (Figure 6). The observed decreases in platform height following mussel removal cannot directly be ascribed to the weathering processes discussed in section “Effects of Mussels on Weathering Regimes” as the height losses mostly occurred during the first 2 months of the experiment and under winter conditions (see Figure 6). Compared to other times of the year, wetting-drying and salt weathering are expected to be less efficient during winter, when desiccation is limited due to relatively low temperatures and irradiation. On the other hand, thermal fatigue due to short-term temperature variations (i.e., second-to-minute fluctuations attributable to winds and clouds; see section “Effects of Mussels on Weathering Regimes”) might continue to operate in winter and may, therefore, have contributed to the observed lowering of the mussel-removal benches. The likelihood that frost weathering occurred during these months is also low as minimum air temperatures fell below the freezing point of seawater (−2.5°C) on only two occasions, when the platform was exposed to such temperatures for less than 2 h (see Table 4). This suggests that the duration of freezing periods and the number of freeze-thaw cycles were well below those required for the weathering of consolidated intertidal substrates (see Robinson and Jerwood, 1987; Dewez et al., 2015). Nevertheless, the potential impacts of biotic cover on the ice-driven breakdown of coastal rocks warrants further investigation, as freezing processes can be important in some situations (e.g., Trenhaile and Rudakas, 1981; Trenhaile and Mercan, 1984; Robinson and Jerwood, 1987; Dewez et al., 2015). Mussels may be especially suited to protect intertidal rocks from frost weathering as they have evolved local adaptations to survive freezing temperatures (e.g., down to −13°C in the case of Arctic populations of Mytilus edulis; Thyrring et al., 2020).

The hydraulic and abrasive impacts of sediment-laden waves are also unlikely to be the cause of the measured lowering of the mussel-removal benches during the first 2 months of the experiment. Although storm conditions conducive to erosion in the study area tend to be more frequent during the winter (i.e., the early phases of our experiment), they occur during all seasons (see Fiore et al., 2009). Within the specific timeframe of this experiment, we found no evidence of increased storm frequency or intensity during the earlier, winter months. Indeed, based on an analysis of positive storm surges (i.e., positive deviations of the observed tide relative to the astronomic tide that exceeded 30 cm and persisted for 6 h or more; see Fiore et al., 2009), storms and wave energy were evenly distributed across the experimental period (see Figure 7) and cannot, therefore, directly explain the temporal pattern of height loss on the mussel-removal benches.

One possible explanation is that mussels weaken the rock they are attached to, for example by enhancing chemical weathering via the retention of water and organic matter (see Moura et al., 2012), but continue to keep these surfaces consolidated as long as they remain in place. Once detached or removed, lowering rates would be relatively rapid at first and then stabilize. This process is analogous to the rapid, catastrophic loss of rock and stone in the terrestrial environment following the removal of protective weathering crusts, which exposes the softened, highly erodible material underneath (e.g., Smith et al., 2010). While it is difficult to confirm such effects, it is possible that the impacts of mussels and other forms of biotic cover on intertidal rock weathering may involve a complex balance of bioprotective (when in place) and bioerosive (once detached/removed) effects.

Another possible explanation is that the experimental removal of the mussels weakened the surfaces of the benches. Mussel detachment involves the failure of the collagenous byssal threads they produce to adhere to the substrate. Typically, byssus failure occurs either because the thread itself or its basal portion breaks (thread and root failure, respectively), the adhesive plaque at the base of the thread detaches from the substrate (plaque failure), or the substrate to which the adhesive plaque attaches is dislodged (substrate failure) (see Bell and Gosline, 1997). Individually, a substrate failure implies the removal of a small substrate chip from the platform surface (i.e., at scales of tens to hundreds of micrometers, based on adhesive plaque size). Yet, a high number of substrate failures per unit area can occur during the mass detachment of mussels. For instance, mussels 15–20 mm in length at our study area have approximately 80 byssal threads on average, and 30% of the failures observed during their detachment occur at the substrate level (Gutiérrez et al., 2018). Considering that mussels within this size range can occur at densities of 200 ind. dm^–2 at this site (Soria, 2020), their mass detachment (either due to experimental removal or storm wave action) could cause approximately 4,800 substrate failures per dm^–2. Although an underestimate is based on a subset of mussel size classes occurring in the field, this calculation illustrates that these highly friable bench surfaces could be densely chipped and weakened during mussel detachment.

While most of the observed decrease in platform height occurred within the first 2 months after mussel removal, with a slower rate of lowering thereafter, it is not possible to rule out the potential for a subsequent period of rapid lowering. Unfortunately, barnacle recruitment prevented us from continuing the experiment over the summer (see section “Field Experiment”) under higher temperature, irradiation, and desiccation conditions. However, our substrate hardness measurements, which were carried out at the end of the experiment in late spring, indicated significantly lower surface hardnesses in the mussel-removal patches compared to the mussel-covered controls (χ² = 12.36, df = 1, p < 0.001; Figure 8). This suggests that progressive surface weathering during the 6 months following mussel removal was linked to the development of weaker bench surfaces relative to those with mussels still attached.