The zombification and reanimation of purple urchins (Strongylocentrotus purpuratus) in response to food availability


 To survive periods of starvation, organisms can reduce their metabolism and/or decrease energy allocation to reproduction. This is especially important for coastal rocky reefs where widespread kelp deforestation has become increasingly common in recent decades. This deforestation often results in the formation of urchin barrens that have high densities of herbivorous sea urchins and little macroalgae for them to consume. While it is clear that these barrens can persist for years to decades, it is unclear how the urchins within them survive such prolonged periods without regular access to macroalgae. Here, we show that urchin metabolism and gonad mass both decrease significantly when the urchins are starved, and that these urchins regain normal metabolic activity and gonad masses when access to food is restored. However, if urchins occur in barren areas that receive drift algae from nearby kelp forests, it appears they can maintain normal metabolic activity and gonad mass. Together, our results provide experimental evidence that reducing metabolism may be a primary strategy for avoiding starvation in urchins occurring within barrens. Our results can be especially important to researchers looking to restore kelp forests and to urchin fishers who seek to harvest these urchins for their gonads but currently cannot because their gonads are of poor quality. Additionally, this has important implications for consumers in other ecosystems where access to energetic resources is spatially or temporally variable and can point to new avenues of research to explain how organisms adjust their energetic needs to survive extended periods of starvation.


Introduction
The ability to survive extended periods of food limitation is essential to survival in habitats where access to energetic resources is temporally or spatially variable. To do this, organisms can reduce their metabolism (Pinter et al 1984;Guidetti et al 2011;Geiser 2004;Careau et al 2014;Crowe 1971) or rely on dormant alternate life stages (Tauber and Tauber 1978;Hochachka and Guppy 1987;Carney and Edwards 2006;Edwards 2000) during periods of resource limitation, and then resume normal metabolic activity or emerge from dormancy once access to resources is restored. The suppression of metabolism can also allow for survival in habitats where environmental conditions, such as temperature and/or rainfall, vary temporally and therefore create periods when conditions are hostile for survival. Indeed, the reliance on dormant alternate life stages has been observed across a wide range of taxa, including terrestrial plants (Venable and Lawlor 1980;Keeley 1987;Leck et al 1989), marine macroalgae (Edwards 2000;Carney and Edwards 2010;Carney et al 2013), microalgae (Hollibaugh et al 1981), terrestrial insects (Tauber and Tauber 1978) and marine crustaceans (Grice and Marcus 1981;Maier 1990).
Alternatively, many birds and mammals exhibit seasonal hibernation or daily torpor to survive unfavorable conditions or periods of food limitation (Pinter et al 1984, Heldmaier et al 2004Tøien et al 2011). However, some species such as the common shrew (Sorex araneus) can starve to death if deprived of food for just a few hours (Crowcraft 1954;Hawkins and Jewell 1962), while others such as the northern sea otter (Enhydus lutris) need to consume large proportions (e.g. 20-50%) of their body mass in prey items per day to survive (Kenyon 1969; Costa and Kooyman 1982;Yeates et al 2007). This (Harrold and Reed 1985), decreased gonad development (Thompson 1983;Dodge and Edwards 2012), and reduced metabolic rates (Spindel et al. 2020) relative to urchins in kelp forests.
