Survivorship and feeding preferences among size classes of outplanted sea urchins, Tripneustes gratilla, and possible use as biocontrol for invasive alien algae

Study animal

The short-spined collector urchin Tripneustes gratilla received its common name from the habit of gathering fragments of coral rubble, rocks, or algae from the benthic environment as camouflage while it forages. Its body is predominantly black but often possesses a pentaradial bluish or reddish hue when its tube feet are retracted close to its body. Its spines are typically black, white or cream. This echinoid is relatively common in shallow waters (0–15 m) across the Hawaiian archipelago (Kay, 1994; Hoover, 2002). Natural densities of T. gratilla range from 2.9–4.4 m⁻², placing it in the top three most abundant urchins in Hawaiʻi (Ogden, Ogden & Abbott, 1989).

Although T. gratilla have shown significant dietary preferences for Kappaphycus spp. in controlled laboratory studies (Stimson, Cunha & Philippoff, 2007), in the wild these urchins are a generalist herbivore that will graze on virtually any algae or sea grass available (Vaïtilington, Rasolofonirina & Jangoux, 2003; Dworjanyn, Pirozzi & Wenshan, 2007; Stimson, Cunha & Philippoff, 2007). The generalist diet and habitat requirements of T. gratilla coupled with high dispersive potential have resulted in an extremely wide, pantropical distribution (Lessios, Kane & Robertson, 2003). It should be noted that T. gratilla was formerly a native resident of Kāneʻohe Bay. Tripneustes gratilla was once thought to be one of the most abundant urchin species within Kāneʻohe Bay (Edmonson, 1946; Alender, 1964; Banner & Bailey, 1970; Kay, 1994). Conversely, T. gratilla is now relatively rare and does not contribute significantly to herbivory on reefs within the bay (Conklin & Smith, 2005; Stimson & Conklin, 2008). However, historical outbreaks of the native alga Dictyospaeria were thought to be controlled by T. gratilla because growth accumulated mostly in calm waters of the bay where the urchin was rare (Banner & Bailey, 1970). Hence, there is considerable interest from both State and local conservation groups to replenish the natural population of T. gratilla in the bay and enhance natural herbivory to control these invasive alien algal species.

Tripneustes gratilla were provided by the DAR urchin hatchery as juveniles. We arbitrarily placed urchins in three non-overlapping size classes: small (17.5–22.5 mm maximum test diameter), medium (29.8–43.8 mm), and large (45.1–65.1 mm) that were then used for each of the experimental trials outlined below. Prior to the experiments, medium and large urchin size classes were raised on diets of all four alien algae. Small urchins were not fed before trials, but instead were placed directly into experiments within a few days of arriving from the hatchery. Therefore, urchins were starved 3–5 days prior to each experiment.

Algae

We chose the four most common species of alien invasive algae found on the patch reefs of Kāneʻohe Bay: Acanthophora spicifera, Gracilaria salicornia, Eucheuma denticulatum, and Kappaphycus clade B. Kappaphycus clade B (formerly identified as Kappaphycus alvarezii, K. striatum, or Eucheuma striatum) and Eucheuma denticulatum were intentionally introduced from the Philippines to Kāneʻohe Bay September 1974 by researchers from the University of Hawaiʻi for scientific research (Doty, 1971; Doty, 1977); fragments apparently drifted away from test sites on the north reef of Moku O Loʻe (Coconut Island) and were also collected and transplanted around the bay by local residents for personal cultivation (Russell, 1983; Batibasaga, Zertuche-González & de San, 2003; Weis & Butler, 2009; Ask et al., 2003). Despite having observed vegetative propagules being released, researchers reported that such propagules were incapable of dispersing over deep water or finding suitable hollows on which to settle (Doty, 1977). This oversight led to the documented proliferation of new eucheumatoid colonies upon the introduction of the algae to test sites around Coconut Island (Doty, 1977). Likewise, the intentional introduction of Gracilaria salicornia by the same researchers to Kāneʻohe Bay occurred September of 1978, specifically for experimental aquaculture aimed at the development of commercial agar production (reviewed by Rodgers & Cox, 1999; Smith et al., 2004). The idea of a commercial agar industry in Hawaiʻi has long since been abandoned, but the introduced G. salicornia has established and spread along the shores of Waikı¯kı¯ and reefs in Kāneʻohe Bay.

