Martinique, a mountainous and volcanic island in the Lesser Antilles archipelago (Figure 1), is the result of subduction between the North American and Caribbean plates, a phenomenon that gives rise to its unique topography marked by massifs in the north, hills in the south and a diversity of soil types [11]-[14].

Figure 1. Location of Martinique in the Lesser Antilles archipelago, Caribbean.

This topographic and pedological diversity, influenced by orographic precipitation and altitudinal climatic variations, generates various bioclimates ranging from dry to hyper humid and various forests such as xerophilous forests at low altitude, mesophilous forests at medium altitude to submontane and montane rainforests [15]-[20].

Martinique, with its diversity of soils, reliefs, hydrography and vegetation, as well as a strong varied anthropization, offers a great heterogeneity of habitats favorable to herbaceous Rubiaceae, which disperse mainly by barochory and zoochory [4][17][21][22]. Forest edges for example, transitional zones between forest and open environments, constitute habitats rich in biodiversity, in particular for pollinators, throughout the year [23]-[25]. These edges constitute ecological corridors that facilitate biological exchanges in a fragmented landscape, offering an ideal habitat for herbaceous Rubiaceae [23]-[25]. Despite its modest surface area (1128 km²), Martinique is home to a diversity of species from the Rubiaceae family, contributing to the biodiversity of the archipelago, a component of the Caribbean hotspot recognized worldwide [4][5][26]-[28]. Around 89 species of the family are present on the island, including 14 herbaceous species divided into 6 genera, but the majority of them remain rare (Appendix1, Figure 2), [2]-[5]. The identification of these plants is possible with the help of floras and reference works containing in particular illustrations facilitating their determination [2][3][8].

Figure 2. Abundance of herbaceous Rubiaceae [2][3].

The data were collected exclusively by non-exhaustive floristic surveys, carried out in 2020 with the help of a few members of the Association of Martinican Botanists (Figure 3). This study uses an approach aimed at analyzing the main trends, as well as the interactions between plant species within largely herbaceous communities [23][29]-[32]. Thus, walking on the hiking trails, in various bioclimates, 29 surveys were carried out targeting the Rubiaceae encountered in the forest edges of the island (Appendix2). One of the main difficulties was then to identify as many species of the family as possible in these very dense edges, during our only passages on each of the hiking trails. Indeed, regardless of the bioclimate, these edges proved to be extremely dense, composed largely of herbaceous plants but including other physiognomic types (shrubs, lianas, liana shrubs, etc.) and therefore constituting multiple populations of very varied surfaces.

Figure 3. Location of surveys.

Wanting to inventory and obtain a representative sample of this diversity of populations was almost impossible. We therefore chose to arbitrarily select a reference surface corresponding to the square meter. This standard methodology consisted of placing wooden sample squares of 1 meter on each side, on non-contiguous surfaces where Rubiaceae were identified (Figure 4 and Figure 5).

Figure 4. Cross-Sectional diagram illustrating the structure of the forest edges traversed.

Figure 5. Some photos to illustrate the different types of forest edges traveled and sample squares made of 1 m × 1 m. (1. Cap Macré, 2. Morne Aca, 3. Fond Baron, 4. Trace des Jésuites).

The objective of the sample squares was to provide an overview of the number of Rubiaceae, their abundances and densities per square meter, and the types of populations to which they can contribute. For each sample square, a set of ecological and floristic descriptors were systematically noted, such as the name of the species, their number of individuals, their height, their physiognomic type as well as their phenological state (flowering and fruiting periods). For lianas or seedlings (less than 10 cm), we simply noted their abundance when their identification was possible (+ very low (number 1 to 5); ++ low (number 5 to 15); +++ average (number 15 to 30); ++++ high (number 30 to 45); +++++ very high abundance (number 45 to 100 and more)). Furthermore, based on the GPS points of the surveys and the QGIS software version 3.36, the biotopes were characterized (Appendix2). The average annual minimum and maximum temperatures, the average annual precipitation were noted using climatic data from Météo-France (French meteorological organization). The slope gradient and altitude were recovered using altimetric data from the IGN (National Geographic Institute, BDALTI 25 meters) and the soil types of the surrounding forests using pedological data from the IRD (Research Institute for Development, soil map of the Antilles at 1/20,000). The level of sunshine received on the ground was noted on site, according to two simple variables (medium (50% sunshine) or high (100% sunshine)). The data obtained were then entered into Excel software and ecological indices and indicators were used for the treatment of Rubiaceae or stations [2][4][15][17][29]-[31][33].