Urchins residing within persistent barren grounds can survive by consuming drift macroalgae that originates from nearby kelp forests (Vanderklift and Wernberg 2008;Harrold and Reed 1985;Konar and Estes 2003;Rodríguez 2003;Britton-Simmons et al 2009;Renaud et al 2015;Quintanilla-Ahumada et al 2018). In doing so, they can exhibit normal feeding and gonad development (Harrold and Reed 1985;Rodríguez 2003;Britton-Simmons 2009;Kelly et al 2012), which we hypothesize will also result in normal metabolic rates. This is especially important for barren grounds that are in close proximity to healthy kelp forests and that receive large amounts of drift algae (Vanderklift and Werngerg 2008;Rodríguez 2003;Kelly et al 2012;Figurski 2010;Filbee-Dexter and Scheilbling 2016). However, it remains unclear how these urchins survive for extended periods without regular access to these resources. Over shorter time periods, they can resorb their gonads and other organs to meet their metabolic demands (Russell 1998;Secor and Carey 2016;Carey et al 2016). Over longer time periods, they might lower their metabolic rates, becoming "zombies" (Spindel et al. 2021), and then regain normal metabolic activity, becoming "reanimated", once food access is restored. To examine this, we conducted a laboratory experiment where we measured the metabolic rates and gonad masses of urchins that were starved for 14 weeks, and then regained access to food for an additional seven weeks. We compared these metabolic rates and gonad masses to those in urchins that received regular access to food during the 21-week period. We hypothesized that 1) urchins deprived of food will reduce their metabolic rates, 2) these urchins will return normal metabolic activity when access to food is restored, and 3) urchin gonad mass relative to body size (GSI) will also decrease during starvation and then recover once feeding resumes. Additionally, we examined whether urchins residing in a barren ground that occurs in close proximity to healthy kelp forests and therefore should receive abundant drift macroalgae, exhibit similar metabolic activity and gonad masses to urchins in the nearby kelp forest. We hypothesized that the urchins in these two habitats would not exhibit substantial differences in either metabolic rates or gonad masses.

Feeding Experiment
To examine the effects of starvation on urchin metabolism and gonad mass, we collected approximately 180 purple urchins (Strongylocentrous purpuratus) while on scuba from a 14.5 m deep area within the Point Loma Kelp Forest, USA during late-January and early-February 2020. The Point Loma kelp forest, which is located near San Diego, California, is approximately 11 km long by 1 km wide and occurs over a low-relief sandstone reef at a depth of 7-24 m. This forest is known for supporting high densities of the canopy-forming kelp Macrocystis pyrifera (Fagerli et al 2013;Neushul 1959;Edwards et al 2004), abundant populations of the understory kelps Laminaria farlowii, Pterygophora californica, Egregia menzeseii, and Ecklonia arborea, and numerous species of understory red and brown macroalgae (Dayton 1984;1992;, all of which contribute to seasonally variable patterns of primary production . The forest also provides habitat and food for numerous invertebrates and shes, especially large populations of purple urchins, S. purpuratus (Parnell 2015;Dayton et al 1999). Like many other kelp forests along the California coast (Ebeling et al 1985;Pearse and Hines 1979;Watanabe and Harrold 1991), parts of the Point Loma kelp forest have historically become devoid of eshy macroalgae due to overgrazing by urchins (Parnell 2015). However, there were no such barren areas at the time of this study in Point Loma and the urchins were collected from an area within the kelp forest that had a high density of M. pyrifera and other foliose algae. All urchins were immediately transported to San Diego State University's Coastal and Marine Institute Laboratory (CMIL) where they were haphazardly allocated to one of twelve 3 L aquaria that received owing seawater from a 1500 gal recirculating seawater system, with seawater kept at a constant 14.45 ± 0.44 ℃ (mean ± SE, n = 59). To prevent spawning during transit, which could have affected their gonad mass, the urchins were placed upside-down on top of damp towels on top of frozen water bottles within a cooler, then covered by separate damp towel before shutting the cooler (Robert Dunn, personal communication). In all, ten haphazardly selected urchins were placed in each aquarium and fed M. pyrifera blades ad libitum over a seven-day conditioning period.
After the urchins had acclimated to the aquaria for seven days, six of the twelve aquaria were randomly selected and designated as "Fed" tanks, and the remaining six aquaria were designated as "Starved" tanks. The aquaria were then arranged in an alternating pattern with every other tank being either "Fed" or "Starved". The urchins in the "Fed" tanks continued to receive food at near ad libitum levels over the following seven weeks while the urchins in the "Starved" tanks were deprived of food. After nearly four weeks, daily access to the CMIL became limited due to COVID-19 restrictions, and the urchins in the "Fed" tanks were therefore fed once per week over the following seven weeks with enough M. pyrifera to last several days, while the urchins in the "Starved" tanks continued to be deprived of food. However, access to the CMIL was restricted during a two-week period (weeks 13 and 14) due to COVID-19, and it is therefore likely the fed urchins spent one to two weeks without access to food. Then, after 14 weeks, access to the CMIL was improved and the urchins in all the tanks were fed approximately twice weekly over the following seven weeks with a su cient amount of M. pyrifera to allow feeding near ad libitum levels.
