Sublethal effects of a rapidly spreading native alga on a key herbivore

Abstract Multiple anthropogenic stressors are causing a global decline in foundation species, including macrophytes, often resulting in the expansion of functionally different, more stressor‐tolerant macrophytes. Previously subdominant species may experience further positive demographic feedback if they are exposed to weaker plant–herbivore interactions, possibly via decreased palatability or being structurally different from the species they are replacing. However, the consequences of the spread of opportunistic macrophytes for the local distribution and life history of herbivores are unknown. The green alga, Caulerpa filiformis, previously a subdominant macrophyte on low intertidal‐shallow subtidal rock shores, is becoming locally more abundant and has spread into warmer waters across the coast of New South Wales, Australia. In this study, we measured (a) the distribution and abundance of a key consumer, the sea urchin Heliocidaris erythrogramma, across a seascape at sites where C. filiformis has become dominant, (b) performed behavioral field experiments to test the role of habitat selection in determining the local distribution of H. erythrogramma, and (c) consumer experiments to test differential palatability between previously dominant higher quality species like Ecklonia radiata and Sargassum sp. and C. filiformis and the physiological consequences of consuming it. At all sites, urchin densities were positively correlated with distance away from C. filiformis beds, and they actively moved away from beds. Feeding experiments showed that, while urchins consumed C. filiformis, sometimes in equal amounts to higher quality algae, there were strong sublethal consequences associated with C. filiformis consumption, mainly on reproductive potential (gonad size). Specifically, the gonad size of urchins that fed on C. filiformis was equivalent to that in starved urchins. There was also a tendency for urchin mortality to be greater when fed C. filiformis. Overall, strong negative effects on herbivore life‐history traits and potentially their survivorship may establish further positive feedback on C. filiformis abundance that contributes to its spread and may mediate shifts from top‐down to bottom‐up control at locations where C. filiformis has become dominant.


