Large‐scale one‐off sea urchin removal promotes rapid kelp recovery in urchin barrens

Sea urchin overgrazing is a leading cause of kelp forest loss and in such cases their removal is increasingly advocated for kelp forest restoration. However, refining removal approaches is needed to improve the efficiency and success of restoration, as most previous removal studies have been small scale and require ongoing removals to maintain low densities and allow kelp recovery. We investigated the effectiveness of one‐off urchin removal from large, semi‐discrete areas of urchin barrens as a tool to promote kelp recovery. We removed sea urchins (Evechinus chloroticus) from four areas (1.6–2 ha) of urchin barrens in northeastern New Zealand. Exposed urchins were reduced to approximately 7% of initial densities and remained low, yet cryptic urchin densities increased after 2 years. Kelp (Ecklonia radiata) and fucoid (Sargassum sinclairii, Carpophyllum spp.) densities increased rapidly in removal areas, but remained constant or declined in adjacent control urchin barren areas. Macroalgal canopy recovery varied among and within removal areas, but increased on average from approximately 5 to 43% in 2 years. Densities of large sea urchin predators did not increase with kelp recovery, likely due to ongoing fishing within removal areas. Our results demonstrate that a single urchin removal from large, semi‐discrete areas of urchin barrens can effectively and efficiently promote rapid multi‐species macroalgal recovery without additional actions (e.g. repeated urchin removals or macroalgae enhancement) for at least 2 years. However, this approach does not restore whole ecosystems and consequently restoration benefits through kelp recovery will be temporary without longer‐term urchin management and/or rebuilding of predator populations.