In contrast to these intentional introductions, a fragment of Acanthophora spicifera was first documented in Pearl Harbor in the fall of 1952, and was believed to have been transported on the heavily fouled hull of the barge “Yon 146” which was towed to Oʻahu from Guam in 1950 (Doty, 1961). By February 1956, A. spicifera had been documented in Kāneʻohe Bay, making it the first documented accidental introduction to the Bay (Kohn, 1961; Coles, DeFelice & Eldredge, 2002). These invasive macrophytes became not only some of the most dominant benthic organisms, but they have also resulted in the most detrimental impacts to marine communities in the bay (Coles, DeFelice & Eldredge, 2002).

The four algae species had widespread distributions in the bay and were readily available for collection. Acantohophora spicifera and G. salicornia were easily collected nearly anywhere around Coconut Island and around the southern portion of Kāneʻohe Bay. The eucheumoids were consistently collected from patch reefs in the central portion of the Bay. We did not included native algae in this study because their abundance is so reduced in the Bay (Stimson, Larned & Conklin, 2001; Conklin & Smith, 2005; Stimson, Cunha & Philippoff, 2007) that we could not collect enough for this experiment without impacting the remaining population.

Growth on single-species diet

Growth rates of T. gratilla were measured while on single-species diets of each A. spicifera, G. salicornia, E. denticulatum and K. alverezii. For each of the four algal species, three T. gratilla were housed in each of three 15 L replicate aerated flow-through tanks (∼1–2 L/min). In order to monitor individual growth rates of urchins without marking the animals, each tank held a single urchin of each size class: small, medium, and large. For each treatment, algae were provided ad libitum to reduce any resource competition, and all aquariums were cleaned twice a week during which freshly collected algae were provided to each tank. Urchin test diameter was measured using Vernier calipers (VWR) to the nearest tenth of a millimeter each week for a month. An analysis of covariance (ANCOVA) was employed to analyze growth, with initial urchin test size used as the covariate. Tukey’s HSD post-hoc comparison was then performed to determine significance of pairwise differences in average growth rates of urchins on each algae diet (Fig. 1).

No-choice feeding trials

No-choice feeding trials provided juvenile T. gratilla in each treatment with only a single species of alga and measured differences in mean consumption among the treatments with different algal species. For each algal species, two urchins of the same size class were added to each of six replicate tanks (15 L tanks with 1–2 L/min flow rate, as above). Algae were blotted on paper towels to remove excess water and weighed before being placed in each tank. Urchins were allowed to graze for ∼5 days and the amount of remaining algae was weighed as before to calculate the amount of each species consumed per urchin per day. In a few cases, we stopped the experiment after the 4th day because we did not want the urchins to consume more than half of the algae offered in any trial. Consumption rates (grams of algae per day) were then compared by analysis of variance (ANOVA). Tukey’s HSD post-hoc comparison was used to identify significant differences between each of the algae (Fig. 2). The assumptions of Normality and homogeneity of variance of the data were tested using the Shapiro–Wilks test and the Levene’s test, respectively. The null hypothesis for the Shapiro–Wilks test was that the data were Normally distributed; therefore, p-values less than 0.05 suggested that the data were not Normally distributed. For the small size class in the no-choice feeding trials the data were Normally distributed (Shapiro–Wilk, W = 0.96, p = 0.82). The data from the medium cohort from the no-choice trial was also normally distributed (Shapiro–Wilk, W = 0.90, p = 0.28). However, the data from the large urchins of the no-choice feeding trial were only marginally non-normal (Shapiro–Wilk, W = 0.80, p = 0.042). To test the homogeneity of variance among our feeding trial data, the Levene’s test was employed. The null hypothesis for the Levene’s test was that the variances are homogenous. For the urchins in the no-choice feeding trials, the data passed the homogeneity test (Levene’s test, F(3, 44) = 0.56, p = 0.64).

Choice feeding trials

The choice feeding experiment provided all four species of algae, in equivalent amounts, simultaneously to urchins. As with no-choice experiments above, each algal species was blotted and weighed before being introduced to the experimental tanks. For this assay, larger tanks (80 L, ∼4 L/min flow-through) were used to allow room to separate algae into the four quadrants of the experimental tank. Four urchins were then introduced to the middle of the tank and allowed to graze at will for ∼7 days. Again, if any species of alga became low relative to the others (less than half the initial amount), we stopped the experiment a day early to avoid biasing results. At the end of each experiment, algae were removed, blotted and weighed as previously to calculate the amount of algae consumed per urchin per day for each species.

