Timing Patterns of Sea Urchin Overgrazing on Nutrient-Rich Coral Reefs

Our study was conducted at Williamstown Beach in Port Phillip Bay (PPB), Victoria, south-eastern Australia where kelp beds (dominated by Ecklonia radiata; (C. Agardh) J. Agardh 1848, Family: Lessoniaceae) are declining [38] and remnant kelp beds are under risk of transition to sea urchin barrens characterised by a lack of kelp and other macroalgae. The sea urchin Heliocidaris erythrogramma (Valenciennes, 1846; Family: Echinometridae) represents the most wide-spread and ecologically important herbivore in PPB and demonstrates active overgrazing of kelp beds [39,40]. H. erythrogramma occurs throughout PPB and at local scales can occur at high densities (up to ~ 120 individuals m^-2).

To facilitate manipulating sea urchin densities and administering elevated nutrient conditions, ‘artificial’ patch reefs were constructed from naturally occurring boulders which were placed on sand substrata adjacent to Williamstown Beach (S 37° 52’ 10.5564”, E 144° 53’ 36.4884”). The site is characterized by shallow sub-tidal rocky reef with large boulders and cobbles interspersed by sand patches. The majority of the sub-tidal reef area (2 to 5.5 m depth) is H. erythrogramma sea urchin barrens, but a small number of remnant kelp patches dominated by E. radiata can be found. The macroalgal community consists of up to 95% cover of E. radiata, the fucoid Sargassum vestitum ((R. Brown ex Turner) C. Agardh 1820, Family: Sargassaceae) and the introduced Japanese kelp Undaria pinnatifida ((Harvey) Suringar, 1873, Family: Alariaceae) with sparse understorey growth, while on the shallowest margin of the sub-tidal area (1.5 m), Sargassum spp. are the dominant canopy-forming macroalgae.

In the study area the sea urchin H. erythrogramma occurred at an average density of 4.6 m^-2 in local urchin barrens and 5.2 m^-2 in kelp beds (at a biomass density of 245 g m^-2 and 539 g m^-2 respectively). Following pilot trials at this site over 6 months, a large sand patch (2,040 m²) surrounded by rocky reef in a depth of 4 m was selected as a suitable habitat upon which to build experimental boulder patch reefs. Pilot trials demonstrated that natural sand barriers were effective in limiting migration of urchins between patch reefs, enabling urchin density treatments to be maintained efficiently. This approach was superior to using artificial fences or cages, which quickly became fouled and required weekly maintenance.

Reefs initiated in the ‘kelp bed’ state demonstrated rapid decline in kelp cover when subject to high densities of grazing urchins (Fig 1a and 1b). The higher the sea urchin density, the more rapid the loss of kelp (Fig 1a and 1b). Decline in kelp cover under moderate to high urchin densities (12–24 urchins m^-2) occurred within 3 months (Fig 1a and 1b; Table 2). After 13 months, any patch reef initiated as a ‘kelp bed’ exposed to urchin densities of ≥ 8 m^-2 (an equivalent urchin biomass density ≥ 427 g m^-2) was completely denuded of kelp. Conversely, intact kelp remained on reefs with ≤ 4 urchins m^-2 (an equivalent biomass density ≤ 213 g m^-2) (Fig 1a and 1b). This clear response of kelp cover to sea urchin density was independent of nutrient levels (Table 2).

On patch reefs initiated as urchin barrens, recovery of kelp after 6 months was only observed for sea urchin densities ≤ 4 m^-2 (equivalent biomass density ≤ 213 g m^-2). Kelp cover increased on reefs with low urchin densities, however no kelp growth was observed on reefs with urchin densities ≥ 8 m^-2 (equivalent biomass density of ≥ 427 g m^-2) (Fig 1c and 1d). Analysis of the effects of ‘reef ecosystem state’ and ‘nutrients’ on the response of kelp cover over the range of sea urchin densities examined showed clearly that the only factor influencing regrowth of kelp canopy was sea urchin density, with nutrient levels having no detectable effect on canopy cover (Table 2). After 13 months, canopy species recovering on reefs starting as the ‘barrens state’ were S. vestitum and U. pinnatifida. On reefs starting in the ‘kelp bed state’, 14 E. radiata recruits were found after 13 months, with 13 of these kelp recruits observed on reefs where adult E. radiata were still present at the end of the experiment (i.e. on reefs that were below the ‘recovery threshold urchin density’ of 4 urchins m^-2, Fig 2). Only a single recruit was found on a reef which started in the ‘kelp bed state’, but which supported urchins at a density above this threshold. Six kelp recruits were found on nutrient enhanced reefs and 8 on reefs experiencing ambient nutrient conditions.

