Effects of shoreline oiling on salt marsh epifaunal macroinvertebrates

The Deepwater Horizon oil spill resulted in varying degrees of oiling in the salt marshes of northern Barataria Bay, Louisiana, USA. This study examines the effects of oiling intensity and recovery on two conspicuous marsh-platform macroinvertebrates, Uca spp., fiddler crabs, and Littoraria irrorata, the salt marsh periwinkle, from 2.5 to 4.5 years after the spill. The dominant fiddler crab within these marshes, Uca longisignalis, was the only species observed in field collections, and no significant difference in burrow density or burrow size was found among oiling levels over the study period indicating recovery from any negative effects of oiling already occurred for this species. The highest density of L. irrorata was found at moderately oiled sites compared to both reference (without visible oiling) and heavily oiled stations. Spartina alterniflora density recovered within two years after the spill at the moderately oiled stations facilitating recovery of L. irrorata approximately one year later. L. irrorata average shell length and length-frequency distributions were equivalent at moderately oiled and reference stations but snails were shorter at heavily oiled stations because of a greater proportion of subadults. Shell length data from the heavily oiled sites indicate that direct mortality due to oiling or oil-induced reductions in recruitment occurred in 2010 and that recovery was starting to occur at 48 months after the spill. The extent and duration of oil in the water during the spill and the biological responses we measured indicates that L. irrorata and Uca longisignalis were both affected in their ability to carry out their life cycle on the marsh and/or in the water column at all stations including the reference stations for some period following the entry of oil into the region.


Effects of shoreline oiling on salt marsh epifaunal macroinvertebrates
Donald R Deis, John W Fleeger, Stefan M Bourgoin, Irving A Mendelssohn, Qianxin Lin, Aixin Hou The Deepwater Horizon oil spill resulted in varying degrees of oiling in the salt marshes of northern Barataria Bay, Louisiana, USA.This study examines the effects of oiling intensity and recovery on two conspicuous marsh-platform macroinvertebrates, Uca spp., fiddler crabs, and Littoraria irrorata, the salt marsh periwinkle, from 2.5 to 4.5 years after the spill.The dominant fiddler crab within these marshes, Uca longisignalis, was the only species observed in field collections, and no significant difference in burrow density or burrow size was found among oiling levels over the study period indicating recovery from any negative effects of oiling already occurred for this species.The highest density of L.
irrorata was found at moderately oiled sites compared to both reference (without visible oiling) and heavily oiled stations.Spartina alterniflora density recovered within two years after the spill at the moderately oiled stations facilitating recovery of L. irrorata approximately one year later.L. irrorata average shell length and length-frequency distributions were equivalent at moderately oiled and reference stations but snails were shorter at heavily oiled stations because of a greater proportion of subadults.Shell length data from the heavily oiled sites indicate that direct mortality due to oiling or oil-induced reductions in recruitment occurred in 2010 and that recovery was starting to occur at 48 months after the spill.The extent and duration of oil in the water during the spill and the biological responses we measured indicates that L. irrorata and Uca longisignalis were both affected in their ability to carry out their life cycle on the marsh and/or in the water column at all stations including the reference stations for some period following the entry of oil into the region.

