2.1. THE COLOMBIAN CARIBBEAN AND THE STUDY SITES

The Colombian Caribbean is a 1760-km-long coastline that extends from the Gulf of Urabá, northwestern Colombia, to the Guajira Peninsula, close to the Gulf of Venezuela (Figure 1). Warm climates characterize the region, whereas the precipitation is modulated by a gradient that goes from wet (mean annual precipitation of 2500 mm, Gulf of Urabá) to arid conditions (mean annual precipitation of 550 mm, Guajira Peninsula). The Intertropical Convergence Zone (ITCZ) modulates the bimodal distribution of precipitation and consequently the fluvial sediment supply to the coastal environments. Our review of the evolution of Colombian Caribbean mangrove forests is based on six sedimentary records from four locations in the Colombian Caribbean: core Calancala from the Guajira Peninsula (^{Urrego et al., 2013}), cores La Flotante and Navío from Cispatá Bay (^{Castaño et al., 2010}), cores Candelaria and El Uno from the Gulf of Urabá (this study), and core Honda Bay from San Andrés Island (^{González et al., 2010}).

Figure 1. General map of location of the study sites along the Caribbean. 

Guajira Peninsula (northernmost Colombia): The Calancala core (11°34’30’’N, 72°52’36’’W) was retrieved from a branch of the Ranchería River delta that is the main source of freshwater of the Guajira Peninsula. The river reaches the coastal plain near the city of Riohacha (Figure 1), and the Caribbean Low Level Jet that blows parallel to the Guajira coast induces coastal upwelling (^{Andrade and Barton, 2005}). The sediments of the area are mostly alluvial, underlain by aeolian sediments. The coastal sedimentary plain is slightly undulated and is mainly fed by alluvial sediments from the Cesar and Ranchería rivers. An estuary that opens only during the short rainy season is formed by the river mouth (Instituto Geográfico Agustín Codazzi, ^{IGAC, 2009}).

Gulf of Urabá: The Gulf of Urabá is the largest semi-enclosed sea inlet along the Colombian Caribbean coastline. The coastline of the gulf is characterized by low intertidal ranges, whereas rivers such as the Atrato at the west and the Turbo at the east exert a strong influence on the geomorphic dynamics because of their high discharge. Thus, the coastline is heavily affected by sediment accumulation and erosion caused by rising sea levels, alluvial sediment transport, and biogenic factors (^{Bernal et al., 2005}). Because of the high fluvial input, soil salinity and interstitial salinity are very low. While the Candelaria Bay core is located on the west side of the gulf close to the Atrato River delta, El Uno Bay core is placed on the east coast of the gulf close to the mouth of the Turbo River (Figure 1).

Cispatá Bay: This bay is located on the northwestern coast of Colombia and encloses a big lagoon complex surrounded by the largest mangrove area of the region. The Sinú River is the main freshwater source, and thus the dynamics of the delta exert high influence on the regional geomorphology. Indeed, the delta has changed its position six times during the last ~350 years. Lagoons La Flotante (9°23’37’’N, 75°48’49’’W) and Navío (9°24’19’’N, 75°51’41’’W), where the cores were retrieved, are part of this system (Figure 1). Regional mean annual precipitation is 1338 mm (^{Serrano, 2004}; ^{Castaño et al., 2010}).

San Andrés Island: The island is part of the archipelago comprised of San Andrés, Providencia, and Santa Catalina Islands, lying about 800 km from the Colombian continental coast and about 150 km off Nicaragua (^{IGAC, 1986}). The Honda Bay (12°33’N, 81°42’W) is located on the eastern coast of the island, and is surrounded by a broad mangrove area, with direct communication with the sea on an emerged marine terrace. The mean annual precipitation is 1900 mm (^{Díaz et al., 1995}).

2.2. SEA LEVEL RISE AND GEOMORPHOLOGICAL DYNAMICS

Following the worldwide trend, sea levels have risen steadily along the Caribbean Colombian coast through the last 200 years. The rates of mean regional sea level rise have been estimated at 0.20 cm/yr from 1907 to 1997, and 0.36 cm/ yr for the last five decades (data from two recording stations: Cristóbal, Panama and Cartagena, Colombia, ^{Torres et al., 2008}). This sea level rise record coincides with the trends of climate indices (SOI and NOI) reported by ^{Mendelson et al. (2005)}.