We distinguish:

Species richness (number of species identified);Abundance (number of individuals per species);Frequency (presence of a Rubiaceae in a specific number of sample squares);Density (ratio between the number of individuals of a species and the area of the surveys), calculated according to the following equation:

where “D” is the density, “n” is the number of individuals of a species, “a” is the area of the sample square.

The distribution index (distribution of Rubiaceae within the surveys, expressed by the product between the frequency and the density per species), calculated according to the following equation:

where “Id” is the distribution index, “F” is the frequency and “D” the density.

Simpson’s diversity index (floristic diversity and numerical dominance of one or more species), calculated according to the following equation:

where “λ” is the Simpson index, “S” is the total number of species, “n_i” is the number of individuals of a species and “N” is the total number of individuals. The index is close to 0 when the specific diversity is high, no species dominates numerically; and close to 1 when the diversity is low and one species dominates numerically. Finally, multivariate statistical processing was carried out using XLSTAT software in order to characterize of our surveys [29][34][35].

The surveys were carried out on various sites, covering bioclimates ranging from dry subhumid to hyperhumid (Figure 3, Appendix2 and Appendix3). The sample squares are distributed in habitats corresponding to the edges of xerophilous, xero-mesophilous (transition zone), meso-hygrophilous (transition zone) and hygrophilous forests (Table 1). The altitudes range from 1 to 635 meters, and the slopes are gentle to moderate (Appendix2). The hiking trails are more or less wide and of a diverse nature, ranging from secondary roads to stone and earth paths. However, the soils of the surrounding forests are mainly composed of vertisols (rarely marine alluvium) in dry subhumid bioclimates, and of red or brown montmorillonitic soils in more humid bioclimates (Appendix2). Most of the surveys are located in semi-shaded areas, with sunlight on the ground limited by the canopy cover of the surrounding trees (Appendix2). In total, the 29 surveys cover an area of 29 m². Approximately 2276 specimens were recorded, of very diverse physiognomic types, belonging to 96 species, divided into 79 genera and from 40 different families (Appendix3).

Two species are invasive according to the DEAL (Directorate of the Environment, Planning and Housing in Martinique): Clerodendronpaniculatum and Dichrostachyscinerea [36]. We find mostly herbaceous species but also lianas, ferns, or even seedlings or young individuals of future shrubs, trees and bushes. Among these specimens, 228 belong to 7 species, divided into 5 genera of the Rubiaceae family (Appendix3).

Table 1. Summary table of floristic surveys carried out between August and October 2020 (Appendix2 and Appendix3).

3.1.1. Forest Edges of Xerophilous Forests: Cap Macré

The species recorded in the sample squares of Cap Macré, present diversified physiognomic types. The flowering of some species, which is conducive to attracting pollinators, indicates an active reproductive capacity (Table 2). The overall low density and abundance of species suggest a relatively stable environment, favourable to ruderal plants able to adapt to water stress and high light. In at least three surveys, the Simpson index approaches 1, reflecting a marked abundance of several species. Only one shrub and two herbaceous plants of the Rubiaceae family are recorded, of which only Diodiaocymifolia (herbaceous) stands out for its particularly high abundance and density.

Table 2. Contents of the sample squares made at Cap Macré in Marin.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (T: Tree; Sh: Shrub; He: herbaceous; L: liana)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

3.1.2. Forest Edges of Xerophilous Forests: Morne Aca

The surveys of Morne Aca, also in a xerophilous zone, also present a diversity of physiognomic types (Table 3).

Table 3. Contents of the sample squares made at Morne Aca in Marin.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (T: Tree; B: Bush; He: herbaceous; L: liana)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

The abundance of flowering plants indicates an environment favorable to ecological interactions, such as pollination, despite the water stress conditions. The abundance of these species is low but the low values of the Simpson index confirm the presence of a diversity of species in the sample squares.

None of them dominate numerically and the densities are therefore very low, which also characterizes these environments as relatively stable for ruderal plants. We also note the presence of a single Rubiaceae, Spermacoceverticillata, with a single individual and therefore a low density in all the sample squares.

3.1.3. Xero-Mesophilic Forest Edges (Ecotone): Pointe Rouge to Pointe À Bibi Hiking Trail

In these surveys of xero-mesophilic edges, we observe a transition towards a slightly more humid environment, with a very marked presence of trees such as Cassipoureaguianensis, Garciniahumilis and Pisoniafragans (Table 4 and Table5).