Introduction
The release of a judge-ruled 3.19 million barrels of oil from the Deepwater Horizon (DWH) oil spill (Malakoff, 2015) exposed the nation's largest and most productive wetland-estuarine environment to an unprecedented potential for environmental damage.Oil spills can cause widespread impacts to the structure, function, resilience, and sustainability of coastal wetlands depending upon the oil type, extent of contamination of the vegetation and marsh soils, exposure to waves and currents, time of year of the spill, and species sensitivity to oiling (Michel & Rutherford, 2014).Although the oil that made landfall in the DWH oil spill was relatively "weathered" and consisted of emulsions of crude oil depleted of its more volatile and toxic components, the spill resulted in the oiling of 796 km of coastal marsh shoreline, as documented by Shoreline Cleanup Assessment Technique (SCAT) teams (Michel et al., 2013).Of that total, approximately 135 km were described as heavy and 165 km as moderate marsh oiling (Michel et al., 2013).
Approximately 95% of the total marsh oiling occurred in coastal Louisiana, and the heaviest marsh oiling was most widespread in northern Barataria Bay marshes dominated by Spartina alterniflora and to a lesser extent Juncus roemerianus (Zengel et al. 2014, Michel et al. 2013).The plant communities in northern Barataria Bay experienced strong responses to oiling, including heavy plant mortality that denuded shorelines (Lin and Mendelssohn 2012;Silliman et al. 2012;Zengel and Michel 2013;Zengel et al. 2014;Zengel et al., 2015).
The marshes of the Mississippi River delta system provide a suite of environmentally and economically important services including marsh platform elevation, the prevention of soil erosion, soil nutrient cycling, food web support both grazing and detrital, and many others (McCall & Pennings, 2012;Silliman et al., 2012).Most, if not all, of these ecosystem services depend upon a healthy, functioning plant-microbial-benthic system.Oil spills can exacerbate wetland loss by destabilizing these interrelated processes that control the capacity of coastal wetlands to function properly.Initial impacts of the DWH oil spill on coastal wetlands have been and are being reported (e.g., Whitehead, 2011, Lin and Mendelssohn, 2012, McCall & Penning, 2012, Silliman et al., 2012).This study is a part of a program documenting the longer term (currently 4 years) impacts and recovery of the oiled marsh systems and their potential for sustainability.
Other members of the study team are analyzing and providing data on plant parameters, soil parameters, soil bacteria, benthic microalgae, and benthic meiofauna (Lin et al., unpublished data;Fleeger et al., 2015).This and other studies have shown that different marsh plant species respond differently to oiling (Alexander & Webb, 1987;Baca et al., 1987: Mendelssohn et al., 1990, 2012;Hoff et al., 1993;Lin & Mendelssohn, 1996, 1998, 2008, 2009;Hester & Mendelssohn, 2000;Pezeshki et al., 2000;Lin et al., 2002;DeLaune et al., 2003;Culbertson et al., 2008;Michel & Rutherford, 2014), demonstrating impacts to coastal wetland vegetation, e.g.reduced plant photosynthesis, transpiration, shoot height, stem density, and above and below ground biomass.These impacts on vegetation and the marsh surface can result in cascading effects on soil bacteria and macro-, meio-, and micro-faunal communities that inhabit wetlands and depend on the plant organic matter as carbon sources or as foundation species.For example, benthic microalgae and meiofauna are, both individually and when monitored together, important indicators of ecosystem function regarding food-web support and a broad range of effects and recovery from oil spills (Fleeger et al., 2015).
