A small step or a giant leap: Accounting for settlement delay and dispersal in restoration planning

Understanding larval duration and hence dispersal potential of the European oyster Ostrea edulis is crucial to inform restoration strategies. Laval duration has an obligatory period of maturity to pediveliger (when larvae are ready to settle), but also an unknown period until metamorphosis is triggered by a settlement cue. The extent to which larvae can prolong the pediveliger period and delay metamorphosis has not been studied. Here we show that O. edulis larvae can delay metamorphosis for a period of 11 days, while retaining the capability to settle in high proportions when presented with a suitable settlement cue. O. edulis larvae are likely to be able to delay metamorphosis even further, since 80% of larvae in the control treatment were still alive when the experiment was terminated at day 14. The results indicate the ability of O. edulis larvae to more than double pelagic duration and probably further delay metamorphosis. We discuss these findings in the context of larval mortality, and the importance of O. edulis’ larval settlement requirements for dispersal potential, recruitment success and connectivity of restoration sites.


Introduction
Ecosystem restoration is increasingly being recognised as a pressing need for counteracting the environmental emergency we are facing. The UN has officially proclaimed the decade of 2021-2030 as the Decade on Ecosystem Restoration following a proposal for action of over 70 countries. Restoring biodiversity and ecosystem services is also a priority in Europe under the European Green Deal and the Biodiversity Strategy to 2030, where the development of legally binding EU nature restoration targets is a key commitment [1]. As ecosystem engineers, oyster reefs and beds play a critical ecological role and provide important ecosystem services [2,3], yet they are also one of the most imperilled marine habitats on Earth [4]. Restoring oyster habitats has therefore emerged as a priority in many countries around the world, particularly in Europe, the USA and Australia.
In Europe, efforts are mounting to restore the European oyster Ostrea edulis and thereby contribute to national and international commitments. O. edulis once formed widespread that O. edulis larvae can delay metamorphosis for a couple of days [31,46], but the extent to which larvae are capable of delaying metamorphosis has not been studied. Biophysical modelling, i.e. combining hydrodynamic models with particle tracking models, which mimic biological traits, has emerged as a powerful tool to simulate and predict larval dispersal. Pelagic larval duration is arguably the most important biological variable that can be included in the biophysical models, as it determines the duration that larvae will be subject to the transport of oceanic currents, a key factor controlling larval dispersal [27,29,47]. Biophysical models have been used to simulate larval dispersal and inform conservation measures in several bivalve species. For instance, in Pamlico Sound (USA) for the Eastern oyster Crassosstrea virginica [47], in Strangford Lough (Ireland) for O. edulis [48], and in the Irish Sea for the common cockle Cerastoderma edule [49] and the horse mussel Modiolus modiolus [50]. The ambition of current O. edulis restoration and conservation efforts is to be able to predict larval dispersal with biophysical models [e.g. 48]. There is, however, a knowledge gap on the extent to which O. edulis larvae can delay metamorphosis once they are competent to settle. Understanding the extent to which delayed metamorphosis can influence pelagic duration will also contribute to answering one of the top 40 questions most important to the policy and practice of native oyster reef restoration in Europe [51] i.e. "How can a map of the connectivity potential of restoration sites (accounting for current populations) be developed?" Since the success of oyster populations is often limited by poor understanding of site-specific dispersal patterns coloured points = laboratory data, with each colour denoting a different source; yellow points = varying food quality in the laboratory prolonged development time. Median values were plotted when ranges where given (see Table A1 in S1 Appendix for original values and data sources). Only data points where < 50% of larvae were reported to be at pediveliger stage were plotted. Black line shows fitted model: pelagic larval duration in days = 59.179-3.918 x + 0.0717 x 2 , where x is temperature in C (see Table A2 in S1 Appendix for more details). Data points in which food quality varied (yellow colour) were excluded from the model.

PLOS ONE
of larvae [48], increasing our understanding of pelagic larval duration and the factors affecting metamorphosis will be crucial to correctly parametrise larval dispersal models and ultimately inform conservation and restoration planning.
The aim of the present research was to study the extent to which O. edulis larvae can delay metamorphosis and thereby influence pelagic duration, dispersal potential and recruitment success, with the goal of informing conservation and restoration efforts. For this purpose, the larvae's ability to delay metamorphosis once competent to settle, while maintaining the capacity to metamorphose and survive as a spat, was examined.