Introduction
"There can be no purpose more enspiriting than to begin the age of restoration, reweaving the wondrous diversity of life that still surrounds us… The next century will, I believe, be the era of restoration in ecology."-TheDiversity of Life, EO Wilson (1992).
Marine restoration is a rapidly expanding field and a key strategy for rebuilding ocean health (Duarte et al. 2020;Saunders et al. 2020).Yet, to date, many initiatives have been hindered by high costs and low confidence in the potential for success (Saunders et al. 2020;Stewart-Sinclair et al. 2020).Research on kelp restoration has increased dramatically in recent years to better understand the methods, costs, and scope for success (Layton et al. 2020;Eger et al. 2022;Miller et al. 2022).Kelp characteristics (e.g.life history, population connectivity) are conducive to effective restoration (Bekkby et al. 2020), enhancing the success, efficiency, and speed of restoration relative to other marine habitats, such as coral reefs or mangroves (Bayraktarov et al. 2016).However, kelp restoration success has been variable and most projects have so far been relatively small scale (<1 ha, 10,000 m 2 ) and short in duration (<2 years; Eger et al. 2022).To be effective, restoration methods must identify and mitigate the cause(s) of kelp decline (Morris et al. 2020), which is becoming increasingly complex with multiple stressors (Strain et al. 2014;Wernberg et al. 2019b;Smale 2020).
Sea urchin herbivory is a primary cause of kelp forest decline on temperate rocky reefs (Steneck 2020).The irruption of sea urchin populations can spur a phase-shift from productive kelp forests to depauperate "urchin barrens" with reduced productivity, biodiversity, and loss of other ecosystem benefits (Graham 2004;Ling 2008;Filbee-Dexter & Scheibling 2014).Once established, urchin barrens can persist in a stable state for decades (Filbee-Dexter & Scheibling 2014; Duarte et al. 2022), displaying hysteresis, whereby it takes a much greater reduction in sea urchin biomass to allow kelp to regrow than the biomass that caused the phase-shift to barrens (71 AE 20 g/m 2 compared to 668 AE 115 g/m 2 ; Ling et al. 2015).These thresholds also depend on nutrient regime (Boada et al. 2017) and sea urchin behavior; in northeastern New Zealand the estimated threshold density for kelp recovery is 1 exposed (non-cryptic and actively grazing) adult sea urchin/m 2 , or approximately 134 g/m 2 (Shears & Babcock 2003).Understanding the dynamics and feedback mechanisms that maintain urchin barrens is essential to planning and designing effective kelp forest restoration strategies.
Large increases in sea urchin populations and subsequent overgrazing of kelp are frequently attributed, at least in part, to overfishing of their predators (Steneck et al. 2002;Steneck 2020).In such cases, protection of sea urchin predators in marine protected areas (MPAs) can result in kelp forest recovery by reducing sea urchin densities (Babcock et al. 1999;Behrens & Lafferty 2004;Edgar et al. 2009), but these processes can take decades (Babcock et al. 2010;Eisaguirre et al. 2020;Peleg et al. 2023).Therefore, alternative and complementary approaches such as active sea urchin removal are increasingly being sought to promote and accelerate kelp recovery (Piazzi & Ceccherelli 2019;Eger et al. 2020;Miller et al. 2022).
Active sea urchin removal from urchin barrens can result in rapid kelp recovery (Ling et al. 2015) and is increasingly being promoted as a tool for kelp forest restoration (Miller et al. 2022).Most sea urchin removal projects have been small scale (<1 ha) and have required ongoing sea urchin removals (typically weekly-monthly) to maintain low urchin densities (Miller et al. 2022).However, for kelp restoration purposes, sea urchin removals are increasingly being carried out at larger spatial scales (>1 ha; e.g.Taino 2010;Sanderson et al. 2016;Lee et al. 2021;Williams et al. 2021), which likely reduces the rate of sea urchin reinvasion seen in small-scale removals (Shears & Babcock 2002;Miller et al. 2022).Despite extensive efforts, large-scale removal studies have had varying success.As such there is no clear guidance on the most efficient and effective approaches to urchin removal and promoting kelp recovery.While reinvasion of removal areas will occur over time through natural recruitment processes, designing removal areas that minimize or eliminate opportunities for reinvasion of adult urchins, via natural barriers and constriction points (e.g.shallow water, sand-reef edges), can optimize removal efforts and maximize the chance of natural kelp recovery (Ling 2008;Kriegisch et al. 2016).
In northeastern Aotearoa New Zealand, urchin barrens dominated by the common sea urchin "kina" (Evechinus chloroticus) typically form a distinctive band on shallow reefs (approximately 2-9 m depth) outside of MPAs (Grace 1983;Shears & Babcock 2004).Primary urchin predators including large snapper (Chrysophrys auratus) and spiny lobster (Jasus edwardsii; Babcock et al. 1999, Shears & Babcock 2002) have experienced major declines in size and abundance due to commercial and recreational fisheries (Parsons et al. 2009;LaScala-Gruenewald et al. 2021).In contrast, nearby fully protected MPAs have shown kelp recovery across the same depth range, pointing to overfishing as the primary cause of urchin barrens outside of MPAs (Shears & Babcock 2003;Peleg et al. 2023).Previous urchin removal studies in New Zealand have resulted in increased macroalgal densities (Ayling 1981;Andrew & Choat 1982;Villouta et al. 2001).Furthermore, urchin reinvasion into smaller removal areas greatly limited kelp recovery despite regular removals (Shears & Babcock 2002).These studies suggest that sea urchin removal may promote kelp recovery, but to be effective, these efforts should be conducted at sufficiently large scales and discrete areas of barrens can minimize urchin reinvasion.
The distinctive band of urchin barrens in northeastern New Zealand provides an opportunity to examine how sea urchin removal from large, discrete areas can promote kelp forest recovery.Furthermore, the proximity of shallow and deepwater algal forests may enhance macroalgal recovery without active enhancement and reseeding.The aim of this study was to examine whether a single sea urchin removal from large, semi-discrete areas of urchin barrens (e.g.bays or areas with constriction points) can be effective at promoting kelp forest recovery.We carried out a single sea urchin removal from four semi-discrete areas (1.6-2 ha) of urchin barrens at sites located across a range of environmental conditions in the Hauraki Gulf, northeastern New Zealand.We measured sea urchins and macroalgal communities in removal and control areas over a 2-year period to estimate (1) how quickly sea urchins recolonize the removal areas and (2) the rate of recovery of kelps and other canopy forming species following urchin removal.Given the importance of predators in controlling urchin populations, we also monitored the abundance and size of key predators over the 2-year period to assess if they responded to kelp recovery.