Our sampling focused on two conspicuous and foundational marsh macroinvertebrates, Uca spp., marsh fiddler crabs (Montague, 1980), and Littoraria irrorata, the salt marsh periwinkle (Silliman & Zieman, 2001).Fiddler crabs are one of the most thoroughly-studied shore crab in North America, with robust literature available that examines individual species population dynamics, life history and ecology (Grimes et al., 1989).Fiddler crabs greatly influence the marsh through burrowing and feeding activities, in effect enhancing effects on vegetation productivity and biomass, altering sediment and nutrient characteristics, altering biogeochemical cycles, altering microbial processes by aerating the marsh sediment, increasing soil drainage, and facilitating nutrient transport.Generally, the presence of fiddler crabs has been noted as indicating greater diversity of other marsh organisms, and crab population densities can reflect the productivity of a wetland (Montague, 1980;Mouton & Felder, 1996).Fiddler crabs have been shown to be sensitive to oil spills, making them a valuable environmental indicator species (Burger et al. 1991(Burger et al. , 1992;;Burger & Gochfeld, 1992).
Various fiddler crab species inhabit the northern Gulf of Mexico coast.Two species in particular are found in the Louisiana marshes, Uca spinicarpa and U. longisignalis.Uca spinicarpa prefers claydominated substrates in brackish marshes ranging from nearly fresh to hypersaline.Uca longisignalis is restricted to sediments of terrigeneous origin ((i.e.mucky soils) and found primarily in lower salinity environments (upper estuaries).The two species can be found in close proximity but have preferred habitats based on elevation, vegetation and sediment character (Mouton & Felder, 1996;Zengel et al., 2014).Although there may be more than one species of fiddler crab, we will refer to them collectively as Uca except when we refer to a specific species.L. irrorata also is an indicator species of the health of the salt marsh habitat (Silliman & Zieman, 2001).In areas dominated by short-to-intermediate form Spartina alterniflora, the species has been noted to occur in densities of at least 100 individuals/m 2 (Silliman & Zieman 2001;Stagg & Mendelssohn 2012).L. irrorata is a rasping detritivore/herbivore specialist, feeding on organic matter on the marsh surface during low tide and ascending the Spartina stems to feed upon standing-dead Spartina and its associated microbial assemblages as the tide rises (Silliman & Zieman 2001).As a detritivore, L. irrorata influences nutrient dynamics by expediting the decomposition of Spartina alterniflora and serves as an important link between primary and secondary production (Stagg & Mendelssohn 2012).The presence of S. alterniflora has been directly linked to increased abundance, growth and survival of L. irrorata in marshes (Kiehn & Morris 2009;Stagg & Mendelssohn 2012).Fiddler crabs (Krebs & Burns, 1977;Burns & Teal 1979;Burger et al., 1991;Teal et al., 1992;Culbertson et al., 2007;Morris et al. 2015) and the marsh periwinkle (Hershner & Moore 1977;Hershner & Lake, 1980;Lee et al., 1981;Bennett et al., 1999;Garner et al. 2015)  study, when they release fertilized eggs into the water and expect return of larvae to the marsh shoreline.
Recovery time following oil-induced impacts for these organisms can vary substantially based on a variety of oiling and habitat conditions, from a year to several decades (Culbertson et al., 2007).
In this study, we present the results of an investigation of the effects of the DWH oil spill on fiddler crab and marsh periwinkle at different oiling levels between 30 and 54 months after the spill.As has been indicated, Lin et al. (unpublished data) studied the response of the plant community including the dominant salt marsh plants S. alterniflora and J. roemerianus in the same sampling stations as used for this investigation.

