CONTRIBUTION TO THE KNOWLEDGE OF THE LESSER ANTILLES FLORA:- STUDY OF THE CASE-NAVIRE, NATURALISTIC AREA OF ECOLOGICAL FAUNA AND FLORA INTEREST (MARTINIQUE)

The complexity of the Lesser Antilles vegetation must be linked to the variety of topographic features that influence the structure of some climatic parameters, particularly the rainfall. The above mentioned environmental factors influence a plurality of biotopes colonized by specific phytocenoses. Using survey transects of different minimal areas, we highlighted the specific, biocenotic, structural and architectural heterogeneity of the survey stations. This spatio-temporal differentiation of plant groupings is strongly guided by the activities of the successive human societies. These elements show that the anthropized vegetation of the Lesser Antilles represents a true laboratory for the study of plant succession using the synchronic approach.

The complexity of the Lesser Antilles vegetation must be linked to the variety of topographic features that influence the structure of some climatic parameters, particularly the rainfall. The above mentioned environmental factors influence a plurality of biotopes colonized by specific phytocenoses. Using survey transects of different minimal areas, we highlighted the specific, biocenotic, structural and architectural heterogeneity of the survey stations. This spatio-temporal differentiation of plant groupings is strongly guided by the activities of the successive human societies. These elements show that the anthropized vegetation of the Lesser Antilles represents a true laboratory for the study of plant succession using the synchronic approach.

…………………………………………………………………………………………………….... Introduction:--
True islands are essential systems for the study of migrations, in particular plant migration and the factorial determinism that cause the spatial and temporal species dynamics and the multiple combinations they form (Leigh et al. 1993 Giambelluca, 1998). This occurs because the plant community components represent resources and also participate in the biological cycles ensuring the planet's homeostasis at all levels of integration (Rispoli, 2014). The Lesser Antilles are no exception from this global phenomenon of deregulation due to human impacts (Joseph, 2012;Cano et al., 2009;Losos and Ricklefs, 2009;Helmer et al., 2008) because they have always faced and are still facing heavy human stress. Rich in a plurality of life forms subject to anthropic erosion, this archipelago is part of 24 world biodiversity Hotspots:-the Caribbean one (Myers et al., 2000; Cincottaet al., 2000; Figure 1). In the light of the regular loss of biotopes and therefore of plant taxa and in order to create protected terrestrial areas, it was essential to implement a program of inventories of the Natural Areas of Ecological, Fauna and Flora Interest (ZNIEFF) in Martinique. In this article we will discuss the main ecological and floristic features of a ZNIEFF located in the South of the Caribbean side of the island of Martinique (Figure 1).

Materials:--
The study area is located in the commune of Schoelcher, on the Northern Caribbean coast of Martinique to the West of Fort-de-France ( Figure 1). Due to a topographic and pluviometric gradient (between 1500 mm and 3000 mm) as well as to the Case-Navire river, this site offers a large enough variety of environments creating multiple floristic and soil potentials (Joseph, 2012). The numerous topographical facets represent as many biotopes which allowed and still allow the establishment of very specific plural floristic communities. There are evergreen seasonal tropical plant formations (mesophilic) on the protected slopes and riparian terraces affected by the sub-humid wet bioclimate, evergreen seasonal tropical plant formations of the lower horizons and with xeric features (xerophilic) on exposed slopes and ridges under the influence of the sub-humid dry bioclimate and tropical seasonal ombroevergreen plant formations (hygro-mesophilic) on the riparian terraces and the protected slopes in the interior of the 1212 land characterized by the sub-humid and wet bioclimates (Joseph, 2013). The vast majority of this vegetation is anthropized but in a manner diversified from the point of view of frequency and intensity. Together they form a floristic mosaic where the components have different ages, physical and floristic compositions. We can observe generalist species of the pioneer or post-pioneer stages in combination with others belonging to the advanced stages of plant succession. However, the very varied geomorphology of this part of the municipality of Schoelcher did not allow us to explore the very specific biotopes which lie within the rock ledges where the vegetation is mostly lithophile (Figures 2 a & b). In fact, this area was deeply modified (changed) by human activities that have increased strongly since 1635, the date when the French took possession of the island. The number and types of selected stations allowed us to acquire a fairly accurate picture of the floristic and ecosystem complexity of the study area ( Figure 1).

