The Southern California Bight (SCB) spans ~1200 km of coastline from Ensenada, Mexico to Point Conception, California and consists of a multitude of subtidal habitats, including rocky reef, kelp forest, and seagrass habitats (Williams et al., 2022). This project was focused along the mainland coast of Santa Monica Bay (SMB) and offshore Catalina Island ( Figure 1 ). Three transplant sites were within coastal SMB, while the two donor bed sites (two largest extant Z. pacifica beds in the region) were located ~ 40km away on the south-eastern side of Catalina Island. Each site was ~0.5km offshore of the local coastline (mainland or Catalina). All transplant sites were established at 10 – 12 m depth, typical of natal Z. pacifica beds within the SCB (Bernstein et al., 2011; Obaza et al., 2022). Z. pacifica transplanting at the three mainland sites occurred in July 2021, during the primary growing season (March through October) established by the 2014 National Oceanic and Atmospheric Administration California Eelgrass Mitigation Policy (CEMP). Donor material was collected at Catalina and transplanting occurred at Redondo Canyon on 20 July 2021, at Malaga Cove on 22 July 2021, and at Dockweiler on 27 July 2021. Dockweiler, Redondo Canyon, and Malaga Cove received 520, 486, and 526 shoots, respectively. A total of 1532 shoots were collected and transplanted to the three sites, constituting the largest Z. pacifica transplant reported to date.

The donor shoots were collected from two robust extant beds on the windward side of Catalina Island, Palisades and East End, with areas and densities of 27.74 hectares and 112 shoots m^-2, and 10.48 hectares and 114 shoots m^-2, respectively (Obaza et al., 2022). The donor bed sites ranged from 11 – 22 m depth. While numerous extant Z. pacifica beds exist closer to the mainland within SMB, their ephemeral nature and small patchy distribution excluded them as adequate donor beds for transplant experiments. We instead selected sites at Catalina Island characterized by extensive and healthy Z. pacifica populations with decadal stability (Engle and Miller, 2005).

Material was collected from donor beds, for both the single shoot and bundle shoot transplant methodologies, in a systematic fashion, utilizing a thinning approach to avoid the creation of noticeable (> 0.5 m^-2) bare patches, and minimizing fragmentation. For the single shoot method, divers selected individual shoots and gently maneuvered the shoot and rhizome out of the sediment such that at least three internodal segments (~100 mm) of rhizome were attached to the apical shoot (Altstatt, 2003). Rhizomes with a single shoot were selectively used for this method, but per Paling et al. (2007) sediment was not actively removed from the rhizomes. Material for the bundle shoot method was collected similarly, although rhizomes with several shoots attached were also included along with a small amount of sediment. Divers selected shoots from the shallow edge, middle, and deep edge of the site to minimize fragmentation impact on the donor bed, while likely increasing clonal genetic diversity following Olsen et al. (2014). To complete the bundle approach (Zhou et al., 2014), multiple shoots were connected with biodegradable twine on the research vessel during the transit to the transplant sites. Donor material was temporarily stored in large coolers with flow-through seawater and kept shaded for up to four hours between collecting and transplanting.

Collecting donor material and transplanting for single shoot and bundle shoot methodology occurred on the same day for each transplant site to minimize exposure stress. At each transplant location, SCUBA divers established a 30 m line at 10 to 12 m depth with individual experimental plots dispersed on either side. Each transplant site consisted of seven experimental plots, four bundle shoot plots (each 12 m^-2) and three single shoot plots (each 9 m^-2), ensuring inter-site replication along a single depth strata. Material was planted in a grid pattern within each plot and meter sticks were used to maintain transplant spacing. For the bundle shoot method, between 10 – 13 bundles (consisting of 80 – 110 total shoots) were planted at 1 m intervals within the four replicate plots at each site (see Figure 1A ). For each plot, divers excavated a small hole in the sediment, and the entire rhizome mass of the bundle was placed in the excavated hole along with a wooden tongue depressor, which was secured to the rhizome mass and positioned parallel to the substrate to act as an anchor. Single shoot planting was conducted using methods adapted from Altstatt (2003). Between 40 – 46 shoots were planted at 0.5 m intervals within the three replicate single shoot plots at each site (see Figure 1B ). Divers used a trowel to excavate sediment and gently maneuvered a single rhizome into the hole, placing a small gardening stake over the rhizome to act as an anchor. Sediment was pushed back over the rhizome so that it was completely buried, leaving the shoot naturally situated above the sediment.