| INTRODUC TI ON
In providing food and physical structure, macrophytes support biodiverse communities and complex food webs (Ellison et al., 2005;Jones et al., 1996;Lloyd et al., 2020). Globally, habitat-forming macrophytes are threatened by a variety of abiotic and biotic disturbances, triggering transformations in the habitats they create (Dijkstra et al., 2019;Filbee-Dexter & Wernberg, 2018;Krumhansl et al., 2016). Warming and heatwaves (thermal stress), changes in the nutrient regimes (i.e., eutrophication), and overgrazing by herbivores can all influence the distribution and abundance of habitat-forming species (Cebrian et al., 2014;Filbee-Dexter & Scheibling, 2014;Steneck et al., 2002Steneck et al., , 2017Wernberg et al., 2016). The loss of habitat-forming macrophytes can result in the transition to alternate stable states in which communities are dominated by previously subdominant or novel species that may be more tolerant to disturbance. Moreover, the shift in macrophyte composition of the new community can result in changes to the strength of biotic interactions, for example, they may experience reduced rates of herbivory, compared with the species they replace, leading to further positive feedbacks even if the disturbance is removed (i.e., the passengerdriver model: Bulleri et al., 2010;Feng et al., 2009;MacDougall & Turkington, 2005). Traditionally, changes in the strength of biotic interactions have focused on the implications for the abundance or the distribution of macrophytes (Boada et al., 2017;Pessarrodona et al., 2019) and much less is known about the consequences for the herbivores, although strong negative effects on herbivores may be predicted if the novel macrophytes are more resistant to herbivory (but see Boudouresque et al., 1996;Felline et al., 2012).
Both the physical and biological properties of macrophytes can mediate changes in the distributions of herbivores via lethal or sublethal effects (e.g., behavioral or life-history traits, Lubchenco & Gaines, 1981). If one macrophyte is replaced with another that is structurally and biologically similar, then consumer-mediated interactions may not change dramatically (Buschbaum et al., 2006;Steneck & Dethier, 1994). If, however, the new habitat is structurally different, this may have large impacts on associated species. For example, scouring or whiplash by algae can limit the access of grazers to the substratum (Konar & Estes, 2003). The growth form of vegetation can affect the abundance and movement of herbivores (Bach, 1981;Lanham et al., 2015;McGuinness & Underwood, 1986) while the quality of vegetation may also influence herbivore behavior (Livore & Connell, 2012b;McArthur et al., 2014;Weterings et al., 2018). Herbivores may avoid a less palatable species or those with lower nutritional quality, thereby affecting their distribution and abundance across land and seascapes . In cases where the less palatable food source is a novel spatially dominant species, consumers may have no option but to consume the less preferred species with sublethal effects (e.g., reduced growth, biomass, or reproductive investment, Gribben & Wright, 2006;Tomas et al., 2011;Wright et al., 2011). Separate studies have shown that the invasive macrophytes can have reduced palatability compared with native macrophytes, which can release the macrophytes from predation pressure (e.g., Cacabelos et al., 2010;Nejrup et al., 2012), but also alter the distribution, behavior, and performance of native herbivores (see Scheibling & Anthony, 2001;Tomas et al., 2011). In other cases, however, native urchin grazers did consume introduced algae at greater or similar rates to native algae (e.g., Cacabelos et al., 2010;Noè et al., 2018). More recently documented is the spread of native macrophytes in response to environmental changes, some of which are having ecological effects that are analogous to those of invasive species. If invasive macrophytes have similar palatabilty to native macrophytes, we predict that the consequences for herbivores in the invaded region will be negligible. Globally, however, the consequences for herbivores in regions where native-invaders (sensu Simberloff, 2011) are colonizing novel regions are unknown.
On the east coast of Australia, the top-down effects of sea urchin grazing create and maintain extensive barrens (devoid of macroalgae). Barrens are typically created by the large diadematid urchin Centrostephanus rodgersii, which grazes on brown algae Fletcher, 1987;Hill et al., 2003). Smaller patches of barrens can also be created by large densities of the echinometrid urchin Heliocidaris erythrogramma, one of the most common herbivores along the Great Southern Reef Keesing, 2020) particularly in shallower, more sheltered waters (Connell & Irving, 2008;Ling et al., 2019;Wright et al., 2005). In smaller densities, H. erythrogramma may not create barrens but can still remove canopy-forming algae such as Ecklonia radiata, Sargassum spp., and Cystophora spp. (Livore & Connell, 2012a). Wright et al. (2005) observed the removal of the articulated coralline Amphiroa anceps and the fleshy brown Zonaria diesingiana by large numbers of H. erythrogramma, while in laboratory feeding trials, they found that the urchin also fed on Corallina officinalis, S. vestitum, and, to a lesser extent, the chemically defended rhodophyte Delisea pulchra. In Western Australia where H. erythrogramma seems to feed primarily on drift algae, the urchin is known to preferably retain E. radiata Vanderklift & Kendrick, 2005). On exposed rocky shores, H. erythrogramma generally stay close to or within crevices to avoid predation or dislodgement by waves. Where crevices are not available, H. erythrogramma, like other urchins, often bore into the rock effectively creating their own custom shelters (Russell et al., 2018).
Urchins generally continue to bore as they grow, often ultimately creating a shelter in which the entrance is smaller than the urchin itself, confining them to their crevice. In this situation, urchins tend to feed on drift macroalgae (Keesing, 2020;Kriegisch et al., 2019). that contributes to its spread and may mediate shifts from top-down to bottom-up control at locations where C. filiformis has become dominant.
The shallow rocky reef habitat of H. erythorogramma is becoming increasingly colonized by the native green alga Caulerpa filiformis in New South Wales, Australia. C. filiformis was previously known to be a subdominant community member and have a distribution restricted to about 300 km (Glasby et al., 2015). It has undergone a pronounced expansion and is now becoming more conspicuous within its range (Glasby, et al., 2015;Voerman et al., 2017) and having spread 500 km outside its known distribution. Unusually, it is spreading northwards into warmer waters (Glasby et al., 2015). At many sites, it forms large dense, monospecific stands and may be the most dominant macrophyte present. Beds of C. filiformis trap sediment (Voerman et al., 2017;Zhang et al., 2014), which occupies the rocky surface and fills in bore holes normally utilized by herbivores. Observations and small-scale experiments suggest that on many rocky reefs, C. filiformis is replacing turf-forming coralline algae and species of Sargassum (Voerman et al., 2021;Zhang et al., 2014). C. filiformis is structurally different from the native algae that it is apparently replacing, having much longer fronds than coralline turf, highly chemically defended, and more compact than species of Sargassum spp. As such, the spread of C. filiformis may affect the behavior or life history of H. erythrogramma. In this study, we combined a variety of disparate approaches (feeding preference assays, field-behavior experiments, and laboratory life-history experiments) to determine the effect of C. filiformis on the local distribution and abundance of H. erythrogramma.
We used large-scale surveys to determine spatial patterns in abundance of H. erythrogramma at sites dominated by C. filiformis, and field (movement) experiments and laboratory (feeding trials) to determine if abundance patterns could be related to any lethal or sublethal effects of C. filiformis on urchins. We tested the hypothe-

