Short‐distance barriers affect genetic variability of Rhizophora mangle L. in the Yucatan Peninsula

Abstract The environmental variability at local scale results in different physiognomic types of mangrove forest. However, this variability has never been considered in studies of mangrove genetic variability. This study analyzed the genetic and morphological variability and structure of Rhizophora mangle at regional and local scales in the Yucatan Peninsula. Thirteen mangrove populations (eight scrub and five tall), located in seven sites, were sampled, and their morphological variability and relationship with the availability of phosphorus and salinity were analyzed. The diversity and genetic structure were estimated at different hierarchical levels with nine microsatellites, also Bayesian inference and Principal Coordinates Analysis were used. We found a great morphological variability of R. mangle that responded to local environmental variability and not to the precipitation gradient of the peninsula. The genetic diversity found in the peninsula was greater than that reported for other populations in Mexico and was grouped into two regions: the Gulf of Mexico and the Caribbean Sea. At a local scale, tall and scrub mangroves had significant genetic differentiation suggesting that ecological barriers promote genetic differentiation within sites. These results need to be considered in future population genetic studies and for mangrove management and conservation.

regulate the population´s structure and the ecological processes behind it, as well as their role in the ecosystem services offered (Salgado-Negret & Paz, 2016). Furthermore, the success of restoration, management, and conservation projects depends on conserving and prioritizing the greater genetic and functional variability (Richards, Wares, & Mackie, 2010). However, studies that correlate morphological and genetic differences for the same species geographically separated are very scarce. Recently, conceptual frameworks have identified local genetic adaptation as an important driver of population genetic structure, but only some studies focus on tree species, and a few focus on mangroves (Farnsworth, 1998;Nosil, Funk, & Ortiz-Barrientos, 2009;Orsini et al., 2013).
Mangroves are intertidal ecosystems located in the tropical and subtropical coasts of the world (Tomlinson, 1986). These forests have a great ecological and economic importance as they shelter a supreme diversity of fauna and provide important ecosystem services (Alongi, 2008). The structure and floristic compositions of mangroves change according to geographical and latitudinal gradients (Lot, Vázquez-Yanez, & Méndez, 1975;Lugo & Snedaker, 1974).
At the local level, their composition, structure, and function vary according to environmental gradients, physiological preference, and flood tolerance of the species (McKee, 1996;Rabinowitz, 1978), which result in different type of physiognomic mangrove forests. Lugo and Snedaker (1974) describe five main types: riverine, basin, fringe, overwash, and scrub. This classification scheme is a supplement to the classical zonation patterns described for mangroves in different parts of the world, which has had a ubiquitous success and strengthens the importance of the distribution of mangrove species and physiognomic units (Lugo & Snedaker, 1974).
The genetic variability of mangroves species, particularly that of R. mangle, began to be studied with the use of biochemical markers by Dodd, Rafii, Fromard, and Blasco (1998), and in the past decade with DNA molecular markers by Arbeláez-Cortes, Castillo-Cárdenas, Toro-Perea, and Cárdenas-Henao (2007).
Moreover, although this region represents more than half of the extension of mangrove in Mexico, no studies of genetic variability of R. mangle have been carried out (Rodríguez-Zúñiga et al., 2013;Spalding, Kainuma, & Collins, 2010).
Accordingly, the principal aim of this work was to evaluate the morphological and genetic structure and variability of R. mangle at regional and local scales in the Yucatan Peninsula, Mexico. In addition, we examined the relation of morphological variability to salinity and phosphorous availability. The following specific questions were addressed which are as follows: (a) How the morphological variability of R. mangle relates to salinity and phosphorus availability in the Yucatan Peninsula? (b) what are the levels of diversity and genetic structure in the Yucatan Peninsula?, and (c) at local scale, do the two physiognomic types of R. mangle, tall and scrub, affect the genetic structure and diversity within a site? We would expect that considering the influence of salinity and phosphorus on mangrove morphology, the sites with the highest salinity and lowest phosphorus availability will correspond to scrub populations. Also, the tall and scrub R. mangle will show genetic differentiation within a site.