All biological monitoring was conducted via SCUBA. Fish and eelgrass structural surveys were conducted at donor beds prior to (June 2021), and after (August 2021, July 2022, March 2023), collection of transplant material. Fish surveys were conducted at each site, where divers swam < 1 m above the substrate and identified the species, counted the abundance, and determined the size of each fish within a 1 m high x 2 m wide survey window along the timed roving diver fish surveys (per Pondella et al., 2006; Looby and Ginsburg, 2021; Obaza et al., 2022). Collection of morphometrics (length, width, and density) of Z. pacifica in donor beds followed methods in McCune et al. (2020) and Obaza et al. (2022).

Divers performed pre-transplant site surveys at each transplant location, and subsequently conducted monitoring to assess the biological characteristics of the transplants at intervals of 1-day, 1-week, 2-weeks, 1-month and quarterly thereafter through June 2023. Survivorship was defined as the number of remaining planting units (i.e., single shoots or bundle shoots) within plots (Wegoro et al., 2022). Inter-plot survivorship was assessed by enumerating single shoots and bundle shoots in each experimental plot at each transplant site during each timepoint (Altstatt et al., 2014). Divers temporarily placed flagging stakes next to each shoot to aid in accurate counts and to increase efficacy during monitoring events.

The establishment of habitat and species-specific threshold values is a critical tool in ecosystem management regimes, as exceedances of threshold values can catalyze abrupt responses, often leading to habitat decline or loss (Uhrin and Turner, 2018). Threshold values of biophysical parameters impacting Z. marina have been expounded in numerous studies across its geographical distribution, both ex situ and in situ (Short et al., 2002; Lee et al., 2007; Moore and Jarvis, 2008; Thom et al., 2008). No critical threshold tolerances have been investigated for Z. pacifica; thus, biophysical thresholds established in this study have been adapted either from data collected near Z. pacifica habitat or from applicable Z. marina literature.

We measured temperature, light, dissolved oxygen, and wave exposure at each site, as these biophysical parameters can result in adverse impact to seagrass health, function, and survival (Fonseca and Bell, 1998; Holmer and Bondgaard, 2001; Johnson et al., 2003; Duarte et al., 2007). Temperature was measured from July 2021 – June 2023. Wave dynamics were measured across four deployments October 2020, July – August 2021, September – December 2021, and February – June 2022. Light and dissolved oxygen sensors were added from May 2022 – June 2023. The number of days with acceptable data in the time series varied among the sites and environmental metrics ( Table 1 ).

Differences in the composition of fish assemblages were compared by creating a Bray-Curtis dissimilarity matrix from average relative species-specific encounter rates across individual surveys completed at transplant and donor beds from 2021 – 2023. All species with < 2% mean encounter rates across all surveys were removed. The dissimilarity matrix was used to create non-metric multi-dimensional scaling plots to visualize changes in fish assemblages across transplant and donor beds. Assemblage differences were tested using permutation multivariate analysis of variance (PERMANOVA) (McCune and Grace, 2002). Differences in group dispersion were tested using the ‘betadisper’ function in the ‘vegan’ package and were not significant; therefore, no additional transformations on these data were conducted. To determine species-specific differences in fish assemblage, a similarity percentages (SIMPER) procedure was performed with treatment as a factor. All multivariate statistical approaches were conducted using the ‘vegan’ package in R (Oksanen et al., 2019).

