The role of sexual reproduction in the maintenance of established Zostera marina meadows

For clonal plants, the role of sexual reproduction in the maintenance of populations can vary widely. Some species are dependent on repeated seedling recruitment. For other species, interactions between adults and seedlings within existing populations can affect seedling survival and limit sexual reproduction in existing populations. Genetic studies of seagrass populations increasingly suggest sexual reproduction is important for the resilience and stability of their populations, but as of yet little observational data support these findings. Because seagrass populations provide important ecosystem services and are threatened with increasing anthropogenic impacts, understanding their reliance on sexual reproduction is evolutionarily and ecologically important. The goals of this study were to determine (a) whether seedlings of a marine angiosperm, Zostera marina, establish and recruit within existing Z. marina meadows and (b) whether interactions between seedlings and surrounding adult shoots influence the survival of established seedlings. To meet these goals, surveys estimated seedling establishment and tracked seedling survival within multiple populations. Manipulative experiments then tested the impact of neighbouring adult shoots on seedling survival and the overall trajectory of experimental plots with and without sexual reproduction. A 3‐year survey identified established seedlings within Z. marina meadows each year. Additionally, concurrent seed addition experiments indicated seed supply could influence seedling establishment rates. A survey tracking the survival of tagged seedlings, as well as the height and density of surrounding adult shoots, showed adult shoots may negatively impact seedling survival. Experiments then demonstrated that seedlings without neighbouring shoots survived longer than those with neighbouring shoots. Lastly, two transplant garden experiments comparing the survival of plots with and without seeds highlighted that seedling recruitment is likely most important to maintain bottom cover where disturbances generate gaps in the adult population. Synthesis. This study demonstrates that seedlings do establish within existing seagrass meadows, and that some survive to recruit into the adult population. Competition with existing vegetation, however, can be a factor compromising seedling survival. Sexual reproduction may thus most likely occur in, and be most important for, clonal plant populations that experience seasonal disturbance.