For each choice and no-choice feeding trials, small urchins were provided with ∼100 g of algae initially, whereas medium and large urchins were offered a starting biomass of ∼150 g of algae. To account for any growth or decline of the algae not attributed to urchin grazing during the experiment, both choice and no-choice experiment tanks had a divider such that one half of the tank housed experimental algae and urchins whereas the other side housed only equivalent amounts of algae to serve as a no urchin control. The consumption rates of algae were then calculated as: ${\displaystyle \text{Consumption }=\left(A_i\left(A{C}_f/A{C}_i\right)-A_f\right)}$ where A_i and A_f were the initial and final blotted masses of algae subject to grazing by urchins; while AC_i and AC_f were the initial and final masses of the algae in the no-urchin control tanks. This equation was used to account for growth of algae over the course of the experiment, but can also account for any unexpected decline in algal biomass unrelated to the grazing trial (Dworjanyn, Pirozzi & Wenshan, 2007; Seymour et al., 2013). Because all species of algae were provided simultaneously during choice feeding trials, the consumption of one species was affected by the consumption of the others, therefore the assumption of independence required to perform an ANOVA was violated. Consequently, choice feeding preference assays were analyzed using a non-parametric Friedman’s rank test, and both parametric and non-parametric analyses are congruent. Relative consumption rates of each algal species were reported (Fig. 3) and ranked. Nevertheless, due to the lack of post-hoc pairwise comparison for Friedman’s rank test, a Tukey’s HSD post-hoc comparison was used to identify significant differences between each of the algae. Again, to test the data’s distribution for Normality and homogeneity of variance the Shapiro–Wilk and Levene’s test were used. The data from the small urchins of the choice feeding trials failed the normality test (Shapiro–Wilk, W = 0.86, p = 0.0036). However, the data for the medium and large sized urchins of the choice feeding trials both passed the normality test (Shapiro–Wilk, W = 0.97, p = 0.27, and W = 0.96, p = 0.17, respectively). For the small urchins of the choice feeding trial, the data passed the homogeneity test (Levene’s test, F(3, 20) = 2.13, p = 0.13). The data from the medium urchins in the choice feeding trial fail to reject the null hypothesis (Levene’s test, F(3, 36) = 0.32, p = 0.81). The data from the cohort of large urchins in the choice feeding trials also passed the test for equality of variances (Levene’s test, F(3, 40) = 1.98, p = 0.13).

Field caging experiment

Cages measuring roughly 50 × 50× 75 cm were constructed from 1 cm² galvanized chicken wire mesh. We constructed both open and closed cages. For closed cages, the mesh extended across all sides including the tops to prevent urchins from being able to crawl out and prevent access by fishes on the reef. In contrast, the sides of the open cages end with back-folded edges and no top to minimize escape of the juvenile urchins from the cage, but still allow open access of predatory fishes. Initial trials with urchins caged in seawater tables indicated that this back-folded edge design (approximating an upside down U) was the most effective for open cages, but urchins still escaped the cages at the rate of 1–2 animals per week. Urchins ranged from 18–22 mm at the start of the experiment. Cages were filled with G. salicornia to provide the juvenile urchins with food and a place to hide, because our initial aquarium trials revealed that urchins were far more likely to escape the open cages in the absence of hiding spots and food in the cage. In the absence of any cover or food, open cages were frequently empty within 24 h in our water table trials (data not shown).

Cages were placed at 6 sites across three habitats, with four cages, three open and one closed control, at each site. 251 urchins were used during these caging experiments. Three habitats surrounding Coconut Island (map in Supplemental Information) were selected to mimic the conditions on the reef to which urchins are being currently outplanted: a protected lagoon, a shallow back-reef and a fore-reef slope each at 1–3 m depth. The protected lagoon had low coral cover, high alien algal cover and minimal water flow, whereas both the back-reef and fore-reef sites had high coral cover, relatively low alien algal cover and relatively high water flow. Each cage was checked three times a week for 30 days to count surviving urchins as well as replenish consumed algae. All studies reported here were conducted under the State of Hawaiʻi, Department of Land and Natural Resources, Division of Aquatic Resources Special Activity Permits sap#2012–63 and SAP#2013–47. Survivorship between treatments was compared using the Kaplan–Meier product-limit method for fitting survivorship curves and comparison by Log-rank (Forsman, Rinkevich & Hunter, 2006), and Wilcoxon non-parametric tests. These statistical tests were done using JMP Pro 11.

Growth on single-species diet

Growth rates of T. gratilla, measured as maximum test diameter (mm), were significantly affected by fixed algal diets (ANCOVA, initial urchin size as covariate, F_(5,84) = 10.80, p < 0.001, Fig. 1). Urchins that fed exclusively on diets of either G. salicornia or K. clade B grew at significantly (p = 0.001 and p = 0.009, respectively, Tukey HSD) higher rates (1.58 ± 0.14 and 1.69 ± 0.14 mm/week TD (Test Diameter), respectively) than those urchins that fed on a diet of E. denticulatum. Urchins that fed on a diet of E. denticulatum had the lowest growth rates (0.97 ± 0.14 mm/week TD) out of the four assays, though not significantly lower to urchins on a diet of A. spicifera (1.23 ± 0.11 mm/week TD).