To examine how the different feeding regimes affected urchin metabolism, their respiration rates (i.e. oxygen consumption) were measured on ve occasions following the different feeding schedules.
Speci cally, respiration measurements were made 1) at the start of the experiment (week 0) when all urchins had been fed at near ad libitum levels for seven days, 2) after seven weeks (week 7) during which the "Fed" urchins were fed at near ad libitum levels but the "Staved" urchins were deprived of food), 3) after another seven weeks (week 14) during which the "Fed" urchins were fed weekly but the "Staved" urchins remained deprived of food, 4) after another ve weeks (week 19 ) during which all urchins were fed weekly at near ad libitum levels, and 5) after another two weeks (week 21) when all urchins continued to receive food weekly at near ad libitum levels. To measure urchin respiration, a haphazardly selected urchin was removed from each tank and placed individually into a 1 L water-jacketed respiration chamber that was connected to the recirculating sea water system. This kept the respiration chamber at the same temperature (14.45 ± 0.44 ℃, n = 59) that the urchins experienced during the feeding schedules. The chamber was then connected to a YSI OBOD probe (self-stirring Optical BOD probe) that recorded the seawater temperature and dissolved oxygen (DO) within the chamber every ve seconds using a ProODO optical DO meter (YSI Inc., Yellow Springs, OH, USA). The urchins were allowed to respire for 15 minutes, and their respiration rates were measured by calculating the rate of change in DO (Δ mg O 2 ind − 1 5 sec − 1 ) within the chamber using linear regressions. The volume of seawater in the chamber was adjusted to account for each urchin's presence by subtracting the urchin's volume (measured according to Ebert 1988) from the 1 L chamber volume. A "blank" (i.e. the chamber was lled with seawater but did not contain an urchin) was also measured to estimate respiration by microbes and microalgae within the seawater. We then scaled the "blank" respiration value by the volume of seawater in the chamber during each trial and this value was subtracted from the measurements for each urchin. Estimates of each individual urchin's respiration (mg O 2 ind − 1 hr − 1 ) were then scaled by its test diameter (mm), and the resulting size dependent respiration (SDR) values were expanded to re ect hourly rates (mg O 2 mm − 1 hr − 1 ). Following each respiration measurement, the urchin was removed from the chamber and its mass determined using a digital scale (to the nearest 0.01g). Also, the urchin's test diameter and height were measured with calipers (to the nearest 0.1mm). Following this, the urchin was sacri ced and its gonads were removed and weighed using the same digital scale. Additionally, we opportunistically repeated these measurements on an unreplicated set of ve purple urchins that were collected from the Palos Verdes peninsula (ca 160 km away) and held in a single tank without food for 16 months. While these urchins were collected for an unrelated study and not included in our formal analyses, we present data on the metabolism and gonad mass for a single individual on each sample date to gain insight into the effects of long-term (> 1 year) starvation.

