Population ecology of the gulf ribbed mussel across a salinity gradient : recruitment , growth and density

Benthic intertidal bivalves play an essential role in estuarine ecosystems by contributing to habitat provision, water filtration, and promoting productivity. As such, changes that impact population distributions and persistence of local bivalve populations may have large ecosystem level consequences. Recruitment, growth, mortality, population size structure and density of the gulf coast ribbed mussel, Geukensia granosissima, were examined across a salinity gradient in southeastern Louisiana. Data were collected along 100-m transects at interior and edge marsh plots located at duplicate sites in upper (salinity ;4 psu), central (salinity ;8 psu) and lower (salinity ;15 psu) Barataria Bay, Louisiana, U.S.A. Growth, mortality and recruitment were measured in established plots from April through November 2012. Mussel densities were greatest within the middle bay (salinity ;8) regardless of flooding regime, but strongly associated with highest stem densities of Juncus roemerianus vegetation. Mussel recruitment, growth, size and survival were significantly higher at mid and high salinity marsh edge sites as compared to all interior marsh and low salinity sites. The observed patterns of density, growth and mortality in Barataria Bay may reflect detrital food resource availability, host vegetation community distribution along the salinity gradient, salinity tolerance of the mussel, and reduced predation at higher salinity edge sites.


INTRODUCTION
Benthic intertidal bivalves play an essential role in several marsh processes, contributing to physical habitat for species while promoting vegetative productivity through suspended nitrogen filtration and biodeposition (Jordan and Valiela 1982).As a foundation species of ecosystem engineering, ecosystem level changes that impact local bivalve population dynamics and distribution may lead to trophic cascades with large ecosystem consequences (Jordan andValiela 1982, Bertness 1984).Given recent and ongoing anthropogenic (i.e., river control, coastal restoration) and natural events (i.e., sea level rise) affecting coastal habitats along the northern Gulf of Mexico, it is important to develop a better understanding of the distribution and ecology of native bivalve populations.Only by understanding the factors controlling their population dynamics can we effectively assess how coastal changes might impact the services they provide.
Ribbed mussels are euryhaline benthic bivalves, native to the western Atlantic coast from the Gulf of St. Lawrence to the northern Gulf of Mexico (Bertness 1984, Watt et al. 2011), although they have also been found in northern Venezuela (Baez et al. 2005) and California (Torchin et al. 2005).Individuals anchor to nearby shells, nearby hard substrate or marsh vegetation with strong byssal thread attachments (Franz 1997).Mature adults average 8 cm in length and have been recorded in population densities of greater than 2000 ind m À2 in New England salt marshes (Chintala et al. 2006).Ribbed mussels are rselected cast spawners producing large numbers of small planktonic eggs (Brousseau 1982).Development and subsequent settlement typically lasts 3-4 weeks, after which juveniles migrate short distances to final attachment sites using a muscular foot and by byssal-drifting within local currents (Widdows 1991).