| Study area and populations sampled
This study was made along the northern and eastern coasts of the Yucatan Peninsula, which is a low elevation karstic platform (slope <1%) with irregular topography (Pannier, 1992). The climate is generally warm and humid and is characterized by three seasons: dry (March-May), rainy (June-October), and early-dry (locally named "nortes"; November-February) (Zaldívar-Jiménez et al., 2010).

| Morphological and environmental characterization
In each population, ten adult individuals (each with at least 10 m distance among each other) were randomly selected. We measured the following: (a) leaf length and width (cm; 3 leaves per individual; the third leaf from the leaf primordium was always selected); (b) length of ten propagules with a graduated ruler (cm); (c) height (m), measured with a Vertex Laser (VL402, Haglöf, Sweden); and (d) stem diameter (cm; performed 20 cm above the last root with a diameter tape or a vernier for the scrub mangrove). At each site, interstitial salinity (~30 cm depth) was measured with an YSI Pro 2030 portable conductivity meter (YSI, Yellow Springs, USA). Also, a soil core with a depth of 30 cm was obtained and transported to the laboratory to measure total phosphorus (P) and extractable P. Total P was determined as orthophosphate after Aspila, Agemian, and Chau (1976) and Parson, Maita, and Lalli (1984). Extractable P was analyzed after Olsen (1954).

| Data analysis
A Canonical Discriminant Analysis was performed to know the morphological differences between populations of Yucatan Peninsula and the most discriminant variables. Also, to distinguish a pattern among the environmental variables to the mangrove type, a Principal Component Analysis was used. Finally, a Canonical Correlation Analysis was applied to identify the correlation among the morphological variables: height, length of the propagule and to include the foliar characters, leaf area was calculated considering the leaf as an oval (Af = π × r 1 × r 2 ; r 1 : leaf width,  DNA was isolated by the modified CTAB/PVP method (Reyes- Medina, 2012). Previously, tissue was macerated in liquid nitrogen, resuspended in 1,000 μl of 2% CTAB plus 10 μl of β-mercaptoethanol, and then incubated at 65°C for 3 hr. Samples were allowed to cool and 30 μl of 20 μg/μl RNAse was then added and incubated at 37°C for 30 min; the enzyme was inactivated at 65°C for 10 min.

| Molecular characterization
Thereafter, it was centrifuged at 18,000 g for 20 min, 700 μl of the supernatant were transferred and extracted with 650 μl of 24:1 (v/v) chloroform-isoamyl alcohol, and centrifuged at 18,000 g for 10 min to separate the phases. This process was repeated twice. For the microsatellite technique, nine loci reported as polymorphic for R. mangle were used as follows: RmBra20, RmBra45, F I G U R E 1 (a) Sampled sites for Rhizophora mangle populations in the Yucatan Peninsula, the physiognomic types are indicated by (T) for tall and (S) for scrub. (b) Ocean currents of the Yucatan Peninsula (Laurindo et al., 2017) RmBra50, RmBra59 (Ribeiro et al., 2013) andRm7, Rm11, Rm19, Rm41, Rm46 (Rosero-Galindo, Aitan-Solis, Árdenas-Henao, Ohme, &Oro-Perea, 2002). Amplification was performed by PCR. Each 20 μl of the amplification reaction consisted of 10× PCR buffer, 1.5 mM MgCl 2 , 0.2 mM dNTPs, 0.15 μM primer (FW, RV), 1 u of Taqpolimerase, and 2 μl of DNA (50 ng). Amplification was performed on an Applied Biosystems Gene Amp PCR SYSTEM 9,700 thermocycler (Thermo Fisher Scientific, USA) at the following temperatures: initial denaturation of 94°C for 2 min, followed by 37 cycles of 15 s at 94°C, 15 s at 50°C (alignment temperatures for each oligonucleotide were according to Rosero-Galindo et al., 2002 andRibeiro et al., 2013), 15 s at 72°C and a final extension of 5 min at 72°C. The PCR products were verified by electrophoresis on 1.5% agarose gels stained with 10 mg/ml ethidium bromide for 40 min at 100 V. PCR products were separated by electrophoresis on denaturing gels of 5% polyacrylamide (19:1 acrylamide bisacrylamide) with 5 M urea and 0.5 X TBE. To the PCR product, 4 μl of a solution of formamide (0.45% bromophenol blue and 0.25% xylene-cyanol, denatured at 94°C for 5 min) was added. The electrophoresis was performed at 60 W for 1.5 hr in a manual sequencer (SQ3 sequencer, Hoefer Scientific Instruments, San Francisco, CA, USA). The fragments were visualized using the silver staining technique (CIRAD, 2002), using a 10 bp marker as a reference and a white light transiluminator for the fragments reading.

