Sea-level rise and landscape change influence mangrove encroachment onto marsh in the Ten Thousand Islands region of Florida, USA
- First Online:
- Cite this article as:
- Krauss, K.W., From, A.S., Doyle, T.W. et al. J Coast Conserv (2011) 15: 629. doi:10.1007/s11852-011-0153-4
- 753 Downloads
The Ten Thousand Islands region of southwestern Florida, USA is a major feeding and resting destination for breeding, migrating, and wintering birds. Many species of waterbirds rely specifically on marshes as foraging habitat, making mangrove encroachment a concern for wildlife managers. With the alteration of freshwater flow and sea-level rise trends for the region, mangroves have migrated upstream into traditionally salt and brackish marshes, mirroring similar descriptions around the world. Aside from localized freezes in some years, very little seems to be preventing mangrove encroachment. We mapped changes in mangrove stand boundaries from the Gulf of Mexico inland to the northern boundary of Ten Thousand Islands National Wildlife Refuge (TTINWR) from 1927 to 2005, and determined the area of mangroves to be approximately 7,281 hectares in 2005, representing an 1,878 hectare increase since 1927. Overall change represents an approximately 35% increase in mangrove coverage on TTINWR over 78 years. Sea-level rise is likely the primary driver of this change; however, the construction of new waterways facilitates the dispersal of mangrove propagules into new areas by extending tidal influence, exacerbating encroachment. Reduced volume of freshwater delivery to TTINWR via overland flow and localized rainfall may influence the balance between marsh and mangrove as well, potentially offering some options to managers interested in conserving marsh over mangrove.
KeywordsCoastal habitat Hydrological change Salinity Wetlands
Temporal oscillations between vegetation boundaries are common in many ecosystems, but especially pronounced in coastal wetlands subjected to sea-level rise, tropical storms, and hydrological alteration (Michener et al. 1997). Management of coastal wetlands for conservation benefit must weigh the balance imposed by natural transgression of one community onto another with changes initiated through direct and indirect human modification. Oscillations between marsh and mangrove habitat provide a prime example; studies from around the world have suggested a somewhat complex interaction between marsh and mangrove distribution that involves sedimentation dynamics (Woodroffe et al. 1985; Rogers et al. 2006), altered rainfall patterns (Saintilan and Wilton 2001), drought (Rogers et al. 2006), facilitation of mangrove establishment by certain marsh plants (McKee et al. 2007b), and local-scale engineering modifications (Saintilan and Williams 1999; Saintilan et al. 2009). The creation of new vectors for mangrove propagule dispersal along dredged channels can be important too (Egler 1952; Sengupta et al. 2005). Collectively, these factors, as well as a host of autecological traits and climatic drivers (Snedaker 1995), have led to encroachment of mangroves onto marsh in some Everglades wetlands (Ross et al. 2000; Doyle et al. 2003).
Identifying the interactions driving the distribution of mangrove and marsh has captured the attention of vegetation ecologists working in the Everglades region of south Florida, USA for decades (Davis 1940; Egler 1952). Some of the very first observations have suggested future dominance of mangroves as sea levels rise, related either to shifts in tidal prisms (Egler 1952) or to the competitive advantage that mangroves have over marsh once established (West 1977; Kangas and Lugo 1990). The Everglades ecosystem itself offers a sentinel example of wetland hydrological change. Starting around 1881, conversion of the Everglades for human use through hydrologic manipulation has been on-going (McCally 1999); the landscape was transformed into a system with year-round discontinuous connections from inshore freshwater wetlands to estuaries (Light and Dineen 1994; Sklar et al. 2001; Grunwald 2006). Water delivery through the Everglades has been reduced tremendously over the past century, while original soils have been largely oxidized or burned; up to 78% of the original peat loss can be accounted for by microbial oxidation as soils were increasingly exposed to air (McCally 1999).