Environmental data were analyzed and graphed in R (RCore Team, 2021, v. 4.1.2). Data corresponding to sensor malfunction and extreme biofouling were removed during quality control and were excluded from analysis (following Ricart et al., 2021). Standardization of the timeseries date range was applied to environmental data for statistical analysis and was based on the latest data start date and the earliest data end date across all sites for each biophysical parameter. Exposure analysis was conducted utilizing the ‘heatwaveR’ package in R (Schlegel and Smit, 2018). Exposure calculations were performed based on daily exceedance (or negative exceedance, i.e., below) of established parameter thresholds. Empirical cumulative distribution function (eCDF) plots were employed to differentiate patterns of the environmental metrics among sites (Beheshti et al., 2022). Environmental data lacked variance homogeneity and were therefore analyzed using a non-parametric Kruskal-Wallis test and Dunn’s post hoc test with a Holm-Bonferroni correction to critical p-value due to application of multiple tests of significance among sites (Pulido and Borum, 2010; Campbell et al., 2018). A p-value < 0.05 represents a significant difference detected between sites, indicating a rejection of the ‘similarity of sites’ null hypothesis.

Transplant sites experienced minor loss immediately post-transplanting, followed by a slight expansion in total shoots observed at two of the three sites around 20 days post-transplant, until complete mortality was observed across all sites after day 315 ( Figure 2 ). Survivorship varied across sites, with poor survivorship at Dockweiler (DW), better survivorship at Redondo Canyon (RC), and longest survivorship at Malaga Cove (MC). At DW and RC, 1.9 and 2.1% of material was lost between transplanting and the 24-hour survey respectively, while MC did not lose any transplanted material until day 15. Highlighting site-specific divergence, survivorship at DW was only 35.8% by day 37, while RC and MC experienced 85.4% and 80.4% survivorship respectively. By day 90, DW, RC, and MC had 23.3%, 67.9%, and 68.1% survivorship, but by day 197 complete mortality occurred at DW and RC, and survivorship at MC had declined to 30%.

Further survival disparity was pronounced among transplant methods, with the single shoot method outperforming bundle shoot method across all sites ( Figure 2 ). Methodological survival differences were most prominent at DW, whereby day 37 bundle shoot survivorship was reduced to 23.1%, while single shoot survivorship was substantially higher at 75.4%. Differences in survival among transplant methods were also evident at other sites, whereby day 90 RC single shoot survivorship was 89.3% while bundle shoot survivorship was 60.7%; and at MC single shoot survivorship was 93.1% while bundle shoot survivorship was 59.8%. At MC (the best performing site) 5.4% of single shoots remained on site until after day 315, while bundle shoots suffered prior complete mortality.

A total of 3,692 fishes from 26 species were recorded over 139 transects across all five sites. The donor bed sites had greater species richness than transplant sites throughout the study period and greater species encounter rates for all fishes as well ( Figure 3 ). No fish encounters were observed during pre-transplant site selection surveys, and there were no significant increases in encounter rate before and after transplant activities at transplant sites. No meaningful relationships of species richness or encounter rate were detected between the three transplant sites. PERMANOVA results indicated that the composition of fish assemblages observed at donor beds and transplant sites were significantly different from one another (F_{1,24 =} 10.14, p = 0.001), with no overlap in the NMDS plot ( Figure 3 ). According to results from the SIMPER analysis, differences were driven by greater relative abundance of sanddab (Citharichthys stigmaeus) and California halibut (Paralichthys californicus) at transplant sites and greater relative abundance of kelp bass (Paralabrax clathratus), rock wrasse (Halichoeres semicinctus), señorita (Oxyjulis californica), and salema (Xenistius californiensis) at donor bed sites.

Temperature exposure above the 20°C threshold was not a crucial indicator of transplant performance as values fell well within, and even below, the exposure conditions at the donor bed sites ( Figure 4 , Tables 1 , 2 ). Transplant site high temperature exposure percentages were DW (2.06%), RC (2.34%), and MC (3.30%) and were less exposed than the donor bed sites Palisades (PAL) (11.54%) and East End (EE) (17.07%). The longest continual exposure period above the threshold for the sites were 10, 11, 11, 11, and 26 days for DW, RC, MC, PAL, and EE respectively. The highest temperature recorded at the transplant sites was 22.39°C and 23.09°C at the donor bed sites. From August 2021 – January 2023 for transplant sites (n = 514 days), DW was significantly different from MC (p < 0.05) and MC was significantly different from RC (p < 0.05), though DW was not significantly different from RC (p = 0.76). For continuity of date range across all transplant and donor sites, from May 2022 – January 2023 (n = 221 days) significant differences were detected between comparisons of transplant sites and donor bed sites (p < 0.05), though temperatures at the two donor bed sites were not significantly different from one another (p = 0.28).

