Predation restricts black mangrove (Avicennia germinans) colonization at its northern range limit along Florida’s Gulf Coast
- 594 Downloads
Climate change-driven range expansion of black mangroves (Avicennia germinans) is predicted along the northern Gulf of Mexico, where sea level rise is also driving conversion of freshwater forest islands to salt marsh. While climate-driven A. germinans range expansion has garnered considerable scientific attention, the role of top-down controls on colonization is largely overlooked. We investigated the effects of abiotic (flooding frequency, soil depth, soil salinity) and biotic (predation, herbivory) controls on A. germinans establishment at its northern range limit along Florida’s Gulf Coast by comparing fates of caged and non-caged propagules across four landscape positions (from creek edge to forest island interior) and at three sites along a tidal flooding frequency gradient. Within 12 days, grapsid crab, Sesarma reticulatum, consumed 99% of non-caged propagules. Among caged propagules, establishment increased with increasing flooding frequency; however, cages did not entirely prevent predation, which remained a primary cause of mortality, except in the rarely flooded island. Propagules that survived to seedlings experienced mild to fatal herbivory across landscape positions and sites. This study revealed that while relict forest islands and surrounding marshes can support A. germinans, predation and herbivory strongly suppress colonization, suggesting that mangrove expansion models should incorporate biotic controls.
KeywordsBottom-up Community reassembly Herbivory Predation Sea level rise Sesarma reticulatum Top-down
Climate change is driving ecological shifts at population, community, and ecosystem scales globally, altering ecological functions via changes in habitat and species composition (e.g., Walther et al., 2002; Parmesan & Yohe, 2003; Parmesan, 2006; Lancaster et al., 2016). Examples range from phenological shifts in flowering times (Fitter et al., 1995; Dunne et al., 2003; Inouye, 2008), bird migration (Møller et al., 2008; Travers et al., 2015), and butterfly migration (Sparks & Yates, 1997) to geographic shifts in floral and faunal ranges (Grabherr et al., 1994; Perry et al., 2005; Kelly & Goulden, 2008). In coastal regions, a notable geographic range shift occurring in response to climate change is the expanded coverage of mangroves into temperate latitudes.
Poleward expansion of mangroves is being observed as warming air temperatures and fewer freeze events promote suitable conditions for mangrove survival (Saintilan et al., 2014; Alongi, 2015; Kilkenny & Galloway, 2016; Osland et al., 2016b; Ward et al., 2016). Sea level rise (SLR) also accommodates landward migration along undeveloped coastlines (Krauss et al., 2011; Nitto et al., 2014). Other effects of climate change, including increasing atmospheric CO2 and changing precipitation regimes, affect the ability of mangrove species to compete with salt marsh communities, further influencing future distributions of mangroves at temperate latitudes (McKee et al., 2012; Osland et al., 2016a). With the effects of climate change becoming more pronounced at local, regional, and global scales, and mangrove ranges being historically sensitive to major abiotic shifts (Saintilan et al., 2014), considerable research is being conducted to better understand the effects of abiotic factors on future mangrove distributions. However, little work has focused on the coupled effects of bottom-up (abiotic) and top-down (biotic) controls on mangrove expansion at their temperate range limits.
In the southeastern US, the frequency and severity of freeze events dictate periods of mangrove population expansion and contraction and, hence, control mangrove coverage at the northern range limit (Lugo & Patterson-Zucca, 1977; Stevens et al., 2006; Osland et al., 2017). Repeated extreme freezes during 1895–1905 and 1977–1989 seriously reduced mangrove coverage along the Atlantic and Gulf of Mexico Coasts (Stevens et al., 2006). Since the 1980s, recovery and expansion has yielded the greatest mangrove coverage in the northern Gulf of Mexico since at least 1893 (Osland et al., 2017). In Florida (USA), northward mangrove migration is occurring on both Atlantic and Gulf Coasts. Along the Atlantic Coast, Cavanaugh et al. (2014) found that mangrove spatial extent has doubled since 1984, and along the Gulf Coast, Giri & Long (2016) found mangrove area had increased by 25–50% at the northern range limit since 1980.