Table 4. Contents of the first three sample squares made on the trail from Pointe Rouge to Pointe à Bibi, in Trinité.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (T: Tree; Sh: Shrub; B: Bush; He: herbaceous; L: liana)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

We note the presence of numerous lianas, shrubs and small trees. The few herbaceous plants present are mainly Rubiaceae, often in flower, indicating a good adaptation to the conditions of the ecotone. The diversity of physiognomic types shows an ecological complexity favorable to pollination and seed dispersal. These plant compositions are however characterized by a low diversity of species as shown by the high values of the Simpson index in the majority of sample squares. Floristic diversity is often low and species also have low abundances and densities, probably linked to a response to the very moderate humidity and light conditions of these sites.

Table 5. Contents of the last three sample squares made on the trail from Pointe Rouge to Pointe à Bibi, in Trinité.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (T: Tree; Sh: Shrub; He: herbaceous)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

Furthermore, we note the presence of three Rubiaceae including Geophilarepens, an herbaceous plant very widespread in the sample squares, forming fairly dense carpets whose individuals were difficult to differentiate and count. The species nevertheless seems very adapted to this ecotone, with an average to high abundance of individuals depending on the squares. Spermacoceassurgens is another herbaceous Rubiaceae present in a single square but with a relatively notable abundance and density. Its individuals were in flower when we visited. Finally, the shrub Guettardaodorata is also present in a single square with a relatively notable abundance and density.

3.1.4. Meso-Hygrophilous Forest Edges (Ecotone): Fond Baron

In the first surveys of this meso-hygrophilous ecotone, herbaceous plants predominate, with species such as Diodiaocymifolia and Hyptisatrorubens, often in flower. Plant diversity, both in terms of species and heights, is notable, even if the Simpson index values reveal high population abundances of the species Hyptisatrorubens or Wedeliatrilobata. Many species were in flower during our visit, ensuring a strong attraction for pollinators and good seed dispersal. These transition zones with a moderate water supply allow species to adapt to humidity variations. The low density of plants seems to promote a balance between competition and survival in this partially shaded habitat (Table 6).

Table 6. Contents of the first three sample squares made at Fond Baron, in Fort-de-France.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (Sh: Shrub; He: herbaceous; F: Fern)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

The last sample squares at Fond Baron also show a diversified vegetation of mainly herbaceous species and mostly in flower (Table 7). The low values of the Simpson index confirm this floristic diversity and that no species dominates numerically with the exception of the high population abundance of Desmodiumincanum in square 6 (Survey 20). The density of species is low, but the majority of these plants thrive in this humid environment, with a high production of flowers and fruits favoring reproduction. We note the presence of two Rubiaceae: Diodiaocymifolia (herbaceous) and Gonzalaguniahirsuta (shrub) whose abundances and densities are low. Diodiaocymifolia is nevertheless better distributed in the squares than Gonzalaguniahirsuta.

Table 7. Contents of the last three sample squares made at Fond Baron, in Fort-de-France.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (T: Tree; Sh: Shrub; He: herbaceous)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

3.1.5. Hygrophilous Forest Edges: Boucher Plateau

Surveys conducted on the Plateau Boucher, in a hyper-humid bioclimate, reveal a hygrophilous environment that is particularly favorable to the development of herbaceous plants, characterized by abundant flowering (Table 8). Some species sometimes display high population density and abundance, while overall floristic diversity remains significant, as shown by the Simpson index. This illustrates the strong resilience and adaptability of species to wetlands. The production of flowers and fruits observed in many plants highlights the essential role of these edges as reproduction and dispersal habitats. In addition, the presence of two herbaceous Rubiaceae is noted. The first, Diodiaocymifolia is well distributed in all the squares but its abundances and densities are low. The second, Spermacocesp. was difficult to determine and only has one individual, present in a single square.

Table 8. Contents of the sample squares made at Plateau Boucher, in Fonds-Saint-Denis.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (B: Bush; He: herbaceous; F: Fern)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

3.1.6. Hygrophilous Forest Edges: Trace Des Jésuites

In the hygrophilous edges of the Trace des Jésuites, herbaceous species, particularly Hyptisatrorubens, largely dominate (Table 9). The Simpson index highlights a numerical dominance of certain species, moderately reducing floristic diversity. Despite an overall low density, the abundant flowering promotes the attraction of pollinators and the dispersal of seeds, demonstrating a balance between competition and survival in this shaded and very humid habitat. This plant diversity contributes to the ecological richness of these edges, which play a key role in maintaining biodiversity in a humid tropical environment. Among the species present, we regularly note Diodiaocymifolia and Spermacoceassurgens, whose densities and abundances generally remain low, although they can be locally high.