Exposure exceeding the wave thresholds was common from July – August 2021 ( Figure 5 , Tables 1 , 2 ). Transplant site exposure percentages above the H_rms threshold were DW (6.82%), RC (0%), and MC (2.27%), while exposure percentages above the U_rms threshold were DW (86.36%), RC (13.64%), and MC (36.36%). The longest continual exposure period above the U_rms threshold for the sites were 28, 3, and 8 days for DW, RC and MC respectively. Over the July – August 2021 deployment (n = 44 days), DW exhibited significantly different H_rms values from RC and from MC (p < 0.05), while RC and MC were not significantly different (p = 0.22). The U_rms data were significantly different from one another for all sites (p < 0.05).

Based on early results, DW was excluded from the subsequent two deployments, opting rather to monitor better performing transplant sites (RC and MC) and donor bed sites (PAL and EE) concurrently. RC, MC, PAL, and EE had continuous in situ wave characterizations from September – December 2021 and February – June 2022. These two deployments were aggregated for analysis (n = 185 days). Again, it is evident that transplant sites were exposed to higher degrees of H_rms and U_rms values and for longer durations than donor bed sites ( Figure 5 , Tables 1 , 2 ). Site exposure percentages above the H_rms threshold were RC (28.50%), MC (31.09%), PAL (2.56%), and EE (2.56%), while exposure percentages above the U_rms threshold were RC (58.55%), MC (83.42%), PAL (34.36%), and EE (55.90%). The highest daily mean H_rms and U_rms values recorded at the transplant sites were 1.46 m and 0.34 m s^-1, while the values at the donor beds were 0.71 m and 0.21 m s^-1. H_rms values at the transplant sites MC and RC were significantly different (p = 0.05), but donor bed sites PAL and EE were not significantly different from one another (p = 0.44). All H_rms comparisons between transplant sites and donor bed sites were significantly different (p < 0.05). The U_rms data shows that RC and EE were not significantly different (p = 0.09), though all other site comparisons were significantly different from one another (p < 0.05).

PAR exposure below the 3 mol m^-2 day^-1 growth limited, non-survival threshold, as well as exposure (or lack thereof) above the upper survival-supportive threshold of 5 mol m^-2 day^-1 was prevalent during the project period ( Figure 6 , Tables 1 , 2 ). Exposure below the growth limited, non-survival threshold is especially prevalent in the late fall and winter months. Transplant site exposure percentages below 3 mol m^-2 day^-1 were DW (96.13%), RC (91.85%), and MC (87.44%) and exhibited greater exposure times than the donor bed sites PAL (62.72%) and EE (47.43%). The longest continual exposure period below the non-survival supportive threshold for the sites were 141, 203, 100, 56, and 75 days for DW, RC, MC, PAL, and EE respectively. Donor bed sites PAL (21.01%) and EE (31.44%) experienced sustained exposure in exceedance of the 5 mol m^-2 day^-1, while MC only had 0.93% and both DW and RC had zero exposure above the survival-supportive threshold. The two donor bed sites each spent considerable time from spring through fall (June 2022 – October 2023) above the survival-supportive threshold (12-day continual exposure at PAL and 29-day continual exposure at EE) and both sites rarely dip below the growth limited, non-survival threshold. The highest daily integrated PAR value recorded at the transplant sites was 6.10 mol m^-2 day^-1 while the highest value at the donor beds was 11.56 mol m^-2 day^-1. Over the timeseries from July 2022 – January 2023 (n = 159 days), DW was significantly different from all other transplant and donor bed sites (p < 0.05), while RC and MC were not significantly different from each other (p = 0.88) but were significantly different from the donor bed sites (p < 0.05). Donor bed sites were not significantly different from each other (p = 0.18).

