Phenological, morphological and agronomic characterization of sixteen genotypes of cotton plant (Gossypium hirsutum L.) in rainfed condition in Benin

In Benin, cotton cultivation is rain fed. There is a need to develop varieties adapted to the current diversity of growing conditions caused by climate disruptions. To identify types of varieties that may be used in crossing to adapt varietal offer to climatic disturbances, sixteen genotypes of diverse origins were characterized with a randomized complete block design with four replications. Fifteen agromorphological variables allowed to describe the genetic variability using descriptive statistics and multivariate analyses. Results showed high genetic variability and a structuration into three groups of genotypes tested. Plant height, length of fruiting branches, height to node ratio, flowering date and opening date of first bolls are the main distinguishing characteristics between groups (p<0.01). The first group consists of compact genotypes with stems, fruiting branches and internodes relatively short. These genotypes were early to flowering and opening bolls. The second group is composed of more vegetative genotypes, with medium size stems with long fruiting branches and long internodes; they are late to flowering and opening bolls. The third group consists of a tall genotype with short fruiting branches and long internodes; it is early to flowering and opening bolls. Compact and early genotypes could be used in crossbreeding to produce varieties adapted to the current climate disruptions.


INTRODUCTION
Cotton (Gossypium hirsutum L.) is the most cultivated fiber plant in the world nowadays.World production of cotton fiber reached 25.74 million tons (International Conference of African Cultures (ICAC), 2017).Cotton is mainly grown for its fiber that is used as raw material for textile industries.But it also produces many byproducts.Indeed, decorticated cotton seed contains a kernel (60% of the weight of the seed) itself composed of 38% oil (Bolek et al., 2016).Cotton oil is used in food after removal of gossypol, a highly toxic alkaloid present in all *Corresponding author.E-mail: emmanuelsekloka@hotmail.com.Tel: (00229): 971 635 98.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License aerial parts except in fibers and seed coat.The kernel also contains 35% protein.This high amount of protein permits the production of cakes occupying an important place in animal feed (12% of world production) and places cotton flour in 2nd place of world plant protein after soybeans (Yue et al., 2012;Camara, 2015).With varieties without gossypol or "glandless", cotton might even become progressively a food plant (Ma et al., 2016;Zhang et al., 2016).
In Benin, cotton sector is the most provider of national currency.It contributes for 14% to Gross domestic product (GDP) and between 30 to 40% of export earnings (Kpadé, 2011;INSAE, 2017).Revenue from cotton cultivation is the main source of cash income for farmers.Unfortunately, for more than ten years, a drastic drop of production was observed (Paraïso et al., 2012).Among other causes, the policy of unique variety so far adopted has shown its limits in view of climatic disturbances becoming recurrent these last years.
Indeed, these disturbances, mainly characterized by rain delays at planting, rather often generate late installations of crop compared to the period recommended by research.This late sowing often limits the operating time of the plant and strongly penalizes the yield.According to Lançon et al. (1989), the potential decline in seed cotton yield is about 20 to 30 kg.ha -1 per day for late sowing.But the H279-1 variety grown all over the country since 2003 has been developed within a cropping system based on more regular rainfall.Today, this variety badly fits to an increasingly reduced operating time because of these climatic disturbances.This leads currently in the extension of three new varieties to replace H 279-1 in different agro-ecological zones of country (Hougni et al., 2014).But these new genotypes are still late and are not yet pronounced on morphological and phenological plans to take into account recent research results which have shown that compact and early varietal types could be an alternative when hydrous conditions limit operating time of the culture in case of late sowings (Sekloka et al., 2008).Thus, current climatic difficulties offer new challenges to which varietal research must continue to face by offering varied range of genotypes adapted to the different growing conditions caused by these disturbances.For that, we must not only re-specify the conditions of use of the current late varieties, but also identify relevant genitors in order to achieve the diversification objectives of varietal offer to match the current diversification conditions of culture.This study fits into this framework and proposes to identify relevant genotypes of interest to the cotton plant breeding program in the current situation of cropping system evolution in Benin.