This study was prompted by our observations of A. germinans seedlings naturally establishing in salt marsh vegetation positioned along creek edges near transitioning forest islands. These observations motivated three initial hypotheses: (1) A. germinans could also establish in forest islands, potentially modifying the forest-to-salt marsh trajectory currently in progress; (2) abiotic factors, particularly tidal flooding frequency, would dictate A. germinans colonization success in forest islands; (3) A. germinans establishment success would be similar in islands with sufficient flooding and in the surrounding Juncus roemerianus Scheele (black needle rush) marsh, and would be higher at creek edges. Creek edges most closely resemble lower elevation, cordgrass-dominated (Spartina alterniflora Loisel.) creek bank habitats where A. germinans colonization has been observed in northern Gulf Coast marshes (Patterson et al., 1997; McKee & Rooth, 2008). After discovering high rates of propagule predation by grapsid crab, Sesarma reticulatum during our initial experimental deployment, we expanded the study to quantify top-down biotic controls on A. germinans colonization through measurements of propagule predation and seedling herbivory.
In tropical regions, propagule predation and herbivory have been investigated extensively to explain regeneration success within mangrove forests, zonation patterns of co-occurring mangal species, and establishment success of mangroves into adjacent habitat (Smith, 1987; Smith et al., 1989; Osborne & Smith III, 1990; McKee, 1995; Farnsworth & Ellison, 1997; Dahdouh-Guebas et al., 1998; Sousa & Mitchell, 1999; Bosire et al., 2005; Cannicci et al., 2008). Avicennia are especially susceptible to top-down pressure in the form of propagule predation due to their soft, fleshy cotyledons that lack a protective coat, relatively high nutrition content, relatively low concentrations of herbivore-deterring compounds, and small size (Smith, 1987; McKee, 1995). These features make them palatable to a variety of fauna including crabs, snails, insects, mammals, and fish (Smith et al., 1989; Steele et al., 1999; Sousa et al., 2003). As seedlings and adults, Avicennia, like other mangals, are susceptible to herbivory by insects and crabs that consume leaves, flowers, and woody tissue, which can lead to defoliation, reduced growth rates, and limited propagule production (Cannicci et al., 2008). However, the roles of propagule predation and seedling herbivory in mediating range expansion of A. germinans in the northern Gulf of Mexico have not previously been addressed. Here we present our findings on the effects of both bottom-up and top-down controls on A. germinans expansion in a region where abiotic conditions are usually considered the primary drivers of mangrove colonization.
Materials and methods
Manipulative field experiments were conducted along Turtle Creek, a tidal creek in Waccasassa Bay Preserve State Park, located along the Big Bend Coast of Florida (Fig. 1). The study area comprises a mosaic of salt marsh and forest islands in various stages of health, ranging from healthy, regenerating tree stands to relict stands now dominated by salt-tolerant shrubs, forbs, and grasses. Continuous freshwater forest dominated by S. palmetto and J. virginiana var. salicicola and upland pine flatwoods occur landward of the study area (Vince et al., 1989; Williams et al., 2007), and an established population of A. germinans is located seaward along the coastal fringe, approximately 3 km to the southwest. Propagules from A. germinans float into Turtle Creek and wash up on creek banks, and small scattered stands (<25 m2) of short (<1 m tall) mangroves are also scattered throughout the marsh.
We established three sites along Turtle Creek, each including a forest island. The sites are approximately 0.7 km apart and are referred to by the relative frequency of tidal flooding each island receives: frequently flooded, moderately flooded, and rarely flooded, based on frequencies reported in Williams et al. (1999). Flooding frequencies in the marshes surrounding each island are approximately the same between sites. The frequently flooded island is a relict forest (non-regenerating) with no live trees and has converted to herbaceous salt marsh. The moderately flooded island supports few live trees and has a mixed understory of halophytic shrubs and herbaceous salt marsh plants. The rarely flooded island supports healthy, regenerating forest with an understory of typical coastal freshwater forest vegetation. Islands at all three sites are surrounded by salt marsh dominated by J. roemerianus. Along the creek edge, vegetation is dominated by S. alterniflora and also includes other plants common in high marsh: Distichlis spicata (L.) Greene (salt grass), Salicornia sp. (glasswort), and J. roemerianus.