DO exposure below the 3 mg O₂ L^-1 tissue degradation threshold occurred ( Figure 6 , Tables 1 , 2 ). Transplant site exposure percentages below 3 mg O₂ L^-1 were DW (23.20%), RC (1.11%), and MC (15.35%) and exhibited consequential exposure times compared to the donor bed sites PAL (0%) and EE (0%). The longest continual exposure period below the tissue degradation threshold for the sites were 24, 3, and 14 days for DW, RC and MC respectively, while PAL and EE experienced zero exposure periods. The lowest daily mean DO value recorded at the transplant sites was 0.01 mg O₂ L^-1 while the lowest value at the donor beds was 6.24 mg O₂ L^-1. A DO timeseries from September 2022 – January 2023 (n = 102 days), indicated transplant sites were not significantly different from one another (DW – RC p = 0.48; DW – MC p = 0.33; RC – MC p = 0.11). Likewise, the oxygen exposure at donor beds were not significantly different (p = 0.05). All comparisons between transplant sites and donor bed sites were significantly different (p < 0.05).

This study conducted the first ever mainland Zostera pacifica transplant, with methods and results that can allow for further technique development of Z. pacifica restoration in the Southern California Bight (SCB). From the in situ biophysical monitoring, this study has developed clear linkages pertaining to Z. pacifica environmental constrains, enabling advancements in site selection procedures that may serve as a model for open coast seagrass restoration. At all three sites the majority of transplants initially established, and at two sites, experienced evidence of growth and expansion. However, beyond 315 days there was complete mortality at the coastal transplant sites, most likely due to predominantly unfavorable environmental conditions associated with high wave exposure, low light, and periods of hypoxia. Thus, in the most conspicuous definition of success, or even that from the NOAA California Eelgrass Mitigation Policy (CEMP) (2014) criteria requiring 85% areal coverage two years post transplanting, the transplants were not successful. Transplanting and restoring seagrass beds is notably difficult, and quantification of seagrass transplant performance is far more nuanced and complex than simply considering total survival as the sole metric for success. Contextualizing, in a meta-analysis of 82 seagrass restoration projects on the West Coast of the U.S., 32 – 60% of projects were unsuccessful (Ward and Beheshti, 2023), yet ‘unsuccessful’ projects certainly can still contribute to advancing methods in restoration science (Zhang et al., 2021). As such, other metrics beyond survivorship (i.e., ecosystem services) should also be considered in the evaluation of transplant success (Ward and Beheshti, 2023).

The assessment of ecosystem services (fish community dynamics, carbon stocks, turbidity, etc.) in restored sites compared to donor beds, can serve as valuable performance metrics and motivations for larger scale restoration projects (Orth et al., 2020; Lange et al., 2022). Zostera spp. (eelgrass) beds have high fish biomass and diversity compared to unvegetated sediments (Duffy, 2006), and fish biomass can be a useful comparative metric between transplant and donor beds (Beheshti et al., 2022). Utilizing fish surveys adapted from Obaza et al. (2022), dissimilarity in fish communities is evident between transplant sites and donor bed sites ( Figure 3 ). The small amount of low-density eelgrass followed by bare soft-bottom in the transplant sites as compared with extensive Z. pacifica donor beds on Catalina Island is likely the driving factor in fish assemblage differences (Pihl et al., 2006). Older aged, structured seagrass habitat offers more robust ecosystem function (i.e., fish diversity) compared to younger aged seagrass beds (i.e., transplant sites) (McGlathery et al., 2012; Orth et al., 2020; Hardison et al., 2023). The species more commonly encountered at transplant sites were flatfish, as would be expected on soft-bottom habitats in the region (Craig et al., 2004; O’Leary et al., 2021), while those more regularly recorded at donor bed sites are usual inhabitants of open coast Z. pacifica (Obaza et al., 2022). Fish surveys associated with this project quantitatively support the conclusion that the eelgrass transplants did not create meaningful structure for nearshore fishes, nor restore that ecosystem function, during the study period.

The only other known Z. pacifica transplant occurred in 2002 at a sheltered cove off Anacapa Island where 450 shoots were transplanted to an area of 300 m^-2 and resulted in a successful and stable Z. pacifica population on the island (Altstatt et al., 2014). Over 95% mortality was recorded on day 155 in that study, suggesting early mortality may be the norm, even in successful efforts. Altstatt et al. (2014) provides context to our study’s ~30% survivorship on day 197 at Malaga Cove ( Figure 2 ) and illustrates the benefit of this effort – affirming the present study represents an advancement on prior restoration efforts for Z. pacifica. Studies of open coast seagrass restoration conducted elsewhere globally report similar findings to those in this study, as open coast environments are inherently dynamic, leading to higher degrees of initial transplant mortality (Paulo et al., 2019; Wegoro et al., 2022).