| Species and study locations
Caulerpa filiformis is a coenocytic green alga found on intertidal and subtidal reefs in Australia, Peru, and South Africa (Glasby, et al., 2015). In New South Wales, Australia, C. filiformis occurs between 0 and 6 m water depth, primarily on rocky substrata in exposed and sheltered locations (e.g., inside Sydney Harbour) where it retains a thick layer of sediment (Voerman et al., 2017(Voerman et al., , 2019. Its root-like rhizomes form dense, entangling mats, which can trap sediment (Voerman et al., 2017), and the flattened blades grow to  Figure 1c) and sections of mixed turf-forming algae consistently of different species with dominance by geniculate corallines that trap sediment (Connell et al., 2014), and the presence of Sargassum spp.

| Patterns of urchin abundance inside and outside C. filiformis
At each of the three locations, in water between 0 and 2 m, we compared the abundance of H. erythrogramma and the number of homing scars on rocky platforms within patches of C. filiformis to those in surrounding algal habitats ("outside"), which consisted primarily of geniculate coralline algae and sparsely distributed browns such as Sargassum spp. and E. radiata. Patches of C. filiformis were a minimum of ~3 m × 3 m and separated by at least 10 m. Within C. filiformis patches, urchins and the number of homing scars were also compared between two positions ("inside" and "edge"). Edges were defined as areas within 1 m of the border of a patch of C. filiformis, while "inside" areas were in central section of patches, >1.5 m from the border. All the urchins found were sitting in self-bored homing scars or rock crevices. Mainly one large monospecific stand of C. filiformis dominates the locations with presence of secondary smaller patches. Abundances of H. erythrogramma and homing scars were counted in n = 5 replicate quadrats (N = 1 per patch; 50 × 50 cm) in each position with respect to patches (inside, edge and outside) at each location. Quadrats were haphazardly placed within patch positions. Quadrats were carefully searched for urchins and homing scars, which involved removing dense vegetation and, in some cases, feeling through sediment among C. filiformis to count the number of homing scars. It is possible that some homing scars were missed among dense C. filiformis.

| Urchin habitat preferences
To determine how C. filiformis influenced urchin movement, we collected ~80 H. erythrogramma of ~5 cm test diameter (TD, without spines) from a shallow rocky reef in Chowder Bay, Sydney Harbour (33°50′24.5″S 151°15′16.1″E). In order to relocate and identify experimental urchins, we tagged them. To do this, we used a hypodermic needle (14 G) to pierce urchins through the test and placed a fishing line through it. The needle was then removed, and the line tied off. This technique has been extensively used to tag sea urchins and does not affect their survival (Boada et al., 2015). On the other end of the fishing line, we attached numbered plastic tags to identify an individuals' starting point in the experiment (i.e., inside, edge, or outside C. filiformis). We placed H. erythrogramma in tanks of freeflowing sea water after tagging and left them to recover for 48 hr.
No urchins died as a result of this process. Urchins were then placed into the field at the Mona Vale location. We haphazardly placed 20 urchin individuals in the middle of a large C. filiformis patch (~500 m 2 , approximate size estimated in situ by divers), 20 on the edge, and 20 outside, among primarily coralline and sparsely distributed browns such as Sargassum spp. For this experiment, the inside position was 1 m from the edge. Urchins on the edge were placed immediately adjacent to C. filiformis, in the interface between the patch and the neighboring habitat and urchins outside were placed at ~1 m from the edge. All urchins were randomly allocated to positions within the patch at sufficient distance (at least 2-3 m apart) to maintain independence among individual urchins. One hour after placing urchins, we recorded the position of all tagged urchins as either middle, outside, or on the edge of a C. filiformis patch. The urchins moved very quickly once placed in the field (up to 0.5 m within the first 2 min) so 1 hr was adequate time to assess movement patterns.