| INTRODUC TI ON
The role of sexual reproduction in the population dynamics of a species can vary widely for clonal plant species, such as seagrasses, which are capable of both sexual and asexual reproduction. Sexual reproduction in clonal plants occurs through seedling recruitment, the germination, growth and survival of seedlings into the reproductive population (Eriksson & Ehrlen, 2008). Determining the patterns and relative importance of seedling recruitment within populations is therefore critical to understand both the population dynamics and the life-history evolution of these species (Eriksson, 1989(Eriksson, , 1993. Clonal species for which seedling recruitment is rare among adult plants exhibit 'initial seedling recruitment' (ISR) patterns with high dependence on asexual reproduction after an ISR cohort (Eriksson, 1993). Seedling recruitment within a population is low or non-existent subsequent to this initial recruitment wave. Conversely, species exhibiting a 'repeated seedling recruitment' (RSR) pattern consistently depend upon seedling recruitment among adult plants for population maintenance. These classifications of recruitment may represent the extreme endmember classification for designating the role of sexual reproduction for a given species, as populations may differentially rely on sexual reproduction for population maintenance and resilience, especially where disturbance alters competition for resources (Bullock, 2000;Eriksson, 1993;McMahon et al., 2017).
Seagrasses are ecologically important aquatic plants that are increasingly subjected to both natural and anthropogenic stresses (Orth et al., 2006;Waycott et al., 2009). As clonal plants able to reproduce sexually and asexually, reproductive plasticity may be an important trait conferring resilience to seagrass populations in the face of these pressures. Because seagrasses produce both asexual and sexual progeny, their seeds may potentially not only germinate among adult plants of the same or different species but also the asexual progeny, or ramets, of their own parent plant. The presence of adult plants surrounding their seedlings may limit the space, light, nutrients and other limiting resources available to seedlings and can therefore influence seedling survival (Bullock, 2000;Bullock & Silvertown, 2003). For seagrasses, the density and height of surrounding shoots may dictate the extent to which adult shoots exploit resources more effectively than seedlings in their vicinity (Bintz & Nixon, 2001;Ralph, Durako, Enriquez, Collier, & Doblin, 2007;Robertson & Mann, 1984;Zimmerman, 2003). The density and height of seagrass shoots can, however, change with seasons and the availability of critical resources, such as light or nutrients (Dennison, 1987;Orth, 1977;Short, 1983;Van Lent, Verschuure, & Veghel, 1995). Changes in the adult population structure could therefore dramatically alter the interactions between existing clones and seedlings in space and time, with high densities and heights of adult shoots likely providing the most significant resource competition with seedlings during peak growth and biomass.
Zostera marina is the dominant seagrass in the temperate waters of the Northern Hemisphere. Across this species' distribution, populations invest and rely variably on sexual reproduction for population maintenance, resilience and dispersal to new areas distant from the parent plant (Jarvis, Moore, & Kenworthy, 2012;Phillips, Grant, & McRoy, 1983;Robertson & Mann, 1984).
Conversely, populations with a 'perennial' life history rely substantially less or not at all on sexual reproduction and largely survive through asexual reproduction (Billingham, Reusch, Alberto, & Serrão, 2003;Reusch, Bostrӧm, Stam, & Olsen, 1999). Asexual reproduction can be less energetically expensive and risky than sexual reproduction and allows perennial populations to maximize clone survival and growth under favourable environmental conditions with limited disturbance (Philbrick & Les, 1996). As a result, some perennial Z. marina meadows growing under favourable conditions are almost entirely monoclonal (Reusch et al., 1999). Sexual reproduction does, however, occur within these populations and seeds have facilitated perennial population recovery from mass die-off events (Jarvis & Moore, 2010;Plus, Deslous-Paoli, & Dagault, 2003). Genetic analyses of perennial Z. marina populations within Long Island, New York also suggest seedling recruitment contributed substantially to the expansion and recovery of a perennial population (Furman, Jackson, Bricker, & Peterson, 2015). Low levels of seedling establishment within perennial Z. marina populations may even occur annually, but with complete (Olesen, 1999) or near total mortality, except in areas on the periphery of the denser portions of meadows or below a critical depth or disturbance threshold (Olesen, Karuse-Jensen, & Christensen, 2017). The high seedling mortality rates in existing meadows are attributed to competition with adult Z. marina shoots. These results suggest sexual reproduction is important for colonization of Z. marina populations but may not be generally important for the maintenance of existing perennial meadows.
Understanding the relative contribution of sexual reproduction and be most important for, clonal plant populations that experience seasonal disturbance.

K E Y W O R D S
clonal plants, competition, plant populations, seagrass, seedling recruitment and seeds to seagrass populations may be more important in the modern era, as efforts to restore seagrasses in regions where anthropogenic disturbances have been ameliorated may rely on seed-based restoration methods rather than traditional whole plant transplantation.
The goals of this study were to determine whether Z. marina seedlings establish and recruit within existing meadows of Z. marina and if interactions between seedlings and surrounding adult shoots influence the survival of these seedlings. The objectives were as follows: (a) to determine the degree to which seedlings establish within existing meadows; (b) to test whether seed availability influences the seedling establishment rate within existing meadows; (c) to quantify the relationship between seedling survival and surrounding adult vegetation and (d) to evaluate the relative influence of sexual reproduction on the maintenance of Z. marina meadows through space and time. This study does not explicitly test density-dependent effects of adult shoots and seedlings on one another. Instead, this study evaluates competition as the outcome of interactions between adult shoots and seedlings that may compromise seedling survival and potentially impact seedling recruitment patterns.  Figure 1). All sites were shallow (<1.0 m at MLW). We define meadows as Z. marina populations that are persistent spatially and temporally within aerial surveys of Chesapeake Bay seagrasses . York River sites were fringing, persistent meadows with similar tidal and thermal regimes while the coastal lagoon site was located on a shoal area within a larger bay that is part of a large-scale seagrass restoration project.