This project highlights insight into the efficacy of exposed mainland coast transplanting methodology, with initial survival differences of the single shoot method over the bundle shoot method ( Figure 2 ). In the exposed mainland coast environment, transplants can be subject to periods of intense wave action and high near bottom velocity ( Figure 5 ; van Katwijk et al., 2009), where the smaller profile single shoots may induce less drag and thus, experience reduced impact compared to bundle shoots (Orth et al., 1999). These insights may contribute to future transplanting efforts, as the bundle shoot method requires a greater degree of harvest investment compared to the single shoot method. Focusing transplanting efforts on the single shoot method allows for a redistribution of the harvest material, spreading the transplanting risk across a greater area and number of ‘planting units’ (Orth et al., 2006; van Katwijk et al., 2009). The results of this study, in concert with findings from Altstatt et al. (2014), reaffirm that methodological approach is critical (Fonseca et al., 1998), as the single shoot method is linked to higher survivorship along exposed coastlines in the SCB. However, during the harvesting and transplanting process, the donor material experiences stress (Lange et al., 2022), catalyzing initial mortality or decreased function. Therefore, future studies on the open coast should aim to gauge the effects of transplanting stress and methodology on survival by also transplanting back into donor beds. These efforts may further disentangle the efficacy of transplant methodologies in the absence of environmental conditions that vary from those experienced at the natal donor beds.

Stochastic events are prevalent in seagrass beds and can result in adverse transplant performance (Lange et al., 2022). Unpredictable deleterious singular events may have influenced transplant performance, as numerous acute events occurred across transplant sites. The Dockweiler transplant site is located ~2km from the largest wastewater treatment facility in Los Angeles, and from July 2021 – May 2022, 60 effluent limit violations occurred because of an unprecedented > 12-million-gallon raw sewage discharge from a pipe 1.6 km offshore, resulting in numerous algal blooms (LA Water Board, 2023). A large body of literature speaks to the negative impacts of eutrophication on seagrasses (Burkholder et al., 2007; Hughes et al., 2013). Further, during a post-transplanting survey at Redondo Canyon, located at the head of an active submarine canyon, divers observed macroalgae detritus engulfing transplant units. Macroalgae inundation is known to decrease health and survival of Zostera spp. by altering light regimes and decreasing photosynthetic capabilities (Huntington and Boyer, 2008), a factor that likely contributed to declines in transplanted Z. pacifica survivorship. The pairing of high wave characterizations (H_rms and U_rms values) and macroalgae inundation at the Redondo Canyon site may have contributed to physical damage – a concept further supported by diver observations of dislodged transplanting units. Hydrodynamical forces and nutrient regimes can negatively impact the biomechanical properties (i.e., shoot strength, elasticity) of seagrasses (Patterson et al., 2001; La Nafie et al., 2013; Risandi et al., 2023), and it is likely that the hydrodynamic conditions and acute eutrophication events impacted the transplant sites, though the quantification of biomechanical properties is outside the scope of this study.

The biophysical characterizations reveal a more compelling and consistent case for deleterious site conditions, as a mosaic of unsuitable environmental conditions occurred at transplant sites. And while biologic controls are undoubtably important to seagrass systems, regionally along the West Coast of the USA, physical drivers are principally responsible for eelgrass transplant success or decline (Ward and Beheshti, 2023). Yet surprisingly, many Zostera spp. restoration projects that offer suspected reasons for transplant success or failure lack quantitative in situ environmental data, instead relying on observational data to infer physical drivers (Ward and Beheshti, 2023). In this study, generally, PAR, DO, and temperature were lower, and wave dynamics (H_rms and U_rms ) higher in transplanted sites compared to donor bed sites ( Figures 4 – 6 ). Indeed, empirical cumulative distribution function (eCDF) curves of transplant sites compared to donor beds depict a pronounced decoupling between transplant and donor bed sites and show only partial overlap, confirming that conditions at transplanting locations were inapt for Z. pacifica long-term survival ( Figure 7 ; e.g., Beheshti et al., 2022). The transplant study approach herein showcases an avenue to discretely monitor in situ data and quantify known physical controls on seagrass habitat function.