In recent years, the Comprehensive Everglades Restoration Plan (CERP) has been initiated by federal and state agencies with at least one goal being the restoration of natural hydroperiods to south Florida wetlands. Project hurdles will be related, in part, to the massive peat loss and resultant subsidence, increased flood duration and depths beyond plant species’ thresholds, and wide-scale human development. Since 1927, in fact, mangrove area has decreased in some portions of the Everglades due principally to water-level manipulations, hurricanes, and tropical storms (Bischof 1995). However, observations of mangrove area increases in the Ten Thousand Islands region (NRCS 2001) supports the trend that chronic loss of mangroves appears to be less evident in some portions of the northern Everglades (Bischof 1995), perhaps due to the influence of greater human modification of the landscape.
In this investigation, we first document the progressive movement of mangroves onto salt and brackish marsh in the Ten Thousand Islands region of Florida by comparing a chronosequence of imagery from 1927, 1940, and 2005. We then discuss that change relative to sea-level rise, synoptic water level records, and pore water salinity differences along three parallel landscape transects through brackish marsh, salt marsh, transitional marsh-mangrove, and mangrove.
Just upstream of TTINWR within the Picayune Strand State Forest is a failed real estate venture (South Golden Gate Estates) that began construction in the early 1960’s. In support of this development, 77 km of canals, 467 km of shell rock roads, and thousands of lots were constructed before the development went bankrupt. While hydrological modification of the headwaters of TTINWR likely began in the early 1900’s (Tabb et al. 1976), water still flowed through the Picayune Strand and into TTINWR at least seasonally as sheet flow prior to development of South Golden Gate Estates (Carter et al. 1973). Water flow was directed into canals which served to channelize water delivery to lower portions of the estuary and eventually through one major conduit, the Faka Union Canal. After the construction of the Faka Union Canal, which itself cut off the Faka Union River, the annual low-water level in the center of Fakahatchee Strand declined from nearly 91 cm above mean sea level in 1972 to 30 cm above mean sea level in 1974 (Swayze and McPherson 1977).
Mangrove boundary mapping was restricted to TTINWR for three time periods—1927, 1940, and 2005—approximating mangrove change over 78 years. Information for the first time period was collected from 1927 topographic sheets (T-sheets). T-sheets are detailed survey maps issued by the National Oceanic and Atmospheric Administration’s (NOAA) National Ocean Service to provide coastline detail for navigation as well as to depict vegetation inland from the coast. All T-sheet maps from TTINWR were produced from photographs taken by the U.S. Army Corps of Engineers at a scale of 1:20,000. T-sheets were previously converted to digital format and georectified (Smith et al. 2002). Maps were published upon completion, some as early as 1931 and others based on supplemental survey data completed in October of 1940 (Smith et al. 2002). Data for the second time period were obtained from 1940s black and white aerial photographs that were downloaded from a U.S. Geological Survey web site (Smith and Foster 2010), and projected at a scale of 1:40,000. True-color aerial imagery from 2005 was acquired from the Collier County Property Appraiser’s Office (Naples, Florida), which was supplied directly to TTINWR personnel. Aerial photographs were taken at an altitude of 5,273 m using a Leica ADS40 digital sensor (Leica Camera AG, Solms, Germany) over a span of 9 days. This imagery was previously imported into a geodatabase by the Property Appraiser’s Office and projected at a scale of 1:400.
All images were digitized into a vector data storage format, or shape file, of mangrove forest coverage using ArcGIS (Versions 8.2-9.3, ESRI, Inc., Redlands, California, USA). Scanned aerial photography for map production of each date was acquired, and all maps were rectified to a common geoid and projection. Photographs for 1940 and 2005 were rectified to the maps they were originally used to create, and were reviewed for signature consistency and checked with current ground features. The range of projection and scales for available imagery has an inherent error of approximately 12.2 m when comparing among dates and coordinate assignment between NAD 27 and NAD 83 (USGS 1999; C.J. Wells, pers. comm.).