| Study sites
Within Chesapeake Bay, Z. marina exhibits significant variability in growth and reproduction with the seasons and over small spatial scales (<5 km; Johnson, Moore, & Orth, 2017;Orth & Moore, 1986;Shields, Moore, & Parrish, 2018). Biomass of F I G U R E 1 The location of experiments and surveys throughout the lower Chesapeake Bay (Bena, Goodwin Neck, Goodwin Island, Allens Island and Sandy Point) and Eastern Shore of Virginia (Spider Crab Bay) Z. marina peaks in the spring and early summer declines dramatically during warm summer months, partially recovers during the fall, and once again senesces in the winter (Moore, Wilcox, & Orth, 2000;Orth & Moore, 1986). Flowering shoot densities can compose up to 19% of total shoots and produce up to 8,000 seeds/m 2 (Silberhorn, Orth, & Moore, 1983). Seeds remain dormant in the sediment during the spring and summer and germinate in the late fall when water temperatures drop below ~15°C (Moore, Orth, & Nowak, 1993;Orth & Moore, 1986). Z. marina in the region has a transient seed bank with all seeds produced in the spring germinating the following fall (Moore et al., 1993). Thus, Z. marina seeds in Chesapeake Bay germinate under environmental conditions favourable for growth and during a period of low adult cover and biomass. Seedlings may thus have an opportunity to germinate and grow within gaps between clones before peak biomass the following spring (here defined as 'establishment'). This growth may allow seedlings to establish carbohydrate reserves necessary to survive the most physiologically demanding warm summer months (Burke, Dennison, & Moore, 1996), and successfully recruit into the adult population the following winter (Moore, Shields, Parrish, & Orth, 2012;Silberhorn et al., 1983).
The Z. marina meadows of Chesapeake Bay are genetically diverse. This high diversity may be a consequence of disturbance events (e.g. hurricanes propeller scarring-Orth et al., 2017;cownose ray activities-Orth, 1977), and largescale diebacks within the region ) that generate bare sediment within meadows that are often meters wide or impact an entire meadow. Populations have recovered from these large disturbances almost exclusively from seedling recruitment within the bare sediments, either from a dormant seed bank or from seeds dispersed from populations meters to kilometres from the disturbed areas (Harwell & Orth, 2002;Orth, Luckenbach, & Moore, 1994), thus providing genetic material from numerous populations (Reynolds, Waycott, & McGlathery, 2013;Reynolds, Waycott, McGlathery, Orth, & Zieman, 2012).

| Study design
This study uses a series of surveys and experiments over 3 years (2016)(2017)(2018) to evaluate the influence of seed and seedling processes on perennial Z. marina meadows in Chesapeake Bay.
Surveys and experiments were first used to evaluate natural levels of seedling establishment in the perennial meadows of Chesapeake Bay and the potential for seed supply to impact these levels. Simultaneously, experimental transplant garden plots were constructed to directly test whether the availability of seeds, and sexual reproduction, was necessary to maintain bottom cover within plots relative to plots without seeds over time. In 2017, a survey and an experiment explored the influence of surrounding vegetation on the survival of seedlings in meadows. A survey of tagged seedlings evaluated whether seedling survival was related to the characteristics of neighbouring adult shoots. Lastly, an experimental manipulation was used to compare the survival of seedlings with adult shoots surrounding them to the survival of seedlings around whom adult shoots had been experimentally removed. We focus on seedling establishment and survival as the primary metrics of sexual reproduction in this study, as these later stages of sexual reproduction are both necessary for successful sexual reproduction but also clearly demonstrate that the previous stages of sexual reproduction (e.g. flowering, seed dispersal and seed settlement) were successful. As a result, measurements of seedling establishment and survival imply successful pollination, seed survival seed germination, and initial seedling growth in the area where individual seedlings can be observed and counted.

