Overview:
Seagrass mapping data from a multitude of previous projects in the Mississippi and Chandeleur Sounds were gathered and combined to provide information on seagrass change from 1940 to 2011. Seagrasses generally occur in three groups: (1) along the Mississippi (MS) mainland coastline dominated by Ruppia maritima, (2) on the north side of MS Sound barrier islands dominated by Halodule wrightii, and (3) on the west side of the Chandeleur Islands dominated by Thalassia testudinum co-occurring with other seagrass species. The study area generally lost seagrasses over the 71-year period, ostensibly due to loss or reduction of protective island barriers, and reductions in water quality. An example of how the time series of maps generated in this project can be utilized to further investigate seagrass change was demonstrated with data from Horn Island, including problems associated with calculating change in seagrass area using data from previous investigations. Comparisons of seagrass area among various studies that used different mapping methods (seagrass extent vs. seagrass coverage vs. vegetated seagrass area) can result in overestimation of area change and misleading conclusions.
Seagrass Loss:
The loss and disappearance of seagrasses have been documented from local and regional studies to worldwide assessments. Waycott et al. (2009) estimated that globally, seagrass decline accelerated from a pre-1940 median rate of 0.9% per year to 7% per year since 1990. Kemp et al. (1983) and Orth and Moore (1983) reported that declines in submerged aquatic vegetation, including Zostera marina, in Chesapeake Bay started from the 1960s, and then accelerated during the 1970s to present. Robblee et al. (1991) and Thayer et al. (1994) documented the die-off of T. testudinum and plant community changes in Florida Bay seagrass meadows during the late 1980s. Duke and Kruczynski (1992) reported 20% to 100% seagrass losses in areas of the northern GOM over the past 50 years. Fifteen years later, Handley et al. (2007) reported that loss of seagrasses in Texas, Louisiana, Mississippi, Alabama, and Florida was much greater than gains during 1940-2002. Although seagrass loss in many parts of the Gulf of Mexico has occurred since 1940, there are areas where seagrass loss has not been as dramatic and increases in cover have actually been recorded. Lewis et al. (2008) reported a less severe decline in seagrass area in 1980-2003 than in the preceding 1960-1980 time period for the Pensacola Bay system, with some increases observed in Pensacola Bay and Santa Rosa Bay from 1992 to 2003. In Mobile Bay and adjacent waters, Vittor and Associates, Inc. (2004, 2005, and 2009) found substantially reduced seagrass area in 2002 compared to 1940, 1955, and 1966. According to their study, an increase in seagrass area found from 2002 to 2008, or from 2002 to 2009 in some of the quadrangles in the northeastern Mississippi Sound, must be considered together with a change in species composition, from monospecific beds of H. wrightii in 2002 to mixed beds of H. wrighitii and R. maritima (an opportunistic species) in 2008-2009. Overall, this compendium of results suggests that a drastic decline in seagrasses at global, regional and local scales has occurred from the mid-20th century to the present.
This study presents an updated and comprehensive evaluation of decadal-scale changes to the seagrass area found in the Mississippi and Chandeleur Sounds. In the state of Mississippi, the MS Sound is the primary body of water that supports seagrasses, while the Chandeleur Islands support the most important seagrass resource in Louisiana. Specific objectives were to: (1) Combine all available data that were acquired in both Sounds from 1940 to 2011 to develop maps showing the general distribution of seagrasses over time, and (2) Determine interannual and decadal-scale changes in seagrass area for the three groups (1) MS mainland, (2) MS Sound barrier islands, and (3) Chandeleur islands within the Mississippi and Chandeleur Sounds. Study sites were placed into three groups representing an inshore to offshore gradient: Group 1 is the MS mainland coastline, specifically the GNDNERR and Waveland locations dominated by R. maritima; Group 2 are the MS barrier islands (Cat, West and East Ship, Horn, and Petit Bois Islands) dominated by H. wrightii; and Group 3 are the northern Chandeleur Islands (NCI) dominated by T. testudinum co-occurring with other seagrass species (Fig. 1).
Figure 1: Study area, which includes the MS mainland coastline (Grp 1), the five barrier islands in the MS Sound (Grp 2), and the northern Chandeleur Islands (Grp 3).
Seagrass Mapping Data:
An overview of the data sources used in the development of the seagrass distribution maps is provided in Table 1. A map of seagrass distribution in the study area was created in ArcGIS ver. 9.3 by drawing outline boundaries of all historic and recent seagrass extents and including points of occurrence where polygon data were not available.
TABLE 1. Data sources obtained for the seagrass maps listed by location and year. Spatial resolution is in meter square and NA indicates a previously completed mapping product (paper or digital) with unknown pixel resolution. Source of data includes A = aerial imagery, G = GPS ground-truthed, and T = transects with points.
There are different definitions of seagrass area that have been used prior to this paper, which can result in confusion and difficulties in calculating change over time. Here we define three terms used to describe seagrass area in this study: (1) Seagrass extent is the area within the outline encompassing all seagrass on a map or in a field survey, (2) Seagrass coverage is based on the area of polygons that each contain multiple seagrass patches (ranging from sparse to dense patch aggregation), (3) Vegetated seagrass area (VSA) is the sum of the area of all the seagrass patches. It is clear that VSA is a subset of seagrass coverage, which in turn is a subset of seagrass extent, and that newer mapping methods have made the increase in spatial accuracy obtained by the VSA method possible. Finally, the term “potential seagrass habitat (PSGH)” was defined by Moncreiff et al. (1998) as the area of everything shallower than the 2 m depth contour, and is generally a larger area than the seagrass extent. The “true” seagrass area is likely to be greater than the vegetated seagrass area but less than the area obtained from seagrass coverage calculations, and substantially lower than either seagrass extent or PSGH.