Polygons were mostly classified in a simple manner as mangrove or non-mangrove; however, marsh, large airboat trails, mud banks, and in some cases, tropical hardwoods, were also discerned. Other tree species displayed habitat preferences similar to those of true mangroves (e.g., buttonwood, Conocarpus erectus L.; Brazilian pepper, Schinus terebinthifolius Raddi), and required ground truthing to reconcile. Ground-truthing consisted of stratified random sampling by transect and foot, as well as spot checks when necessary (Barry 2009). We made no distinction among red (Rhizophora mangle L.), black (Avicennia germinans (L.) Stearn.), and white (Laguncularia racemosa (L.) Gaertn. f.) mangrove species.
Vegetation, water level, and salinity
Distinct vegetation assemblages on TTINWR include brackish marsh, salt marsh, transitional marsh-mangrove, and mangrove. In 2006, we established three transects, each approximately 5 km in length, to document hydrologic and salinity characteristics of the four vegetation assemblages (Fig. 1). Each transect (labeled West, Center, and East) began in a brackish marsh zone dominated by Cladium mariscus (L.) Pohl, with some Spartina bakeri Merr., located just south of U.S. Highway 41, then ran through a salt marsh zone dominated by Juncus roemerianus Scheele, Sesuvium portulacastrum (L.) L., and/or S. bakeri. Both of these zones also had individual mangrove trees in the vicinity, especially R. mangle and L. racemosa, but mangroves were not yet a dominant vegetation type of these zones and tended to occur along small waterways. The third vegetation zone represented an abrupt transition between salt marsh and mangrove (Fig. 1), with salt marsh dominating on one side (especially Distichlis spicata (L.) Greene and Salicornia virginica L.) and mangrove dominating the other. The fourth vegetation zone was represented by continuous mangrove coverage, and extended to the Pumpkin, Little Wood, and Wood Rivers.
Individual water level recorders (Infinities USA, Inc., Port Orange, Florida) were placed within each zone on each transect for a total of 12 recorders, all logging at 1-hour intervals over approximately 3 years. Salinity near each recorder was sampled eleven times from the soil surface over that same time frame, and at depths of 15, 30, and 45 cm into the soil using a plastic insertion tube and extraction syringe. Pore water samples were injected into a scintillation vial and measured for salinity with a portable conductivity meter (Model 30, YSI Inc., Yellow Springs, Ohio, USA) within 60 s of extraction.
Sea-level rise trends for Key West were obtained from long-term NOAA tide gauge records (NOAA 2011) and rainfall data were extracted from Carter et al. (1973), Thomas (1970), and through personal communication (Robert Sobczak, Big Cypress National Preserve, Ochopee, Florida).
Mangrove area over a 78 year period was related graphically to mean sea-level trends from Key West. Salinity data were averaged across sampling periods and analyzed with a three-way ANOVA, using sampling depth, zone, and transect as independent variables. Data were subjected to a square-root transformation to meet parametric assumptions, and analysis was conducted using SAS Version 9.1 (SAS Institute, Inc., Cary, North Carolina, USA).
Historical rainfall record (from Thomas 1970) versus the past 38 years (1970–2008: from Robert Sobczak, unpubl.) for the Ten Thousand Islands region of South Florida. Data are reported in cm
Our documentation of a 35% increase in mangrove coverage at TTINWR at the expense of marsh is within the range reported from an Australian estuary (25%–80%) undergoing similar wetland habitat conversion (Saintilan and Williams 1999). Indeed, marsh replacement by mangrove has been described in north Florida (Stevens et al. 2006), southern Texas (Sherrod and McMillan 1985), Louisiana (McKee et al. 2004; Perry and Mendelssohn 2009), Trinidad (Ramcharan 2004), and Hawaii (Chimner et al. 2006), as well as in other locations of south Florida (Davis 1943; Ross et al. 2000). Mangroves have also been found to colonize salt flats in Mexico (López-Medellín et al. 2011) and accreting mud banks in New Zealand (Lovelock et al. 2007), precluding colonization by alternate vegetation types.