Study Sites
This study was performed at four sites across the Hauraki Gulf (Tīkapa Moana/Te Moananui-a-Toi), northeastern New Zealand.The inner Gulf is characterized by high anthropogenic sedimentation and turbidity, while the outer Gulf and offshore islands have relatively high water quality, clarity, and wave exposure (Seers & Shears 2015).This environmental gradient is mirrored with changes in macroalgal communities whereby with increasing turbidity macroalgal distributions vertically truncate and sediment tolerant species (especially Carpophyllum flexuosum) become more widespread (Blain et al. 2019).All sites showed depth-based zonation of communities typical for this region.Shallow mixed algal zones (including Ecklonia radiata, Carpophyllum maschalocarpum, C. plumosum, Sargassum sinclairii, Cystophora retroflexa, foliose seaweeds) extended from approximately 0 to 2 m depth (below mean low water springs [MLWS]), followed by urchin barrens from approximately 2 to 9 m, below which are deep kelp (E.radiata) or, where reef is limited, sand flats (Grace 1983;Shears & Babcock 2004).
Study sites were selected using satellite imagery to identify large semi-discrete areas of barrens across a gradient of increasing water clarity and wave exposure (Table 1; Fig. 1).The study sites were predominately sloping rocky reefs of moderate complexity with areas of bedrock and boulder (Miller & Shears 2023).Evechinus chloroticus was the dominant sea urchin species at all sites.While relatively few long-spined sea urchin Centrostephanus rodgersii were present in the study areas, their densities are regionally increasing (Balemi & Shears 2023).Each site included a removal and control area (Fig. 1).

Experimental Removal of Sea Urchins
Permission was granted from local M aori iwi (tribes) Ng ati Manuhiri and Ng ai Tai ki T amaki, and Fisheries New Zealand (Ministry of Primary Industries Special Permit 679-3) to undertake removals at these locations.Given the generally small size and poor roe quality of sea urchins within urchin barrens and the scale of the removal areas, culling was the most practical method and deemed appropriate for this study (Miller & Shears 2023).Most urchins were crushed by SCUBA divers using a 3-4 cm diameter metal pipe or a large hammer with remains left in situ, adapted from methods in California (House et al. 2018; K. Rootsaert 2020, Giant giant kelp restoration (G2KR), personal communication).Some sea urchins were collected and distributed to the local M aori community when condition was suitable (Miller & Shears 2023).
Sea urchin removal was conducted sequentially across sites from September 2020 to March 2021 (spring to late summer; Table 1).Removal rates were similar across sites at approximately 50.5 diver hours/ha (Miller & Shears 2023).Removal area boundaries were marked with permanent subsurface floats at each corner and during the removals, side boundaries were marked with weighted lines.Divers aimed to clear all kina ≥40 mm test diameter (approximately 90% effectiveness; Miller & Shears 2023).This approach targeted large exposed individuals with the greatest grazing impacts, while avoiding cryptic juvenile kina (<40 mm) hidden in crevices or under rocks (Shears & Babcock 2002).Sea urchins were systematically culled across sites including approximately 3 m buffer past where barren habitat transitioned into deep contiguous kelp habitat.Culling continued as far as practical to reduce incursion from the shallow mixed algal zone.Sites were systematically cleared using lanes designated by weighted transect lines or lines of divers working side-by-side.