Examination of the community composition of fleshy algae growing on experimental patch reefs at 13 months (2-way fixed effects ANOVA on species richness and Shannon Diversity) revealed no overall effect of nutrients, while the starting state of the reef and sea urchin density significantly influenced algal cover (Fig 3b and 3c, Table 2). Increasing sea urchin densities resulted in declining macroalgal species richness and diversity while, unsurprisingly, reef patches initiated in the ‘kelp bed state’ showed higher algal species richness and diversity. Yet, as was evident for E. radiata, a precipitous decline in species richness occurred at urchin densities between 4 and 8 m^-2 (Fig 3a and 3b). Interestingly, algal species richness as well as Shannon diversity was highest at 4 as opposed to 0 urchins m^–². The impoverished algal community on reefs with ≥12 urchins m^-2 chiefly consisted of the early colonising species Ulva sp. and the ephemeral Stenogramme interrupta ((C. Agardh) Montagne) which rapidly monopolize available space when grazing pressure is low.

In this study the density of the sea urchin H. erythrogramma required to overgraze kelp on experimental patch reefs was ≥ 8 urchins m^-2 (equivalent urchin biomass density of 427 g m^-2). In contrast, kelp recovery on patch reefs initially starting as ‘barrens’ occurred at urchin densities ≤ 4 m^-2 (equivalent biomass density of 213 g m^-2). These observations might be interpreted as the existence of a single tipping point between 4 and 8 urchins m^-2. However, consistent with other observations worldwide [42] it is more probable that there are different tipping-points in H. erythrogramma density that either drive destructive overgrazing or enable kelp bed recovery, and thus indicate a hysteresis in this system (reviewed by [22,43]). The hysteresis would then be found in between 4 and 8 urchins m^-2 (equivalent to 213 and 427 g m^-2 urchin biomass density respectively) with recovery closer to 4 and overgrazing closer to 8 urchins m^-2.

Many previous studies have demonstrated that sea urchins cause overgrazing and that kelp bed recovery is possible by simply removing urchins from the system (e.g. [21,39,44]). However, in these ‘all or nothing’ studies, gradients in the density of sea urchins are not explored and typically only a single initial state (i.e. urchin barrens) is considered so that tipping-points that precipitate phase-shift in either direction are elusive. Few studies have ever examined both the response of kelp beds and urchin barrens to gradients in sea urchin density [35,36,45,46]. To date, Hill, Blount et al. [35] performed the most detailed experiments on density-dependent urchin overgrazing by separating ‘collapse’ and ‘recovery’ shifts. However, they did not estimate transition thresholds and the recovery and overgrazing experiments were conducted as separate experiments at marginally different scales, in different locations and using different urchin densities across ‘kelp bed’ and ‘barrens’ reef states, so it is difficult to identify hysteresis from their experiments.

Additionally, most studies using sea urchin manipulations report sea urchin density, but not the biomass required to shift from kelp to barrens or vice versa. While Filbee-Dexter and Scheibling [43] and Ling, Scheibling et al. [22] both give estimates of densities and biomasses for tipping-points derived from studies worldwide, there remains a lack of definitive studies within particular systems that determine local phase-shift dynamics. It is important to acknowledge that urchins of the same species not only occur at highly variable densities across different sites, but their size structure and thus biomass can also differ substantially among sites [22]. Only by considering biomass density (g m^-2) can variability in the abundance and size of herbivores be taken into account to estimate tipping-points across grazer-driven systems, thus enabling meaningful comparison between different study areas, times and species. Furthermore, it is important to note that the mean urchin biomass measured within remnant kelp beds in northern PPB was ~ 539 g m^-2 which is 112 g greater than the experimentally estimated tipping-point in urchin biomass of kelp bed overgrazing. This observation is consistent with the observed steady decline of remnant beds in this region from 2009 to 2015 (Authors pers. obs.; S. Reeves, unpublished data).