Methods
The 21 shoreline sampling stations were established in 2011 within an approximate 8 km by 5 km area in Wilkinson Bay and Bay Jimmy in northern Barataria Bay, Louisiana, USA (between coordinates N 29.44060 -29.47459,W 89.88492 -89.94647) that had been impacted to varying degrees by the DWH oil spill (Fig. 1).Sampling stations were randomly selected using SCAT data and our own field observations, and included seven locations each that received either no oiling (i.e., reference) (RF), moderate oiling (MD), or heavy oiling (HV).Sampling occurred at various occasions (generally bi-annually) from 30 months (November 2012) after the spill to 54 months (October 2014) after the spill.Sampling was generally in the spring (April -June) and fall months (September -November).Total petroleum hydrocarbon and plant community data were collected at the time of sampling (Lin et al., unpublished data).

Total Petroleum Hydrocarbon (TPH) analysis
The 0−2 cm surface soils were collected from each station on each collection date, transported to the Louisiana State University laboratory on ice, extracted with dichloromethane (DCM), and analyzed gravimetrically (Lin & Mendelssohn, 2012).DCM extracts were transferred to pre-weighed dishes, where the DCM was evaporated, and the unevaporated oil remaining in the dishes was weighed to the nearest 0.0001 g.TPH concentration was calculated and expressed as mg g −1 dry soil.The results of the analysis for 30 to 54 months after the spill are provided in Figure 8.

Aboveground Biomass and Stem Density
Plant aboveground biomass and stem density were taken at each station during each sampling within a haphazardly located 0.25 m 2 quadrat.All plants rooted within the quadrat were then clipped to the ground surface and separated into live and dead components by species.Stem density was determined by counting the number of intact living stems for each dominant plant species.All aboveground biomass was then dried to a constant mass at 60°C and weighed.Parameters reported included total live and dead above ground biomass and above ground biomass and stem density for each of the dominant species (Spartina alterniflora and Juncus roemerianus).