| Data analysis
First, to explore how genetic diversity of R. mangle is organized in the Yucatan Peninsula, two approaches were used as follows: (a) An individual assignment test was done with a Bayesian approach implemented in the Program STRUCTURE v.2.3.4 (Pritchard, Stephens, & Donnelly, 2000). The program was run with the Admixture model and the LOCPRIOR option (Hubisz, Falush, Stephens, & Pritchard, 2009). A period of burning of 200,000 and 400,000 iterations after the burning period was used. We performed 20 replicates for each K value (K = 1-15) and checked the consistency of results. The optimal K value was calculated using the ΔK method described by Evanno, Regnaut, and Oudet (2005) and implemented in the STRUCTURE HARVESTER program (Earl & Von Holdt, 2012). 2).
A Principal Coordinates Analysis (PCoA) with mean population genetic distance was performed with the GenAlex v6.5 program (Peakall & Smouse, 2006); this analysis allows the spatial rec-  Note. For key populations: T = tall and S = scrub; diameter (measured 20 cm above the last aerial root); PT: Total Phosphorus; PE: Extractable phosphorus (% of extractable Phosphorus, PE/PT*100). significance was tested for all levels from 1,000 permutations in the Arlequin program V. 3.5.2.2 (Excoffier & Lischer, 2010). The Nei's pairwise population genetic distance matrix was calculated with GenAlex V6 program (Peakall & Smouse, 2006). Also, to evaluate the distance isolation hypothesis, Mantel test was done for sites and populations using the geographic distances, obtained as the coast distance between sites (Km) and Nei's genetics distance matrix (Nei, 1978), with 1,000 permutations by the GenAlex V6 program (Peakall & Smouse, 2006 and allelic richness (Na), all estimators were obtained with GenAlex V6 program (Peakall & Smouse, 2006). Comparisons were made with estimators calculated between regions and mangrove types with a validation of 1,000 permutations with the FSTAT program v.2.9.3.2 (Goudet, 2001). Also, inbreeding coefficient (F IS ) was estimated for all populations, regions, and mangrove types, as well as for all the Yucatan Peninsula. The evaluated estimators were compared among populations, regions, and type of mangrove, using 1,000 permutations with FSTAT program ver. 2.9.3.2 (Goudet, 2001).

| Morphologic and environmental variability
The study populations had high morphological variability within and among sites (Table 1)  Interstitial salinity varied from 0.8 ppt in scrub-Mahahual to 122 ppt in scrub-Progreso (Table 1). The population with the highest total phosphorus (P) content was scrub-Celestún (1.13 mg/g) and the lowest total P content was in tall-Progreso (0.23 mg/g). The extractable P was the highest in scrub-Nichupté (25.06 mg/kg) and the lowest in scrub-Mahahual (1.01 mg/kg). The percentage of extractable P in the sites was less than 6% in all cases (Table 1), and no direct relationship was found between tree height and salinity or extractable P (EP; Figure 3a,b). However, the relationship between environmental and morphological variables was supported by the Canonical Correlation Analysis. The first factor explained 92.45% of the total variation, and a significant correlation among morphology variability, salinity, and EP was also observed (0.36, p < 0.05).
However, the λ Wilks showed a poor predictive power (0.85). This analysis also showed that salinity was the variable that contributed the most to the variability in morphology and had a negative correlation with tree height and leaf area, but EP had a positive correlation with these latter variables (Figure 4).
The response of the two physiognomic types of mangrove to salinity and EP correlation was supported by the Principal Component Analysis; we found that: Sites with high salinity and low or high EP had scrub mangroves (Sia-S1, Cel-S, Pro-S); sites with low salinity and high EP values had tall mangroves (Sia-T, Cel-T, Rio-T, Nic-T); and sites with low salinity and EP values had scrub mangrove (Mah-S, Rio-S, Pue-S, Sia-S2) ( Figure 5). Two sites did not respond to these trends: scrub-Nichupté with the highest values F I G U R E 2 Canonical Discriminant Analysis of the morphologic variables measured in Rhizophora mangle of the Yucatan Peninsula of EP and low salinity, and tall-Progreso with high salinity and low EP values.