While kelp was removed relatively quickly at sea urchin densities ≥ 8 m^-2, in the treatments where it occurred, the rate of kelp recovery was comparatively slow over the 13 month duration of the experiment. Reductions in urchin densities (usually complete urchin removal) have been shown to be effective in rehabilitating kelp beds in manipulative and ‘natural’ experiments [20,44,47,48], however by considering a range of urchin densities this experiment explored possible thresholds of urchin density to allow recruitment and growth of kelp E. radiata. Importantly, recruitment was notably greater in the presence of adult sporophytes. This is contrary to the studies of Carnell and Keough [49] and Flukes, Johnson and Wright [50], who showed greater recruitment of E. radiata into patches cleared of canopy-forming algae. There are 2 principal mechanisms that might account for the divergent results among the difference studies. The first addresses the effects of the scale of the kelp bed patch and extent of intraspecific competition. Carnell and Keough [49] and Flukes, Johnson and Wright [50] worked in large beds of dense kelp where the benthos was heavily shaded (>100% canopy cover) and there was (ostensibly) an abundance of spore production. In contrast, since the small patch reefs initiated in the ‘kelp bed state’ supported only 3 adult sporophytes defining 50% canopy cover, light levels on these patches were likely higher than on larger kelp patches with a similar density of individuals (C. Layton, unpublished data), and thus the effects of intraspecific competition are likely much reduced relative to larger kelp bed patches. Secondly, it is likely that there is an interaction between the timing of the availability of substratum to which new sporophytes can recruit, and the effect of adult kelps sweeping the substratum in surge. Kennelly [51] showed that timing of canopy clearance has a large influence on recruitment of E. radiata, and both Carnell and Keough [49] and Flukes, Johnson and Wright [50] cleared patches within established kelp canopies just before the seasonal peak (during autumn) in E. radiata sorus development and spore release. Studies with other kelps have also clearly demonstrated that the timing of canopy disturbance relative to seasonal availability of spores is a major determinant of recruitment (e.g. [48,52–54]). In the current study, experimental reefs were constructed in late spring and therefore were overgrown by understorey algae before E. radiata was reproductive, which is likely to have affected the ability of zoospores to settle and gametophytes and sporophytes to develop and gain a foothold on the reef [51]. Given the higher rate of kelp recruitment on patch reefs supporting adult sporophytes compared to those devoid of kelp, it appears that the presence of adult E. radiata can enhance recruitment not only through supplying a local source of propagules (which was controlled across all experimental reefs), but also by influencing the surrounding benthos. A likely mechanism is that the sweeping action of E. radiata fronds on the substratum as a result of water motion [55,56] reduces the abundance of other algae (particularly turf-forming species) that compete with kelp sporelings or gametophytes. In the current experiment cover of turf-forming algae on experimental patch reefs declined with increasing cover of adult E. radiata sporophytes (Fig 4). Overall, these results suggest that rehabilitation of kelp beds on urchin barrens will be most readily facilitated by actively transplanting adult kelp, once urchins have been reduced to densities below the recovery threshold. The adult sporophytes therefore appear to play a dual role by providing both propagules and also reducing the cover of competing understorey in the form of turfing algae.

Nutrient measurements for the current study site throughout 2013 showed an average nitrogen concentration (nitrite, nitrate and ammonia) of 0.76 μmol l^-1 (± 0.11), which suggests nitrogen may be limiting algal growth [57,58]. This concentration is lower than that reported for other temperate reef systems (e.g. North-West America, [59], South-West Finland, [60]), slightly higher than published levels for western Australia, and below measurements from New South Wales [23]. Given this, it was expected that nutrient enhancement would have a strong effect on algal growth. However nutrient enhancement did not have any detectable effects on either rates of kelp loss or recovery on the experimental reefs. Seeing that the concentrations of ammonia were significantly higher on reefs with enhanced nutrients but neither phosphorus nor nitrogen-oxides showed higher concentration suggests that at least ammonia was enhanced sufficiently to promote algal growth if this was a limiting factor for production. The technique of nutrient addition was in accordance with other experiments and even exceeded the amount which was added to experimental units in those studies [23,26,41], so the fact that elevated nitrogen-oxide concentrations were not detected on ‘nutrient enhanced’ reefs in the current study suggests rapid uptake by microbes, phytoplankton, and microphytobenthos. PPB experiences high rates of denitrification as a result of metabolism of microorganisms [61] and therefore nitrogen-oxides in the water disappear very quickly. This is evident in the discrepancy in the input of nitrogen and what can be found in the water column [62]. A large sewage treatment plant to the west (i.e. Western Treatment Plant) contributes ~3400 tons of Nitrogen year^-1 and the Yarra River in the North adds ~ 1050 tons year^-1 [61], but concentrations of nitrogen are still low in the water column in PPB [62].

Species richness and species diversity of fleshy algae was observed to be highest at an urchin density of 4 m^-2 (equivalent to 213 g m^-2 biomass), which implies that grazing in moderation, or selective grazing on canopy-forming kelp, may release the growth of subordinate understorey algal species leading to a more diverse algal community than in the complete absence of grazing [63]. H. erythrogramma shows preferential grazing on E. radiata in eastern Australia (for PPB N. Kriegisch, unpublished data; for NSW see Wright et al. [45] and Hill et al. [35]), suggesting that the cause of increased algal diversity at moderate grazing is the result of increased space for the recruitment and growth of less palatable algal species. Due to this selective grazing, this grazer-driven stimulation of both algal diversity and richness may ultimately increase overall resilience of the ‘kelp bed’ community when kelp production exceeds grazing rates.