Macroinvertebrate sampling
Sampling included fiddler crab (Uca) burrow density and size and marsh periwinkle (Littoraria irrorata) density and size for both components of the study.Sampling was conducted using three 0.25 m 2 quadrats placed at random approximately 1 m inward from the marsh edge.Data was converted to m -2 basis for analysis.Each quadrat was sampled for both L. irrorata abundance and Uca burrow abundance.
L. irrorata were found on the marsh surface or attached to the vegetation within the plot; juvenile L. irrorata were often found hiding in the leaf bracts of Spartina alterniflora.L. irrorata shell length and Uca burrow size (diameter) were measured to the nearest millimeter using calipers and a transparent ruler, respectively.Shell lengths were taken as a total height, a measurement from the base of the aperture to the top point of the shell.Once measured, L. irrorata were returned to the marsh within their specific sampled quadrat.As fiddler crabs are mobile and difficult to quantify within a certain area, their burrows were used as a proxy for their abundance on the marsh.We were not able to measure crab burrows at two sampling occasions (36 and 42 months after the spill) because water was present on the marsh platform, covering the marsh sediment in the sampling area.Photos were taken of the general site setting and each quadrat, and general notes were made concerning site flora and fauna composition and characteristics.Uca adults were identified to species level and released if found in the sampling site vicinity.Measured L. irrorata were grouped into putative age classes based on their size: 0-6 mm were classified as juveniles, 7-13 mm were classified as sub-adults, and 14+ mm were classified as adults (Hamilton, 1978;Stagg & Mendelssohn, 2012).Sizes were examined as a percentage of the total abundance for each site type (RF, MD, and HV; Figs.5-7).Size frequency distributions of L. irrorata are often bi-or tri-modal with the modes representing age class (Hamilton, 1978;Zengel et al., 2014).The largest snails in most populations range from 20 -26 mm in shell length (Emerson and Jacobson, 1976;Morris, 1975), though the largest reported size is 32 mm (Kaplan, 1988).Based on growth estimates (Stiven and Hunter, 1976), juveniles would be <1 year of age, subadults would average about 1 year of age, and adults would be 2 years and older in age.These designations were used in analyses to examine life stage-specific parameters of L. irrorata abundance and distribution throughout the study.Evaluating size distribution by comparing size frequency histograms over time is helpful in determining how populations were impacted and tracing recovery trajectories (Zengel et al., 2014;Zengel et al., 2015;Pennings et al., in review).

Statistical Analysis
All statistical analyses were conducted using SAS (Statistical Analysis Systems, version 9.2, SAS Institute, Cary, NC).L. irrorata abundance data were tested for normality and found to be skewed [Shapiro-Wilk Test (alpha = 0.05)].Several transformations were tested and the natural log transformation was the best fit; therefore, all L. irrorata abundance data were transformed as the natural log prior to analysis.We used repeated measures, two-way mixed-model Analysis of Variance (ANOVA) with oiling level, sampling period, and their interactions to test for variation in L. irrorata abundance.We used Mauchly's Test of Sphericity to examine the form of the common covariance on the assumptions of the repeated measures ANOVA and found that the data do not meet sphericity assumptions (Pr>ChiSq = 0.0057).An adjusted Tukey's test was used to determine differences between L. irrorata abundances and oiling level at each sampling period.We tested the L. irrorata shell size data for normality and found these data to be skewed [Shapiro-Wilk Test (alpha = 0.05)].All attempted transformations (log, natural log, square root) were unsuccessful.
Kruskal-Wallis test was therefore used to examine shell size by oiling level and sampling period, e.g.
comparing shell size for the RF sites by sampling period.The nonparametric Kolmogorov-Smirnov two-sample test was used to compare the total abundance (total, adult, and subadult) and average shell length data between oiling levels across all sampling periods with the null hypothesis that the two oiling levels have the same distribution.
Uca burrow abundance data were tested for normality and found to be skewed [Shapiro-Wilk Test (alpha = 0.05)].Transformations and removal of the 36 and 40 month data sets which were mainly zeros (periods of high water) were unable to normalize data.Oiling levels and abundance of burrows were therefore tested using Kruskal-Wallis/Mann-Whitney (non-parametric) tests, both leaving in the 36 and 40 month datasets and removing those datasets from the analysis; however, no significant results were found.
Correlations were performed, using Kendall's Tau non-parametric rank correlations with dependence on variables, with oiling level and sampling period to 48 months after the spill because of the lack of availability of plant and TPH data at the time of the analysis; L. irrorata abundance; Uca burrow density; S. alterniflora stem density and biomass; J. roemerianus stem density and biomass; total live biomass; total dead biomass; and TPH.Significant difference was defined as p ≤ 0.05.P-values are reported to 2 decimal places and, in cases where p = 0.00, the p-value is reported as p < 0.01.All error terms are expressed in standard error (SE).Raw data can be found at the following link: doi:10.7266/N7FF3Q9S.