Table 9. Contents of the sample squares made at the Trace des Jésuites, in Morne Rouge.

Legend. S: Survey/Ph: Physiognomic type in maximum morphogenetic development (He: herbaceous)/H: Height in cm/A:Abundance/D: Density (cm²)/Fl: Flowers (Y: yes)/F: Fruits (Y: yes).

3.2.1. Ascending Hierarchical Classification (HAC)

A hierarchical ascending classification (HAC) was carried out using a presence/absence table of all species. It is composed of 96 columns (species of all families, all physiognomic types included) and 29 rows (surveys or sample squares), (Appendix 3). The general dissimilarity and the “complete link” aggregation method offered the possibility of showing the differences in floristic compositions between our surveys, in the form of distinct groups or classes. The dendrograms obtained thus show five main groups or classes (Figure 6 and Table 10). These groups demonstrate a great heterogeneity of floristic compositions. While the majority of our edge surveys reflect the typical vegetation of adjacent forests, some have particularities that distinguish them from this general model (Table 10). The surveys of groups or classes 3 and 5, initially different in their location, nevertheless show a convergence of their floristic composition. Among the species that compose them, a significant number could be described as “erratic”, reflecting an ecological flexibility that could be linked to the dynamic conditions of these edges.

Figure 6. Dendrogram of dissimilarity of the 29 surveys according to their floristic composition.

Table 10. Typology of stations according to their floristic compositions.

This analysis suggests that forest edges play a role as transitional habitats, serving as intermediate zones between forest ecosystems and other types of environments. They thus shelter a diverse set of species with varied ecological profiles. These areas could therefore contribute to regional diversity by supporting populations of species with varied ecological needs, some of which do not find favorable conditions in dense forests or adjacent open spaces.

3.2.2. Factorial Correspondence Analysis (FCA)

A factorial correspondence analysis (FCA) was performed using the same species presence/absence table, composed of 96 columns (species from all families, all physiognomic types included) and 29 rows (surveys or sample squares). Only the Hellinger distance allows to obtain a link between rows and columns, with the p-value (1.000) almost similar to the alpha significance level (0.9999), but the F1 and F2 axes support more than 30% of the information and give the best possible quality of data representation (Figure 7).

These axes demonstrate that the floristic compositions of our surveys are very heterogeneous (Figure 7, Table 10). While the floristic composition of the majority of our surveys reflects the adjacent forest type, some surveys are exceptions.

Figure 7. Factorial correspondence analysis (Appendix 2and Appendix3).

Survey 28 carried out in a hyper-humid bioclimate at the Trace des Jésuites, for example, stands out from other surveys of the same bioclimate, and the surveys of the dry subhumid bioclimate, xerophilous forests, are split into two distinct groups (Figure 7). Forest edges therefore appear to constitute transitional habitats, sheltering large pools of species with different ecological profiles, including a significant number of so-called “erratic” species.

3.2.3. Principal Component Analysis (PCA)

A principal component analysis (PCA) was also carried out, this time using a cross-tabulation of data, composed of 29 rows (surveys or sample squares) and 7 columns (eco-climatic factors), (Appendix2).

The eco-climatic factors that we took into account were as follows: altitude (meter), slope (%), annual rainfall (mm), average minimum annual temperature (˚C), average maximum annual temperature (˚C), soil type and level of sunshine. For these last two parameters, we had to assign numerical values to each variable in order to be able to take them into account. Thus, the numerical variables are as follows for the types of soils: Marine alluvium (1), Vertisols (2), Fersiallitic soils (3), Red or brown montmorillonitic soils (4). The numerical variables are as follows for the level of sunshine: medium (1), high (2). The links between the variables are very well taken into account, because the F1 and F2 axes of the PCA obtained using the Person Coefficient, support more than 90% of the information and give an excellent quality of their representation (Figure 8).

Figure 8 shows four groups of coherent surveys according to the ecological differences between the areas studied. Axis F1 organizes the surveys mainly according to humidity, altitude and shade gradients. It contrasts areas adapted to dry and open environments (edges of xerophilous and xero-mesophilous forests)

Figure 8. Principal component analysis of the eco-climatic factor matrix (Appendix 2). Legend. T_min = Average annual minimum temperatures, T_max = Average annual maximum temperatures.

with areas adapted to humid and shaded environments (edges of meso-hygrophilous and hygrophilous forests). Axis F2, on the other hand, allows us to single out the surveys of ecotones (transition zones). This PCA attests to the plurality of forest edges inventoried in various bioclimates.