Figures 2 a & b:--
Modelling of the Fond Rousseau ZNIEFF from aerial photographs.

Methods:--
We conducted surveys which show the diversity of plant species and the associations they form. We have listed 13 stations on the basis of their topographic characteristics. The purpose was to compare the latest changes and understand the spatial distribution of the main species. The method used for this study is part of the macro-ecology.
The objective was to decipher the structural, architectural and functional vegetation dimensions. We used transects subdivided in quadrats, depending on the minimal survey area (between 400 to 1000 m²) 1 we have generated data that are both ecological and floristic descriptors ( Figure 3):-species, number of individuals in populations of regeneration plant species to mature specimens (bio-demographic aspect), diametric classes (sections measured at 1.33 m from the ground in accordance with the international standards), total height and the first branching classes. These elements have enabled us:---to assess the distribution of the tree sections, the formations' architecture, the characteristics of the canopies, -to evaluate the phytomass or biovolumes using the basal area that corresponds to the sum of the circle areas consisting in sections measured at 1.33 m from the ground.
-to know the species distribution between the transect quadrats and stations using the Distribution Index that corresponds to the following formula:-Id = fr × d (fr being the relative frequency and d (nb/ sr) the density corresponding to the number of individuals of the species (nb) divided by the survey area (sr).
-to know the relative dominance of plant species in relation with each other through the Dominance index (ID). ID = Id × St (basal area).

1213
Aside from the dominance index (ID), the comparison between stations and the affine floristic groups of the latter was carried out due to:--.
-a factorial analysis of correspondences (AFC) using the XLSTAT software (new version).
-the use of a datamining descriptive classification technique:-the clustering. It regroups the species in a limited number of groups or clusters. These clusters are not predefined by the analyst but discovered after the implementation of the algorithms. These clusters include species with similar ecological characteristics and species with different environmental characteristics (internal homogeneity and external heterogeneity). Unlike the factorial analysis component (AFC), this method has the big advantage of taking into account all variables without a lower dimensional subspace projection. No information is lost as a result. The results below were obtained using the SimpleKmeans and EM (Expectation, maximization) algorithms which are two algorithms commonly used for the implementation of clustering techniques.