Comparing urchins from barrens and kelp forests
To compare the metabolisms and gonad masses between urchins living in an urchin barren and a nearby kelp forest, we collected 40 purple urchins from Stillwater Cove, located near Paci c Grove, California, USA in February 2020. Stillwater Cove is characterized by moderate-relief granite, sandstone and conglomerate terraces that are separated by cobble and sand channels at depths from 6-14 m. The kelp forest has historically supported a dense surface canopy of M. pyrifera, an abundant subsurface canopy of P. californica, and a diverse assemblage of understory red and brown algae (Edwards 1998;Clark et al 2004), all of which contribute to seasonally variable patterns of primary production . The cove has recently (ca within last 5 years) experienced substantial losses of kelps and understory algae due to large increases in urchin abundances, as large swaths of the cove have been deforested and become urchin barrens. Stillwater Cove was opportunistically selected as our sample site because it has a barren with high urchin density that is near areas of high kelp density. Additionally, its proximity to Moss Landing Marine Lab (MLML) greatly reduced transport time and chances of urchin mortality among the individuals that we collected. Twenty of the urchins were collected from a 7.5 m deep area of Stillwater Cove that had abundant M. pyrifera and understory algae, and the other twenty urchins were collected from a large nearby (ca 200 m away) 12 m deep urchin barren that was surrounded by abundant macroalgae. The urchins from each habitat were placed in separate mesh bags upside down over a damp towel in dark cooler and immediately transported to the aquaculture facility at MLML. At MLML, the urchins from each habitat were divided into two groups and placed into separate oating baskets (n = 10 urchins per basket). All four baskets were placed in a ow-through seawater tank and the urchins allowed to acclimate to the seawater conditions for 16 hrs. After this acclimation period, urchin respiration was measured using the same methods described above for the laboratory measurements, with the respiration chamber's water jacket connected to a water chiller that maintained the seawater at ambient temperatures (13.28 ± 0.59 ℃, n = 20).

Statistical Analysis
All statistical analyses were done using the statistical software R-Studio Version 1.2.5001, R Version 3.5.2., and Primer-E Ver. 6. Before analyses, all data were checked for normality using Shapiro's test, and for equal variances using Bartlett's and/or Levene's tests. We used Box-Cox transformation to satisfy assumptions of parametric statistics when necessary. However, data for size dependent respiration (SDR) within the starved treatment were both non-normal (Shapiro test: p = 0.043) and heteroscedastic among measurement weeks (Levene's Test: p = 0.02), and these problems could not be corrected by transformation. Therefore, we used a two-way xed factor permutation analysis of variance (PERMANOVA) on a Euclidean distance-based resemblance matrix to evaluate differences in SDR between the fed and starved urchins (i.e. treatment), and among the ve sample dates (i.e. weeks) (Anderson et al 2006). Data were square-root transformed prior to analysis. We then used individual parametric t-tests as post hoc tests to evaluate a priori hypotheses regarding differences in SDR among feeding treatments during each sample week separately. Here, all data met the assumptions of parametric statistics. We used Benjamin-Hochberg adjusted p-values to control for Type I error in ation due to the multiple tests (Benjamin and Hochberg 1995). We similarly used a two-way xed-factor ANOVA to evaluate differences in GSI between the fed and starved urchins, and among the ve sample dates, and followed this with individual parametric t-tests to evaluate a priori hypotheses regarding differences in GSI among treatments during each sample week separately. We again adjusted the resulting p-values to control for Type I error in ation due to the multiple tests (Benjamin and Hochberg 1995). Here, data for GSI were heteroscedastic between the treatments during week 7, and we therefore used a Welch's t-test for unequal variances for this comparison. We used separate 3-factor ANCOVAs to evaluate differences in the relationship between individual urchin metabolic rate and urchin test diameter, and between GSI and urchin test diameter, among feeding treatments and weeks. We followed each of these with separate 2factor ANCOVAs as post hoc tests to evaluate a priori hypotheses regarding differences in the relationship between individual urchin metabolic rate and urchin test diameter, and between GSI and test diameter, between the two feeding treatments within each week separately. We then used ANCOVA to evaluate differences in the relationship between SDR and GSI between the two feeding treatments. Lastly, we used separate t-tests to evaluate differences in SDR and GSI between urchins collected from the kelp forest and urchins barren areas within Stillwater Cove. Data for SDR were heteroscedastic and therefore square-root transformed prior to analysis.