| Genetic variability
The 130 individuals of R. mangle sampled in the Yucatan Peninsula were divided into four groups according to the optimal K ob-  Table 2). The tall and scrub populations within a site (with distances of few kilometers) had equal genetic distances, or even higher, than distances between populations from different sites. For instance, the genetic distance between Cel-S and Cel-T (0.12) was the same distance than that between Cel-S and Rio-S (0.12); or the genetic distance between Sia-T and Sia-S2 (0.2), which was higher than the distance between Sia-T and Pue-S (0.17; Table 2).
F I G U R E 3 Relationship among height, salinity, and extractable phosphorus for 13 populations of Rhizophora mangle in the Yucatan Peninsula (see Table  1 for abbreviations). Bars are mean height ± standard deviation; lines are mean ± standard deviation of (a) interstitial salinity and (b) extractable phosphorus The genetic structure of R. mangle in the Yucatan Peninsula showed by the AMOVA indicated that the studied populations differed 13% (p < 0.001). Also, the genetic differentiation between regions was 6.5% (p < 0.001, Table 3). The analysis performed by sites (including indistinctly tall and scrub individuals within sites) showed a lower genetic differentiation (9%, p < 0.05) than by populations. However, the differentiation between regions was similar (6%, p < 0.05) (Table 3). At local scale, the genetic differentiation of tall and scrub populations within sites was high and for some sites even higher than for all the Yucatan Peninsula, with values from 3% to 13.11% (Table 4) F I G U R E 5 Principal Component Analysis of interstitial salinity and extractable phosphorus (measured at 30 cm depth) in 13 Rhizophora mangle populations (see Table 1 for abbreviations) genetic differentiation between physiognomic types than those from the Caribbean Sea (Table 4). The relation among geographic distance and Nei's genetic distance demonstrated by Mantel test among sites of the Yucatan Peninsula showed a significant relation (R 2 = 0.52, p < 0.001); while the analysis made by populations presented a weak but significant relation (R 2 = 0.21, p < 0.001).  Table 1 for abbreviations), the pie charts represent Q value for K = 4 F I G U R E 7 Principal Coordinate Analysis in three dimensions, performed withnine microsatellite loci and 13 populations of Rhizophora mangle. Axis 1 (36.11%) divides the Gulf of Mexico region and the Caribbean Sea; Axes 2 and 3 (38.97%) indicate the genetic difference within each region. Populations of the Caribbean Sea are indicated as closed black symbols and populations of the Gulf of Mexico as closed gray symbols; circles represent tall populations and triangles scrub populations (see Table 1 for abbreviations) Within regions, the Mantel test did not present significant relationship.

| Genetic diversity
The nine microsatelite loci used detected a total of 35 alleles. With the exception of the RM46 locus, all loci were polymorphic. The Yucatan Peninsula had an expected (H E ) and observed heterozygosity (H O ) of 0.37 ± 0.003 and 0.27 ± 0.003, respectively, and an allelic richness (Na) of 2.41 ± 0.02. Also, the H O was higher for the Caribbean Sea region (0.37 ± 0.01) than for the Gulf of Mexico region (0.23 ± 0.02), but their differences between H E and H O were lower (Table 5). The Na was similar in both regions, 2.38 ± 0.05 in the Gulf of Mexico and 2.49 ± 0.08 in the Caribbean Sea (Table 5) Table 5). Tall populations had lower F IS (0.20 ± 0.03) than scrub populations (0.31 ± 0.02), but values were not significantly different.