No-choice feeding trials

When presented with no choice, urchins consumed different species of algae at different rates, but the effect varied by urchin size (Fig. 2). On average, large urchins ate G. salicornia at the highest rate of all algal species offered (15.24 ± 0.001 g day⁻¹), and K. clade B at the lowest rate (12.43 ± 1.51 g day⁻¹) although these trends were not significant (ANOVA, F_(3,12) = 1.94, p = 1.78, Fig. 2C). Medium sized urchins showed similar trends, but with significant differences in the amounts of algae they consumed on a daily basis (ANOVA, F_(3,12) = 8.49, p < 0.05, Fig. 2B). For the medium size class, Gracilaria salicornia was eaten at a significantly higher rate (12.60 ± 0.08 g day⁻¹) than either A. spicifera (10.33 ± 0.36 g day⁻¹) or E. denticulatum (9.35 ± 0.90 g day⁻¹) (p = 0.034 and p = 0.003, respectively, Tukey HSD); K. clade B was also eaten at a higher rate (11.87 ± 0.27 g day⁻¹) than E. denticulatum (p = 0.018, Tukey HSD), but not A. spicifera (p > 0.05, Tukey HSD). Among the small collector urchins, feeding rate patterns were comperable to those of the medium urchins, but with more significant disparities among algal species, (ANOVA, F_(3,12) = 51.30, p < 0.001, Fig. 2A). Small urchins offered only G. salicornia had a significantly higher mean consumption rate (6.08 ± 0.19 g day⁻¹) than any other algal assay in the non-choice feeding trial (p < 0.05 for each pairwise comparison, Tukey HSD). Small urchins likewise consumed Eucheuma denticulatum at the lowest rate (2.32 ± 0.39 g day⁻¹), which was significantly lower than both A. spicifera and K. clade B (p < 0.05, Tukey HSD).

Choice feeding trials

Feeding trials in which urchins were offered multiple species of algae simultaneously revealed different patterns than those observed in the non-choice feeding assays. Small urchins significantly preferred to feed on A. spicifera than any of the other three available algae species (Friedman’s rank test, p < 0.05; p ≤ 0.01, Tukey HSD, Fig. 3A). Small urchins did not display any patterns of preference among G. salicornia, E. denticulatum or K. clade B (p > 0.05, Tukey HSD). Likewise medium urchins showed a significant preference (Friedman’s rank tests, p < 0.001, Fig. 3B) for A. spicifera over G. salicornia and K. clade B (p < 0.05 and p < 0.001, respectively, Tukey HSD), as it was consumed at the highest rate (3.69 ± 0.21 g day⁻¹), whereas G. salicornia and E. denticulatum were consumed at intermediate rates (0.95 ± 0.1 and 1.19 ± 0.13 g day⁻¹, respectively). In contrast, medium urchins consumed K. clade B at the lowest rate (0.51 ± 0.10 g day⁻¹). Unlike the small and medium urchins, which avoided K. clade B in the choice trials, large urchins exhibited significant preferences for both A. spicifera and K. clade B (4.30 ± 0.09 g day⁻¹ and 4.31 ± 0.14 g day⁻¹, respectively) in the choice feeding trials (Freidman’s rank test, P < 0.001, Fig. 3C). Large urchins significantly preferred A. spicifera to both G. salicornia and E. denticulatum (p < 0.0001 and p < 0.001, respectively, Tukey HSD). Kappaphycus clade B was also significantly preferred to G. salicornia and E. denticulatum (p < 0.01 and p < 0.05, respectively, Tukey HSD). Whereas G. salicornia and E. denticulatum were consumed at intermediate rates by small and medium urchins, these species tend to be avoided by the large urchins (3.11 ± 0.20 g day⁻¹ and 3.20 ± 0.20 g day⁻¹, respectively) when given a choice of algae on which to feed (Fig. 3).

Caging experiment

Considerable differences in urchin survivorship among the three habitats that they were caged in were found to be significant (Table 1). On the reef flat, 84.4% of urchins deployed in closed cages remained and 75% of urchins remained in the open cages after 29 days (Fig. 4A). Urchins caged in the lagoon had a substantially higher rate of loss than those on the reef flat, with only 56.3% of urchins remaining after 29 days, regardless of cage type (Fig. 4B). After 29 days, 55.2% of urchins remained in closed cages, but only 20% survived in the open cages, giving urchins deployed on the reef slope the highest rate of loss (Fig. 4C).