Mangroves occupied intertidal locations in the southeastern USA at least as far back as the early Eocene Epoch (ca. 45 Million years bp), but those fossil deposits were associated with a vastly different coastline boundary driven by a warmer climate and higher sea level (Berry, 1916, 1924; Westgate & Gee, 1990). Mangrove forests from the Eocene Epoch likely occurred at densities similar to those seen in modern-day Neotropical mangrove forests, just much farther north (Sherrod & McMillan, 1985; Gee, 2001). The first fossil evidence of Avicennia in the Caribbean appeared in the late Miocene Epoch (ca. 10 Million years bp), and by the mid-Pliocene Epoch (ca. 3.5 Million years bp) multiple mangrove genera were evident (Graham, 1995). A prominent lack of mangrove fossil evidence along the northern Gulf just preceding the Pleistocene Epoch (ca. 11 700 years bp) until 3000–4000 years bp (from Holocene peat deposits in south Florida) suggests an eradication event for mangroves along the northern Gulf of Mexico, perhaps related to colder temperatures when mangroves were aligned in distribution closer to the equator (Sherrod & McMillan, 1985).
At the northern limits of present-day mangrove extent in the Gulf of Mexico, population extent has, in the recent past, been periodically reduced by frost (McMillan & Sherrod, 1986), with heavy frost in 1983 and 1989 leading to 95–98% loss amongst several of the northernmost populations (Lonard & Judd, 1991; Everitt et al., 1996; Montague & Odum, 1997). This observation prompted Snedaker (1995) to suggest that periodic heavy frost would limit northern expansion for some time. Ecotypic differences in cold tolerance among natural mangrove populations in the Gulf do have the potential to buffer this impact somewhat. This is especially true for populations of Avicennia germinans (McMillan, 1971); those populations growing along the Texas coast were especially tolerant to freezing among others surveyed in the wider Caribbean region (Markley et al., 1982). However, in more than 20 years since the 1989 freeze event, winters have been sufficiently mild to allow rapid expansion of mangroves at their northern limits into salt marsh, documented in Texas (Comeaux et al., 2012; Bianchi et al., 2013), Louisiana (Perry & Mendelssohn, 2009; Alleman & Hester, 2011; Pickens & Hester, 2011) and Florida (Stevens et al., 2006).
Avicennia germinans coverage increased from 57 ha in 1986 to 1182 ha in 2006 in Louisiana, but fluctuated from a maximum documented coverage of approximately 2180 ha in 1983 before the freeze of that same year (Giri et al., 2011). By another account, A. germinans increased in abundance by nearly fivefold between 2002 and 2009 within the Louisiana deltaic plain (Michot et al., 2010). Populations of A. germinans seem to be regulated strongly by air temperatures of −6.7 to −8.9 °C or less (Lonard & Judd, 1991; Stevens et al., 2006; Osland et al., 2013). This threshold is more restrictive for other Neotropical mangrove species (Lugo and Patterson Zucca, 1977; Krauss et al., 2008). For instance, there was no reported survival of transplanted Rhizophora mangle seedlings after the 1983 freeze in Texas (Sherrod et al., 1986), and embolism is a common consequence of temperatures slightly below 0 °C in the same species (Fig. 2a and b). Likewise, Laguncularia racemosa trees are highly susceptible to repetitive freeze-induced dieback events (Fig. 2c), although re-sprouting from the base is a common response in both L. racemosa and A. germinans.
Figure 2. (a) Air temperatures (°C) for the Ten Thousand Islands region of Florida, USA from November 2006 through April 2007, with days having subzero temperatures highlighted (inset graphs). These subzero temperatures were responsible for (b) branch tip mortality from vascular embolism in Rhizophora mangle, and (c) complete stem dieback in many Laguncularia racemosa trees growing in open environments. Avicennia germinans trees in the Ten Thousand Islands region were generally unaffected by this freeze. (Temperature data source: DBHYDRO Browser, South Florida Water Management District, www.sfwmd.gov/dbhydro, Station SGGEWX, accessed 11 April 2013).
Download figure to PowerPoint
Indeed, after extensive losses during the 1983 freeze, mangroves have extended in many Gulf study sites since 1984 (Giri et al., 2011) although have not reached pre-1983 extent (C. Giri, unpublished results). Mangrove trees have been documented visually in Louisiana as early as 1938 (Penfound & Hathaway, 1938) and in Texas as early as 1853 (cited in Sherrod & McMillan, 1981). Using an historical time-series of aerial photography extending back to 1956, Perry & Mendelssohn (2009) were able to demonstrate that mangroves first occupied their Louisiana site in 1995. Along with a reduced incidence of freeze-induced mortality, recent expansion of mangroves in Louisiana has been assisted by widespread dieback of S. alterniflora resulting from drought; Avicennia germinans was unaffected by drought and proliferated (McKee et al., 2004). Environmentally mediated competition between S. alterniflora and mangroves also occurs along latitudinal gradients in Florida (Kangas & Lugo, 1990) and was probably of importance during post-Pleistocene recolonization of mangroves toward northern latitudes. A recent analysis applied to the northern Gulf suggests that short-statured A. germinans vegetation has an overall lower requirement for water use in early growing season assessments than S. alterniflora (Krauss et al., 2013). This may help to explain the differential survival of A. germinans over S. alterniflora during drought, and suggests an interaction between climate variability in both temperature and rainfall (Krauss et al., 2013).