Biological Monitoring
Kina, macroalgae, and benthic habitat were surveyed in removal and control areas pre-removal, and approximately 6, 12, and 24 months post-removal.An abbreviated survey immediately post-removal (sea urchin density, size, and behavior) in removal areas only ensured densities of adult urchins were less than 1 urchin/m 2 .
Four semi-permanent transect lines inside and outside removal areas were marked with weights and subsurface floats and global positioning system-marked start and end points (Fig. 1).These ran roughly perpendicular to shore from the Table 1.Site description and summary information for large-scale sea urchin removals at four sites in the Hauraki Gulf, northeastern New Zealand.Wind fetch (as a proxy for wave exposure) was calculated using the windfetch webapp (Seers 2021) with vectors calculated at 10 intervals, and distances truncated at 300 km and averaged across all vectors.Secchi depth measurements from Hansen (2019).Kina (exposed only) and canopy cover (kelp and fucoid) data show values for removal areas only (mean AE SE). ).Patches of high densities of kelp, fucoid, or sand/ gravel within the boundaries of initial barren were retained.

Site
To understand the potential effects of kina removal and kelp restoration on snapper (Chrysophrys auratus) and spiny lobster (Jasus edwardsii and Sagmariasus verreauxi), sites were surveyed using underwater visual census pre-removal, 6, 12, and 24 months post-removal.Six transects were conducted inside and outside removal areas between approximately 4 and 8 m.Fish transects were 25 m long, 5 m wide, and 5 m high (125 m 2 area or 625 m 3 ) following Department of Conservation protocols (Haggitt 2011), with legal snapper size considered ≥30 cm fork length (FL).Lobsters were surveyed with larger transects (30 Â 10 m) and size was estimated visually; legal size J. edwardsii are ≥95 mm carapace length.

Statistical Analyses
Data analyses were undertaken in R (v4.0.3,R Core Team 2020).The brms package (Bürkner 2017) was used to fit Bayesian models using the Hamiltonian Monte Carlo algorithm implemented in the programming language Stan (Carpenter et al. 2017).All models were ran using weakly informative priors and chain convergences using the b R statistic (Gelman & Rubin 1992).To determine the effect of the kina removals on total canopy cover of all kelp and fucoids over a 2-year period, a hurdle mixed effects model with a gamma distribution was used.The hurdle model was chosen due to excess zeros due to absence of a macroalgal canopy in the barrens at the start of the experiment, while the gamma distribution could effectively model percentage (non-integer) data.For the effects on kelp, fucoid, or other densities, hurdle mixed effects models with a Poisson distribution were used for count data.
Models (specified in brms syntax below) tested fixed effects treatment (control, removal), time (0, 6, 12, and 24 months), and depth with random effects site, transect (nested within site) and quadrat ID.Quadrat ID was included to manually generate an estimate of overall residual error.For the hu (zero) components of the model, only treatment and time effects were modeled.Posterior pairwise differences between treatment and time combinations were assessed by testing the 95% highest density intervals of group means.Differences were considered significant if the 95% CI of group differences did not cross zero. Model: To determine if treatment (control, removal) effects persisted through time on kina density (cryptic and exposed) postremoval, a hurdle mixed effects model with Poisson distribution was used only post-removal (time at 6, 12, and 24 months).These models follow a similar design to the model specified above but with kina density (cryptic and exposed, modeled separately) as the response variable.Depth was not expected to have an effect on kina densities and was excluded from these models.For densities of kina predators (snapper and spiny lobster, both total and legal-sized, modeled separately), a hurdle mixed effects model with Poisson distribution was used with predator density as the response variables and treatment, time, and site as fixed and random factors as above.

Kina Density
Across all sites, approximately 403,000 kina and 166 Centrostephanus rodgersii were removed from 7.1 ha of urchin barrens (Table 1).Exposed kina densities were reduced by approximately 93% within the removal areas at all sites (Table 1; Fig. 2).Modeling showed exposed sea urchin densities remained consistently low in removal areas from 6 to 24 months post-removal (0.3 AE 0.03 kina/m 2 ; Table S1