Since the end of the Little Ice Age (~1850 AD) to present, mangroves have shown an increasing representation at several locations of the Colombian Caribbean (Figure 2): El Uno and Candelaria bays in the Gulf of Urabá, La Flotante and Navío in the Cispatá Bay, and Calancala in the Guajira Peninsula. Such expansions have been mainly characterized by a steady increase of Rhizophora mangle pollen and to a lesser extent Avicennia germinans and Laguncularia racemosa. The mean sedimentation rates of these cores have been high since 1850 AD (Figure 3). In El Uno and Candelaria bays the mean annual sedimentation rates have been 0.79 and 0.89 cm/yr, respectively, and in the Cispatá bay these rates were 0.33 and 0.26 cm/ yr in La Flotante and Navío cores, respectively (^{Castaño et al., 2010}). In contrast, a sedimentation rate of 1.44 cm/yr was recorded for the Calancala core (^{Urrego et al., 2013}). High sedimentation rates are related to the high sediment loads of the rivers, a common feature of deltas (Turbo River, 403.29 t km² yr^-1; Atrato River, 315 t km² yr^-1; and Sinú River, 167 t km km² yr^-1). It has been suggested that Caribbean mangroves could tolerate rates of sea level rise of about 0.08 - 0.09 cm/yr, and would be at risk of disappearing with increases between 0.09 and 0.12 cm/yr (^{Ellison and Stoddart, 1991}; ^{Ellison, 1993}; ^{Field, 1995}). However, our data show that mangroves in the Colombian Caribbean have not only survived high mean rates of sea level rise (between 0.20 and 0.36 cm/yr) but have expanded in territory. Regional mangrove expansion has probably benefited from high fluvial sediment inputs that provide materials to offset sea-level rise (^{Parkinson et al., 1994}; ^{Krauss et al., 2014}).

Figure 2. Diagrams of pollen percentages of mangroves from analyzed sediment cores from the Colombian Caribbean during the Anthropocene. A generalized increase of mangrove forests since 1850 AD and especially after 1950 is recorded in all of them. 

Figure 3. Comparison of sedimentation rates of mangroves from analyzed sediment cores from the Colombian Caribbean during the Anthropocene. Note the increase of sedimentation rates in the continental records located close to deltas, derived from changes in river discharges. 

Despite the high sedimentation rates recorded at the Atrato River delta, there were differences in both sedimentation rates and mangrove ecological succession between the two localities sampled in the region. Differences between these locations (Candelaria and El Uno bays) are likely modulated by geomorphology, the specific location of the coring sites within the delta, and the sediment load that each river branch may carry according to the size of the drainage basin. At Candelaria Bay, north of the Atrato River Delta (Figure 2), mangrove pollen shows a decreasing trend towards the present, despite the high sediment load transported by the Atrato River. However, the bay is in direct contact with the sea, receiving high impact from waves (^{Suárez et al., 2015}). This marine influence and the increasing sea level have caused coastal erosion and local mangrove losses. Contrastingly, El Uno Bay, located on the eastern side of the Gulf of Urabá (Figure 2), shows an expansion of mangrove forests during the last two centuries, probably related to a local reconfiguration of the coastal geomorphology. At this site, the Turbo River spit is advancing, causing a natural closure of the bay, which in turn is becoming a coastal lagoon more suitable for mangrove expansion. The eastern side of the Gulf of Urabá, where El Uno Bay lies, receives more sediment than the western side, mostly related to littoral drift and a lesser impact of strong winds (^{Suárez et al., 2015}). This is also the case of Cispatá Bay where changes in mangrove and marsh vegetation reflect at least four migrations of the Sinú River delta. Hydrological and sedimentological dynamics changed the relative position of the lagoons with respect to the shorelines and the river delta. These examples show the critical role of coastal protection and/or geomorphological position for mangroves coping with sea level rise.