The high tolerance of ribbed mussels to environmental stressors including extreme temperature (0-458C; Hilbish 1987, Jost andHelmuth 2007) and salinity (3-48;Pierce 1970, Neufeld andWright 1998) is well-documented, and allow these mussels to exist across large environmental gradients (Baez et al. 2005, Torchin et al. 2005).Although the mussels have a wide tolerance to environmental gradients, changes or extremes in environmental variables (i.e., salinity, temperature) affect basic demographic processes, such as the timing of gametogenesis, spatial and temporal patterns of juvenile recruitment, adult growth and mortality (Bayne et al. 1983, Widdows 1991, Baez et al. 2005, Thompson et al. 2012, Le Corre et al. 2013).For example, physiological responses to osmotic stress at the extremes of their salinity tolerance result in lower growth and increased mortality (Strange andCrowe 1979, Wang et al. 2011).
Ribbed mussels along the U.S. Atlantic coast provide important ecosystem services as critical facilitators of nutrient cycling in coastal ecosystems, contributing to estuarine filtration, fertilization of host vegetation and soil strengthening (Jordan andValiela 1982, Culbertson et al. 2008).These populations have been extensively studied in Spartina alterniflora saltmarshes, quantifying population responses to host-vegetation density, substrate characteristics, and nutrient and flooding regimes (Jordan and Valiela 1982, Bertness 1984, Klaus and Crow 1985, Franz 1996, Chintala et al. 2006).However, latitudinal differences in population dynamics and distribution suggest potential regional population differences related to environmental variation, and differences in predation dynamics (Lin 1991).
Research on ribbed mussel dynamics along the Atlantic coast of North America have focused exclusively on Spartina alterniflora salt marshes, providing little insight on U.S. Gulf coast communities, where other vegetation species are equally dominant to S. alterniflora, where salinity regimes differ dramatically, and where a putative different species of ribbed mussel exists (Sowerby 1914).The Gulf ribbed mussel (Geukensia granossisima), like its Atlantic cousin, Geukensia demissa, is a benthic intertidal bivalve forming large aggregations within dense salt marsh vegetation, and exists along the U.S. gulf coast.Few studies in this region have examined ribbed mussel populations, particularly within Mississippi River deltaic marshes.Understanding these basic distribution and population dynamics is critical to predicting effects of environmental change on local ecosystems and their survival.
We examined the recruitment, growth, mortality and distribution of the ribbed mussel (G.granossissima) within Barataria Bay, Louisiana.We quantified ribbed mussel distribution and population structure in low, mid and high salinity (;4, 8, 15 salinity mean) zones at interior and marsh edge sites in Barataria Bay.Among the salinity regimes and marsh zones present in Barataria Bay, it was hypothesized that midsalinity marsh edges would host the greatest ribbed mussel densities, coinciding with peak S. alterniflora density.Such a hypothesis is supported by similar patterns documented in temperate Atlantic populations (Bertness 1984, Franz 1997).
In the low-salinity upper regions of the estuary, reduced mussel densities were expected due to hypo-osmotic stress and reduced resource availability (bottom-up control) potentially resulting in reduced growth rates and greater mortality.In the high salinity zone, it was hypothesized that predation (top-down control) would limit mussel densities due to greater densities of known bivalve predators associated with the higher salinity and halophytic vegetation density (i.e., Callinectes sapidus; Williams et al. 1990).