However, as with SDR, when a prior hypotheses regarding differences between feeding treatments were examined within each week separately, these patterns varied among sample weeks (Table 2B). Speci cally, and similar to SDR, GSI did not differ between the urchins that would later be starved (2.15 ± 0.22 %, mean ± SE, n = 6) and those that would be fed regularly (2.87 ± 0.37 %, n = 6) at the start of the experiment (week 0: t-test: t = 1.682, df = 10, p-adjust = 0.163). However, GSI had decreased in the starved urchins after seven weeks without food (0.38 ± 0.11 %, n = 6) but not in the fed urchins (2.53 ± 0.78 %, n = 6), although these differences were only marginally statistically signi cant (week 7: Welch's t-test: t = 3.180, df = 5.21, p-adjust = 0.090). After an additional seven weeks of starvation, GSIs remained lower in the starved urchins (1.42 ± 0.66 %, n = 6) than in the fed urchins (3.99 ± .69%, n = 6), (week 14: t-test: t = 3.080, df = 10, p-adjust = 0.057). Following this, the starved urchins began receiving regular access to food over the following ve weeks and their GSIs increased (4.53 ± 0.62 %, n = 6), but they remained lower than in the urchins that had received regular access to food (6.49 ± 0.56 %, n = 6) (week 19: t-test: t = 672, df = 10, p-adjust = 0.070). Then, after an additional two weeks of feeding, GSIs in the starved urchins continued to increase (5.57 ± 0.84 %, n = 5) such that they were no longer different than those in the fed urchins that had received regular access to food during the 21-week experiment (6.55 ± 1.70 %, n = 6) (week 21: t-test: t = 0.484, df = 9, p-adjust = 0.620). Together, this indicated that GSI followed similar, albeit less pronounced and perhaps slightly delayed, responses to starvation and feeding as SDR.
Additionally, examination of the ve urchins that were collected from Palos Verdes, CA revealed that these urchins had no identi able gonads (GSI = 0%) on any of the ve measurement dates.
We observed an overall positive relationship between individual urchin respiration (mg O 2 ind -1 hr -1 ) and urchin test diameter (mm) when examined across both feeding treatments and all sample weeks (ANCOVA: F 1,39 = 9.264, p = 0.004) (Table 3A, Figure S1). However, when a prior hypotheses regarding differences in the relationship between individual respiration and test diameter were examined between the two feeding treatments within each week separately, these patterns varied among sample weeks (Table 3B). Speci cally, similar to patterns observed for SDR, individual respiration rates did not differ between the feeding treatments at the start of the experiment (ANCOVA, week 0: F 1,8 = 0.011, p = 0.918), but they were signi cantly lower in the starved urchins than the fed urchins after both seven (week 7: F 1,8 = 22.829, p = 0.001) and fourteen (week 14: F 1,8 = 4.919, p = 0.057) weeks of starvation across the range of observed urchin sizes. Then, after the starved urchins began receiving regular access food, individual respiration increased such that it was only marginally lower in the starved urchins than the fed urchins (week 19: F 1,8 = 4.432, p = 0.068) after ve weeks of feeding, and it was not different between the two feeding treatments after an additional two weeks of feeding (week 21: F 1,8 = 0.054, p = 0.824) (Table 3B, Figure S1). Interestingly, the relationship (slope) between individual respiration and test diameter did not differ between the feeding treatments in any of the sample weeks (treatment*diameter interaction: p > 0.1 for all ve weeks, Table S3B).
Unlike respiration rates for individual urchins, we did not observe an overall signi cant relationship between urchin GSI (%) and urchin test diameter (mm) when examined across both feeding treatments and all sample weeks (ANCOVA: F 1,38 0.112, p = 0.734) (Table 4A, Figure S2). However, when a prior hypothesis regarding differences in the relationship between GSI and test diameter were examined between the two feeding treatments within each week separately, these patterns did vary among sample weeks (Table 4B). Speci cally, GSI did not differ between the feeding treatments at the start of the experiment (ANCOVA, week 0: F 1,8 = 2.332, p = 0.165) or after seven weeks of starvation (week 7: F 1,8 = 0.655, p = 0.442). Then, after an additional seven weeks of starvation, GSIs were signi cantly lower in the starved urchins than in the fed urchins (week 14: F 1,8 = 6.120, p = 0.038). GSIs remained lower in these urchins even after they began receiving food over the following ve weeks (week 19: F 1,8 = 11.023, p = 0.011), but were not different between the two feeding treatments two weeks later (week 21: F 1,8 = 0.367, p = 0.564) (Table 4B, Figure S2). However, the relationship between GSI and test diameter did not differ between the feeding treatments during any of the sample dates during the rst fourteen weeks of the experiment (treatment*diameter interaction: p > 0.5 for all three sample weeks), but it did differ between the feeding treatments on both sample dates during the last seven weeks (treatment*diameter interaction: p < 0.05 for the last two sample weeks) (Table 4B). Lastly, SDR was positively related to GSI when examined across all urchins used in this study (ANCOVA: F 1,41 =51.711, p < 0.001), but the relationship between SDR and GSI did not vary among the urchins that were fed consistently throughout the study, those that were starved for the rst fourteen weeks of the study, or those that regained access to food during the last seven weeks of the study (GSI*treatment interaction: F2,41 = 1.180, p = 0.328) (Table 5, Figure 3).