Results:--
General descriptions of the stations:--Station 1:--Evergreen seasonal tropical forest formation (765 m², Riparian terrace, Figure 1). This phytocenosis also called mesophilic exhibits average anthropization and ranks at barely structured secondary stage (two defined strata). The most competing species appear to be in order of ecological importance ( Table 1)      Pimenta racemosa (Bois d'Inde) is the dominant species and has a population characterised by good site distribution and bio-demographic balance (good distribution of age classes). Together with the Pimenta racemosa, the Calliandra tergemina, Eugenia monticola and Myrcia citrifolia they represent the most widely distributed species (Table 3). This anthropized plant formation consists of eco-units belonging, by order of importance, to the shrub, pre-forest and young little stratified forest stages:-average low diameters and heights respectively between 2.5 and 5 cm and between 1 and 8 m (Figures 8 and 9). In fact, the floristic matrix consists of shrubs (Calliandratergemina) as well as specimens of tree species and regeneration trees of the later dynamic stage (Pimenta racemosa, Coccoloba swartzii, Pisonia fragrans, Bursera simaruba, Cordia sulcata, Inga laurina, Zanthoxylum caribaeum, Inga ingoides).   This anthropized plant community is dominated by a Eugenia monticola population with small sections (mostly between 2.2 and 5 cm, Figure 10). The Eugenia monticola population in morphogenetic expansion (morphological growth)associated with that of Calliandra tergemina form a dense low matrix (mainly between 1 and 8 meters, Figure 10) where species such as Pisonia fragrans (Mapou) and Pimenta racemosa (Bois d'Inde) regenerate in a later stage. A few mature Courbaril specimens (Hymenaea courbaril Figures 10 and 11) are relics of a former forest more advanced from the point of view of vegetation dynamics.      Figure  1).
Funtumia elastica is the allochthonous speciesin most competition in relation to the biophysical factors and is present in all quadrats( Figure 14). In addition, the age classes show an almost balanced distribution. In general, other species have lower distribution indices and basal surfaces. These result from the dominance of sections and heights which distinctively and essentially range between 2.5 and 5 cm and between 1 and 8 m (Figures 14 and 15).   The predominant formation consist of three species (Myrcia splendens, Inga ingoides and Funtumia elastica) and, here and there we also see individuals with different basal areas and stratigraphic status (Figures 16 and 17), species of the understory, middle and upper strata).    Figure 1).
In order of importance we see Myrcia splendens, Simarouba amara, Andira inermis, Sapium caribaeum. The structure of diameters and heights of this formation consisting of shrub, pre-forest and young forest units results mainly from the recurrent disturbances due to Man's actions. The majority of individuals have weak sections (between 2.5 and 5 cm) and rise to heights between 1 and 8 m (Figures 18 and 19).      Figure 1).
The populations of species that comprise this tropical seasonal evergreen forest floristic association are sparse and also exhibit low cumulative basal surfaces. The heights and diameters range mainly in the first classes:-2.5 to 5 cm for the sections and 1 to 8 m for the heights (Figures 22 and 23). The Cupania americana is the only species that dominates from an ecological point of view through a distribution index well above that of other species. We must mention the existence of very strong installation (seedlings and regeneration) of Brosimum alicastrum.  Station 11:--Tropical seasonal evergreen forest formation on a slope exposed to the wind, in the pre-forest to young foreststage(500 m², Figure 1) The same above mentioned conclusions are also valid here except that the collection of species (the floristic potential) and the predominant floral formations are not identical. Myrcia citrifolia forms a matrix within which we find by order of dominance:-Guettarda scabra, Coccoloba swartzii Cornutia pyramidata, Bursera simaruba, Eugenia monticola Bourreria succulenta, Croton corylifolius and Pimenta racemosa. No species populations is truly a balanced structure. This happens both at the level of diameters as heights. For the latter, the majority classes are respectively 2.5 to 5 cm and 1 to 8 m (Figures 24 and 25).All these taxa are post-pioneer ones, except Pimenta racemosa which settles very early in the plant succession and participates in all the phases of intra-forest evolution until the climax with notable demographic variations.   (Figures 26 and 27).    Figure 1).

This station is characterised by specific (in this case) notable diversity, an average to high basal area for Simarouba amara, Funtumia elastica, Inga ingoides, Sapium caribaeum, Cupania americana, Myrcia splendens, Tabernaemontana citrifolia, Hirtella triandra, Andira inermis and Zanthoxylum caribaeum
Simarouba amara, Myrcia splendens and Cordia alliodora are the dominant species. This is due to their modes of height and diameter distribution (Figures 28 and 29) as well as to their means of distribution and their population basal areas. Although having low densities and low biomass, some species of the later succession stages are present either in the form of mature specimens which are relics of an earlier dynamic phase, or in the form of regenerations:-Hymenaea courbaril, Cupania americana Andira inermis, Sapium caribaeum, Homalium racemosum, Hirtella triandra, Brosimum alicastrum. We also note the presence of an invasive species, Funtumia elastica, which is also well distributed in the various quadrats despite its low basal area.

Speciessurvey stations:-Differences between stations with regard to their collections of species using CFLs (XLSTAT):-
Considering the species' bio-demographic structures per station, the CFLs (Factorial Correspondences Analysis) allowed us to differentiate between survey stations. We use only the results of axes 1 and 2 which combine the maximum of inertia in other words 38.33% (Figure 30). Axis 1 shows environmental features specific to the environments while axis 2 differentiates between levels of ecosystem evolution. Stations 1, 2, 6, 7, 8, 10, 12 & 13 match mesophilic environments or meso-hygrophile environments in some cases where the containment is important (Figure 30). However, within stations 3, 4, 5, 9 & 11, the eco-climatic conditions are typical for xeric environments (Figure 30). The mesophilic stations include forest groups on slopes protected from the wind (station 10) and respectively riparian terraces (1, 2, 6, 7, 8, 12 & 13). Irrespective of the considered biotopes, either mesophilic or xeric, the stations are placed at different stages of plant succession (Figure 30). Stations 1, 2 & 6 are among the most regressive and consist of matrix shrub eco-units with some remaining mature trees with important biovolumes (only in stations 1 & 2). The stations associated with the xeric environments are colonized by phytocenoses between the shrub, pre-forest (stations 3, 4 & 5) and young unstructured forest with no defined stratification (9 and 11) stages.