| Natural seedling establishment and potential seed limitation
To quantify seedling establishment within meadows of Z. marina and to determine whether the size of the seed bank at a given location may limit seedling establishment, in situ plots were constructed within the middle of seagrass meadows at GI, SP, AI and BE in the York River. At each location 6, 2 m 2 plots were constructed in three distinct blocks, each containing two plots, at similar depths (±10 cm) in autumn 2015. Two thousand viable seeds, collected in spring, 2015 (according to Marion & Orth, 2010), were then broadcast evenly onto the sediment surface within one, randomly selected plot in each block (three plots at each location). The remaining plot in each block did not receive additional seeds and was considered a control plot that would maintain the natural level of seedling establishment in the meadow that year. Four, 0.02 m 2 cores were taken from each 2 m 2 plot in late May 2016, 5-6 months after seed germination. The number of seedlings, the total shoot number, the maximum adult shoot height and a random adult shoot height were then recorded for each core. Seedlings were identified as having a heavily rooted and curved rhizome base (Setchell, 1929;Figure 2a).
Additional seeds were added and the sampling procedure was repeated in the fall and spring of 2016-2017 and 2017-2018 (Table S1).
Because these plots were constructed within existing Z. marina meadows, any seedlings identified from control plots, those not receiving 2000 supplemental seeds, were likely from seeds naturally settling in that area. As such, cores taken within these plots acted as a survey of natural seedling establishment within the meadow. A generalized linear model (GLM) fit to a Poisson distribution was used to determine whether the number of seedlings naturally establishing in meadows of the York River varied by location and/ or year (Table S2). A generalized linear mixed-effects model, also fit to a Poisson distribution, was then used to evaluate if the seed addition treatment significantly increased the number of seedlings within cores taken from plots with additional seeds relative to control plots at a given location in a given year. For this model, the block, location and year of sampling were treated as nested random variables. A mixed-effects model was used to test whether mean seedling heights were significantly shorter than the mean height of surrounding, adult shoots within cores. The year and location from which shoot heights were measured were considered random variables. The heights of shoots were log 10 transformed to meet model assumptions.

| Adult shoot neighbours and seedling survival
To evaluate whether the characteristics of surrounding adult shoots influence the survival of seedlings in Z. marina meadows of the lower Chesapeake Bay, seedlings were identified and followed monthly at GI, SP, AI and BE in the York River. Seedlings were identified within a single, 4 m 2 plot at each location in April 2017. At this time of the year, Z. marina seedlings have germinated, but the growth of Z. marina in Chesapeake Bay has not consolidated Z. marina meadows to the extent that seedlings are indistinguishable from adult shoots. Seedlings were identified as spatially isolated and lacking clonal integration with surrounding shoots. Once a seedling was identified, its position was recorded using 1 m 2 North-South orientated grid quadrats gridded into 100, 100 cm 2 cells that were placed over each 1 m 2 of the 4 m 2 plot. A stainless steel 19 mm washer was then slid down the leaves to the base of the shoot so that the shoot would grow through the centre of the washer and anchor the washer A first-order autoregressive model fit to a Gaussian distribution was used to test whether the mean numbers of seedlings surviving in an area was related to the mean estimated LAI of Z. marina or R. maritima shoots surrounding the seedling. The estimated LAI of Z. marina and R. maritima were considered fixed variables, but the month and the location at which seedlings were evaluated were considered nested, random variables. Due to the radically different scales between the LAI of Z. marina and R. maritima, these variables were scaled with the scale function in the base R package. Because bottom cover was not recorded at Goodwin Island in July, the bottom cover for that month was estimated by averaging bottom cover taken in June and August.