Comparing urchins from a barrens and nearby kelp forest
Urchins collected from within a barren area of Stillwater Cove did not show the same effects of starvation as those observed in the feeding experiment, even though no macroalgae were observed within the barren at the time of collection. Speci cally, SDR did not differ between the urchins that were collected from within the Stillwater Cove urchin barren (0.018 ± 0.006 mg O 2 mm -1 hr -1 , n = 10) and those that were collected from the nearby kelp forest (0.020 ± 0.006 mg O 2 mm -1 hr -1 , n = 10) (t-test: t=0.981, df=18, p=0.340) ( Table 6, Figure 4A). Likewise, GSI also did not differ between the urchins collected from within the kelp forest (4.03 ± 2.69 %, n =10) and those collected from within the nearby urchin barren (2.66 ± 1.56 %, n = 10) (t-test: t = -1.386, df = 18, p= 0.183) ( Table 6, Figure 4B).

Discussion
The primary result of our study is that purple urchins decrease their metabolism when access to food is eliminated and then restore normal metabolism once access to food is regained, and that this occurs over a time frame of a few weeks (Fig 1). This may allow urchins occurring within barren grounds that are devoid of macroalgae algae and that do not receive adequate drift algae to survive prolonged periods of starvation, which has led to some urchin species being characterized as "zombie urchins" (Spindel et al. 2021). Further, our data show that these urchins regain normal metabolic activity (i.e. they become "reanimated") if they regain access to food, such as would be expected when the kelp forests recover or if the urchins receive drift algae from nearby kelp forests. Thus, patterns of kelp deforestation and recovery, and of proximity to nearby forests are likely important drivers of urchin metabolism. Speci cally, although the urchins used in this study exhibited similar metabolic activity at the start of the experiment prior to any manipulation, the urchins that were subsequently starved for seven to fourteen weeks exhibited substantially lower metabolic rates than urchins of similar size that were fed consistently during this period. Then, when these starved urchins regained regular access to food during the subsequent ve to seven weeks, their metabolisms returned to levels that were similar to those at the start of the experiment, and to the urchins that were fed regularly during the 21-week experiment. In contrast, the urchins that were fed regularly throughout the experiment did not alter their metabolisms, except for a smaller decrease that occurred in week 14 following a period when all urchins were presumably deprived of food for one to two weeks because of limited access to them due to COVID-19 restrictions. Additionally, we opportunistically examined the effects of long-term starvation on a group of ve purple urchins that were collected from Palos Verdes peninsula as part of an unrelated study. All ve urchins survived 16 months without access to food and exhibited substantially reduced metabolic activity. Thus, it is likely that metabolic activity in our starved urchins from Point Loma would have continued to decrease if the experiment had run longer. We also observed that urchin metabolism was, in general, positively related to urchin size, but this was not signi cant in all feeding treatments and weeks (Fig S1). We believe this lack of signi cance was likely due to low sample sizes (n = 6 for each treatment*week commination), as other studies have found urchin metabolism to be signi cantly linked with size (Spindel et al. 2021, M. Edwards, unpublished data).