Littoraria irrorata abundance
L. irrorata abundance varied among oiling levels when data were combined across all sampling periods and size classes.L. irrorata were found to have the highest average density at MD (mean = 92.5 ind m -2 , n=3,888) compared to RF (mean = 38.1 ind m -2 , n =1604, Kolmogorov-Smirnov test, p < 0.01) and HV sites (mean = 33.5 ind m -2 , n = 1404, Kolmogorov-Smirnov test, p < 0.01).RF sites were also found to be statistically different than HV sites (Kolmogorov-Smirnov test, p = 0.02).This same trend was found within adults as both RF and MD differed from HV sites (Kolmogorov-Smirnov test, p = 0.02 and <0.01, respectively, Fig. 2).No difference was found among oiling levels for subadults.Juveniles were cryptic and inconsistent in their distribution, and there was no significant difference between oiling levels and abundance of juveniles.L. irrorata abundance also exhibited differences between size classes at all of the oiling levels.Adults are significantly more abundant than both sub-adults (Kolmogorov-Smirnov test, p < 0.01) and juveniles (Kolmogorov-Smirnov test, p < 0.01), while sub-adults are significantly more abundant than juveniles (Kolmogorov-Smirnov test, p < 0.01).
When sampling period and oiling level were analyzed together with ANOVA, oiling level had a significant effect on L. irrorata total abundance (n = 7 and p < 0.01), but the sampling period and interaction effect were not significant (Fig. 3).We examined within-subjects main effects tests to examine changes in snail abundance over time with the null hypothesis that mean abundance did not change over time.We rejected the null hypothesis using Wilk' test (Pr>F = 0.0139) and concluded that abundance varied over time.We used the same method to test the interaction between time and oiling level.We again rejected the null hypothesis and concluded that change in mean abundance across time depended on oiling level (Pr>F = 0.0037).Significant differences in L. irrorata total density were noted between oiling levels at different sampling periods.Significant differences existed between both MD (Tukey's test, p = 0.02) and RF (Tukey's test, p < 0.01) and HV sites at 36 months after the spill; between the MD and RF sites (Tukey's test, p = 0.05) at 40 months after the spill; and MD and RF (Tukey's test, p = 0.05) and HV sites at 36 months after the spill.

Littoraria irrorata shell size
L. irrorata body size also varied among oiling levels when data were combined across all sampling periods.L. irrorata average shell length was significantly less at HV sites (mean = 15.44 mm) when compared to both MD sites (mean = 17.37,Kolmogorov-Smirnov test, p < 0.01; Fig. 4) and RF sites (mean = 16.62 mm, Kolmogorov-Smirnov test, p < 0.01).RF and MD sites showed no statistical difference in shell length (Kolmogorov-Smirnov test, p = 0.43).L. irrorata shell length was also used to examine size-frequency distributions at all oiling levels.RF and MD sites exhibited similar proportions of juveniles, subadults, and adults, while the HV sites contained a lower percentage of adults, in particular, large adults.This low proportion of adults at the HV sites likely corresponded to the direct mortality of all size categories of L. irrorata at the HV sites due to the oiling in 2010.As indicated in methods, adults are two years or older; therefore, even at 30 months, smaller adults would just be entering the populations and replacing these losses at the HV sites.
The null hypothesis of no difference in shell length across sampling periods was rejected (Kruskal-Wallis test, p < 0.01) indicating that there was a difference in shell size among oiling levels (figure 8).
Periwinkles in all oiling categories averaged less than 15 mm in shell size 30 months after the spill.The HV sites continued to average less than 15 mm in shell size up to 48 months after the spill; whereas, the RF and MD sites achieve an average shell size greater than 17 mm at 48 months after the spill.

Uca burrow density and size
All of the fiddler crabs collected during all sampling periods were identified as Uca longisignalis.
No significant difference was found between oiling level and Uca burrow density.Because no significant difference was found with density, burrow size was not tested.

