Numerical taxonomy of Berlinia species (Caesalpinioideae: Leguminosae) and their distribution in Nigeria

A morphometric study was performed on six Berlinia species of the Nigerian flora. Individual specimens deposited at Forest Herbarium Ibadan and University of Ibadan Herbarium formed the units of study.  Twenty morphological characters were assessed and analyzed following conventional taxonomic practice. Results showed a close resemblance between B. craibiana and B. grandiflora, which share affinity with B. bracteosa and B. confusa , while B. coriacea appears to be distantly related.  Species distribution studies revealed that B. grandiflora exists in savanna and forest regions while other members of the genus are found only in forest zones. While suggesting practical measures for conserving endangered Nigerian flora, we also recommend that further taxonomic studies be carried out on the genus Berlinia and its related allies to ascertain their placement within the Caesalpiniodeae.


INTRODUCTION
Berlinia is a member of the angiosperm family Leguminosae.It was first described by Hooker & Bentham in 1849 from three different collections (Smeathmann s.n., Ansell s.n. and Heudelot 886), which were ascribed to a single species (B.acuminata Sol.ex Hook.f.& Benth.).The currently accepted narrow circumscription of Berlinia was established by Leonard (1957) based on Harms' sections Euberlinia and Macroberlinia.The genus belongs to the tribe Detarieae sensu lato of Caesalpinioideae, and comprises 10 species in West Africa, with eight found in Nigeria (Hutchinson & Dalziel 1958;Keay 1989;Soladoye & Lewis 2003;Ayodele & Yang 2012).
Leaflets 2-6 pairs, usually fairly large and symmetrical at the base.Flowers are borne on stout stalk and there are 10 prominent free stamens with small anthers attached in the middle.Petals of B. bracteosa and B. occidentalis are more or less subequal in length and inflorescence is a stout terminal simple raceme while those of other West African species are very unequal in length and their inflorescence in a cluster of racemes or sparingly branched panicles (Hutchinson & Dalziel 1958;Keay 1989).Of the eight Nigerian species, B. congolensis and B. hollandii have been reported to be rare within the region (Keay 1989).Hutchinson & Dalziel (1958) and a recent monograph by Mackinder & Pennington (2011), reported only four specimens of B. congolensis in Nigeria, and these were collected from two locations (3 from Shasha Forest reserve on 07 0 05'N & 04 0 30'E and 1 from Eket on 04 0 39'N & 07 0 56'E) as far back as 1935.These areas however, have been greatly disturbed and availability of this species can no longer be guaranteed.Similarly, only two collections of B. hollandii were gathered (from Cross River state), between 1897 and 1898 by John Henry Holland who was the Curator of The Calabar Botanic Garden (Hepper & Neate, 1971); and as suggested by Mackinder & Pennington (2011), this species may have disappeared.Mackinder (2001) noted that identification and curation of Berlinia collections can be problematic due to lack of consensus over genus and species limits, in some cases confounded by nomenclatural confusion which have led to conflicting taxonomic treatments in regional African floras.A poor understanding of the phylogenetic relationships within and among Berlinia species and its supposed allies is also a big problem.Nevertheless, a number of generic systems have been proposed for Berlinia sensu stricto and its presumed allies (Leonard 1965;Cowan & Polhill 1981), and this has rendered the taxonomic position of Berlinia uncertain.
The need to classify plants based on morphological characters (e.g., of leaf, flower, fruit and seed) while assigning equal importance to such attributes as proposed by Adanson (1763) brought about the establishment of Numerical Taxonomy (Sneath & Sokal 1963).Over the last 20 years the field has developed rapidly, to the extent that we now distinguish between traditional morphometrics (Marcus 1990) and recent geometric morphometrics (Adams et al. 2004), which are not widely used in plant systematics (Jensen 2003;Henderson 2006).Plant morphometric studies (Gomez-Campo et al. 2001;Sonibare et al. 2004;El-Gazzar & Rabei 2008;Soladoye et al. 2010) have been very useful in delimiting species at generic and infra-generic levels.
This study employs a morphometric approach to the delimitation of West African Berlinia species occurring in Nigeria, based on external morphology of some vegetative and reproductive characters.We also report the availability of species within the region and thus present the basis for a distribution map (Fig. 1).(Holmgren et al. 1990) were used for this study.Some of these are detailed in Table 1.Vouchers for B. congolensis and B. hollandii were not deposited in these herbaria by the collectors.Several efforts to locate fresh specimens were also futile due to continuous deforestation and other human activities within the reported locations of original collections.Prior to data collection, the available specimens were carefully examined and the choices of characters were determined.