Our ndings have important implications not only for urchins within barrens but also for those occurring in areas where kelp forests recover following extended periods of deforestation, such as has been observed in numerous areas of the world (Ebeling et al 1985;Estes et al 1998;Pearse and Hines 1979;Scheibling et al 1999;Bolton et al 2012;Fagerli et al 2013). Indeed, our study suggests that these urchins can restore normal metabolic activity within a few weeks once the kelp forests recover and access to food is restored. Our study further suggests that urchins residing in barren grounds that receive adequate drift algae from nearby kelp forests (Vanderklift and Wermberg 2008;Harrold and Reed 1985;Konar and Estes 2003;Rodríguez 2003;Britton Simmons et al 2009;Renaud et al 2015) can maintain normal metabolic activity. Indeed, the urchins from within the barren in ground Stillwater Cove did not exhibit signi cantly different metabolic rates than the urchins from within the nearby kelp forest, both of which were similar to the rates observed within the urchins that had been fed regularly throughout the 21-week feeding experiment (Figs. 1, 4A). We believe that this was because the urchin barren ground in Stillwater Cove was surrounded by healthy kelp forests that provided large amounts of drift algae (Figurski 2010;Edwards unpublished data). Additionally, the primary substrates in Stillwater Cove are dominated by crustose coralline algae (Edwards 1998), which can serve as an alternative food source for the urchins (Harold and Reed 1985), but it is unclear if this is su cient to meet their energetic needs. Thus, future research should focus on the importance of allochthons transport of drift algae to the barren grounds and on the energetic value of crustose coralline algae.
Much like with urchin metabolism, we observed a signi cant decrease in urchin GSIs within the urchins that were starved but not in the urchins that received food on a regular basis (Fig. 2). GSIs were similar among the urchins at the start of the experiment but decreased in the starved urchins over the next seven to fourteen weeks. In contrast, GSI did not change in the urchins that received regular access to food over this period. This is not surprising given that previous studies that have found GSIs to be in uenced by food availability (Thompson 1983;Dodge and Edwards 2012). However, while signi cant decreases in metabolism in the starved urchins occurred within the rst seven weeks of the experiment (i.e. by week 7), we did not observe signi cant decreases in GSIs in these urchins until the second seven weeks of the experiment (i.e. by week 14) (Figs. 2, S2). This again is not surprising given previous studies that have shown urchins can resorb their gonad tissue and thereby reduce their GSIs when starved (Russell 1998;Secor and Carey 2016). We interpret this to mean that when deprived of food, the urchins rst decrease their metabolic demands, and then resorb their gonad tissues if starvation continues. Indeed, no identi able gonads were found in the urchins from Palos Verdes after they were starved for sixteen months. However, once the urchins from Point Loma began feeding, we observed a complete return of normal metabolic activity within ve weeks, but we did not see a return to normal GSI until two weeks later, which we interpret to mean that gonad recovery began after metabolism was restored. This was further supported by the observation that GSIs were positively related to metabolism regardless of whether they had access to food or not, though the strength of this relationship varied depending on the urchin's feeding history (Fig. 3). Additionally, GSIs in the urchins within the Stillwater Cove barren ground were not different from those in the kelp forests (Fig. 4B), which also supports our hypothesis that access to drift algae can maintain both normal metabolism and reproductive condition.