Correlations
With all oiling levels and all stations, L. irrorata abundance was positively correlated with most of the plant parameters including S. alterniflora above ground biomass (p = 0.05); J. roemerianus above ground biomass (p < 0.01) and stem density (p < 0.01); and total live (p < 0.01) and dead above ground biomass (p = 0.06).Uca burrow density was negatively correlated with S. alterniflora stem density (p = 0.03).
For the RF sites, the only significant correlation occurred at 30 months after the spill between L.
irrorata abundance and Uca burrow density (p = 0.04).The MD sites showed a significant positive correlation between L. irrorata abundance and J. roemerianus above ground biomass (p = 0.02) and stem density (p = 0.03) at 40 months after the spill; a significant negative correlation between L. irrorata and Uca burrow density (p = 0.01) 42 months after the spill; and a significant negative correlation with S.
alterniflora stem density (p = 0.03) and positive correlation with J. roemerianus stem density (p = 0.02) at 48 months after the spill.The HV sites showed no significant correlations.TPH did not correlate with either L. irrorata or Uca abundance; however, there was a positive significant correlation between TPH and S. alterniflora stem density (p = 0.04, all data; p < 0.01, HV sites).
There were significant negative correlations between J. roemerianus above ground biomass (p < 0.01) and stem density (p < 0.01) and total live and dead biomass (both, p < 0.01).

Littoraria irrorata
As stated in the Results, L. irrorata were found to have the highest average density at MD compared to RF and HV sites.RF sites were also found to be statistically different than HV sites.This same trend was found within adults as both RF and MD differed from HV sites.The reduced abundance at HV compared to MD was due to the oiling, either by direct mortality or effects on recruitment by the loss of vegetation at the HV sites.Many studies (Kiehn & Morris, 2009;Silliman & Zieman 2001;Silliman & Bertness 2002;Silliman & Newell, 2003;Silliman et al., 2005. Kiehn & Morris (2009) have noted that marsh periwinkle density is positively correlated with S. alterniflora stem density.Stagg & Mendelssohn (2012), in a study of marshes restored using sediment from dredging operations, found that L. irrorata growth, survival, and productivity were positively correlated to increasing S. alterniflora canopy cover.High levels of J. roemerianus, however, may be detrimental to L. irrorata density.Alber et al. (2008) noted that densities of L. irrorata are generally low on J. roemerianus in their analysis of the possible reasons of saltmarsh dieback.There may be a limit to the amount of J. roemerianus (biomass or stem density) within the marsh that is beneficial compared to the amount that suppresses the L. irrorata population density.
Above ground biomass and stem density of J. roemerianus was equal to or greater at times than S. alterniflora density at the RF stations (Lin et al., unpublished data), potentially limiting the abundance of L. irrorata.J. roemerianus has been shown to be more sensitive to oiling than S. alterniflora (Lin et al., unpublished data, Lin & Mendelssohn, 2012), and its density and biomass was expectedly suppressed at the moderately and HV stations (Lin et al., unpublished data).
Location may have also contributed to lower L. irrorata density at the RF sites.All of the RF sites are located northwest of the moderately and HV stations in a cove that is connected to larger marsh islands.This location could have a larger predator population than the smaller marsh islands where the other stations are located.The location is also further north into the estuary and may be influenced by freshwater, potentially altering the recruitment of L. irrorata and/or the population dynamics of J.
roemerianus and S. alterniflora (Pennings et al., 2005).Alternatively, abundances in the MD sites may have been higher than the RF sites because of an elevated stem density of S. alterniflora (Fleeger et al., 2015).Increased stem density at the MD sites may have been a response by S. alterniflora to reduced competition from J. roemerianus or a compensatory response associated with oiling.
Although the density of L. irrorata was low at RF sites, length-frequency data are similar to those found in other gulf coast areas not impacted by oiling (Pennings et al., in review), suggesting that these  Pennings (2012) reported no difference in the population density at 4 or 16 months after the spill.A similar comparison of the average shell length (Fig. 4) between the RF and MD stations and the HV stations indicates that body size was much lower at the HV stations and that they did not fully recover within the time period of the study.Shell length also indicates that all of the oiling levels may have initially been affected by oiling within Barataria Bay because the average shell length was initially smaller than that which was eventually achieved at the RF sites (17.71 mm).These data indicate that recovery may interesting result was that the reference station had less than 50 m -2 L. irrorata in 2011 (17 months after the spill) and greater than 150 m -2 in 2012 (29 months after the spill).Zengel et al. (2015) mentions two potential reasons for the differences in densities between the sampling periods; one, under-sampling the juvenile L. irrorata in the first sampling period due to not searching leaf bracts for juveniles and, two, the possible effect of widespread oiling in the water surface within Barataria Bay during the summer of 2010 and the effect that may have had on the larvae of L. irrorata.We find the juvenile L. irrorata very patchy in their distribution on the marsh platform because of the juveniles' preference, described in Zengel et al. (2015) to aggregate and hide in areas on S. alterniflora, such as within leaf bracts.We also find the juveniles sporadic between years potentially because of their broadcast life cycle and we do mention below that both L. irrorata and Uca were affected in their ability to carry out their broadcast life cycle in 2010 throughout northern Barataria Bay because of oil on the water.This caused a potential recovery effect at all of our stations including the RF stations.
Additionally, Fleeger et al. (2015) found that the infauna at our study sites followed the recovery of S.
alterniflora.Of particular interest is the correlation of L. irrorata with J. roemerianus parameters (above ground biomass, stem density, and total live and dead above ground biomass).This may indicate the importance of J. roemerianus to the recovery of L. irrorata within the marsh.Hughes (2012) describes the potential importance of J. roemerianus as alternative refuge for L. irrorata; as described above, however, Alber et al. (2008) note that the densities of L. irrorata on J. roemerianus is generally low.We, therefore, may be underestimating the importance of J. roemerianus within a diverse marsh system.The correlation of L. irrorata and J. roemerianus at 40 months (above ground biomass and stem density) and 48 months (stem density) at the MD stations, however, may indicate the coincidental recovery of the two species with the reduction of oil at those stations (Figure 9).