Morphological studies
For each species, 20 morphological characters comprising 10 quantitative and 10 qualitative macrocharacters were assessed from the available specimens (Tables 2 and 3).All quantitative characters were measured using a thread and rule.Measurements were obtained from specimens deposited at FHI and UIH, thus individual specimens formed the units of study (Image 1).Leaflet length and width were obtained by spreading the middle leaflet on a flat surface on the laboratory bench to ensure uniformity (Olowokudejo 1999;Soladoye et al. 2010).Some characters (e.g., number of lateral nerves, number of leaflet pairs) were counted and carefully recorded while others (qualitative) were observed either with the naked eyes or hand lens where necessary.The mean and standard error for each measured character were calculated; all data were coded and thereafter subjected to statistical analysis using Minitab 17.0 and supported by IBM SPSS Statistics 20.0.

Species distributional study
This was based on existing collections deposited at the aforementioned herbaria.For each Berlinia species, place and date of collection were adequately recorded from all available specimens, and the derived data used to produce a distributional map for Nigeria using Arc GIS 9.3 at the Department of Geography, University of Ibadan, Ibadan, Nigeria.

RESULTS AND DISCUSSION
Herbarium collections showed B. grandiflora to be more abundant than the other species studied.While it extends from the wet zones to the drier parts of Nigeria, others are confined to the rain forest region except for B. confusa, which was also collected from the savanna of Nassarawa State (Fig. 1  noted that the Nigerian species of Berlinia are found only in the rain forest zones, except B. grandiflora which occurs both in the savanna and forest regions of the country.In fact, the species of Berlinia studied in this work can be said to be sympatric.
Results from the macro-characters assessed showed that mean leaflet length and leaflet width ranged from 10.6cm and 4.9cm in B. confusa to 22.7cm and 10.7cm in B. coriacea.This variation was also observed for lamina length and distance between nodes, which ranged from 13.5cm and 3.3cm in B. confusa to 28.0cm and 5.8cm in B. coriacea, respectively.Petiolule length was least in As observed in Table 4, there is high positive correlation between leaflet length and lamina length, leaflet width and lamina length, fruit length and fruit width, lamina length and distance between nodes.Others include leaflet length and leaflet width, leaflet width and distance between nodes, leaflet length/leaflet width ratio and length of bract, amongst others.High negative correlation also exists between lamina length/ petiolule length ratio and leaflet base, distance between nodes and number of leaflet pairs, petiolule length and fruit texture, leaflet width and fruit length etc.While characters with positive correlations may be more useful in distinguishing the Berlinia species studied, those with negative correlations when combined may not be suitable for species delimitation.
Principal component analysis (PCA) revealed that only three components contributed 92% of the total variance.Other extracted components were ignored as uninformative.Component 1 accounts for approximately 48%, while the other two constitute the remaining 44% (Table 5).These three components all have eigen values greater than 1, reflecting their reliability in the present study.Furthermore, the first component is most highly correlated with leaflet width and petiolule length.Leaflet width is a better representative, because it is less correlated with the other two components.For the second component, length of the bract is a better choice while pedicel length may be preferred for the third component (Table 6).This suggests that these three characters (leaflet width, bract length and pedicel length) could be focused on in further studies.