At each site, we tested propagule survival and seedling establishment in four landscape positions: island interior, island edge, marsh plain, and creek edge. We selected island interior positions representative of the overall island habitat and island edge positions that were on the creek side of the islands. Marsh plain positions were located between island edge and creek edge positions. Creek edge positions were within 1 m of the creek bank slope. We calculated tidal flooding frequencies (number of weeks during which flooding occurred over the course of the experiment) in forest island interiors and edges using a model developed by Williams et al. (1999) specifically for these islands. The model requires elevation data, which we obtained for island interiors from Williams et al. (1999) and estimated for island edges using 2007 Light Detection and Ranging (LiDAR) data for Levy County, downloaded from the NOAA Digital Coast website (http://coast.noaa.gov/digitalcoast/). For marsh plain and creek edge positions, tidal flooding was assumed to occur daily during high tide based on observations made during the monitoring visits and previous field work in the area.
In addition to tidal flooding frequency, we quantified soil salinity, soil depth, percent cover of plant species, and crab burrow density for all landscape positions at each site. Soil salinity was measured from five soil samples collected in June 2016 at each landscape position at each site. Samples were air dried for 12 days, then oven dried at 40°C for 3 days, sieved, and analyzed using the saturated paste method (Rhoades, 1996; Reddy et al., 2013) with a 1:5 soil weight per DI water volume ratio. Salinity of the soil slurry was measured using a conductivity meter (Thermo Scientific Orion Star A215). Soil depth was measured at 10 random locations by inserting a pin flag into the soil until it made contact with underlying limestone bedrock, then measuring the length that was inserted with a metric ruler. Percent cover of plant species was measured in four 1-m2 quadrats. We counted S. reticulatum burrow density as a proxy for herbivorous crab density within eight randomly placed 0.25-m2 quadrats. Sesarma burrows were identified by features described in Holdredge et al. (2009).
The initial experimental deployment consisted of placing 10 non-caged A. germinans propagules approximately 25 cm apart in four 0.25-m2 plots in each of the four landscape positions, for a total of 16 plots per site and 48 plots total. Propagules were set out in October 2015 and had hypocotyls extending 0–1 cm. Plots were marked at each corner with a flag and had no physical barriers separating them from the surrounding area. We planned to count the number of propagules present and record propagule fate approximately every 2 weeks in each landscape position at each site; however, during the first monitoring event on day 12, we observed 99% propagule predation. Based on this finding, the non-caged experiment terminated and the caged experiment (see below) was initiated.
To determine the relative importance of landscape position and flooding frequency on propagule and seedling fates, we attempted to control for S. reticulatum predation by placing propagules in cylindrical cages (20 cm height, 10 cm diameter) made of galvanized steel cloth (0.6-cm2 mesh). Each cage was placed on top of a square of steel cloth buried in the soil and covered by another square of steel cloth secured to the top of the cage with zip ties. Cages were anchored in place with landscape stakes. Ten cages were placed in each landscape position. We placed one propagule each in five cages and five propagules each in the remaining five cages to additionally test for density effects on propagule establishment, survival, and growth. Given observations of equivalent predation and propagule viability across density treatments, data for all caged propagules were later pooled, resulting in n = 30 propagules per landscape position per site. Propagules were placed in the cages in December 2015 and typically had hypocotyls extending between 1 and 3 cm but had no budding roots. All propagules used in caged and non-caged experiments were collected from the established mangrove forest southwest of Turtle Creek (Fig. 1) and had already dropped from parent trees. Only propagules with no signs of predation or decay were used.