Ecklonia radiata
Sea urchin removal had a positive effect on the density of the kelp Ecklonia radiata (Table S3; Figs. 4 & S6).Within the first year, E. radiata density in removal areas increased significantly from an average of 2.0 AE 0.4 to 13.8 AE 1.3 plants/m 2 and remained at similar densities after 2 years (12.5 AE 0.9 plants/ m 2 ; Table S3; Fig. 4).In the control areas, densities showed a small but significant decrease after 2 years (Table S3; Figs. 4 &  S6).
The timing of peak E. radiata recruitment varied among sites post-removal, occurring between 0 and 6 months post-removal at Nordic and Hauturu-o-Toi and between 6 and 12 months post-removal at V Bay and Otata (Fig. 4).In control areas, E. radiata density increased in spring/summer (September 2021-January 2022) at Nordic, Otata, and Hauturu-o-Toi, but then declined to zero or low numbers in subsequent surveys (Fig. 4).
Within the first year, most of the kelp that had recruited into and survived within removal and control areas were small (<25 cm) juveniles or sub-adults.However, while the growth rate varied by site, all sites showed numerous fully grown (≥75 cm) E. radiata by 2 years (Fig. 5).In removal areas, E. radiata canopy cover was relatively low (8.8 AE 1.4 to 21.1 AE 3.6%) at 12 months, but more than doubled between 12 and 24 months post-removal (27.5 AE 3.9 to 46.7 AE 3.9%; Figs. 5 & S1).E. radiata density increased across all depths, but models showed a weak positive depth effect (Table S3; Fig. S2).

Sargassum sinclairii
Sargassum sinclairii density increased rapidly in removal areas and continued to increase through time, with little or no increase in control areas (Figs. 4, 5, & S6); however, seasonal fluctuation was seen for both density and canopy cover.Models showed a weak negative depth effect (Table S3).The highest densities were generally recorded on shallow reef in the removal areas (<6 m depth; Fig. S3); however, at Hauturu-o-Toi, densities of approximately 30 plants/m 2 were seen between 6 and 9 m in summer (November 2021; Figs. 5 & S3).Overall, percent canopy cover of S. sinclairii remained less than 10% in removal areas (except for Hauturu-o-Toi, approximately 18% in November 2021) and less than 3% in control areas (Fig. S1).
Large-scale urchin removal for restoration Carpophyllum spp.
The density of Carpophyllum spp.increased in removal areas through time and was variable in control areas (Fig. 4), but modeling showed no effect of removal (i.e.no treatment Â time interaction) on Carpophyllum spp.densities (Table S3; Fig. S4).Although this was also shown in the pairwise contrasts of the posterior marginal effects (Fig. S5), the hu component of this model suggests an interaction with the probability of density being 0 is lower in the removal areas after 6 months (Table S3; Fig. 4).

Sea Urchin Predators
Densities of snapper (Chrysophrys auratus) and spiny lobster (Jasus edwardsii, Sagmariasus verreauxi) did not show a consistent effect of urchin removal across the 2-year study period (Table S4; Figs. 6 & S7).While total snapper densities showed a significant treatment Â time interaction, these were variable over time and there was no clear increase through time or difference between removal and control areas (Table S4; Figs. 6 & S7).Pairwise comparisons showed a significant removal effect for total snapper densities 1 year after removal, which corresponds to a large spike in the removal areas at V Bay and Otata (Figs. 6 & S4).This effect was not evident after 2 years.The vast majority (95%) of snapper recorded were juveniles and legal-sized snapper were rare (only recorded at two sites) and did not vary in relation to urchin removal (Table S4; Figs. 6 & S7).

Discussion
This study demonstrates that a single sea urchin removal from large, semi-discrete areas of urchin barrens can provide an efficient and effective approach to promoting rapid kelp recovery.This removal design was effective at reducing sea urchin reinvasion for the 2-year duration of the study, although densities of cryptic urchins had increased at some sites.At all sites, sea urchin densities remained below the estimated threshold that could maintain barrens (<1 exposed kina/m 2 ; Shears & Babcock 2003).Following urchin removal, multiple forest-forming macroalgae species (kelp and fucoids) rapidly regenerated whereas control areas remained largely barren.However, no consistent increases in sea urchin predator abundances were observed in removal areas after 2 years.These findings have important implications for understanding the effects of urchin grazing and kelp recovery and for informing future kelp restoration projects.