| Morphologic variability
Although salinity and phosphorous (P) are two of the main factors that have been related to mangrove forest structure, especially in scrub TA B L E 2 Nei's (1978) pairwise genetic distances between populations of Rhizophora mangle in the Yucatán Peninsula  (Feller et al., 2003;Lin & Stenberg, 1992a;Lovelock et al., 2006Lovelock et al., , 2004Naidoo, 2006Naidoo, , 2010, in our study, no direct relation of these variables with height was observed (Figure 3a,b). Instead, the mangrove height responded partially to the interaction of the effects of salinity and P (Figure 4). This agrees with the relationship observed among P assimilation and water availability in mangroves (Mckee, Feller, Popp, & Wanek, 2002;Medina, Cuevas, & Lugo, 2010).

Mah-S Nic-S Pue-S Sia-S
As salinity increases and water availability diminishes, P assimilation can also decrease with consequent growth reduction (Naidoo, 1987).
Nevertheless, given the low correlation and poor predictability of the environmental variables, other variables can interact in the synergy that controls morphological structure. One of them can be the hydroperiod, which regulates the resources and modifies stress factors of a site (Feller et al., 2010;Lin & Sternberg, 2007;Twilley & Rivera-Monroy, 2005 High phenotypic plasticity is related to species adapted to heterogeneous environments and it has important ecological and evolutionary implications (Crispo, 2008;Gianoli, 2004). Mangroves are characterized by a high heterogeneous landscape with environmental settings characterized by high stress factors that control structure and function of populations (Feller et al., 2010;Twilley & Rivera-Monroy, 2005). Similarly, other aquatic species also show a great phenotypic plasticity at very fine scales (Kudoh & Whigham, 1997;Millet, Kristjánsson, Einarsson, & Räsänen, 2013). This plasticity, resulting from the environmental variability, can be an important factor in evolutionary diversification, because it plays a significant role in the relationships among divergent selection, adaptive divergence, and gene flow (Crispo, 2008;Nosil et al., 2009).  (Fruet et al., 2017;Lu, Wang, Li, & Liu, 2016;Nosil, 2007;Spurgin, Illera, Jorgensen, Dawson, & Richardson, 2014).

| Regional genetic variability
Additionally, for conservation biology, it is important to consider possible sources of greater genetic diversity because this represents relevant variability for species persistence and their evolutionary potential (Piñero et al., 2008). For mangroves, for instance, In addition, our study showed, for the first time, that the genetic diversity of R. mangle in the Yucatan Peninsula is clearly arranged in two regions: the Gulf of Mexico and the Caribbean Sea coasts (Figure 6a, c), which correspond to two oceanographic regions (Wilkinson et al., 2009). Based on other molecular markers, differences on allele frequency among these regions have also been observed for populations of R. mangle (Núñez-Farfán et al., 2002) and Avicennia germinans (Nettel & Dodd, 2007) and this differentiation can be explained by the geomorphologic history of the peninsula.  At the most recent glacial period, an extreme geographic isolation took place for many species, with subsequent species-specific patterns of postglacial expansion (Kennedy et al., 2016). In mangroves, the influence on genetic variability attributable to postglacial establishment has been reported in America for R. mangle and A. germinans (Cerón-Souza et al., 2015;Pil et al., 2011). Ocean currents are another factor that plays a role in the genetic structure of the hydrochoerus species R. mangle (Cerón-Souza et al., 2015;Pil et al., 2011). In Brazil, the northern and southern populations of R. mangle have a strong relationship between geographic and genetic distance (R 2 = 0.5) due to currents that interrupt continuous genetic flow along the Brazilian coast (Pil et al., 2011). and keeping genetic differentiation between both regions (Martínez & Pares, 1998;Wilkinson et al., 2009).
In our study, no morphological differences between regions were found, which reflect that the genetic variability did not respond to historical differences between populations, but to local adaptation to environment. For instance, European oaks do not show any association between genetic divergence connected to colonization events and those associated with local selection pressures (Kremer et al., 2002). Therefore, it is important to consider different scales, which can include local adaptation traits to environment and reveal the colonization history of populations. Several factors regulate population genetic structure in natural landscapes and also provide insights into the complex interactions between the environment and the genome that influence the distribution of species, and mediate phenotypic adaptation to local conditions (Bragg, Megan, Andrew, & Justin, 2015;Orsini et al., 2013).