Methods
Larvae were obtained from adult Ostrea edulis originating from the Danish Limfjord. Adult O. edulis oysters were conditioned to spawn at the Danish Shellfish Centre according to the FAO guidelines [52]: their holding tank temperature was raised daily by 1-1.5˚C up to 21˚C, after which the temperature was maintained at 20-22˚C. Adult oysters were fed daily with a mixture of Chaetoceros muelleri, Rhodomonas salina, Tisochrysis lutea and Pavlova gyrans (volume ratio 20:7:1:1 respectively). Newly spawned O. edulis larvae were transferred into 15 L flowthrough holding tanks with 1 μm filtered seawater (FSW) at 25˚C. Larvae were raised at concentrations of 10 larvae/ml and fed daily a microalgae mixture consisting of Chaetoceros muelleri, Tisochrisys lutea and Pavlova gyrans (volume ratio 5:1:1) at a concentration of circa 100 cells/μl. Every 1-2 days larvae were monitored with a binocular microscope for developmental stage. After 9 days, 60% had developed to pediveliger stage, in which larvae are mature to settle.
On day 9, a subsample of larvae was examined under a binocular microscope, to selectively pipette pediveliger for experimentation (N = 153). Selected larvae were divided into three 250 ml beakers at a concentration of 0.4 larvae/ml. The beakers were covered by a lid and fitted with slow aeration filtered to 0.2 μm (Fig 2A). Larvae were maintained in 0.2 μm FSW at an average (± SE) room temperature of 24.2 ±0.01˚C and continued to be fed according to the schedule above. The beakers were maintained and monitored for the duration of the experiment: every 3 days the beaker water was changed [52] and dead or settled larvae were counted and removed during water changes.
Settlement viability of pediveliger larvae was examined on days 0, 4, 7 and 11 (each called a 'settlement test' hereafter). In every settlement test, 41-48 larvae from the beaker culture were subjected to two treatments, which had previously been determined to be 'highly suitable' (conspecifics) and 'unsuitable' (control sea water) for settlement [31] (Table 1; Fig 2B). Each treatment was replicated six times and randomly allocated to a well in two 6-well culture plates. Four larvae were assigned into each well replicate with 24 larvae per treatment, except for the last settlement test, where remaining larvae (N = 41) where divided between the treatments (five wells with 4 larvae/well and seven with 3 larvae/well). After three days, the status of each larvae ('alive'/ 'dead' and 'settled'/ 'not settled') was examined with a binocular microscope. Living spat were identified by the brownish and yellowish colouration of their organs, particularly around the umbo area (see Fig 3). Dead spat were white-translucent with open valves and dead larvae were identified by a faded colour of their inner organs as well as prolonged immobility and open valves (see Fig 3 for an example of the inner colouration of living larvae). Settled spat were measured and placed into a growing tray and after three days, survival and growth were examined under a binocular microscope. Spat were measured with a measuring ruler of 1mm certified length that had been fitted into the eyepiece of a binocular microscope.

Statistics
Fisher's Exact test was used to test for significant differences between the number of living larvae, which had settled or not settled in each settlement test and treatment. Significant differences in mortality per settlement test and treatment were also tested with Fisher's Exact test. All data analysis was performed in R v.3.4.0 [53].

Results
Ostrea edulis larvae were able to settle 11 days after reaching pediveliger stage (Fig 3): 95% (19/20) when presented with a 'highly suitable' settlement treatment on day 11 metamorphosed successfully compared to 11% (2/18) in the 'unsuitable' treatment (Fig 4). 80% of all larvae (16/20) from the 'unsuitable' treatment, were still alive and had not metamorphosed 14 days after reaching maturity to settle (pediveliger stage), when the experiment was terminated due to the low number of remaining 'non-settled' larvae (N = 17 out of 41) (Fig 5). When comparing live larvae, the number of settled larvae versus not-settled larvae did not  Table 1 for specifications of the two treatments). Scale bars: 3 cm.  4) and 0-9% in the beakers (Fig 2A) with larvae for future settlement tests ( Table 2). The number of mortalities (both settled and not settled) was significantly different between settlement tests within the 'highly suitable' settlement treatment (Fisher's Exact test, df = 3, p < 0.001) and the 'unsuitable' treatment (Fisher's Exact test, df = 3, p = 0.047). In the 'highly suitable' treatment, 37.5% (9/24) of larvae died and 58% (14/24) were not found in the first settlement test (day 0-3), while only 4.2% (1/24) died and 4.2% (1/24) were not found in the 'unsuitable' treatment ( Fig 5). In subsequent settlement tests, all larvae in the 'unsuitable treatment' were accounted for and nearly all in the 'highly suitable' treatment ( Fig 5). No further dead larvae were found in any of the subsequent 'highly suitable' treatments. In the 'unsuitable' treatment, there were no further dead larvae either until the last settlement test (day [11][12][13][14], when 10% of larvae died (2/20; Fig 5). Only non-settled larvae were found dead at the end of each settlement test, all settled spat were alive. Mortality of larvae in the pediveliger culture beakers (Fig 2A) was <4% until the end (day 11, see Table 2).
Survival of settled spat was 96-100% within the three days of monitoring. The spat grew from average ± se length x width of 458 ± 2.9 x 459 ± 3.8 μm to 625 ± 12 x 582 ± 12.9 μm. Spat survival and growth was not compromised with increasing delay of metamorphosis (Table 3).