Effectiveness of a Single Sea Urchin Removal
At all sites, a single removal reduced exposed sea urchin densities to roughly one-third of the threshold density necessary for maintaining barrens with exposed densities remaining low for at least 2 years.By clearing large discrete areas and utilizing natural boundaries, sea urchin reinvasion from surrounding barrens was also minimal over 2 years.These results align with other studies where urchin reinvasion was limited by either removing sea urchins from small discrete patches of barrens (<100 m 2 ; Andrew & Underwood 1993;Ling 2008) or very large areas within expansive barrens (>18 ha; House et al. 2018;Williams et al. 2021).However, increasing cryptic kina densities highlights that over time sea urchin numbers and grazing pressure will increase, thus repeated removals, or other measures, would be required.While cryptic sea urchins preferentially feed on detached drift kelp rather than actively grazing on live kelp (Harrold & Reed 1985;Kriegisch et al. 2019), as kina densities increase we would expect to eventually see more active grazing.
While sea urchin densities stayed low at all removal areas, removal effectiveness varied spatially within each site, and all sites had patches of high urchin densities post-removal (>5 kina/m 2 ).Sea urchin ingress across removal boundaries also varied among and within sites; some boundaries showed a clear demarcation between recovered kelp and the barren control while others showed sea urchin diffused across the boundary.However, even within fully discrete areas, sea urchin densities are expected to increase over time through larval settlement.Therefore, without additional measures to control sea urchins such as ongoing removal/harvest or protection of predators, high sea urchin densities and urchin barrens are expected to eventually reestablish.Longer-term monitoring of this experiment is needed to elucidate how long these "restored" kelp forests may persist before returning to barren areas.