Study site
Barataria Bay is a well-mixed, microtidal coastal plain estuary in southeastern Louisiana.Located between the Mississippi River and Bayou Lafourche, the bay lies atop remnants of an abandoned deltaic lobe culminating in a chain of barrier islands separating Barataria Bay from the Gulf of Mexico.Water temperature ranges from 58C in January to 338C in August (Feng and Li 2010) and there is a distinct salinity gradient ranging from freshwater to a high salinity of 25 psu near the Gulf of Mexico.Water levels increase in the spring from riverine discharge and runoff, and remain high as marine influx and salinity increase through the late summer before falling to a winter low as northerly cold fronts rapidly flush Bay waters (Feng and Li 2010).

Field survey
Sampling design.-Twotransects were established at each of the six sites (3 sampling areas 3 2 sites), with transects located parallel to the water's edge in two marsh zones; one along the marsh edge (,1 m from the water edge), and one in interior marsh (.5 m from the water edge).Sample areas represented differences in salinity while zones represented potential differences in site exposure and flooding rates.Each transect was 100 m in length, with 0.25 m 2 quadrats placed every 10 m for a total of 120 plots (3 areas (salinity) 3 2 sites 3 2 zones (edge, interior) 3 10 sample plots ¼ 120; hereinafter called ''zone plots'').
Environmental variables.-Salinity,temperature and water level data were downloaded from Coastwide Reference Monitoring System (CRMS) stations located near each study site.Stem density of each vegetation species was quantified within each quadrat.
Mussel distribution.-Withineach quadrat, all mussels were excavated to 30 cm depth and mussel densities (ind m À2 ) calculated from abundance counts.Mussels were sized (mm) along the greatest distance between the shell's umbo and anterior edge with hand calipers.Surveys were conducted from May through July 2012.
Growth and mortality.-Fiveframed grids for mussel attachment were placed at both edge and interior marsh (1, 5 m from edge) at each of the six sites to determine the effects of salinity and marsh zone on mussel growth and mortality (3 areas 3 2 sites 3 2 zones 3 5 grids).Grids consisted of plastic mesh stretched over 0.25 m 2 PVC bases placed in the marsh.Five randomly collected mussels from mid-salinity marsh (HB) were placed within each quadrat across all sites in accordance with previously observed mussel densities in Barataria Bay (Spicer 2007).The mussels were evenly spaced between mesh spaces, with both shells and quadrats embedded in the marsh.Once placed, mussels were checked after 1 week to confirm byssal attachment within PVC quadrats, and immediate survival.Individual mussels were identified by their unique, anchored placement within quadrat mesh, and were not observed to move during the experiment.
All experimental quadrats were deployed in March and sampled in October 2012.The initial and final size of all live mussels was recorded by measuring the greatest distance between the shell's umbo and anterior edge (mm).Mean size (6 SE) of deployed mussels was 81.5 6 5.1 mm.Dead mussels were classified as either predation morality if shells were broken or exhibited clear signs of forced entry, while intact shells were classified as ''dead,'' implying mortality related to non-predation events.While it is possible that shells were broken by crabs after non-predation mortality, valves weaken considerably post-mortem, allowing for thorough scavenging without shell destruction (Lin 1991).
Recruitment.-Threeclay flower pots (15 cm diam.) were embedded at both the marsh edge and interior (1, 5 m from channel) at each of the six sites.The pots were filled with local sediment that was filtered through 13 mm mesh to remove vegetation and infauna.Five intact mussel shells were randomly collected in Barataria Bay, cleaned of all fouling organisms and secured within plastic mesh anchored within each pot.The pots were placed in the marsh in June, and sampled monthly with replacement through September 2012.Juvenile recruitment was deter-mined by recording the number of recruits attached to sampled pots (shells, pots, mesh) each month using a dissecting microscope in the laboratory, and recorded as number of recruits per pot per month.

Statistical analyses
For all tests, a significance value of p , 0.05 was used.Temperature, salinity, total and dominant vegetation density and flooding rates were examined by sample area (TB, HB, RL) and marsh zone (edge, interior) using a general linearized mixed model.Interactive and single effects were examined.Significant results were examined using LSMeans post hoc test.
A general linearized mixed model was used to examine differences in mussel density, size, growth, mortality and recruitment between sample areas (TB, HB, RL) and marsh zone (edge, interior).Vegetation density and flooding v www.esajournals.orgrates were tested as co-variates in all models.Significant results were examined with LSMeans post-hoc test.Extremely high mussel densities were observed in several experimental plots, while no mussels were found in the majority of plots (64%).Due to overdispersion, ribbed mussel density was assessed using a GLM negative binomial distribution function.Mortality data were examined using a binomial distribution function.All results are presented as mean 6 standard error unless otherwise indicated.
Vegetation stem density differed significantly by area only.Specifically, mid-salinity sites (HB) had the highest density of vegetation which was greater than both the high salinity site and the low salinity site (F ¼ 13.03, p , 0.01; RL: 629.3 6 12.4, HB: 1146.2 6 60.9, 464.1 6 13.6).Species composition also varied by site.S. alterniflora was found at only 50% of TB sites, while occurring at 82% and 98% of the HB and RL sites.J. roemerianus occurred more frequently at HB and RL sites (TB: 5% of sites; RL: 25% of sites, HB: 42% of sites).TB sites were dominated by fresher mixes of species including Spartina patens, Schoenoplectus americanus and Distichilis spicata, with some Paspalum vaginatum, Batis maritima and Avicennia germinans.
Mean mussel size differed significantly by salinity with significantly greater sizes at the two higher salinity sites (RL, HB) as compared to the low salinity site (TB) (RL:53.6 6 3.1, HB: 42.5 6 3.6, TB: 11.76 2.9 mm; Fig. 5).Mean size varied by salinity and marsh zone with significant interaction (F ¼ 4.9, p ¼ 0.02).Covariates of dominant vegetation and flooding rates were not