Together, our study shows that purple urchins are indeed capable of becoming "zombies" by reducing their metabolic activity, resorbing their gonad tissue, and surviving long periods without food. These urchins can then become "reanimated" by restoring normal metabolic activity and increasing their gonad masses once access to food is regained, or if they have access to drift algae. This may explain how urchins living in persistent urchin barrens are able to survive for years without apparent access to food, such as those within the long-standing barren grounds of the Aleutian Islands (Estes et al 1998;Konar et al 2014;Konar and Estes 2003), and it may help predict their survival in the newly established barren grounds in northern California (Rogers-Bennett and Catton 2019). Unfortunately, COVID-19 related restrictions reduced access to the CMIL for activities other than feeding the urchins on all but a few dates when metabolisms were measured, and therefore prevented us from examining ner scale timedependent changes in metabolism and GSI. Future studies should examine variation in metabolic activity in shorter time intervals, which will also provide a clearer understanding of how quickly increasing respiration can drive gonad tissue recovery. In addition, future studies should examine the viability of the gonad of reanimated urchins. Furthermore, it is important to understand how starvation affects other species of urchins that form barrens. In particular, red urchins (Mesocentrotus franciscanus) are an important barren forming species in central and northern California (58,68), and in British Columbia (Spindel et al. 2021) that similarly exhibits reduced metabolism relative to when they live in kelp forests (Spindel et al. 2021). Crowned urchins (Centrostephanus coronatus) are expanding their range expansion northward along the southern California coast and are becoming more in uential in the ecology of California's kelp forests (Freiwald et al 2016). Green urchins (Strongylocentrotus drobeacheinsis) form extensive urchin barren grounds in Alaska's Aleutian Islands (Estes et al 1998), the east coast of Maine (Steneck et al 2013), Nova Scotia (Scheibling et al 1999), and northern Europe (Fagerli et al 2013 Tables   Table 1. A) Results of a two-way xed-factor PERMANOVA testing differences in size-dependent respiration (SDR) in purple sea urchins between feeding treatments (fed, starved) and among the ve sample weeks (weeks 0, 7, 14, 19, 21). B) Results of separate t-tests examining a prior hypotheses regarding differences in SDR between feeding treatments during each sample week separately. Both uncorrected (p-value) and Benjamin-Hochberg corrected (p-adj) p-values are given, the later of which were adjusted according to avoid alpha error in ation due to the multiple tests (102) (n = 6 for each feeding treatment and each sample week).  Table 2. A) Results of a two-way xed-factor PERMANOVA testing differences gonad somatic indices (GSI) in purple sea urchins between feeding treatments (fed, starved) and among the ve sample weeks (weeks 0, 7, 14, 19, 21). B) Results of separate t-tests examining a prior hypotheses regarding differences in GSI between feeding treatments during each sample week separately. Both uncorrected (p-value) and Benjamin-Hochberg corrected (p-adj) p-values are given, the later of which were adjusted according to avoid alpha error in ation due to the multiple tests (102). A Welch's t-test for unequal variances was used in week 7 (n = 6 for each feeding treatment and each sample week).  Table 4. A) Results of a three-way ANCOVA testing differences in the relationship between urchin test diameter (mm) and gonad somatic indices (GSI) in purple sea urchins between feeding treatments (fed, starved) and among the ve sample weeks (weeks 0, 7, 14, 19, 21). B) Results of separate two-way ANCOVAs examining a prior hypotheses regarding differences in the relationship between urchin test diameter and GSI in purple sea urchins between feeding treatments during each of the ve sample weeks.
(n = 6 for each feeding treatment and each sample week).
A  Table 5. Results of two-way ANCOVA testing differences in the relationship between gonad somatic indices (GSI) and respiration (mg O 2 individual -1 ) in purple urchins from three different feeding treatments (fed regularly through throughout the entire experiment, starved during the rst 14 weeks of the experiment, and fed during the last seven weeks of the experiment following a period of starvation) (n = 24, 12, and 11, respectively).  Figure 1 Boxplots showing differences in size dependent respiration (SDR [mg O2mm-1hr-1]) in purple urchins among both the fed and the starved treatments throughout time. Black horizontal lines above pairs of boxes within weeks 0, 19, and 21 denote non-signi cant difference in respiration rates between the feeding treatments. "Week 0" measurements were taken before feeding manipulations began. "Week 7" and "Week14" measurements occurred while the "Starved" urchins had no access to food while "Week 19" and "Week 21" measurements occurred while both "Starved" and "Fed" urchins were receiving equal amounts of food.

Figure 3
Regressions showing the relationships between size dependent respiration (SDR) and gonad somatic indices (GSI) in purple urchins within each of the feeding treatments. Urchins in the "Fed" group were measured on each of the sample dates (weeks 7, 14, 19 and 21) during which time they received regular access to food over the 21 week experiment (n = 24). Urchins in the "Post-Starvation" group were urchins that were measured during the last seven weeks of the experiment (weeks 19 and 21) after they began receiving regular access to food following being starved for the rst 14 weeks of the experiment (n = 11).
Urchins in the "Starved" group were urchins that were measured during the rst 14 weeks of the experiments (weeks 7 and 14) during which time they were deprived of food (n = 12).