Mangrove transgression onto marsh can certainly be determined by a number of factors, including local geomorphic change and hydrological characteristics, climatic extremes and potential shifts in those extremes, biogeochemical characteristics of soils, and biotic interactions (Saintilan et al. 2009). For TTINWR, we focus on two primary entities: geomorphic change and sea-level rise. We did not attempt to link transgression of mangroves in TTINWR to temperature shifts because there were no discernable long-term changes in the mean number of freeze days in at least one south Florida location from 1948–1989 (Duever et al. 1994).
The greatest alteration to TTINWR over the 78 year mapping period was the construction of the CR 92 and Faka Union Canals, both of which lay mostly outside of refuge boundaries (Fig. 1). The Faka Union Canal traverses several tidal creeks in Port of the Islands at the upper reaches that serve to vector tidal waters into the northeastern portion of TTINWR. Indeed, tidal influences are most evident in east transect hydrographs (Fig. 3); based upon data from west and center transects, tidal reach would almost certainly not have extended as distinctly to the brackish and salt marsh zones to the east without the canal’s influence. Saintilan et al. (2009) suggested that mean alterations in tidal prism extent as a result of engineering works can confound the rate and process of mangrove transgression. Altered tidal regimes through construction of a seaway and tidal barrage along the Pimpama River associated with Moreton Bay, Australia was responsible for an approximate 10% expansion in mangrove coverage to the detriment of marsh over a 46 year period (Morton 1994).
We also see mangrove encroachment in the extreme western portion of TTINWR along the CR 92 canal outside of transect boundaries. Rhizophora mangle propagules can float and remain viable for up to 300 days (Davis 1940; Allen and Krauss 2006), giving ample opportunity for dispersal to nearly all locations on TTINWR, but especially along CR 92 and Faka Union Canals. In fact, local extensions of tidal range likely facilitate propagule dispersal of all TTINWR mangrove species. Yet, while hydrology was altered, salinity was unaltered along the east transect (24.4 ppt) closest to the Faka Union Canal relative to the other two transects (24.5 ppt). What is most interesting is that current salinities did not differ between salt and brackish marshes, suggesting that salt marsh may also be on the move.
The majority of water delivered to the Everglades is used by the vegetation in a years’ time, which creates an ecosystem heavily dependent upon annual rainfall (Kushlan 1990), with relatively little runoff routing to the ocean (Duever et al. 1994). Because much of the rainfall to the Everglades originates from in situ evapotranspiration, and evapotranspiration is responsible for exporting upwards of around 70%–90% of rainfall entering the Everglades (Duever et al. 1994), the development of south Florida for human infrastructure since 1940 would have influenced this balance. In fact, modeling efforts suggested a 9% decrease in summer rainfall in 1973 and an 11% decrease in summer rainfall in 1993 relative to rainfall depths from 1900 (Pielke et al. 1999). Mangrove expansion has been linked to freshwater inputs from rain in the upper intertidal zone of an estuary in Australia (McTainsh et al. 1988). It is inconclusive whether a near-century-scale drying trend for areas surrounding TTINWR is real (see Table 1); however, it is important to note that much variability can exist in rainfall even within the refuge boundary. For example, rainfall from three locations on TTINWR as close as 3–5 km apart indicated a mean discrepancy of 21 mm within any given month (unpubl.). Given this variation, it is difficult to discern clear differences in the rainfall amounts reported in Table 1.
Expansion of mangroves in TTINWR tracked annual mean high water (mhw) trends from a tide gauge in Key West, Florida over the past century (Fig. 5); this is especially evident for transgression from 1940 to 2005. The increase in mangrove coverage was not linear. In fact, if we do assume a linear relationship, extrapolations to 1927 from 1940 to 2005 place mangrove area in 1927 approximately 11% higher than the T-sheets indicate. It is possible that water extraction up-stream between 1927 and 1940 influenced changes in the balance between mhw and mangrove area in TTINWR. Water table depths certainly decreased throughout the Everglades region from 1914 to 1953, regardless of potential temperature or rainfall changes (Thomas 1974).