| Seedling competition experiment
To directly test whether adult Z. marina plants influence the survival of Z. marina seedlings within an established meadow, the survival of seedlings growing among adult shoots was compared to the survival of seedlings around whom adult shoots were experimentally removed. In all, 21 seedlings were identified and tagged with plasticcoated wire bent around the base of the shoot in May 2017 within an 11 m 2 area at BE in the York River, VA (Figure 2c). The location of each seedling within a North-South orientated 1 m 2 grid split into F I G U R E 2 Seedlings and adult shoots of Zostera marina: (a) the distinctive hook and 'hairy' end to the rhizome used to identify seedlings in the study, (b) a seedling identified in situ with a 19 mm lock washer about to be placed around its rhizome, (c) a dead seedling dug up from the competition treatment after defoliating and (d) a mixed Z. marina (longer leaves, foreground) and Ruppia maritima (shorter leaves, middle ground) meadow at Goodwin Island [Colour figure can be viewed at wileyonlinelibrary.com] 100 cm 2 cells was recorded. For 10 haphazardly selected seedlings, all other shoots and rhizomes within 15 cm of the seedling were removed to eliminate any resource competition between seedlings and neighbouring shoots. The shoots surrounding the remaining 11 identified seedlings were counted and left to grow around the seedlings.
After waiting 2 weeks to account for any mortality resulting from the application of treatments, the survival of each tagged seedling was recorded weekly through October 2017.
Kaplan-Meier survival curves were generated for seedlings growing with and without neighbouring, adult Z. marina shoots. A log-rank test was then used to compare these survival curves to determine whether the duration of seedling survival differed between these two treatments.

| Transplant garden experiments
To test whether the establishment of seed banks by sexual reproduc-

| Natural seedling establishment and potential seed limitation
Naturally established seedlings were identified in control plots at all locations over the course of the three study years (Figure 3). A significant increase in the number of established seedlings was detected between plots with an additional 1,000 seeds/m 2 (Table S5; β = 2.0 ± 1.1, z = 6.7, p < .001) relative to control plots over the three study years and across all four locations. On average, more seedlings were found at GI (126 ± 65 seedlings/m 2 ) than at AI (55 ± 14), SP (50 ± 16) or BE (24 ± 7). More seedlings were also found in 2018 (125 ± 49 seedlings/m 2 ) than in 2016 (45 ± 13) or 2017 (21 ± 5). The effect of location on seedling establishment varied significantly with the year of sampling (Table S3; p < .001). All locations except GI demonstrated higher seedling establishment in 2018 relative to 2016 and 2017. Goodwin Island exhibited high seedling establishment in both 2016 and 2018. The significant interaction between the location and the year of sampling on seedling establishment likely stems from this high seedling establishment at GI in both 2016 and 2018. Seedlings also made up the largest fraction of the total shoots present at GI (19 ± 30%) relative to SP (16 ± 26%), AI (10 ± 20%) or BE (3 ± 6%).

F I G U R E 4
Adult shoot neighbours and seedling survival. The number of seedlings surviving during monthly sampling at Allens Island (n = 24), Goodwin Island (n = 23), Bena (n = 12) and Sandy Point (n = 11). While sampling in May 2017, an additional seedling was identified and tagged at Sandy Point. Similarly, a tagged seedling, that was not found in September, was discovered alive in October at Goodwin Island

| Seedling competition experiments
The survival of seedlings without neighbouring shoots was significantly higher than the survival of seedlings with neighbouring shoots (χ 2 = 12.4, p < .001). In fact, the only seedlings to survive the experimental period were seedlings without neighbouring shoots within a 15 cm radius (Figures 2c and 5).

| Transplant garden experiments
The per cent bottom cover of plots with sexual reproduction (the transplanted and seeded or solely seeded treatments) did not significantly differ from plots with exclusively asexual reproduction (the solely transplanted treatment) across all 3 years of the study (F 2,286 = 0.68, p = .5; Figure 6). Bottom cover within seeded plots did, however, have significantly (F 2,190 (Table S7;