| Feeding preferences of H. erythrogramma
We conducted a series of mesocosm experiments at the Sydney Institute of Marine Science (SIMS) to test the feeding preference of H. erythrogramma. All the urchins used in the experiments were collected from Bare Island (33°59′36″S, 151°13′54″E), where H. erythrogramma occur in high densities (50-100 m −2 , see Wright & Steinberg, 2001). We did not starve urchins prior to the start of the experiment. We first conducted no-choice (i.e., a single food item) experiments and then a set of preference feeding assays with two species at a time. We used experimental containers (30 × 30 × 15 cm; ~4 L) supplied independently with filtered free-flowing sea water (300 µm). For no-choice (i.e., single food item) experiments, we placed one urchin (~5 cm TD) together with a ~5 g piece of either C.
Algae were blotted dry to constant wet weight. We also visually assessed all algae to ensure they were free of epiphytes before being offered to urchins in containers. Experimental controls (n = 10 per treatment) to test for natural degradation of algae were created by placing five grams of each seaweed in containers without H. erythrogramma. After 72 hr, we removed any remaining seaweed from Sydney containers, blotted dry to constant wet weight, and recorded the final weights. We used the same methods for preference (choice) feeding assays except we crossed the different algae into pairwise treatments (C. filiformis × E. radiata, C. filiformis × S. vestitum. and E. radiata × Sargassum spp. n = 20 pairs/experiment). When presented with two species, H. erythrogramma were placed in the center of the container with algal species at either end. Water flow in the containers was directed as such to cause minimal interference to algae upon initiation of the experiments. After 48 hr, any remaining algae were collected and reweighed as above.

| Effects of C. filiformis on urchins
To test the effects of C. filiformis consumption on urchins, we per-

| Statistical analysis
We analyzed urchin abundance, number of homing scars, and percent homing scar occupancy using generalized liner mixed models with position in the seascape (middle, edge, and outside C. filiformis patches) as a fixed factor and location as a random factor.

| Patterns of urchin abundance inside and outside C. filiformis
The abundance of H. erythrogramma differed among all three habitats and was highest outside (22.13 ind/m 2 ± 2.61 SE) patches of C.
filiformis, followed by patch edges ( 3.1.2 | Effects of C. filiformis on urchin feeding preference and performance The amounts of C. filiformis, E. radiata, and Sargassum spp. consumed by H. erythrogramma did not differ in no-choice feeding trials ( Figure 4, p > .05), although there was a tendency for reduced consumption of C. filiformis (compared with other algae, Figure 4).
Across all preference trials, H. erythrogramma consumed approximately four times as much E. radiata as either C. filiformis or Sargassum spp. When presented with a choice, urchins consumed significantly more E. radiata than C. filiformis ( Figure 5). There was, however, no difference in consumption of C. filiformis and Sargassum spp. Similarly, urchin consumption of Sargasssum spp. and E. radiata did not differ, although there was a tendency for urchins to consume more E. radiata ( Figure 5).