Macroalgal Recovery
Total macroalgal canopy cover increased greatly in all urchin removal areas over the 2-year period with increases in multiple species of kelp and fucoids.Timing of kelp and fucoid recruitment, however, varied among sites over the first year.This was likely due to variation in the timing of urchin removals among sites (which occurred over a 6-month period from spring 2020 to late summer 2021) relative to the seasonal reproductive patterns of key species.Other studies have suggested avoiding removing sea urchins in late spring as competition from seasonal algae could restrict kelp recovery (Kriegisch 2017).Nevertheless, after 2 years, canopy cover had increased to similar levels across all sites (approximately 43% on average), suggesting the seasonality of urchin removal is less important over longer time periods in this system.
Ecklonia radiata exhibited the greatest response to sea urchin removal and comprised greater than 80% canopy cover in removal areas after 2 years.E. radiata density increased rapidly in the first year but declined slightly in the second year; however, canopy cover continued to increase through time, largely due to increasing sizes of individual plants.In contrast, E. radiata densities in control urchin barren areas declined slightly over the 2-year period.There was high seasonal recruitment of E. radiata into some control areas (up to 200 plants/m 2 at Hauturu-o-Toi), but these recruits were short-lived due to Large-scale urchin removal for restoration severe grazing by sea urchins.Fucoids generally exhibited a smaller increase in canopy cover in response to urchin removal.The annual species Sargassum sinclairii exhibited a significant increase in density in removal areas, but this response varied seasonally with increases in spring followed by senescence of adult plants during late summer.
Kelp recovery in removal areas was more widespread than anticipated across all depths.As most kelp species are considered to have limited spore dispersal ranges (approximately 8 m for 75% of E. radiata recruits, and 3-4 m for fuculean algae; Schiel 1981Schiel , 1988)), we expected that initial kelp regrowth would be greatest adjacent to existing macroalgae populations (especially deeper kelp forests), with slower regrowth further away.While the greatest increase in E. radiata density was seen in 6-9 m (often the deepest depth range), all depth ranges showed at least fourfold density increases.The greatest increase in fucoid density was seen in the shallowest two depth ranges (0-6 m) in closest proximity to existing populations, but recruitment was also seen across all depth ranges.These results are consistent with other studies that have shown kelp recruitment far from existing plants and may indicate longer-distance spore dispersal (Reed et al. 1988;Gaylord et al. 2006) or existing spore banks (Schoenrock et al. 2021;Veenhof et al. 2022).Adult drift kelp were observed in removal areas after large storms, providing another possible spore dispersal mechanism.
This study confirms that sea urchin herbivory is a primary determinant of macroalgal forests on shallow reefs (approximately <10 m depth) across a range of environmental conditions in the Hauraki Gulf.The removal of sea urchins alone can allow for kelp regeneration without requiring additional kelp enhancement (e.g.environmental mitigation or seeding).The natural recovery was likely supported by the proximity of all sites to nearby deeper kelp forests which could provide a supply of spores.Kelp recruitment was observed within control areas, but these were consumed by sea urchins (K.Miller 2022, University of Auckland, personal observation), demonstrating that overgrazing was responsible for the lack of kelp in barrens rather than kelp recruitment failure.Furthermore, similar levels of kelp recovery were observed across the four sites despite varying environmental stressors and disturbance events, such as higher turbidity at Otata where kelp display reduced productivity compared to the other sites (Blain et al. 2021), or a large marine heatwave across the study region in 2022 (Bell et al. 2022).Notably, E. radiata and other macroalgal species are not at the end of their thermal tolerances in this region and not likely to be limited by temperature (Chapman 1966;Wernberg et al. 2019a;Cornwall   In this study, we saw no consistent evidence of increases in the densities of two primary predators of sea urchins (snapper and spiny lobsters) in urchin removal areas, particularly of large, legal-sized individuals which were consistently low.Only two previous urchin removal studies have examined the response of fishes, with one showing positive and one showing mixed results (Mooney 2001;Williams et al. 2021).While kelp-associated species would be expected to increase with kelp restoration, any benefits will be limited if these species are targeted by fishing in removal areas.Recreational fishers (including anglers, SCUBA divers, and spearfishers), who target both snapper and lobster, were regularly observed at all the removal sites and commercial lobster pots were commonly set in the removal area at Hauturu-o-Toi (K.Miller, 2022, University of Auckland, personal observation).Consequently, the observed density and size of snapper and spiny lobster were typical of heavily fished reefs and far lower than in protected areas (Allard 2020;LaScala-Gruenewald et al. 2021).This confirms that kelp recovery alone is unlikely to restore key sea urchin predators when fishing pressure is high; without predators to moderate and maintain low sea urchin densities, the observed benefits of urchin removal on kelp will be temporary.

The Role of Sea Urchin Removal in Kelp Restoration
Our results and emerging large-scale urchin removal studies underway in California, British Columbia, Japan, and the Mediterranean (Taino 2010;Guarnieri et al. 2020;Lee et al. 2021;Williams et al. 2021) demonstrate that in regions where sea urchins are a major driver of kelp loss, active removal of sea urchins over large-scales can promote kelp and macroalgae recovery with limited need for repeated removals or reseeding.While active removal can therefore provide a useful tool in kelp forest restoration, it does not address the underlying cause of sea urchin overpopulation and thus on its own provides only a temporary recovery.Where historic loss of predators are the main driver of elevated urchin densities and kelp loss such as in northern New Zealand, long-term marine protection can promote kelp recovery but this can take decades (Babcock et al. 1999;Shears & Babcock 2003).Until overfishing of predators is addressed at large scales, active sea urchin removal can only provide limited benefits in both extent and duration, unless additional measures are implemented to control urchin populations and restore stable and resilient kelp forest ecosystems (Peleg et al. 2023).For example, combining urchin removal and marine protection may provide an opportunity to accelerate kelp forest recovery and increase long-term resilience through the protection of predators (Eger et al. 2020;Miller et al. 2022).
While culling on SCUBA provides, an efficient method of removing sea urchins (Miller & Shears 2023), the authors recommend careful consideration of ecological, socio-economic, and cultural implications of removal methodologies.In New Zealand, kina is a taonga (treasured species) to M aori people who have relied on it as a food source for centuries.While sea urchin size and/or roe quality is often reduced in urchin barrens (Claisse et al. 2013), it is favorable to harvest and utilize urchins when possible (Cresswell et al. 2023).Large-scale sea urchin removal in Gwaii Haanas, British Columbia led by Indigenous leaders and the government of Canada have demonstrated restoration approaches that combine harvesting when viable (commercial or community) and culling of poor quality urchins to support restoration aims (Lee et al. 2021).The costs and benefits (e.g.practicality, effectiveness, and cultural-appropriateness) of different removal strategies should be assessed and potential uses of harvested sea urchins considered.
If sea urchin removal is deemed an appropriate tool for kelp restoration, we demonstrate how selecting suitable removal strategies that reduce sea urchin reinvasion will optimize efforts and improve restoration efficacy.Sea urchin removal from large (>1 ha) and discrete areas of barrens can reduce sea urchin reinvasion and maintain sufficiently low sea urchin densities to allow kelp recovery, leading to significant cost-savings for practitioners.While additional kelp enhancement may be necessary in some locations, natural kelp recovery can occur over large scales.This study demonstrates that active sea urchin removal can reset reef ecosystems and jumpstart kelp recovery when sea urchins are the primary cause of kelp loss, but additional management measures to restore predators and control urchin populations long term are required.