Much of what we are now documenting in the Southern USA is the northern boundary of the post-Pleistocene recolonization (sensu Sherrod & McMillan, 1985). Currently, mangroves (primarily A. germinans) have also extended north on the Florida Atlantic coast at least as far as St Augustine, occupying back-barrier intertidal flats as scattered clusters of individuals (29°57′59″N), and have expanded within this estuary since the early 1990s. In fact, A. germinans has expanded into salt marsh at several other sites on the Atlantic coast, including the Indian River lagoon (Harris & Cropper, 1992). To the south, Rhizophora mangle has expanded landward more than a kilometer into previously Cladium and Eleocharis marshlands in the Everglades (Ross et al., 2000), possibly in response to higher sea levels, changing water levels, and shifting fire regimes (Smith et al., 2013). Similar landward expansion has been noted on the Pacific coast of Mexico at Magdalena Bay, Baja California. Here, a 20% increase in mangrove extent through landward encroachment into sparse halophytic shrubland was attributed to sea-level rise, and was particularly pronounced during El Nino seasons (Lopez-Medellin et al., 2011).
On the Gulf Coast of Florida, mangroves increased coverage in the Ten Thousand Islands National Wildlife Refuge by 35% since 1927, principally at the expense of salt marsh (Krauss et al., 2011). Over a similar time period, oscillations between marsh and mangrove area have been documented in other Gulf coastal areas of Florida (Egler, 1952; Bischof, 1995; Smith et al., 2013); sometimes to the detriment of marsh and sometimes to the detriment of mangroves. In the absence of any discernable change in mean number of freeze days over the period, encroachment of mangroves onto marsh was attributed primarily to the increase in sea level over the period (2.24 mm yr−1 at the Key West station: Krauss et al., 2011). After comparing mangrove extent at three sites in Cedar Keys between 1995 and 1999, Stevens et al. (2006) predicted that all three sites would develop complete mangrove cover within 25–30 years, if not impacted by frost. Our assessment of the same sites using 2012 aerial photography (Google Earth imagery, 19 January 2012) suggests that this outcome has been realized in less than half the predicted time.
There are insufficient historic data on the southeast Japanese coast to unequivocally argue for an extension in natural range of Kandelia obovata (syn. K. candel). The northern limit of K. obovata in Japan was reported by Wakushima et al. (1994) to be Kiire, Kagoshima Prefecture (31°30′N), although they note the long-term survival of a planted population in the estuary of the Aono River in the Shizuioka Prefecture at 34°38′N.
Determining changes in northern limits of mangroves in China and Taiwan is complicated by extensive clearance. A further complication in China is the introduction of mangroves north of their natural limits: K. obovata in Zheihang (Li & Lee, 1997); and Sonneratia caseolaris and Bruguiera sexangula in Guangdong (Li et al., 1998). One of the few locations where mangroves and salt marshes coexist in near natural state on the Chinese mainland coast is in the Zhanjiang Mangrove National Nature Reserve on the Leizhou Peninsula of Guangdong Province (21°34′N; 109°45′E). The reserve is a Ramsar-listed wetland of international significance and supports nearly one-third of China's mangroves. Regionally, mangroves have declined due to agricultural developments, and extensive dyking restricts landward encroachment (Leempoel et al., 2013). However, within the reserve mangroves, dominated by A. marina, Aegiceras corniculatum and K. obovata, have expanded fourfold, including encroachment on salt marsh (Durango-Cordero et al., 2013). Mangroves have also proliferated in the Zhuhai Qi'ao Provincial Nature Reserve (22°26′N; 113°37′E), established in 2000 to encourage the rehabilitation of mangroves (Peng et al., 2009). Spread in the extent of the native mangrove K. obovata as well as Sonneratia apelata, introduced from the Sunderban (Ren et al., 2009), has led to a decline in Spartina alterniflora saltmarsh (G. Lei, personal communication).
The northernmost mangrove community in Taiwan is located in the Danshui River estuary (21°09′N; 121°26′E) and is the largest K. obovata forest in the world (Lee & Yeh, 2009). The mangrove and associated Phragmites communis salt marsh community has been protected in the Danshui Mangrove Reserve since in mid-1980s. Mangroves have doubled in extent since the establishment of the reserve, and in detailed satellite imagery analysis Lee & Yeh (2009) were able to demonstrate landward encroachment of mangrove on nonmangrove vegetation, presumably Phragmites salt marsh.