Growth and mortality
There was a significant salinity by marsh zone interaction for mussel growth (F ¼ 13.7, p , 0.001; Fig. 6).Growth rates of mussels in edge plots at the two higher salinity sites were greater than in lower salinity edge plots (RL: 1.3 6 0.3 mm mo À1 ; HB: 1.4 6 0.3 mm mo À1 ; TB: 0.3 6 0.1 mm mo À1 ) and all interior plots (RL: 0.5 6 0.1 Fig. 5. Size class distribution (mm) as percentage of the mussel populations located along the salinity gradient (low salinity, TB; mid-salinity, HB; high salinity, RL).Fig. 6.Growth rate (mm mo À1 ) 6 standard error at edge and interior marsh sites located across the salinity gradient at low salinity (TB), mid-salinity (HB) and high salinity (RL).Different letters above the bars indicate significant differences in growth rates.v www.esajournals.orgmm mo À1 , HB: 0.5 6 0.1 mm mo À1 , TB: 0.25 6 0.07 mm mo À1 ; Fig. 6).

DISCUSSION
Ribbed mussel distribution in Barataria Bay extended from low (;salinity 4 psu) to high salinity (;salinity 15 psu) marsh with larger mussels and higher densities of mussels closely associated with mid-salinity marsh and dense vegetation stands dominated by J. roemerianus.
The peak in mussel densities and size at the midsalinity sites is likely explained by reduced recruitment and growth observed at low salinity sites, and greater predation mortality observed at the high salinity sites.These results support our original hypotheses, with the exception that higher densities of mussels were associated with J. roemerianus as opposed to S. alterniflora.Overall, the observed patterns of mussel population dynamics likely reflect detrital food resource availability and predation related mortality due to local site flooding rates and host vegetation community along the salinity gradient, as well as salinity tolerance of the ribbed mussel.Changes in salinity regimes and concomitant marsh vegetation communities may significantly affect the distribution and density of ribbed mussels in this region, and ultimately, their contribution to overall estuarine and marsh ecosystem services.
Ribbed mussels were found to be fairly ubiquitous across Barataria Bay, with densities across the salinity gradient similar to other mean densities reported for Gulf ribbed mussel populations.In this study, mean mussel densities among the three salinity regimes ranged from 3.9 6 0.4 (low salinity) to 66.6 6 16.3 ind m À2 (midsalinity).These densities are within the range reported in past studies of ribbed mussels in Barataria Bay (82 6 18 ind m À2 ; Spicer 2007) and Alabama (5 ind m À2 ; West and Williams 1986),  Lin 1991).These densities however are surprisingly low in comparison to densities of G. demissa reported from northern temperate marshes that ranged from a mean of 200 6 1 ind m À2 to a high of 9,227 6 731 ind m À2 (Fell et al. 1982, Nielsen and Franz 1995, Evgenidou and Valiela 2002, Culbertson at al. 2008).In Rhode Island, mussel densities ranged from 470-1412 ind m À2 , and were often found in beds 2-3 mussels deep, covering 90% of exposed surfaces in S. alterniflora salt marshes (Bertness 1984).
Ribbed mussel densities within estuarine subtropical marshes such as in Louisiana may be lower compared to the Atlantic east coast populations due to osmotic stress resulting from the large range in salinity and high variation in salinity.While ribbed mussels can withstand large variations in salinity (3-48; Bertness 1984), they occur in high salinity or ''salt marsh'' (18-30 psu) areas (Bertness andGrosholz 1985, Chintala et al. 2006).Only one study that we could determine based on site description data documented high densities (470-1412 ind m À2 ) in an area with salinity ranging from 10 to 15 psu (Bertness 1984).This salinity range covers the range between our mid and high salinity sites, but densities were considerably lower than the minimum of this range.Examining annual salinity variation and ranges may provide insight in comparing these two locations; Louisiana estuaries are marked by enormous variation and ranges in salinity.Other factors, such as flooding rates, or vegetation composition and density may explain some differences.In this study, flooding rates ranged from 15% to 45%, similar to the 20-49% range reported in Rhode Island with high densities of mussels (Chintala et al. 2006).
While salinity may impose physiological limits on mussel populations, vegetation stem density may provide valuable attachment substrate, and possible protection from predation.In previous studies along the Atlantic coast, vegetation stem density has been associated with increased mussel growth rates, reduced mortality and increased ribbed mussel population densities (Lin 1991, Chintala et al. 2006).Results from several studies suggest a strong mutualistic relationship between ribbed mussels and the dominant host vegetation (S. alterniflora) where mussels promote shoot growth and strengthen root mass through nitrogenous biodeposition while S. alterniflora provides mussels with stable v www.esajournals.organchoring substrates, detrital food resources, defense from predation, ultimately strengthening host soils (Lin 1989, Bertness and Leonard 1997, Watt et al. 2011).In this study, there was a clear positive relationship between mussel density and vegetation density; more specifically, mussel density was most strongly correlated with J. roemerianus, a co-dominant in this region with S. alterniflora.