Figure 30:-Environmental and temporal differentiation between stations
Irrespective of the dominance relationship between the taxa, Figure 31 shows clear differences between wet biotope species (A, annex 2) and those of dry environments (B, annex 2). This differentiation allows us to distinctively connect different species in the secondary shrub and pre-forest stages (A3 and B2, Figure 31), or in the more advanced unstructured or barely structured forest stages (A1, A2 and B1, Figure 31). Logically, the more generalist species are strongly associated with stations 1 and 2 (riparian terraces) and stations 3, 4 and 5 (xeric coastal slopes). The treatments performed using the EM algorithm starting from ecological descriptors of the study stations show us quite significant differences. There are differences both in topography (Figures 32 & 34), biotopes (Figures 33 &  34), dynamic stages (Figures 32, 33 & 35) and the degree of anthropization (the qualitative descriptor corresponds to the physiognomic type of the stations, Figure 35). There are two station groups:-S1, S2, S6, S7, S8, S10, S12 and S13 are mesophilic and S3, S4, S5, S9 and S11 are xerophytic (Figure 32). The latter as well as S10 is located on 1227 variable gradient slopes while others occupy portions of riparian terraces (Figure 32). With less precision, we can also differentiate between the anthropization degree of the floristic station formations and associate them with dynamic stages (Figure 35). Generally speaking, the xerophytic stations are less developed on the scale of plant succession and therefore, they are logically more anthropized. In Figure 36 the blue cluster allows us to identify the predominant species affected by a dominance index (ID) > 0.0314, an Index of distribution (Id) > 0,01436 and a basal Surface (basal area) > 1,0572 (Annex 2). On the x axis we see a ratio of 10 4 between the species with the largest and the lowest basal area and on the y axis a factor of 10 4 between the species with the greatest and lowest Index of distribution (Annex 2, Figure 36). The species with a high index of distribution and a relatively high basal area (blue cluster) mean that the green and red clusters are little differentiated and grouped in the vicinity of the origin (Annex 1, Figure 36). Accordingly, we propose to remove these species from the analysis file as well as those with very low values close to 0 for the Index of distributionand basal area:-those are species with an Dominance index(ID) < 1.26 10 -5 . Due to their very low biomass, the latter cause important reports on the x and y axes between the highest and lowest values:-and this occurs both for the basal area and for theIndex of distribution 2 .
1228 Having operated the above mentioned changes, we observed that the clusters can be differentiated using the index of distribution (Id) and in a less obvious way using the basal Surface ( Figure 37). The green cluster represents the species with a very low index of distribution (Id) and very low Basal Surface (Figure 37). The red cluster represents the species with a distribution of index (Id) and a Basal Surface ranging from very low to low ( Figure 37). Finally, in this new context apart from Sapium caribaeum and Samanea saman, the blue cluster shows a wider distribution of species with a Basal Surface ranging from low to medium ( Figure 37). We propose to remove the two little distributed high basal surface species from the data file:-Sapium caribaeum and Samanea saman. We also remove the species with an Id < 10 -5 :-Artocarpus altilis and Swietenia mahagoni (Annex 1).
In this last configuration, the Red Cluster (Id > 0.0085725 and Basal Surface > 0.13246875) includes moderately distributed species with a low to medium basal surface (Figure 38). The blue Cluster consists only of the following species with an ID<0.0071233 and a basal surface >0.48915314. Finally, the green cluster represents the species with low ID (Dominance index), a low index of distribution (Id) and low Basal Surface out of which two are characteristic:-Odontonema nitidum and Homalium racemosum (Annex 1). The first species is the only characterised by an Id (index of distribution) higher than that of a species of the Red cluster and an Id greater than all the species of the blue cluster. However, its Basal Surface is lower than that of all species of the blue and red clusters. Therefore we consider that this species could be included in the Red cluster but it would have the lowest Basal Surface. compositions and physiognomies which correspond to the following dynamic stages:-mature shrub, pre-forest, young barely structured forest, secondary forest. Despite the fact that we can associate the survey stations with evolution stages the latter are often composed of a conglomerate of shrub, pre-forest and forest eco-units with variable frequency ratios. For example, the stations colonized mostly by shrub units belong to the shrub dynamic stage. The same applies for the stations consisting either in mature shrub units surpassed by some tree taxa or in forest units, which respectively can be connected to the pre-forest and forest successional stages. Everything seems to indicate that the eco-units of the surveyed formations correspond to interlocked succession stages. This phenomenon is an ecosystem response to old and new anthropization terms:-recolonization after the abandonment of vegetable crops on small intra-forest plots, the recovery of vegetation on the ruins of precarious dwellings, selective samples of species for the needs of everyday life (this form of resource utilization was and remains very significant). Ultimately the structural differences of the diameter and height histograms of the xerophytic and mesophilic stations testify to a human influence that occurred throughout many centuries 3 . Depending on the biotope and in the light of the Dominance index (ID), the site diversity of the dominant formations results from combinations of species linked to the structure of the biophysical factors in which Man plays a major role. In other words, the temporal dynamics of the collection of affine species distinctively characterizing the moderately wet and dry environments has been and is being guided by human activities to a degree. We can state that today's species and the phytocenoses they form are anthropophilic since the structure of their population and their dominance is under anthropogenic stress.