| Local genetic variability
Tall and scrub populations of R. mangle showed significant genetic differences that could be caused by their contrasting environments within a site (Table 1). These environmental differences could represent a reproductive barrier, because of the influence of salinity and nutrient availability on autogamy rate, flowering season, fruit production and ripening, and size of propagules (Coupland, Paling, & McGuiness, 2006;Klekowski, Lowenfeld, & Hepler, 1994;Lowenfeld & Klekowski, 1992;Proffit & Travis, 2010;Sánchez-Núñez & Mancera-Pineda, 2011). Thus, the genetic distance between tall and scrub populations within a site could be explained, in part, for the high autogamy rate characteristic of the species and also to an asynchronous phenology between adjacent populations that interrupts gene flow (Sánchez-Núñez & Mancera-Pineda, 2011). On the other hand, the intricate root system of R. mangle, especially in the scrub mangrove, represents a physical barrier for propagules dispersion (Tonné, Beeckman, Robert, & Koedam, 2017;Van der Stocken & Menemenlis, 2017;Van der Stocken et al., 2015). This suggests that scrub populations have more limited dispersion, which promotes genetic differentiation within a site (Table 4). Also, in scrub mangroves, the smaller size of propagules indicates fewer stored reserves, growth rate and resistance to wind and wave force that could produce a negative selection against them (Boizard & Mitchell, 2011;Dissanayake et al., 2014;Farnsworth & Ellison, 1996;Huxham et al., 2010;Lin & Sternberg, 1995;Nosil, Vines, & Funk, 2005;Proffit & Travis, 2010;Tomlinson, 1986).
The environmental and ecological differences between tall and scrub mangroves, which coexist at short distances, could be generating breeding barriers that limit the genetic flow between them and promoting a greater genetic differentiation even more than among sites (Tables 2 and 4). The higher F IS found in the scrub than in the tall mangrove populations in the Caribbean Sea indicate that the autogamous nature of the R. mangle has accentuated the differences between populations at local scale by genetic drift. The main adaptive advantage of gene drift, despite the low diversity associated with inbreeding, denotes reproductive safety and adaptation to a particular niche (Loveless & Hamrick, 1984). Genetic differentiation is a consequence of the combined effects of natural selection and gene drift, which could fix certain genotypes linked to environmental variables that result in greater adequacy, and whose effect is counteracted with the gene flow (Bragg et al., 2015;De Kort et al., 2014). Although this work was performed with neutral markers, natural selection at a locus can affect the frequency of alleles in the loci attached to them due to the hitch-hiking effect (Smith & Haigh, 1974). Moreover, the morphological plasticity responds to environmental variables, but it may also be under genetic control and respond to pressures of natural selection; thus, genotypes with greater morphological and functional plasticity will be more advantageous in wide environmental ranges (Crispo, 2008;Gianoli, 2004 responses to divergent natural selection may impact genomewide population structure has gained momentum and needs to be considered in future studies (Nosil et al., 2009;Orsini et al., 2013).
The genetic structure of R. mangle, observed at local scale in the Yucatan Peninsula, suggests that each site is probably composed of demes, where the ecological limits could be generating reproductive barriers. In continuous populations, recurrent processes, such as gene flow, genetic drift, and selection, act in concert to shape the genetic structure (Latta, 2003;Lenormand, 2002). Fine-scale genetic differentiation has often been reported in other plant populations, mammals, fishes, and insects (Dewoody et al., 2015;Fruet et al., 2017;Nosil, 2007;Vekemans & Hardy, 2004), even under substantial gene flow, suggesting that strong selective pressure promotes local adaptation at small scale in heterogeneous landscapes (Linhart & Grant, 1996;Parisod & Christin, 2008).
The two different types of R. mangle considered in our study could probably allow us to find a greater genetic structure in the

DATA ACCE SS I B I LIT Y
Morphological and environmental data and microsatellite genotypes are available at https://doi.org/10.5061/dryad.1578ks0.