By using a very conservative sea-level rise trend of 0.61 mm/yr, Egler (1952) suggested that all coastal vegetation bands, including mangroves, might be expected to move a distance of about 4.1 m/yr inland over a 100 year period. In TTINWR, the rate of transgression was faster than this in some areas; however, the true sea-level rise trend for the Key West station was also greater (2.24 mm/yr) than Egler assumed. As it turns out, mangroves moved up-slope, but there was less observable loss of TTINWR mangrove area fringing the bays or Gulf of Mexico between 1927 and 2005; small losses were confined to the erosional shores of the outermost islands (Fig. 2). This is not always the case documented in the literature (see López-Medellín et al. 2011). On the other hand, an analysis of differential mangrove species loss and replacement on the islands, especially loss of those species of mangroves and mangrove associates with horizontal root structure (e.g., A. germinans, C. erectus) in lieu of vertical root structure (e.g., R. mangle), may provide a different metric of change (Snedaker 1995). Mangrove associates, such as C. erectus, have experienced large die-offs in back swamp locations of TTINWR, with many of those die-off areas being replaced by true mangroves (Barry 2009).
While surface accretion of sediments was correlated with tidal range in southeast Australia, mangrove encroachment onto salt marsh was linked to long-term trends in surface elevation (Rogers et al. 2006). It is unknown how much subsidence has occurred in TTINWR marshes over the past century. Mangroves are generally very efficient at adjusting soil elevations in response to sea-level rise (sensu Parkinson 1989; McKee et al. 2007a). But, if marshes in TTINWR are less efficient at vertical adjustment, especially within unvegetated or sparsely vegetated mudflats (comprising 110 ha in 2005), then subsidence may decrease the elevation of the marsh relative to mhw. It is common for soil surface elevation to drop by 9–15 cm when transitioning from mangrove to marsh in TTINWR (unpubl.). Also, the depth of standing water within marshes increases with distance inland (Fig. 3), perhaps reflecting greater elevation loss to subsidence during dry conditions annually (sensu Rogers et al. 2006), perennially reduced surface water delivery to the Everglades over the past century, or greater surface accretion or root zone expansion in mangroves relative to marsh. Groundwater stage strongly influenced surface elevation change in an Everglades mangrove swamp (Whelan et al. 2005); it is likely that the TTINWR marsh responds similarly given the persistence of shallow groundwater.
Mangroves have expanded coverage within Ten Thousand Islands NWR by approximately 1,878 ha between 1927 and 2005. This 35% increase in mangrove coverage poses consequences for waterbirds that rely on marsh for foraging. Mangrove transgression is facilitated by tidal expansion in TTINWR, not only as a consequence of sea-level rise, but also as a result of the construction of the Faka Union and CR 92 Canals. As part of CERP, plugs and culverts have been installed to distribute additional freshwater more uniformly across U.S. Highway 41 north-to-south akin to historic flows. This has altered the stage, discharge, timing, and distribution of flow across the marsh-mangrove coastal margin of TTINWR, and perhaps will even reduce the rate of mangrove encroachment to the remaining marsh while loading the system with greater amounts of freshwater.
We gratefully acknowledge Layne Hamilton, Joyce Palmer, Matthew Martin, Larry Richardson, and Ben Nottingham for assistance with many items over the past decade while working on the refuge; Thomas J. Smith III and Anne M. Foster for providing much of the imagery; Rebecca J. Howard and Richard H. Day for assistance with field and project planning, and Pat O’Donnell (and the staff of the Rookery Bay National Estuarine Research Reserve) for assistance with accommodations and data mining. Thomas J. Smith III, Rebecca J. Howard, and Hongqing Wang provided helpful reviews of earlier drafts, and Christopher J. Wells helped us assess the accuracy of the mapping. We thank Laura Brandt and G. Ronnie Best for funding to support this work through the USGS Priority Ecosystems Research Program and the National Park Service’s Critical Ecosystems Initiative. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the U.S. Fish and Wildlife Service. Furthermore, any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.