The clarifications brought by the statistical approaches (AFC and Clustering):-
The AFC and Clustering results fully match the above mentioned interpretations. The EM algorithm has allowed us to differentiate between stations and thus species which are endemic due to the (moderately wet or dry) environments, the stages of evolution, topographic facies and the degree of anthropization (Figures 32, 33 34 & 35). Among the statistical treatments, those derived from the SimpleKmeans algorithm are a valuable contribution to the multifactorial analysis (AFC) and the classical ecological description based on indicators such as the Basal Surface, the Index of distribution (Id) and the Dominance index (ID). In fact, this algorithm highlights the population gaps between the different species in principal with regard to their strength and their total Basal Surfaces. First, we were able to differentiate between the dominant species whose representatives have an average to high index of distribution and also an average to high basal Surface ( Figure 36). These species are present in a large number of stations and occupy a fairly wide range of diameter classes. Secondly, when the previous species are removed from the file we get three new groups which are not totally disjointed due to the existence of two sparsely distributed species with a high basal Surface ( Figure 37). Thirdly, after removing these species from the file data, the population differences are far more clear (Figure 38). Ultimately, the statistical analyses match the descriptions of the survey stations and allow us to better decipher the ecological strategies of species whose population structure is a relevant indicator. Within the different stations, some species are considered balanced because their individuals occupy many diameter classes:-they are often dominant ( Figure 36, annex 1). Others have individuals concentrated in low diameter classes:-these are often species in regeneration or small size trees and therefore they have low aboveground biomass ( Figure 36).

Conclusion:-
The ecological study of the Case Navire ZNIEFF stations highlighted the deeply heterogeneous nature of its vegetation. In reality the environmental differences are added to those caused by human activities. The diversity of species, of the phytocenoses and the floristic units that make up the survey stations result in a dense mosaic both within the stations and between them. From one formation to another and irrespective of the environment, the dominant species formations vary, the different dynamic stages affine species intermingle. This heterogeneity is also structural (distribution of diameters) and architectural (distribution of heights). Generally speaking, very few individuals have high basal areas and reach heights greater than 30 metres. Large trees can often be found in matrices of trees in full morphogenetic development or in matrices composed of mature shrubs. In reality, the ecological descriptors analysed in this study show a diversity of phytocenotic responses in connection with a variety of anthropization means taking place throughout several decades. From this point of view, this type of ecosystem context is specific to the study of vegetation dynamics.