Minimum colonization depth is often restrained by wave orbital velocity and large wave exposure directly correlates with transplant mortality (Bos and van Katwijk, 2007; de Boer, 2007). Wave characterizations were of paramount importance for transplanting on the exposed mainland coast of the SCB. Inter-site comparisons amongst solely transplant sites from initial nascent stages of the transplant revealed disparities between sites, namely at Dockweiler where exceedances of threshold values corresponded with large drops in survivorship ( Figure 5 ). Coupling survivorship and wave characterizations allow for insights to be gleaned, as periods of similar H_rms and U_rms values at transplant and donor bed sites reveal stabilization in survivorship (September – October 2021), while deviations above threshold values (November 2021 – June 2022) correspond with transplant mortality ( Figures 2 , 5 ). The paucity of wave characterization for exposed mainland coast seagrass beds makes quantitative comparisons challenging. Hansen and Reidenbach (2012) investigated significant wave heights in extant Z. marina beds, finding maximum values around 0.18 m – far smaller wave heights than those values reported from extant beds in this study. Further, the wave characterizations of our study are conceptually supported by Hansen and Reidenbach (2013), who investigated near bottom wave velocity inside and outside Z. marina beds in lower Chesapeake Bay, finding flows to be significantly higher in the unvegetated site (0.09 m s^-1) than within the seagrass site (0.026 m s^-1). Additional studies have found, while uncommon in extant beds, maximum velocities of 0.5 m s^-1 were incurred for brief periods by Z. marina (Fonseca et al., 1983; Koch, 2001). Though perhaps the most supportive of our results comes from a study conducted on survival of two western Mediterranean seagrasses at water depths of 12 and 18 m, concluding that near-bottom velocities tolerated were between 0.18 – 0.39 m s^-1 (Infantes et al., 2011). These depths and velocities more aptly compare to those experienced in an exposed mainland coastal shelf environment occupied by Z. pacifica, evidencing that this species inhabits areas with sustained exposure above values typically tolerated by Z. marina, and other seagrasses, in protected environments.

Light availability constrains seagrass growth at the deep edge of the bed, but sufficient quantities are required to maintain health throughout the entirety of the local distribution (Drew, 1979; Dennison, 1987; Duarte et al., 2007). Based on the adapted non-survival light threshold from Thom et al. (2008), transplant sites did not receive sufficient light to satisfy growth requirements. The values (< 3 mol m^-2 day^-1) experienced at the transplant sites, and the ensuing mortality, are supported by results from a field study that found drastic mortality in Z. marina transplants exposed to PAR below 3 mol m^-2 day^-1 (Moore et al., 1997), as well as by a meta-analysis reporting similar deleterious impacts of values below the established PAR threshold (Dunic and Côté, 2023). Our study also shows that the donor bed sites average below those thresholds from November 2022 – April 2023 and did not induce mortality. During light limiting conditions, Zostera spp. can facilitate survival through the metabolization of carbohydrates (sugar and starch content) stored in the rhizomes and leaves (Zimmerman et al., 1995; Burke et al., 1996; Ralph et al., 2007; Vichkovitten et al., 2007; Bertelli and Unsworth, 2018). Research from Baja, Mexico supports this notion that carbohydrate reserves allow Zostera spp. to survive in short-term adverse light conditions but noted that mortality occurred when carbohydrate reserves decreased ~85% following three weeks of nearly zero PAR values (Cabello-Pasini et al., 2002). It follows then that lower light limits are not binary, rather eelgrass light requirements are nuanced, requiring sufficient light during growing seasons to build carbohydrate reserves needed to draw down during light limiting conditions encountered when irradiance levels are naturally lower (i.e., winter months) (Govers et al., 2015; Wong et al., 2020; Wong et al., 2021). The donor bed sites in our study did not experience near zero PAR conditions for an extended duration and did receive sufficient growing season light levels, and thus, Z. pacifica at these sites were likely able to draw on ample reserves to survive during brief light limiting conditions. Further, while acclimatization of transplanted Z. marina is documented (Krumhansl et al., 2021), the ability to build carbohydrate reserves is diminished under 3 mol m^-2 day^-1 light regimes (Eriander, 2017). The transplant sites in our study may have adapted to a new light regime initially, but when PAR values were reduced to ~ 0 mol m^-2 day^-1 for extended periods of time, an insufficient carbohydrate storage likely prevented survival in these adverse light conditions.