We analyzed the effects of flooding frequency and landscape position on soil salinity, soil depth, percent cover of salt marsh vegetation (i.e., all halophytes), and burrow density using analysis of variance and Tukey’s HSD for post hoc multiple comparisons. Fates of caged propagules and seedlings were compared by landscape position at each site using Pearson’s χ 2 tests. Because propagule survival was confounded by inconsistent caging effectiveness, we combined viable and predation fates in our statistical analysis of abiotic controls on propagule survival. Additionally, mild and moderate herbivory on seedlings were combined for analysis. We assessed the correlation between caged propagule predation and burrow density using univariate linear regression. To correct for an increased chance of Type I error from multiple ANOVA and χ 2 tests, we compared P-values to α = 0.0167, instead of a standard α = 0.05.
Heights of caged seedlings between monitoring intervals were compared using a repeated measures, mixed model ANOVA, with site and landscape position set as fixed effects. Only live seedlings were included. Because differences in seedling establishment times and mortality created variation in sample sizes between time intervals and overdispersion in the data within landscape positions, the ANOVA model was a poor fit. Therefore, we analyzed the dataset using ANOVA and Tukey’s HSD in separate analyses for each site and sampling interval. The effects of environmental variables and burrow density on the survival of pooled post-cage removal seedlings were evaluated using multiple regression. All statistical analyses were performed in R (R Core Team, 2016).
Characteristics of landscape positions across sites
At the frequently flooded site, model estimates indicated that tidal flooding occurred during every week of the caged experiment (28 weeks) in the island interior and island edge. At the moderately flooded site, tidal flooding occurred during 21 weeks in the island interior and 23 weeks at the island edge. At the rarely flooded site, no flooding occurred in the island interior during the caged portion of the experiment (though a hurricane flooded the island interior in September 2016), and the island edge flooded during 23 weeks. Marsh plain and creek edge positions at all sites were assumed to have flooding events during every week at all three sites.
Sesarma reticulatum burrows were present everywhere except the interior of the rarely flooded island, demonstrating the wide distribution of S. reticulatum across sites (Fig. 3d). At the frequently flooded site, mean burrow density ranged from 22.5 to 35.5/m2 and was not significantly different between landscape positions (F 3,28 = 1.46, P-value = 0.25), likely due to similarities in tidal flooding and vegetation cover across the relatively homogeneous, low-elevation site. At the moderately flooded site, mean burrow densities in the marsh plain and creek edge were more than twice as high (32/m2) as in the island edge and interior (9–13.5/m2), given more frequent flooding and deeper soil depths more characteristic of salt marsh closer to the creek (F 3,28 = 19.3, P-value <0.0001). Burrow densities were most variable at the rarely flooded site, where burrow density increased from 0 in the island interior to 41/m2 in the creek edge (F 3,28 = 26.3, P-value <0.0001).
Biotic pressure on non-caged propagules
Non-caged propagules experienced extremely low survival (<1%) regardless of site or landscape position after 12 days, providing a measure of ambient predation pressure. We identified predation by S. reticulatum as the primary cause of disappearance based on the moderate to high crab burrow density present across sites and landscape positions (with the exception of the island interior of the rarely flooded site), and video footage of S. reticulatum taking a propagule and pulling it into a burrow that we recorded during a separate experiment at the site (Supplemental File). Propagule disappearance in the island interior of the rarely flooded site was likely depredation by other fauna such as small mammals or wild boars, which are present on regenerating forest islands, though we were unable to confirm this assumption.
Caging experiment: propagule establishment success and seedling growth
When we disregarded predation due to the confounding results of inconsistent caging effectiveness, we found that the fate of caged propagules depended on landscape position at the frequently flooded site (χ 2 = 35.7, P-value <0.0001), where rot and undetermined causes of mortality reduced viability in the island interior by 20% and creek edge by 37%. Landscape position was also important at the rarely flooded site (χ 2 = 112.5, P-value <0.0001), where desiccation prevented propagule survival in the island interior, further demonstrating propagules’ reliance on adequate tidal flooding. At the moderately flooded site, non-predation causes of mortality were low and consistent across landscape positions (χ 2 = 5.44, P-value = 0.49). Across sites, seedling establishment at creek edge positions was not higher compared to other positions, contrary to our hypothesis. Overall seedling establishment on day 200 was correlated with more frequent tidal flooding and salt marsh vegetation cover and decreased with deeper soils and higher burrow density, but was not affected by soil salinity (F 5,90 = 28.3, P-value <0.0001, r 2 = 0.59).