The cluster analysis (Table 7) shows that the distance measure between species 5 (B.craibiana) and species 6 (B.grandiflora) is the least while that between species 1 (B.auriculata) and species 4 (B.coriacea) is the highest.This result implies that the degree of affinity existing between species B. craibiana and B. grandiflora is the highest (with a very strong similarity level -83.1601) while species B. auriculata and B. coriacea are the most distantly related taxa (similarity level -13.3992).This is     further illustrated in Fig. 2 showing the dendrogram of the examined species.B. coriacea could be considered as the outlier, as members of the genus are divided into three main groups: B. craibiana, B. grandiflora, B. bracteosa and B. confusa as a more united group (group 1), B. auriculata as a monospecific group having some common ancestral characteristics with group 1, and B. coriacea existing as an entirely different mono-specific group with less degree of affinity with other members.The above observations reflect the statistical rule that, the less the coefficient value, the more the degree of affinity existing between any two species.Nonetheless, the extent of similarity measured by the correlation coefficient of cluster existing between B. craibiana and B. grandiflora may suggest a monophyletic origin of these species.Furthermore, the descriptive chart presented in Fig. 3 supports the result from the PCA and dendrogram generated from analysis.This result tends to clearly separate B. coriacea and B. auriculata from other members of the genus, with uniquely different polygonal shapes while uniting B. craibiana, B. grandiflora, B. bracteosa and B. confusa with somewhat similar shapes.
We observed that leaflet shape and sizes varied within populations, which may be attributed to variations in light intensity affecting growth (Aborg 1943), or other environmental and genetic factors.In general, variations in the vegetative and floral organs are important diagnostic tools, which could be used in the delimitation of taxa, and the importance of these morphological features in taxonomic classification of plant species was noted by Nwachukwu (1997) and Stern (2000).Numerical taxonomy has also been reflected in parallel taximetrics and orthodox studies in Crotalaria (Bisby & Polhill 1973), the phenetic classification of   Baphia species (Soladoye 1982), and delimitation of Nigerian species of Sterculia and Eribroma (Soladoye et al. 2011) to mention but a few.Thus its application in the present work has in no little way contributed to the existing taxonomic information regarding Berlinia species occurring in Nigeria.However, it is important to note that morphometric analysis is not sufficient to delimit taxa, even though it has greatly benefited plant systematic studies.Further studies in Berlinia are recommended in order to overcome challenges in the identification of species, especially in sterile or fragmentary states.

CONCLUSION
All the Nigerian Berlinia species studied can be said to be hygrophilous, except B. grandiflora, which is often found in forest-savanna boundaries.This work supports the co-existence of the Berlinia species examined, since they share many overlapping vegetative and floral characteristics.Some of these include: number of leaflet pairs (2(3)-5), leaf type (paripinnate), leaf margin (predominantly entire), leaflet arrangement (opposite/ sub-opposite) and fruit shape (oblong).
The conservation of our rich but endangered forest regions, particularly in Cross River State which is home to many indigenous plant species (regarded endemic), is also of great concern as this would ensure the availability and sustainable collection of many indigenous flora species including those of Berlinia.
). Observations from this study revealed that B. bracteosa is restricted to the rich forests of Cross River State and B. coriacea in Edo and Delta states.Nonetheless, B. auriculata spreads from Port Harcourt (Rivers State) to Enugu-Ogbeti (Enugu State) and to Calabar (Cross River State).In general, all of the Berlinia species studied except B. coriacea are found in Cross River State.This state has been known for its rich biodiversity.The occurrence of B. grandiflora in both savanna and rain forest ecosystems is also an indication that it can survive in either environment unlike other members of the genus examined.This result also agrees with the earlier work of Ayodele & Yang (2012) who

Figure 1 .
Figure 1.Distribution of Berlinia species in Nigeria.

Table 8 . Conservation status of the Berlinia species studied
* -This species is the most widely distributed member of the genus.