Community structure is driven by biological interactions and physical processes that can vary across environmental gradients and spatial scales. Early ecological models focused on the role of resource availability (i.e. bottom-up effects), predicting that the strength of top-down control varied along gradients of primary productivity and that local species interactions determined community structure. However, the role of regional scale oceanographic processes in determining species interactions and community structure is now widely recognized, with bottom-up effects such as coastal upwelling driving regional scale patterns of resource availability. Such nutrient subsidies can significantly alter primary production and drive changes in algae-herbivore interactions in rocky intertidal habitats. However, despite the potential for upwelling to alter these interactions, studies investigating the effects of upwelling and grazing pressure are scarce, particularly for warm-temperate systems, and generally cover narrow geographical ranges. Using in-situ herbivore exclusion experiments replicated across multiple upwelling regimes, we investigated the effects of both grazing pressure and upwelling, as well as their interactions, on the sessile invertebrate community and primary production of macroalgal communities in a warm-temperate system. Invertebrate cover remained consistently low at upwelling sites and was reduced at non-upwelling sites when grazers were excluded. Macroalgal cover was greater at upwelling sites when grazers were excluded and there was a strong effect of succession throughout the experimental period. Grazing pressure was greater at upwelling sites, particularly during winter months. There was a non-significant trend towards greater grazing pressure on early than later successional stages. Our results show that the positive impacts of bottom-up effects of nutrient supply on algal production do not overwhelm top-down control in this warm-temperate system. We speculate that global increases in air and sea-surface temperatures in warm-temperate systems will promote top-down effects in upwelling regions by increasing herbivore metabolic and growth rates.

Experimental design. At each site, 15 experimental plots were established in 5 blocks around the mid-shore level across approximately 100 metres of shoreline, with a minimum of 2 metres between blocks. Each block contained 1 replicate from each experimental treatment. The experiment had a fully factorial design with two factors: 1) upwelling regime (fixed, two levels: upwelling and non-upwelling) and 2) grazing pressure (fixed, three levels: total exclusion, partial exclusion, total access). Grazer exclusion plots were established by enclosing plots with fences (35 x 35 cm, 10 cm high) constructed from stainless steel mesh (0.9 mm wire diameter, 5.16 mm aperture) fixed to the substratum with screws and washers. Fences were chosen instead of enclosed cages with lids to prevent light limitation (Appendix S1: Section S2). This allowed the exclusion of all benthic grazers from the fenced  plots. To test for experimental artefacts affecting on macroalgal production as a result of the presence of the fences, procedural controls were established using fences around half of the plot in an L-shape to allow partial grazer access. Experimental controls were marked at the corner with screws only, allowing complete grazer access. Fences were cleaned of any epibiota during every sampling period. Every experimental treatment was replicated five times at each site but adverse weather and sea conditions meant some cages were lost and sample sizes varied from 3-5 replicates per treatment per site. Prior to the beginning of the experiment, all experimental and control plots were scraped clear to ensure all living organisms were removed and algal and sessile invertebrate cover was 0 %. Plots were left for a 4-week period before surveys began. The experiment started in January 2021 and was surveyed monthly for 12 months.

Data collection. To quantify benthic grazing pressure, grazer density was recorded monthly at each site. Ten 25 x 25 cm quadrats were placed haphazardly throughout the mid-shore level and the identity (species level) and density of grazers were recorded. As grazer size has been shown to be an important factor in determining the outcome of algal-grazer interactions, grazers were classified into meso- and macro-grazers. On South African shores, dominant macro-grazers in the mid-low intertidal, including the gardening limpet Scutellastra longicosta and the limpet Cymbula oculus, are large individuals, > 20 mm, and well adapted to wave-exposed conditions (Díaz & McQuaid 2011, Branch & Odendall 2003; Nakin & McQuaid 2014). Meso-grazers are also abundant at the mid-shore level and include pulmonate limpets such as Siphonaria spp. and gastropods including Oxystele spp. approximately 15-20 mm in size (Whittington-Jones, 1997; Hodgson 1999, Díaz & McQuaid 2011). As these were the dominant grazer species identified in the current study, a 20 mm cut-off between grazer size classes was chosen to distinguish between meso- and macro-grazers.   

To quantify the effects of grazing pressure and upwelling regime on the colonisation of sessile invertebrates, percentage cover for each sessile invertebrate species in each plot was estimated monthly using the point intercept method with a 25 cm x 25 cm quadrat with 36 intersections, positioned carefully within each cage to avoid sampling edge effects. Although sedentary species such as mussels were present during the experiment, abundances were very low (> 1 individual per plot) and invertebrate communities were dominated by sessile invertebrates (predominantly barnacles). We therefore refer to sessile, not sedentary, invertebrate communities. A similar method was used to test for the effects of grazing pressure and upwelling regime on macroalgal percentage cover (as a proxy for macroalgal production). Sessile invertebrates and macroalgae were identified to species level where possible and species present within the quadrat but not occurring directly beneath an intersection were assigned a percentage cover value of 1 % (O’Connor & Crowe 2005). Total percent cover values of macroalgae sometimes exceeded 100 % owing to the multi-story nature of macroalgal communities. Grazers found in grazer-exclusion plots during monthly surveys were removed. At the end of the experiment, biomass was destructively removed from all plots, separated into species and dried at 60°C until constant weight.