Seedling survival and growth after cage removal
We found that top-down biotic pressure exerted significant control over A. germinans propagule and seedling survival, establishment, and growth from creek edges to forest island interiors across a range of tidal flooding regimes. In the absence of caging, propagules suffered nearly 100% predation. Sesarma reticulatum was the primary predator, but propagules in the island interior of the rarely flooded site, where S. reticulatum burrows were absent, also suffered intense predation, likely by small mammals or wild boars. Herbivory also decreased survival of established seedlings. Caging reduced but did not exclude predation on propagules or herbivory on seedlings as they grew to cage capacity. Soil disturbance from tidal flooding and burrowing activity (by S. reticulatum and other crab species present in the area, including Uca spp., which do not consume propagules) likely shifted the sides of cages off their bases, creating subsurface gaps through which S. reticulatum could access propagules. Soil disturbance is likely most intense in creek edge positions due to stronger effects of tidal flooding and generally higher burrow densities (Fig. 3), which may explain high predation rates on caged propagules in that landscape position.
Where caging successfully prevented predation, propagules were able to survive in forest islands and develop into seedlings, provided there was sufficient tidal flooding, as hypothesized. Only in the rarely flooded island interior, where propagules suffered desiccation in the absence of tidal flooding during the experiment, did abiotic conditions exert greater pressure on caged propagules than did top-down biotic controls (Fig. 4). Rot and undetermined cause(s) of mortality reduced propagule survival in other landscape positions at each site, and resulted in significant differences in propagule viability between landscape positions at the frequently flooded site, but no clear pattern emerged across sites that could be explained by differences in landscape position characteristics.
Relatively fast rates of seedling establishment in the island interior and edge of the frequently flooded site suggest that conditions at these landscape positions might be more favorable than in the surrounding marsh, though slower establishment in the marsh plain did not result in lower total establishment compared to other positions at the three sites (Fig. 5). Unfortunately, the inconsistent exclusion of S. reticulatum from cages undermines our ability to clarify the relationships between increased seedling establishment and increased flooding, salt marsh vegetation, and decreased soil depth and burrow density across all positions and sites. Consequently, it is unclear from these results whether sufficiently flooded forest islands provide bottom-up conditions that support A. germinans colonization success at rates similar to the surrounding marsh or whether creek edges, in the absence of predation, provide more (or less) favorable abiotic conditions for seedling establishment compared to other landscape positions. Ambient predation intensity observed at the frequently, moderately, and rarely flooded sites is comparable to Avicennia propagule predation reported in tropical and subtropical mangrove forests: 72% of A. germinans propagules were non-viable from predation within 4 days in south Florida (Smith et al., 1989); 60% of A. germinans propagules were depredated within 9 days in Belize (McKee, 1995); 100% of A. marina propagules in northern Australia were consumed within 20 days (McGuinness, 1997); and 98–100% of A. schaueriana propagules were consumed within 18 days in Brazil (Souza & Sampaio, 2011). Among the more extreme cases, 100% of 200 A. marina propagules were non-viable from predation within 60 min in Kenya (Van Nedervelde et al., 2015). Whether intense predation pressure and generally high burrow densities of S. reticulatum along Turtle Creek are representative of other potential mangrove colonization areas along the Big Bend requires further investigation. Critically, studies investigating A. germinans colonization in temperate regions focus largely on abiotic drivers (Stevens et al., 2006; McKee & Rooth, 2008; Comeaux et al., 2012; Osland et al., 2013; Cavanaugh et al., 2014; Saintilan et al., 2014). The nearest study conducted in a similar habitat that addressed predation took place in Louisiana, where Patterson et al. (1997) found only 11% mortality of A. germinans from propagule predation [by S. cinereum and Littoria irrorata (snail)] in a Spartina-dominated marsh, suggesting there may be significant spatial variability in predation pressure across the northern mangrove range limit.