| Lethal and sublethal effects of C. filiformis on urchins
Although results were not statistically significant, probably due to missing replicates for the E. radiata treatment, mortality was higher for starved urchins (10% ± 5.77 SE) and urchins fed with C. filiformis (5.56% ± 4.01 SE) compared with those fed with E. radiata (1.67% ± 1.36 SE) (Figure 6a). Of those urchins that survived, in local abundance of C. filiformis (Voerman et al., 2021). Here, we show that the sea urchin H. erythrogramma, an abundant herbivore in the region, actively avoids dense areas of this expanding habitat and that prolonged consumption of C. filiformis results in reallocation of energy, reduced gonad size, and a tendency for increased mortality. Although gonad condition is known to relate to food quality in H. erythrogramma and other urchins (Andrew, 1986;Livore & Connell, 2012b), we found little difference between effects of starvation and a diet of C. filiformis. Thus, C.
filiformis likely experiences positive reinforcing demographic feedbacks from reduced herbivory, facilitating its stability and spread.
The physical properties of habitat-forming species can have strong effects on associated organisms (Dijkstra et al., 2017;Gribben et al., 2020;Jones et al., 1996;Uyà et al., 2020;Wright & Gribben, 2017). Hori (2006), for instance, found that dense surfgrass beds (which are similar in structure to C. filiformis patches) were able to confine sea urchins to rock pools by limiting their movement.
Specifically, vegetation can influence the foraging behavior of organisms living in the habitat they generate or alter the abiotic conditions of the environment (Bach, 1981;Underwood & Jernakoff, 1981). In our study, increasing abundances of urchins away from the center of C. filiformis patches appeared to be, at least in part, determined by the responses of urchins to C. filiformis. In our movement experiment, we found no urchins actively entering into C. filiformis patches. Instead, urchins concentrated at the edge between the two habitats possibly because the edge offers a good trade-off between although longer experiments will add further insights in this respect (see Farina et al., 2014). The fast movement rates contrast with previous literature suggesting the limited movement of this species (see Keesing, 2020). Thus, even the dense C. filiformis beds did not act as physical barriers to movement. Instead, urchins may avoid remaining within patches because homing scars are filled in by either rhizomes or sediment trapped by C. filiformis (Voerman et al., 2017). Indeed, most homing scars in C. filiformis were filled with sediment and were unoccupied by urchins. This supports Voerman et al. (2017) who suggested that C. filiformis may have overgrown areas of turfing algae, which were previously home to H. erythrogramma. In addition, reduced access to homing scars and increased sedimentation may force urchins to be located higher off the surface, making them more prone to predators or dislodgement via wave action (Pagès et al., 2013;Walker, 2007).
In addition to the physical effects, the presence of C. filiformis seems to also shape the distribution and abundance of urchins across the seascape through behavioral and physiological responses. H. erythrogramma is able to switch feeding modes in relation to food quality (from drift to grazing, Livore & Connell, 2012a).
Similarly, the decrease in high-quality food resources across a seascape and its substitution by less preferred species would force urchins to move to neighboring areas foraging for preferred food items. In this case, the preference for the more appealing Ecklonia radiata would stimulate these herbivores to move away from patches of C. filiformis, increasing pressure on the adjacent seaweed forest and indirectly enhancing the success of the spread of C. filiformis. However, when urchins remain confined to homing scars (Russell et al., 2018) or much of the reef is colonized by C. filiformis, urchins may be forced to feed on this food resource, resulting in sublethal (e.g., reduced reproductive capacity) and potentially lethal effects due to obliged consumption of drifting C.
filiformis. This set of lethal and sublethal effects, probably associated with the secondary compounds typical of species from the genus Caulerpa, generally affect the feeding of different types of herbivores, (Boudouresque et al., 1996;Davis et al., 2005;Felline et al., 2012;Gollan & Wright, 2006;Miranda et al., 2019). Indeed, similar patterns have previously been described in the context of invasive species including species of the genus Caulerpa (see Scheibling & Anthony, 2001;Tomas et al., 2011).
Finally, we found that urchins reallocate energy from reproduction (gonads) to structural traits (test mass), suggesting negative consequences to the population, possibly via algal toxicity, of C. filiformis consumption, at least when compared to E. radiata. These results align with previous investigations that found clear links between resource allocation of H. erythrogramma and the habitat it occupies, and a clear hierarchy of resource allocation prioritizing gut and lantern then test and spines and finally gonads (see Keesing, 2020). Constable (1990)  to algal communities is a key process allowing the initial establishment and expansion of C. filiformis. Our study suggests that its physical structure and reduced palatability may then contribute to reinforcing demographic feedbacks for C. filiformis. If herbivores reliant on macroalgae generally avoid C. filiformis as a food source then its continued expansion and increase in local abundance could cause a critical shift in the ecosystem functioning to bottom-up control, although herbivorous fish abundances do not appear to be strongly affected by the presence of C. filiformis (Bradley et al., 2018), possibly because of their higher mobility. A recent global focus has been on describing the massive impacts range-spreading herbivores have on algal communities through strengthening top-down processes (Vergés et al., , 2016. Here, we show that the opposite can occur, although the overall consequences for urchin populations and ecosystem function remain to be further determined. The stability of subdominant macrophytes after disturbances seems globally very high with foundation macrophytes having limited capacity to recover their original extension unless active management actions are applied (i.e., restoration, Bulleri et al., 2020;Wood et al., 2019). Therefore, if these findings are the rule rather than the exception, we may expect further global declines in the abundance of foundation macrophytes and an important shift in the energy flow in these ecosystems.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used in this study are publicly available at the Dryad repository (https://doi.org/10.5061/dryad.gmsbc c2nr).