Low DO values impact Z. marina growth and can accelerate mortality (Holmer and Bondgaard, 2001). Hypoxic values (< 2 mg O₂ L^-1) and values nearing anoxia (< 0.01 mg O₂ L^-1) are confirmed in other systems containing submerged aquatic vegetation (Tassone et al., 2022). DO values in donor bed sites did not drop below 6.24 mg O₂ L^-1, but transplant sites, primarily Dockweiler and Malaga Cove, underwent multiple periods of extended hypoxia, and at Dockweiler, even values approaching anoxia ( Figure 6 ). Low dissolved oxygen concentrations are reported as contributors of Zostera decline in other systems in the USA, such as the lower Chesapeake Bay (Moore and Jarvis, 2008), an Oregon estuary (Magel et al., 2023), and two separate California estuaries (Walter et al., 2018; Beheshti et al., 2022). A buildup in concentration of sulfides in the rhizome can occur during periods of hypoxia (Goodman et al., 1995), and may have significantly impacted transplant performance, though a quantification of sediment composition in our study is absent. The rapid impact of low DO can manifest in 12 hours and increases with duration, with absolute mortality occurring with anoxic exposure for 36 hours (Pulido and Borum, 2010; Raun and Borum, 2013). Continuous hypoxic conditions for 24 days at Dockweiler and 14 days at Malaga Cove is a salient contributor to transplant failure.

Heat stress induced mortality, decreased photosynthetic efficiency, or reduced growth rate, are common occurrences amongst Zostera spp. (Moore et al., 2014; Hammer et al., 2018; Serrano et al., 2021). Our study shows that temperature stress at the transplant sites was not a key variable influencing transplant site performance, as extant Z. pacifica donor beds experienced higher overall temperatures and for longer periods of time. The limited temperature threshold exposure in our study is supported by data taken in the proximity of a Z. pacifica bed off Santa Cruz Island, California (Kapsenberg and Hofmann, 2016), highlighting that extant open coast Zostera spp. proliferate within the range of temperatures reported in our study. Climate change models predict warming ocean temperatures will impact seagrasses (Carr et al., 2012), though the extant Z. pacifica beds in our study do not appear to be growing near their physiological temperature tolerances and may be less exposed to climate change related temperature stress compared to other temperate foundational species (i.e., Macrocystis pyrifera, see Cavanaugh et al., 2019 and Leichter et al., 2023). Yet, Zostera spp. are susceptible to both pulse extreme heat events and episodic heat exposure which can alter local resilience to additional stressors (DuBois et al., 2022), therefore it follows that the increasing magnitude, frequency, and duration of episodic events (i.e., ENSO) may coincide with baseline climatic warming to induce deleterious impacts on seagrasses in the SCB (Johnson et al., 2003; Echavarria-Heras et al., 2006).

An amalgamation of synergistic stressors, whether short-term stochastic or chronic, ultimately resulted in conditions unfavorable for Z. pacifica proliferation at transplant sites. While the survivorship of this project initially performed better than the Anacapa Island transplant conducted by Altstatt et al. (2014), fundamentally, the hydrodynamic data in our study from November 2021 – June 2022, coupled with water quality data from Spring 2022 – Spring 2023 portray an unambiguous unsuitability for long-term Z. pacifica survival at these sites. In opposition, the Anacapa transplant site characteristics proved hospitable for Z. pacifica expansion following a reduction in grazers responsible for the initial mortality (Altstatt et al., 2014). The present study represents the first quantification of physical parameters within Z. pacifica beds on the open coast, distinguishing from the less turbulent conditions of shallow, protected, estuarine Z. marina habitat.