Propagules that survived predation continued to face mild to intense herbivory as seedlings (Fig. 6). Seedling mortality from herbivory was not as prevalent as crab predation of propagules, though most seedlings suffered some degree of herbivory. Sesarma reticulatum may have contributed to seedling herbivory by continuing to consume fleshy cotyledons of young seedlings. Then, as seedlings matured and produced true leaves, insects most likely replaced crabs as the main seedling consumers, limiting survival and growth (Fig. 7). Insects are important herbivores in mangrove forests, affecting all stages of tree development (Robertson et al., 1990; Sousa et al., 2003; Minchinton, 2006; Cannicci et al., 2008). We suspect that grasshoppers were responsible for most seedling herbivory at our study site, as they are common grazers of J. roemerianus in marshes along the Gulf Coast (Parson & De La Cruz, 1980; Sparks & Cebrian, 2015) and were present in spring 2016 when seedlings were leafing out.
In addition to top-down biotic controls, competition with salt marsh may have also affected seedling growth. Seedlings grew best at the edge of the rarely flooded island, appearing healthier and developing faster than all seedlings at the frequently and moderately flooded sites. This was most apparent at the end of the caging experiment (April–June, Fig. 7) and post-cage removal (Fig. 8). The tallest seedling (115 cm) had developed visible pneumatophores by December 2016 (day 372). Salt marsh vegetation cover was much lower than at all other landscape positions where seedlings established (Fig. 3c), suggesting these seedlings may have faced less pressure from competition with salt marsh plants. Patterson et al. (1993) and McKee & Rooth (2008) found that growth rates of A. germinans decreased in the presence of S. alterniflora, and Guo et al. (2013) found that biomass of A. germinans seedlings was lower when grown adjacent to salt marsh plants, all of which were attributed to resource competition. However, other studies have found that salt marsh plants can benefit seedling establishment (Lewis & Dunstan, 1975; McKee et al., 2007). Further work is needed to better understand the role of competition as an additional biotic control on mangrove range expansion.
This study demonstrates that as remnant forest islands convert to salt marsh and become more suitable for mangrove colonization, they also become more suitable for fauna that commonly regulate mangrove populations. Sesarma reticulatum is common in salt marshes in the eastern and southeastern US (Bertness et al., 2009), and grapsid crabs are among the most pervasive and intensive propagule predators worldwide (McGuinness, 1997; Smith et al., 1989; Van Nedervelde et al., 2015). Though not observed depredating propagules at Turtle Creek, snails and salt marsh insect grazers may also become common fauna in relict islands, creating additional potential predation pressure on propagules and seedlings as islands convert to salt marsh. The flooding regime of the marsh and transitioning islands along Turtle Creek may also contribute to predation intensity. Time between daily tidal inundation is greater than the period of time in which the marsh, including creek edge, is flooded (personal observation) and transitioning forest islands are more often not flooded than flooded. Time between flooding events could allow plenty of opportunity for foraging, a hypothesis posed by Osborne & Smith III (1990) to explain higher predation in high versus low intertidal zones in Australia.
As the effects of climate change drive increased tidal flooding and fewer freeze events in the region, favorable habitat for mangrove colonization will increase in the northern Gulf of Mexico and US Atlantic Coast. However, our findings suggest that natural colonization of persisting and encroaching mangrove populations will require high propagule density to overcome biotic pressure. Currently, natural propagule density along Turtle Creek is relatively low (which we mimicked in our study), and crab densities were sufficiently large to prevent propagule establishment in non-caged experiments. Delivery of high densities of propagules from coastal storms may be required to sustain encroaching mangrove populations, and we are currently working to identify propagule density thresholds (if present) required to overcome predation pressure. Critically, the long-term persistence and expansion of mangrove populations in any location will be driven by the interactive effects of environmental variables (SLR, stochastic freeze events, coastal storms) and biotic controls (propagule density, predation, competition). Thus, both bottom-up and top-down drivers need to be considered when projecting future mangrove range expansion.