Decadal Changes in Seagrass Area:
More data are available during 1990 – 2010 than in previous decades, a reflection of advances in geospatial technology (remote sensing and GIS), which promote increasingly accurate natural resource mapping. Horn and Petit Bois Islands have the most consistently available information, only data for either the 1970s or 1980s are missing (Table 2). Most other sites do not have information prior to 1969, and this study includes no mapping data prior to 1992 for the Chandeleur Islands (Table 2), even though earlier imagery does exist. Decadal-scale changes in seagrass area are demonstrated in Figures 2a and 2b using data from 1969, 1992, and 2005/2006. The difference between the seagrass area estimated in the 1969 map and other maps in this paper is very large (Table 2 and Fig. 2a). The 1992 and 1999 maps created by USGS-NWRC also show substantially larger calculated seagrass area compared to maps in other years. These differences are largely the result of different investigators using different mapping criteria. Seagrass area was calculated in 2005/2006 based on vegetated seagrass polygons, whereas in 1992 it was calculated from seagrass coverage, and in 1969 the data used was seagrass extent. Seagrasses have continued to be supported in all the locations that were first mapped in 1969, except for Waveland on the mainland. There was an overall decline in VSA from 1940 to 2011 in Mississippi-Chandeleur Sound region. The Chandeleur Islands, which are furthest offshore, are home to a much greater area and diversity of seagrasses when compared with the MS sites, but may have also experienced the largest decline to protective barrier island erosion and loss.
TABLE 2. Seagrass area (ha) calculated for each site by year. Values in bold are seagrass extent, values in italics are seagrass coverage, and all other values represent vegetated seagrass area.
Figure 2: (A) Change in seagrass in the MS Sound (Group 1 and 2) from 1969 to 2006. The apparent location shift of seagrasses at Waveland is due to a mapping error in the early paper maps rather than a geolocation change. (B) Change in seagrass on the Chandeleur Islands (Group 3) from 1992 to 2005. Data in 1969 was not available for the Chandeleur Islands.
Case Study of Horn Island:
To demonstrate habitat change and the subsequent shift in location and total area of seagrass beds at a particular site, Horn Island was chosen as a case study. A set of seagrass maps on Horn Island spanning the study period from 1940-2008 is shown in Figure 3. The island underwent thinning in the north-south and shortening in the east-west direction over time. The 1940 seagrass area showed the furthest offshore distribution of vegetated patches in the north-south axis (data from 1969 is of insufficient spatial resolution to be considered very accurate), whereas the 2006-2008 seagrass area showed a shift onshore into shallower waters compared to the earlier data. There was land loss on the eastern end at a much higher rate than land gain on the western end of the island. Horn Island is gradually moving westward and the shape of the eastern tip keeps changing. Seagrass distribution on Horn Island shrunk in both north-south and east-west directions mirroring the geomorphological changes of the island shoreline.
Figure 3. Time series maps showing changes in seagrass and island size of Horn Island from 1940 to 2008. The star indicates a common geo-reference in all years. Map dates and the associated total seagrass area (ha) are shown on the right.
Problems Comparing Maps from Different Sources:
There are numerous possibilities for misinterpretation when comparing among seagrass maps from different sources. For a given site, seagrass occurrence in one map but not in the others may be due to lack of survey information rather than any seagrass loss/gain. Yet another source of error comes from comparing data collected in different seasons, as leaf biomass and shoot density can vary substantially during the year. A less obvious, but potentially major source of misinterpretation, is due to differences in mapping objectives and methods among studies. Eleuterius (1973), Eleuterius (1979), and Vittor (2014) generated maps of seagrass extent, while Moncreiff et al. (1998) and USGS-NWRC (1998a, 1998b, 2003) generated maps of seagrass coverage, both approaches mapped general seagrass locations using relatively coarse-scale polygons that included unspecified large areas of unvegetated sand bottom. This approach may generally be more appropriate for estimating the extent of seagrass habitat rather than vegetated seagrass area and does not allow for the quantification of patch size and shape. More recent studies (e.g., Carter et al., 2011) have employed object-based mapping with manual pixel editing to accurately identify the boundaries of individual seagrass patches from vertical aerial image data with a horizontal resolution of 1 m or better and enable calculation of VSA. Such present-day methods, enabled by advances in computer processing power and software, are much more effective in quantifying the presence, shape, and area of the many small patches that often comprise a seagrass population. Direct comparisons of seagrass area among studies using the different techniques (seagrass extent vs. seagrass coverage vs. vegetated seagrass area) may yield erroneous conclusions of dramatic change over time (e.g., compare between 1969 and 1971 Horn Island data, Table 2), simply because of the various mapping methods employed.
Conclusions:
In summary, studies in the MS Sound indicate both seagrass loss and gain depending on location, but decline has remained the major overall trend consistent with earlier report. The early 1970s was suggested as the beginning of seagrass decline in the MS Sound, largely due to prolonged low-salinity incursions from frequent Bonnet Carré Spillway openings. The compilation of various data in this study confirms the previously reported trend of seagrass decline (both in potential habitat and seagrass area), but the rate of change is not as high as reported previously based on inappropriate comparisons of maps created using different mapping methods. On the one hand, as shown in our case study of Horn Island, direct estimation of seagrass area change can be highly misleading if accepted prima facie without understanding the different mapping techniques use. On the other hand, development of new mapping techniques allows us to more thoroughly analyze patch fragmentation and obtain more accurate rates of change in vegetated seagrass area over time, which is needed for better management of this declining natural resource.
This work was published as: Pham, L.T., P.D. Biber, and G.A. Carter. 2014. Seagrasses in the Mississippi Sound and the Chandeleur Islands, and problems associated with decadal-scale change detection. Gulf of Mexico Science 32: 24-43.