Figure 1 .
Figure 1.Map of Hauraki Gulf in northeastern New Zealand, with four experimental sea urchin removal areas from inner to outer Gulf: (A) Otata Island/Noises (inner Gulf); (B) V Bay (Leigh coast); (C) Nordic Bay (Leigh coast); and (D) Hauturu-o-Toi/Little Barrier (offshore island).In insets, yellow lines show removal area boundaries; turquoise lines show approximate location of semi-permanent transects in removal and control areas.Top right photo shows extensive urchin barrens at Hauturu-o-Toi/Little Barrier (photo credit Paul Caiger).

Figure 2 .
Figure 2. Sea urchin (kina, Evechinus chloroticus) density (mean kina/m 2 AE SE) at sea urchin removal (green lines) and control areas (peach lines) in the Hauraki Gulf, northeastern New Zealand.Densities were surveyed before and 6 (8 months for Hauturu-o-Toi), 12, and 24 months after sea urchin removal; removal areas had an additional sampling following removal.Gray vertical line denotes date removals were completed.Sites are arranged in chronological order of removal.Densities shown separately for exposed (on reef) and cryptic (hidden in crevices) kina.

Figure 3 .
Figure 3. Mean canopy cover (% AE SE) of forest-forming macroalgal species (Ecklonia radiata, Sargassum sinclairii, and Carpophyllum spp.) in sea urchin removal (green line) and control urchin barren (peach line) areas at four sites in the Hauraki Gulf, northeastern New Zealand.Times represent pre-urchin removal, and 6 (8 months for Hauturu-o-Toi), 12, and 24 months post-urchin removal.Gray vertical line denotes date removals were completed.Sites are arranged in chronological order of removal.

Figure 4 .
Figure 4. Mean density (plants/m 2 AE SE) of Ecklonia radiata, Sargassum sinclairii, and Carpophyllum spp., in sea urchin removal (green line) and control urchin barren (peach line) areas at four sites in the Hauraki Gulf, northeastern New Zealand.Times represent pre-urchin removal, and 6 (8 months for Hauturu-o-Toi), 12, and 24 months post-urchin removal.Gray vertical line denotes date removals were completed.Sites are arranged in chronological order of removal.

Figure 6 .
Figure 6.Mean density (individuals/transect AE SE) of total Australasian snapper (Chrysophrys auratus), legal-sized snapper (≥300 mm FL), and total spiny lobster (Jasus edwardsii and Sagmariasus verreauxi) in sea urchin removal (green line) and control urchin barren (peach line) areas at four sites in the Hauraki Gulf, northeastern New Zealand.Times represent pre-urchin removal, and 6 (8 months for Hauturu-o-Toi), 12, and 24 months post-urchin removal.Gray vertical line denotes date removals were completed.Sites are arranged in chronological order of removal.