| D ISCUSS I ON
Our results demonstrate seedling establishment consistently occurs within the perennial Z. marina meadows of our study system, but the magnitude of establishment varies substantially over space and time.
Seed availability or the size of the seed bank may partially explain the observed variability in the seedling establishment rate. Unexplored factors, such as local seed predation (Fishman & Orth, 1996) or winter storms (Jarvis & Moore, 2015;Marion & Orth, 2012), can also, however, markedly influence seedling establishment rates. Established seedlings contributed to the structure of meadows during peak bio-  (Reusch, 2006). Intermediate or minor disturbances may potentially lower intraspecific competition precluding seedling recruitment among adult shoots and enhance seedling survival (Yang, HilleRisLambers, & Ruesink, 2016). whether ISR or RSR (Eriksson, 1993), capable of maintaining and recovering populations in the face of disturbances.
Observations of seedling establishment and the increased importance of sexual reproduction at locations and in years with low adult cover suggest a reproductive 'bet hedging' strategy within the perennial Z. marina meadows of Chesapeake Bay. During years of high disturbance, for example, years with high summer water temperatures and/or poor water quality, low numbers of adult shoots will survive, and seedlings germinating after environmental conditions improve will likely constitute more of the shoots the following spring (Jarvis et al., 2014;Jarvis & Moore, 2010). During years with less seasonal disturbance, for example, years with cooler summer water temperatures and/or better water quality, more adult shoots can survive throughout the entire year and the relative influence of seedlings will likely be low compared to the growth of existing adult shoots the following spring . During these years, shoots of existing plants likely outcompete seedlings germinating within the meadow. The relative role of sexual reproduction in meadow maintenance would thus change along a gradient of disturbance to the existing canopy so long as seed production and environmental conditions suitable for seedling establishment exist.
Such selection for reproductive strategies to maintain populations likely exists more broadly. Sexual reproduction may serve an important role in Z. marina population survival at locations with discrete seasonal or cyclical disturbances, such as ice scour (Robertson & Mann, 1984), seasonal light disruptions (Kim et al., 2014) or high temperatures (Jarvis et al., 2012;Santamaria-Gallegos et al., 2000;this study), to the existing population, but which subside or disappear by the time seeds germinate.
Conversely, at locations without acute seasonal disturbances or a high potential for disturbance at some point during seedling establishment, asexual reproduction may play a larger role in maintaining individuals and populations (Billingham et al., 2003;Reusch et al., 1999). The variability and importance of sexual reproduction in this study and across the distribution of Z. marina suggests the plasticity of reproductive traits among Z. marina populations may facilitate the species' colonization of diverse environments and habitats and enhance its resilience to natural and anthropological environmental stresses.

| Potential impact of seed supply and establishment
For sexual processes to accelerate meadow recovery from disturbances, seed production and banks must be sufficient at locations where a disturbance occurs and the disturbance itself cannot also disturb the seed or seedling bank. Flowering intensity and seed bank densities fluctuate in both space and time (Harwell & Orth, 2002;Phillips et al., 1983;Silberhorn et al., 1983;Van Lent et al., 1995).
Results from seed addition experiments in our study suggest seed availability or supply may influence the number of established seed- where, seedlings will generally flower for the first time in their second year of growth (Jarvis & Moore, 2010;Setchell, 1929). Multiple acute disturbances, for example, consecutive years of highly stressful summers, may thus dramatically reduce the flowering population and seed supply. With simultaneous low shoot survival and low seed supply, seagrass populations will struggle to recover (Jarvis et al., 2014;Jarvis & Moore, 2010;Kuusemäe et al., 2018;Valdemarsen, Wendelboe, Egelund, Kristensen, & Flindt, 2011). Similarly, if seed-based restoration were to remove a high proportion of seeds and lower the supply of seed available to the donor meadow, the donor meadow could experience lower seedling establishment. A previous study, however, did not recorded significant impacts from seed harvesting on donor meadows, likely because the adult population did not experience any disturbance after the harvesting, the removal did not significantly lower seed supply, the donor meadow was selected for its high fecundity or dispersal from adjacent populations still provided adequate seed supply (Marion & Orth, 2010). Meadows of Z. marina in Chesapeake Bay generally produce high densities of flowering shoots, so results from that study may not be applicable to regions where flowering shoot densities are much lower or flowering populations more isolated.
The intensity and timing of disturbances may also alter the relative benefit of sexual or asexual reproduction. Unlike in the York River, bottom cover within the seeded plots in Spider Crab Bay transplant garden experiments was consistently lower than in plots with adult transplants. The lower cover in seeded plots in Spider Crab Bay likely resulted from consistently higher wave and tidal current energy at this location relative to the York River during seedling establishment ( Figure S4). Colder and clearer water within coastal lagoons of the Delmarva Peninsula may also favour shoot survival and asexual reproduction in Spider Crab Bay relative to the York River location . The combined impact of higher energy during vulnerable seedling establishment periods and higher water quality may select for higher clonal than seedling survival in Spider Crab Bay. Of course, disturbances during seed settlement or seedling establishment, even if away from competing adult vegetation, may also diminish the role of sexual reproduction in meadow maintenance (Marion & Orth, 2012;Yang et al., 2016). Should disturbances increase in frequency and intensity before or during seed production or seedling establishment, the resilience of seagrass populations reliant on sexual reproduction may decline and population collapse may occur.