MATERIALS AND METHODS
The study was conducted in 2015 in Benin on the experimental farm of the Faculty of Agronomy of the University of Parakou (9°18'56.87"North latitude and 2°42'4.87"East longitude).The soil type is tropical ferruginous poor in organic matter with C content of 1.43%, C / N ratio of 8.46, clay and silt content of 22.40% (Azontondé et al., 2009).Annual rainfall has been abundant (1100 mm) and well distributed.August and September were the wettest months, corresponding fairly well to active periods of production of cotton plants (Figure 1).The average daily temperature varied between 20 and 25°C with a daily average of 22°C over the period of the study.
The experimental design was a randomized complete block with four replications.Basic experimental plots (48 m²) were set up with three 20 m rows.Seedlings were thinned to one plant per hole.The stand density was 42 000 plants/ha.We also implemented the crop management sequences generally recommended for cottongrowing areas in Benin (CRA-CF 2015).The observations were carried out on the center lines of the basic plots.They focused on: (1) First flower opening date (FF), determined by counting the number of flowers daily after flowering onset.This corresponds to the date (expressed in days after planting) when the sum of the daily counts is equal to the number of plants in the row.
(2) First boll opening date (FB), determined by counting the number of open bolls daily after opening of the first bolls.This corresponds to the date (expressed in days after emergence) when the sum of the daily counts is equal to the number of plants in the row.
(3) Production earliness ratio (R1/RT=first harvest/total harvest) (4) Morphological and boll distribution indicators measured at harvest time on 10 individual plants randomly selected on the center lines of basic plot, using plant mapping technique (Bourland et al., 1990): (i) Height at harvest (HH), measures the height of the main stem (in cm) from the first cotyledonary node to the tip.(ii) Height to node ratio (HNR), the ratio of the plant height (in cm) to the total number of nodes counted above the cotyledonary node on the main stem.(iii) Number of vegetative branches (NBV).(iv) Length of fruiting branch (LFB), measured (in cm) on the third fruiting branches of the plant, and the length of vegetative branch (LVB) measured (in cm) on the second vegetative branch of the plant, as described by Hau and Goebel (1987).(v) Height of first fruiting node (HFFN), measures the height on the main stem (in cm) from the cotyledonary node to the first fruiting branch.(vi) Height of last fruiting node (HFFN), measures the height on the main stem (in cm) from the cotyledonary node to the last fruiting branch carrying a harvestable boll.(vii) Boll retention at first positions of fruiting branches (RP1) is the ratio of the number of bolls harvested at the first positions of the fruiting branches to the number of fruiting branches.
(5) Boll weight (BW), calculated as mean weight of 3 first position bolls per plants (harvested in low, medium and top part of plant) calculated from 10 plants randomly selected from the central rows of the elementary plots.(6) Seed cotton yields (Rdt) were also calculated on the basis of boll harvests from the central rows of the elementary plots.The mean productivity of varieties was analyzed.
Variance analyses were performed with the R software version 3. 1.3 (2015-03-09).Tukey test (TukeyHSD) was used for comparison of means when differences are significant.Phenological and morphological data were subjected to a Principal Component Analysis (PCA) and a Discriminating Factorial Analysis (DFA).Wilks' Lambda test was then used to extract the quantitative variables most discriminating the groups obtained.

Variability of traits studied
Significant differences were observed between extreme values for most characters and the differences between varieties were highly significant.The differences were more than 10 days for first flower opening date (FF) and nearly one week for first boll opening date (FB).Length of fruiting branch (LFB) and length of vegetative branch (LVB) varied from simple to more than double.Height at harvest (HH), and height to node ratio varied of almost 50% between the two extremes.It was the same for yield parameters like bolls number on fruiting branches (BFB), average boll weight (BW), seed cotton yield (Rdt) (Table 1).

Structure of the genetic diversity tested
The first two axes of the PCA carried out with earliness and morphology variables explained 65.48% of the variability.The first axis is more correlated with the morphological variables, height at harvest (HH), height of first fruiting node (HFFN), height to last fruiting node (HLFN) length of vegetative branch (LVB) and height to node ratio (HNR).So it can be considered as an axis of vegetative development.
The second axis is highly correlated with first flower opening date (FF) and first boll opening date (FB).It can be considered as an axis of precocity (Table 2).The varieties projection in the factorial plan formed by the two axes allowed to distinguish three groups (Figure 3).Group 1 consists of five genotypes (31.25% of the total).These are early varieties with low vegetative growth.The genotypes of this group are of short height.They are earlier for boll opening, seed cotton production, but medium for flower opening.The first fruiting branches are inserted lower on the plant.
They produce less vegetative branches, short in length, and short internodes on the main stem (Table 3).
Group 2 made up of ten genotypes (62.5% of the total), includes late maturing genotypes with high vegetative growth: plants had average height at harvest, many and long vegetative branches; they were last ones to bloom   and to open bolls (Table 3).Group 3 consisted of only one typical genotype (Oultan).It is characterized by early flowering cotton.Plants are very tall, with longest internodes and many long vegetative branches.But fruiting branches are shorter (Table 3).

Discriminant analysis
Discriminant analysis was performed using the three groups obtained from the PCA as categorical variable.
Result confirm varieties categorization obtain from PCA    at 93.57% and offers a reclassification of certain genotypes analyzed (Table 4).The F-test of Wilks Lambda revealed that five of the eleven characters used allowed to better discriminate genotypes studied (Table