2.3. INCREASED SOIL AND PORE-WATER SALINITY AFFECTING MANGROVE SUBSTRATE

Precipitation increases have contributed to the resilience of Caribbean mangrove communities to sea level rise. During periods of high rainfall, mangroves face the indirect effects of increased erosion rates, high continental sediment loads and lower soil salinity. In contrast, when rainfall decreases for longer than the regular dry season (e.g., during strong El Niño events), mangrove forests might die off or change their species composition. Apart from the obvious consequences of rainfall reduction and temperature increases, droughts cause a substantial increase in both soil and interstitial salinity. When the substrate becomes more saline, mangrove community composition switches towards the dominance of Avicennia germinans, which is the most salt tolerant mangrove species (^{Sobrado, 2000}). Despite its low representation in mangrove pollen spectra, during prolonged droughts this species increases at the expense of Rhizophora mangle, sometimes becoming the dominant species of the forest stand (^{Ellison, 1996}; ^{Monacci et al., 2009}), a process that has been observed in some instances in the Colombian Caribbean (Figure 2, La Flotante, Calancala and Honda Bay; ^{Castaño et al., 2010}; ^{González et al., 2010}; ^{Urrego et al., 2013}). Similar high-salinity conditions can also be produced by increases in sand sedimentation that cause relief increases, but that are easily drained. This process combined with high evaporation rates and shallow water causes increases of soil salinity, and therefore the turnover from Rhizophora mangle to Avicennia germinans (^{Twilley et al., 1999}; ^{López-Medellín et al., 2011}; ^{Monacci et al., 2011}). At El Uno Bay, high deposition of sand was recorded in the 1950s when the Turbo River bed was artificially deviated. This anthropogenic intervention changed the local sedimentary dynamics and the composition of the mangrove community, turning a typical fringe mangrove stand into a typical inland community (^{López-Medellín et al., 2011}). Thus, the dominance of Avicennia germinans can either be a signal of mangroves adjusting to current climate change (e.g. López-Medellín et al., 2011), or alternatively a local change in the geomorphology and the sediment dynamics as shown for El Uno Bay.

Several palynological records have shown strong reduction of Caribbean mangroves since the Medieval Warm Period, possibly as a result of decreasing rainfall. Extreme events such as storm surges and hurricanes that have been recorded in several sites have also changed the mangrove forest composition reflected in the sedimentary sequences (^{López-Medellín et al., 2011}; ^{McCloskey and Liu, 2013}). Increases in Avicennia germinans have been reported as indicators of early colonization after hurricanes and windstorms at Honda Bay (Figure 2) (^{González et al., 2010}) and in La Guajira Peninsula (^{Bernal et al., 2016}). The climates of both regions were characterized by a strong and prolonged dry season after the occurrence of the extreme event. Increases in Avicennia germinans were probably the result of the high resprout capacity of damaged stems and the increase of light availability caused by massive defoliation (^{Smith et al., 1994}; ^{Baldwin et al., 2001}). Avicennia germinans has been regarded as the best-adapted mangrove species to conditions associated with current climate change (increasing temperatures, salinity, and drought frequency; ^{Field, 1995}; ^{Snedaker, 1995}). Thus, its expansion during the last century in San Andrés Island (^{Urrego et al., 2009}; ^{González et al., 2010}) and other localities (^{Behling et al., 2001}) is related not only to droughts and high salinity, but also the increased frequency of extreme storms and winds.

2.4. HUMAN DISTURBANCES

Human activities have affected the extension and composition of most Caribbean mangrove forests, obscuring the signals of current climate change on these ecosystems and therefore on their pollen spectra. Through the last few decades, substantial losses of mangrove forest extensions have been attributed to human disturbances (^{Rull et al., 1999}; ^{Martinuzzi et al., 2009}). Nevertheless, clear evidence of the long-term changes, such as pollen records, are rather scarce. Human disturbances to mangrove ecosystems are represented by either the expansion of Laguncularia racemosa (^{Benfield et al., 2005}) and Avicennia germinans at the expense of Rhizophora mangle, or mangrove replacement by herbaceous vegetation (grasslands or crops), especially the fern Acrostichum aureum (^{Medina et al., 1990}). However, these responses have proven highly dependent on species’ differential tolerance to soil salinity and local conditions.

Around 1850 AD, at Honda Bay, San Andrés Island, Colombian Caribbean, mangroves and beach vegetation were replaced by coconut plantations (^{González et al., 2010}), a practice that expanded widely through the Caribbean since that time (^{Parsons, 1985}). The proliferation of Laguncularia racemosa after 1850 AD is also an indication of disturbances, especially during the last few decades, not only in San Andrés Island but also on the continental coastlines of the Colombian Caribbean (La Flotante and Navío localities from the Cispatá Bay, ^{Castaño et al., 2010}; and Calancala in the Guajira Peninsula, ^{Urrego et al., 2013}). While in San Andrés Island, Zea mays pollen indicating its cultivation was detected in the surroundings of the mangrove forest, in Cispatá, the pollen spectra revealed swampy areas covered by rice crops. Although after 1900 AD these crops were abandoned because of a marine incursion, Laguncularia racemosa has prevailed so far as an indication of pervasive anthropogenic disturbances. In the Gulf of Urabá, mangrove conversion into grasslands for cattle raising is indicated at El Uno Bay (Figure 2). Also, assessments of current vegetation in the Gulf of Urabá have reported expansions of Avicennia germinans, likely as a result of human disturbance resulting in higher sedimentation rates (^{Suárez et al., 2015}).