| Impact of adult shoots on seedlings
Although previously hypothesized (Olesen, 1999;Olesen et al., 2017), our study observationally and experimentally demonstrates a relationship between surrounding vegetation and seedling survival within existing meadows. Adult shoots do compete with each other and seedlings for common resources, such as space, light or nutrients (Gopal & Goel, 1993;Gustafsson & Bostrӧm, 2013;Invers, Zimmerman, Alberte, Pérez, & Romero, 2001;Williams, 1987). Interestingly, we observed the site with the lowest measured LAI, Sandy Point, was smothered in extensive and unexpected algal mats during the summer of 2017 ( Figure S3).
For this location, interactions with macroalgae, not surrounding seagrass shoots, likely compromised seedling survival. Although resource competition, whether inter or intraspecific, may not immediately or directly threaten seedling survival, the lower resource levels left for seedlings after adults draw down common resource pools can negatively impact seedling growth (Bintz & Nixon, 2001;Robertson & Mann, 1984;Zhang, Zhang, Niu, Sun, & Tian, 2014;Figures S5 and S6).
For Z. marina in Chesapeake Bay near the southern limit of its distribution, growth during optimal environmental conditions is important for accumulating non-structural carbohydrate reserves needed to survive during stressful environmental conditions (Burke et al., 1996). Because seeds germinate in late fall in Chesapeake Bay and have the potential to grow in gaps before interacting with adult shoots in the spring, the timing and growth of seedlings prior to experiencing stressful summer environmental conditions could be critical to their survival and should be investigated more thoroughly Figures S7 and S8). By slowing growth and the accumulation of reserves, resource competition could dramatically hamper seedling survival for Z. marina, as observed here, but also potentially other seagrass seedlings growing among established clones of their own or different species in other systems.
Some seedlings did, however, survive the stressful summer pe-

| CON CLUS IONS
The relative role of sexual reproduction in seagrass meadow maintenance is likely a function of both seed supply and the survival of propagules competing with surrounding clones. Seedling establishment is an important demographic process within the perennial Z. marina meadows of our study system but varies across locations.
Seed availability may partially explain this variability in establishment among locations. Interactions between surrounding shoots and established seedlings appear to limit the survival of seedlings within existing meadows through the stressful summer period in our system. Sexual processes will likely be important for Z. marina and other seagrass population dynamics where seed production is high and clonal survival is low at some point in either space or time, due to biotic (e.g. ray or crab holes, macroalgal mats) or abiotic (e.g. stressful water temperatures, ice scour or strong wave energy) disturbances. These disturbances to the population may provide windows of opportunity for seedlings to germinate and survive, so long as they are offset in time from seed germination and seedling development. Sexual reproduction and seedling recruitment may thus be required to play an increasingly important role in the persistence and resilience of seagrass populations disturbed by increasing anthropogenic pressures. Finally, these processes may only become more important to understand as efforts to restore seagrass begin to incorporate seed-based approaches.

ACK N OWLED G EM ENTS
We would like to thank J. Richardson and F. Holbert for their invalu-