Discussion
The aim of this study was to examine the ability of Ostrea edulis larvae to delay metamorphosis and hence to increase their larval duration and dispersal potential. O. edulis larvae were able to

PLOS ONE
delay metamorphosis for at least 11 days under experimental conditions. The proportion of larvae metamorphosing in response to the 'highly suitable' treatment (95-100%) did not differ significantly over the 11 days and it was similar to the 100% settlement observed in a previous study in response to conspecifics [31]. Settled spat survival was 96-100% during the three days of monitoring. Spat mortality is highest during the first few days after settlement [54][55][56], it is therefore likely that the O. edulis spat would have continued growing healthily, despite a delay in metamorphosis.
At day 14, 80% of larvae (16/20) in the 'unsuitable' treatment were still alive and had not settled, indicating that O. edulis larvae are likely to be able to delay metamorphosis even further. Related bivalve species were able to delay metamorphosis for 28-46 days (blue mussel Mytilus edulis [34]) and for at least 30 days (pacific oyster Crassostrea gigas [57]) suggesting that O. edulis may be similar. M. edulis and C. gigas larvae have slightly longer pelagic durations than O. edulis [cf. ref [52] for C. gigas and ref. [58] for M. edulis] because fertilization of O. edulis larvae occurs inside the mother oyster and larvae are brooded for about 7-10 days before being released into the water column [37]. However, the overall developmental time from fertilisation to pediveliger was similar for O. edulis and C. gigas [52, p.102]. The capacity

PLOS ONE
to delay metamorphosis is thought to depend on the rate of development, which, in turn, depends on seawater temperature and the amount and quality of the food available [39][40][41]: the longer the pre-competent period, the greater the capacity for delaying metamorphosis [34]. In addition, the amount and quality of the food available during development may also affect the period of competence and proximate post settlement mortality when the metabolic demands of the transition from a planktonic to a sessile life are likely to be most acute. In the present experiment, larvae were fed three microalgae species (Chaetoceros muelleri, Tisochrisys lutea and Pavlova gyrans) according to the hatchery procedures of the Danish Shellfish Centre.  Table 2. Percent of settled, living and dead larvae in the pediveliger culture beakers (Fig 2A)

PLOS ONE
This mix of microalgae represents a good food source with fast growth (on average 7 days at 25˚C), but under natural conditions the food regime may be more appropriate since O. edulis larvae develop faster in the sea than in the laboratory experiments (see Fig 1). The quality and quantity of food in the sea is likely to be variable [59] and this may be exacerbated by mismatches in the timing of spawning and larval development with phytoplankton blooms as a result of climate change [60]. Food quality and quantity is of critical importance for larval development to pediveliger [40,41], but it is not clear to what extent it affects the larvae's capacity to delay metamorphosis. For instance, M. edulis larvae's ability to delay metamorphosis was barely affected by differences in quality and quantity of food [58]. More studies are therefore needed to elucidate the extent to which the feeding regime can affect O. edulis larval ability to delay metamorphosis. The period of time larvae delay their metamorphosis is strongly correlated to their specific substrate and habitat requirements [57]. Some species have an extensive target habitat and can settle almost anywhere in the benthos (e.g. the crab Cancer magister [27]), whilst others have restricted target habitat and require cues specifically indicative of surfaces where those species occur [61]. Larvae delay metamorphosis if required settlement cues are absent, as well as if they sense the presence of dominant competitors [34] or predators [62]. As metamorphosis is delayed, larvae can become more sensitive to environmental stimuli, which trigger metamorphosis [34,57]. For instance, in previous experiments, O. edulis larvae settled preferentially in response to particular substrates [43][44][45] and immediately in response to conspecifics, but it took them almost two days to start settling in response to a biofilm cue from a natural habitat [31].
When habitat-associated cues were absent in the present and previous studies, the proportion of larvae settling under low larval concentrations (i.e. <2 larvae ml -1 ) was remarkably low and constant both over time and across experiments (Table 4). Larvae metamorphosed consistently at rates <15% despite the presence of hard substrate (shell or stones but lacking an appropriate cue [31]) and despite having delayed metamorphosis for 11 days (present experiment). This indicates that O. edulis larvae are specialist settlers [cf. 61], finely-tuned to target conspecifics and their own habitats [31,44]. Specialist settlers metamorphose at low rates on