As most studies have examined ribbed mussels where S. alterniflora is the dominant vegetation species, it is unclear if it is simply the structure provided by the stems, or the species that is important.Atlantic coast S. alterniflora stem densities in past studies ranged from 100 to over 2000 stems m À2 (Morris and Haskins 1990, Havens et al. 1995, Dai and Wiegert 1996, Altieri et al. 2007).Similar overall stem densities were quantified at the coastal Louisiana plots; however while S. alterniflora was present in the majority of plots (.90%), the highest stem densities and mussel densities were associated with a different species, J. roemerianus.In this region, stem densities are similar to those reported in the mid-Atlantic, but J. roemerianus stem densities are on average 4 times greater than those recorded for S. alterniflora (i.e., this study, Nyman et al. 1995, Lin andMendelssohn 2012).In Mexico, invasive ribbed mussels were found to be positively and significantly associated with their native cordgrass, Spartina foliosa (Torchin et al. 2005).Determining if other vegetative species, including J. roemerianus, may provide similar benefits and have similar mutualistic relationships would yield important insights into mussel ecology across vegetatively diverse marshes, and within different salinity zones, and is important as salinity and vegetation zones are altered in the Mississippi delta region.
Mussel growth rates were greater at the mid and high salinity marsh edge sites (RL, HB) than all interior and low salinity sites.The growth rates at the two higher salinity edge sites (1.3-1.5 mm mo À1 ; April-November) are similar to those found in the Atlantic ribbed mussel during their shorter growing seasons (1.5-20 mm yr À1 , Bertness and Grosholz 1985, Stiven and Gardner 1992, Culbertson et al. 2008, Hillard and Walters 2009).The growth rates in this study reflect areas with, on average, lower salinity but greater salinity variation, higher temperatures, different dominant host-vegetation, and a longer growing season than most reported in the literature for ribbed mussel species.These environmental differences may contribute to other population dynamic differences which explain differences in mussel populations between past studies and this gulf coast study.
Understanding both rates and causes of mortality of a species are critical in determining environmental limits and thresholds, and understanding overall population dynamics.In this study, the low salinity sites experienced over 75% mortality, most from environmental stress (salinity), and likely a combination of low flooding rates resulting in decreased access to detrital food resources (Jordan andValiela 1982, Stiven andGardner 1992).At the high salinity sites, predation mortality was highest, although still lower than that found by Lin (1990) who estimated over 50% predation mortality.This lower predation mortality overall may be due to dense host vegetation which was equally dense on the edge as the interior.In particular, root masses of J. roemerianus were difficult to break apart, and extract mussels while most S. alterniflora vegetation could be examined by hand suggesting J. roemerianus may be preferred as it offers better refuge overall.It seems likely that such dense vegetation may serve as spatial refuge from crab predation and may explain the high mussel J. roemerianus association found in this region.
Along with mortality, differences in recruitment are critical in controlling population densities; relative flooding rates, availability of substrate, and proximity to spawning populations are thought to control recruitment in many bivalve species.This study found highest recruitment at edge marsh, and in areas with higher mussel densities.Similar patterns of recruitment have been observed in salt marsh in the northeast U.S., where recruitment was greatest at marsh edge as compared to interior marsh plots (Nielsen and Franz 1995).Greater flooding at the marsh edge may increase larval access to recruitment substrates such as conspecifics and vegetation shoots by suspended mussel larvae (Bertness andGrosholz 1985, Nielsen andFranz 1995), while greater vegetation densities reduce tidal velocity and limit access to invertebrate predators, increasing larval settlement during slack high-water (Watt et al. 2011).
Equally important in recruitment is the presence of reproductively mature mussels.Ribbed mussels typically reach sexual maturity at 20 mm in length, after two growing seasons (Brousseau 1982), although this is highly dependent on relative temperature, salinity, flooding rates and food quality (Franz 1996, Honig et al. 2014).The low salinity site had less than 20% of the mussel population over 20 mm, and was located nearly 25 km up estuary from the mid and high salinity areas with higher mussel density, and populations with greater than 90% larger than 20 mm.Consequently, mussel populations contributing to larval supply may be concentrated towards higher salinity sites in Barataria Bay.As a result, recruitment was likely affected by larval transport within Barataria Bay.The most significant physical forces affecting fluid transport in the central areas of the bay, where ribbed mussel population densities were the greatest, were southerly winds occurring midsummer, when spawning likely takes place (Baumann 1987).
With relatively high recruitment and growth rates, and reduced mortality rates within mesohaline areas, it is likely that the gulf ribbed mussel contributes to marsh nutrient cycling and soil structure, but these services have yet to be quantified in this region.Mussel biodeposition may contribute to a mutualistic relationship with local vegetation where mussels promote vegetative growth and strengthen rootmass through byssal attachment while host cordgrass provides mussels with stable anchoring substrates, detrital food resources, shade protection, defense from predation and flow reduction, ultimately strengthening host soils.Ongoing restoration projects exploiting such environmental services conferred by ribbed mussel populations include the Delaware Living Shoreline Initiative, where stabilized fibrous bio-logs increase mussel recruitment within planted S. alterniflora clusters (Kreeger et al. 2011).Given the results of this study, restoration techniques exploiting ecosystem services of ribbed mussels may be most effective within mid-salinity sites, particularly in association with J. roemerianus.