As sea level rises and global warming accelerates, more coastal habitat will become suitable for mangrove colonization. Along the Big Bend coastline in the northern Gulf of Mexico, we showed that forest islands that are currently transitioning to salt marsh and adjacent marsh landscapes can also support A. germinans colonization. The current freshwater forest-to-marsh trajectory may be thus modified to include the eventual replacement of a freshwater forest community by a mangrove community, shifting the temperate coastal landscape to one dominated by tropical saline forest. However, as forest islands convert to salt marsh, they also support fauna that regulate mangrove populations via herbivory. Without large influxes of propagules, mangroves may be unable to overcome the resulting strong, top-down predation control present in these transitioning habitats. While bottom-up controls like temperature and tidal flooding are known to dictate large-scale patterns and rates of mangrove expansion, our study supports the recommendation by He & Silliman (2016) that top-down controls need to be evaluated alongside abiotic controls when investigating and modeling mangrove expansion in coastal systems.
We thank Waccasassa Bay Preserve State Park for permission to conduct research at Turtle Creek. We also thank University of Florida Seahorse Key Marine Lab for providing transportation support, especially Captain Kenny and Rose McCain. We acknowledge Shawn Taylor and Nathan Reaver for much appreciated field assistance. AL acknowledges funding from the University of Florida Graduate School Fellowship and H. T. Odum Graduate Fellowship. CA acknowledges funding from National Science Foundation Department of Environmental Biology (Award 1546638).
Supplementary material 1 (M4 V 17755 kb)
- Castaneda, H. & F. Putz, 2007. Predicting sea-level rise effects on a nature preserve on the Gulf Coast of Florida: a landscape perspective. Florida Scientist 70: 166–175.Google Scholar
- Cavanaugh, K. C., J. R. Kellner, A. J. Forde, D. S. Gruner, J. D. Parker, W. Rodriguez & I. C. Feller, 2014. Poleward expansion of mangroves is a threshold response to decreased frequency of extreme cold events. Proceedings of the National Academy of Sciences of USA 111: 723–727.CrossRefGoogle Scholar
- FNAI, 2015. Florida Land Cover Classification System Definitions for the Cooperative Land Cover Map v3.1.Google Scholar
- Lewis, R. & F. Dunstan, 1975. The possible role of Spartina alterniflora Loisel in establishment of mangroves in Florida. Proceedings for Second Annual Conference on the Restoration of Coastal Vegetation in Florida May 17, 1975. Hillsborough Community College, Tampa: 82–100.Google Scholar
- Lugo, A. E. & C. Patterson-Zucca, 1977. The impact of low temperature stress on mangrove structure and growth. Tropical Ecology 18: 149–161.Google Scholar
- Patterson, C. S., I. A. Mendelssohn & E. M. Swenson, 1993. Growth and survival of Avicennia germinans seedlings in a mangal/salt marsh community in Louisiana, USA. Journal of Coastal Research 9: 801–810.Google Scholar
- R Core Team, 2016. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna. www.R-project.org.
- Reddy, K. R., M. W. Clark, R. D. DeLaune & M. Kongchum, 2013. Physicochemical characterization of wetland soils. In Delaune, R. D., K. R. Reddy, C. J. Richardson & J. P. Megonigal (eds), Methods in Biogeochemistry of Wetlands. Soil Science Society of America, Madison: 41–53.Google Scholar
- Rhoades, J., 1996. Salinity: electrical conductivity and total dissolved solids. Methods of Soil Analysis, Part 3: Chemical Methods. Soil Science Society of America and American Society of Agronomy, Madison: 417–435.Google Scholar
- Vince, S. W., S. R. Humphrey & R. W. Simons, 1989. The Ecology of Hydric Hammocks: A Community Profile. U.S. Department of the Interior, Fish and Wildlife Service, Research and Development.Google Scholar
- Williams, K., M. MacDonald, K. McPherson & T. H. Mirti, 2007. Ecology of the coastal edge of hydric hammocks on the Gulf Coast of Florida. In: Ecology of Tidal Freshwater Forested Wetlands of the Southeastern United States. Springer, Dordrecht: 255–289.Google Scholar