Uca
This study started 30 months after the DWH oil spill, and our data clearly show that that any effect of the spill on fiddler crab, Uca, density passed prior to the beginning of this effort.We found that average density of crab burrows within the area of our sampling (approximately 1 m from the shoreline edge) was low, between 5 and 10 burrows m -2 .This is not surprising, as Mouton & Felder (1996) noted the variation in burrow density of U. longisignalis along a 15-m transect from the edge into the marsh.The burrow density was lowest near the marsh edge and highest in the middle and upper reaches of the transect with no marked change in elevation noted beyond 3 m from the water's edge.Several studies have found similar results.Silliman et al. (2012) sampled Uca sp.crab burrow density at the three reference and three impacted sites described above and found no difference between the density of crab burrows (approximately 10 m -2 ) at the two site types.McCall & Pennings (2012) sampled Uca burrows at the sampling sites described above.Because they were sampling further back in the marsh, McCall & Pennings (2012) reported greater numbers of burrows (approximately 20 m -2 , see Mouton & Felder 1996 above) and found a significantly lower density of burrows at the oiled sites 4 months after the spill and no difference at 16 months after the spill.Zengel et al. (2015) monitored Uca burrow densities in relation to shoreline treatment options after the DWH spill (see above).Uca burrow densities in heavily-oiled and treated plots were similar to reference at both 17 and 29 months; however, the heavily-oiled sites that were not treated were significantly different at 17 months, but similar at 29 months.Zengel et al. (2014) revisited the treatment sites at 41 months after the spill (September 2013) and found significant differences between the oiled control sites (no treatment) relative to the references sites.Maximum average burrow density was approximately 10-15 burrows m -2 similar to densities in our study.We also identified only Uca longisignalis within our study sites.The study at the treatment sites (Zengel et al., 2014;Zengel et al., 2015) identified U. spinicarpa occurring at the heavily-oiled plots where there was reduced vegetation coupled with areas of surface oil residue overlaid with thin algal mats and clay-like sediments.Morris et al. (2015) conducted studies on adult male and female Uca longisignalis placing the individuals in Total Polycyclic Aromatic Hydrocarbon (TPAH)50 concentrations from 0.07 (reference sediment) to 26 mg/kg in the upper 2 cm of sediment.During the 10 day exposure period several females became gravid.The gravid females were removed after 10 days and placed in clean water until the embryos hatched.The zoea were collected, held in clean water, and exposed to varying levels of ambient sunlight.They found substantial toxicity (calculated LC20 value of 0.62 mg/kg TPAH50 in the upper 2 cm of sediment) to Uca zoea at relatively low concentrations of oil in the sediments.Fig. 9 provides the total petroleum hydrocarbon concentrations from 30 to 54 months after the spill at the 21 stations from this study.It is difficult to compare the two hydrocarbon parameters and much of the aromatic portions of the oil have weathered reducing the toxicity; however, the levels were found to be significant higher at the HV sites compared to the RF and MD sites (Lin et al. unpublished data).
The negative correlations between Uca sp.burrow density and S. alterniflora stem density (all oiling levels /all stations) and L. irrorata (MD stations 42 months after the spill) may be indicative of the conflict between the burrowing and feeding activity of the fiddler crab in and on the marsh and the density of S. alterniflora.Mouton & Felder (1996)  were affected in their ability to carry out their life cycle on the marsh (Morris et al. 2015) and in the water column at all stations including the reference stations for some period following the entry of oil into the region (Fodrie et al., 2014;Pennings et al., unpublished data).This study found that the dominant fiddler crab within these marshes, U. longisignalis, recovered at all oil levels prior to sampling at 30 months after the DWH spill.There are indications that L. irrorata started recovery at the MD stations at 40 months after the spill and population density is beginning to demonstrate seasonal cycling.Recovery, however, has not occurred at the HV stations.
data adequately represent a true reference population for comparison with oiled sites.Average L. irrorata shell lengths (Fig. 4) of L. irrorata population at the MD and RF stations were equivalent indicating that recovery at MD sites was achieved within the time period of this study.As an example of recovery of lightly oiled areas, McCall and Pennings (2012) estimated L. irrorata density, not shell length frequency, at oiled and control sites in Louisiana and Mississippi four and 16 months (August 2010 and 2011) after the DWH spill.The sampling sites were located 1-2 m behind the HV zone in areas dominated by Spartina alterniflora, and often partially coated with oil, with a very light (sheen) on the soil surface.McCall and have been occurring at the RF and MD sites approximately 36 months after the spill and, at the HV sites, 48 months after the spill.Other studies have investigated more heavily-oiled sites associated with the DWH spill using population density.Silliman et al. (2012) sampled L. irrorata density at three reference and three impacted sites, approximately 3 m from the shoreline, in Barataria Bay approximately 6 months (October 2012) after the DWH spill.Their group found no live individuals at impacted stations and approximately 50 m -2 at the reference stations.Note that their reference stations had relatively low densities (slightly greater than the density at our RF stations) and less than the average density of our MD stations.In a study of shoreline treatment options on a marsh island in northern Barataria Bay, approximately 1 km south of our study sites,Zengel et al. (2015) monitored L. irrorata population densities and shell size frequencies in relation to shoreline treatment options (manual, mechanical, and no treatment; see Zengel and Michel (2013) 17 and 29 months (September 2011 and 2012) after the DWH spill.This study provides data at reference and heavily-oiled stations between the time period of the Silliman et al. (2012) and our study.They found minimal recovery of L. irrorata at treatment sites.An indicate that U. longisignalis burrow density increased landward of the 4-m into the marsh where the vegetation became more open and scattered.They report the maximum density of burrows near the 8-m to 12-m marks into the marsh suggesting a potential relationship between tidal flooding and the deposition of wrack contributing to food resources.The extent of shoreline oiling within northern Barataria Bay (Zengel & Michel 2013) and the extent and duration of oil in and on the water indicated in the Environmental Response Management Application (ERMA) (http://gomex.erma.noaa.gov/erma)would indicate that the populations of L. irrorata and Uca

Fig. 5 .
Fig. 5. Reference (RF) sites Littoraria size distribution by percentage of total abundance.X-axis = shell length in mm, Y-axis = percentage of total abundance represented by each specific shell length across all sampling times.

Fig. 6 .
Fig. 6.Moderately-oiled (MD) sites Littoraria size distribution by percentage of total abundance.X-axis = shell length in mm, Y-axis = percentage of total abundance represented by each specific shell length across all sampling times.

Fig. 7 .
Fig. 7. Heavily-oiled (HV) sites Littoraria size distribution as percentage of total abundance.X-axis = shell length in mm, Y-axis = percentage of total abundance represented by each specific shell length across all sampling times.