CONCLUSIONS
The relative size distribution, growth rates and mortality rates of mussels across salinity, vegeta-tion and flooding regimes in Barataria Bay suggest that population densities may be primarily limited by bottom-up control in lowsalinity, high elevation, S. patens-dominated marsh sites with lower flooding rates and by top-down control in high-salinity, low elevation, J. roemerianus/S.alterniflora-dominated marsh sites with high predation mortality.Greater recruitment at high-salinity sites driven by relative proximity to dense adult spawning populations in lower Barataria Bay may reinforce existing differences in mussel densities in southeastern Louisiana salt marshes.Changes in salinity, marsh vegetation and marsh extent may impact the ribbed mussel population, and affect their contributions to ecosystem services.With many areas of deltaic Louisiana facing reduced salinity regimes due to proposed freshwater and sediment diversions, the density and distribution of these mussels may be significantly affected.However, with relatively high recruitment and growth rates, and reduced mortality rates within mesohaline areas, it is likely that the gulf ribbed mussel contributes to marsh nutrient cycling and soil structure, and could be important in helping maintain marsh integrity within these mesohaline areas; however these services have yet to be quantified in this region.Understanding population ecology of a native bivalve is critical to informing management on the effects of their activities on native populations, and ecosystem functioning.

Fig. 1 .
Fig. 1.Location of study sites in Barataria Bay, Louisiana, USA.Study sites were located along a salinity gradient from low salinity (annual salinity mean ;3 psu) at Turtle Bay (TB), to mid-salinity (annual salinity mean ;8 psu) at Hackberry Bay (HB) to high salinity (annual salinity mean ;16 psu) at Raccoon Lake (RL).

Fig. 3 .
Fig. 3. Mean mussel density (ind m À2 ) 6 standard error at edge and interior sites at low salinity (TB), midsalinity (HB) and high salinity (RL).Different letters above the bars indicate significant differences in mussel density.

Fig. 7 .
Fig. 7. Mortality (%) of mussels at edge and interior marsh sites located across the salinity gradient at low salinity (TB), mid-salinity (HB) and high salinity (RL).Mortality estimates from predation (crushed shells) versus environmental factors (no evidence of predation) are split out.Different letters above the bars indicate significant differences in mortality.

Fig. 8 .
Fig. 8. Monthly recruitment estimates (no.individuals pot À1 mo À1 ) 6 standard error at edge and interior marsh sites located across the salinity gradient at low salinity (TB), mid-salinity (HB) and high salinity (RL).