PLOS ONE
surfaces which provide a habitat that is neither optimal nor deleterious for juveniles [63]. This is reflected in the observed low settlement rates of O. edulis on such surfaces ( Table 4). The benefits of being able to delay metamorphosis are clear: it increases the larvae's chance of finding a suitable settlement location, which will support survival and reproduction [34]. However, as larvae delay metamorphosis, the risk of predation in the water column increases [37], and if larvae do not find a suitable settlement location, many may eventually die without metamorphosing [57,65,66]. In the sea, the largest proportion of O. edulis larvae may therefore be lost through mortality if their settlement preferences are not fulfilled.
In this study there were significant differences in mortality between both treatments. During the first settlement test (day 0-3), many larvae died in the 'highly suitable' spat treatment (38%), but not in the control treatment (4%). This indicates contamination originating from the spat (e.g. their faeces resulting in increased bacterial growth), which may have been exacerbated by warm temperatures (~24˚C). In addition, a large proportion of larvae from the 'highly suitable' treatment (58%) were unaccounted for at the end of that settlement test. These larvae may have also died and subsequently been ingested by the spat through its filtering-feeding behaviour. No further mortality was observed until day 14, when 10% of control larvae died. Natural mortality was thus low throughout the 14 days of observation. However, if the experiment had continued, O. edulis larvae would presumably have eventually died or metamorphosed spontaneously, with metamorphosis of O. edulis larvae being able to occur in response to chemical stimuli [67] and without the need of prior attachment to a substrate (personal observation at Danish Shellfish Centre). Future experiments may investigate where that balance lies, and whether, in the absence of suitable settlement cues, most O. edulis larvae would eventually die (despite the absence of predation), or whether they would spontaneously metamorphose.
Pelagic larval duration is the most widely used proxy for dispersal potential. However, while a species with short pelagic duration will inevitably have a short dispersal, species with long pelagic duration do not necessarily disperse more widely [27]. This is because larval behaviour can contribute to retention or return to natal sites with their behaviours [68,69]; thus breaking the otherwise direct relationship between pelagic larval duration and dispersal distance [25]. In the laboratory, O. edulis larvae displayed a strong benthic preference throughout their development [32]. If this behaviour is not overridden by local hydrodynamics [33], it would markedly reduce dispersal distances despite potentially long development times. This study showed that larval behaviour can also considerably prolong pelagic durations if settlement preferences are not met. Potential dispersal time was increased by 2.2-fold in the present study: from 9 days, in which larvae developed to pediveliger stage, to an additional 11 days, in which larvae delayed metamorphosis due to a lack of suitable settlement cues.
The selectivity of O. edulis larvae at settlement indicates that there is a need to better understand the settlement cues, their efficiency, and the manner in which they are dispersed and sensed to increase restoration success. For instance, if the primary settlement cue is chemical and derived from conspecifics, do beds have to be of a minimum critical size/concentration of living oyster biomass to release sufficient cue to trigger settlement effectively in an open sea environment? What is the strength of the cue of a spat versus an adult oyster in the open sea: would it be possible to lay spat-on-shell on top of dead cultch material to increase settlement success? Alternatively, would it be possible to identify the cue, synthesize it and release it in association with the provision of dead cultch material to increase natural settlement in the initial stages of restoration? These are questions that could be investigated further as they have the potential to significantly increase settlement and thus restoration success.
In conclusion, pelagic duration of O. edulis larvae-and hence its dispersal potential-can vary considerably. Obligatory development times last from 6 to >20 days (Fig 1), but once competent to settle, larvae can prolong their pelagic duration for at least 11 days, and probably longer. Larvae can therefore remain pelagic for considerable time, if suitable settlement sites are absent, and this is likely to also depend on seawater temperature and the abundance and quality of phytoplanktonic food. For this reason, connectivity of oyster restoration sites may be greater at greater distances than might otherwise be expected. Conversely, prolonged pelagic larval duration also increases larval mortality, which can strongly diminish recruitment strength. In this study the potential dispersal time was more than doubled through the delay in metamorphosis, indicating the important role that the larvae's settlement requirements can play when planning for dispersal potential, habitat connectivity and recruitment success in restoration programmes.