DISCUSSION
Suitable cotton variety selection is imperative to cope with climatic variations for yield enhancement and sustainability under unpredictable climatic conditions (Habib ur Rahman et al., 2016).Under rainfed conditions where climatic variations are unpredictable like that occurs in Benin, the demonstration of a genetic variability for the morphological and phenological characters in selected genotypes is a guarantee of future genetic progress (Mergeai, 2006;Hajjar et al., 2008).Our study, showed a strong phenological and morphological heterogeneity of the studied collection, thus providing usable genetic variability to achieve the objectives of adaptation of the cropping system to current evolutions on the climatic conditions.The results distinguish three groups of which one consisted of compact and early genotypes.These could be used in crossbreeding to produce varieties adapted to limiting hydrous conditions.Compact and early genotypes were found able to adapt to a more reduced cycle of precipitations (Sekloka et al., 2016;Lu et al., 2017) and their low spatial extent allows for high planting density (Sekloka et al., 2008(Sekloka et al., , 2016;;Sahito 2016).These genotypes could be backcrossed to cultivated varieties already adapted to local growing conditions in order to improve the precocity and plant shape in the new varieties.
Results also showed that the Beninese selections, late maturing varieties with high vegetative growth, gave the best yields in cotton seed.In a previous study comparing varieties of different geographical origins and different agro morphological characters, Beninese varieties H279-1, Stam 18 A also gave the best yields in cotton seeds when the water conditions are not limiting (Sekloka et al., 2008(Sekloka et al., , 2016)).Our results, consistent with these, validate that in African rainfed conditions, late maturing varieties with high vegetative growth continues to be interesting when rains are regular, abundant and well distributed (Lancon et al., 2007) as was during our study.However, previous work had shown that when water conditions are limiting (late sowings for example), these indeterminate varieties were capable of the best and the worst: they maintain irregular yields that can be found both in the low yield classes and in high yield classes (Sekloka, 2008).Furthermore, it is known that in rainfed, water deficit is the most limiting abiotic factor for productivity and yield in several crops (Loison, 2015).Several authors have shown that water stress particularly affects flowering and boll formation and consequently the fiber yield (Kouakou et al., 2008;Loison 2015;Huang, 2016).The present studies must be repeated in more northern areas of the country where water stress may be stronger.This would allow to better specify the responses of these different genotypes to changes in the environment and to better justify the value of their use in selection.
The most structuring criteria of the genetic variability have been the plant height at harvest, the length of fruiting branches, the height to node ratio, the first flowers opening date and the first boll opening date.Although the number of varieties studied is not very large, the result is similar to those fairly often reported in the literature with respect to the analysis of genetic diversity in the species cotton G. hirsutum L. In an earlier study on three years of collection, Djaboutou et al. (2000) have also highlighted three genotype groups contrasted by the same criteria of morphology and precocity.On the other hand, our works differ from those of Bourgou et al. (2014) that highlighted six diverse groups at the end of the evaluation of 336 accessions collected across Burkina Faso.Indeed, the collection described by these authors was larger and contained ecotypes of all grown species, diploid (G.arboreum and G. herbaceum) as tetraploid (G.hirsutum and G. barbadence).The variability described by these authors was therefore necessarily larger.The varieties tested in our study, all of the species G. hirsutum, were not enough representative of the genetic pool potentially available for improving the cultivated varieties in cotton.Wild cotton previously neglected may be useful in the current context of climate change and continuous narrowing genetic base of cultivated varieties (Mergeai, 2006;Sarr and Mergeai, 2009;Bourgou et al., 2014).

CONCLUSION AND SUGGESTIONS
The different analyzed genotypes have variability for all characters used, particularly those related to phenology, morphology and distribution of production throughout the plant.Compact and early varieties described could be used in crossing to create new improved varieties for earliness and compactness of the port, adapted thus to limiting water conditions.However, this variability is far from being representative of the genetic pool potentially available to improve the varieties grown in cotton.

Figure 1 .
Figure 1.Rainfall data during the test.

Figure 2 .
Figure 2. Variation of seed cotton yield of varieties tested.
d.a.p.: days after planting; FF: First flower opening date, FB: First boll opening date, R1/RT: the production earliness ratio, HH: Height at harvest, HFFN: height of first fruiting node, HLFN: height of last fruiting node, LFB: Length of fruiting branch, LVB: Length of vegetative branch, HNR: Height to node ratio, NBV: Number of vegetative branches, NBF: Number of fruiting branches.

Figure 3 .
Figure 3. Projection of the varieties in the factorial design formed by the two axes of the principal component analysis (PCA).

Figure 3 :
Figure 3: Projection of the varieties in the factorial design formed by the two axes of the Principal Component Analysis (PCA)

Table 1 .
Minimal, maximum value and coefficient of variation of the quantitative characters.
d.a.p.: days after planting; min: minimum value; max: maximum value; CV: coefficient of variation, FF: First flower opening date, FB: First boll opening date, HFFN: height of first fruiting node, HH: Height at harvest, LFB: Length of fruiting branch LVB: Length of vegetative branch, HNR: Height to node ratio, NBV: Number of vegetative branches, NBF: Number of fruiting branches; BFB: Bolls number on branches fruiting CBV: Bolls number on vegetative branches, BW: Boll weight, yield: seed cotton yield, RP1: Boll retention at first positions of fruiting branches, RBFB1_7: Boll retention over the first 7 fruiting branches.

Table 2 .
Eigenvalues and percentage of variance expressed by the first five axes.

Table 3 .
Characteristics of different genotypes groups from principal component analysis (PCA).First flower opening date, FB: First boll opening date, R1/RT: the production earliness ratio, HH: Height at harvest, HFFN: height of first fruiting node, HLFN: height of last fruiting node, LFB: Length of fruiting branch, LVB: Length of vegetative branch, HNR: Height to node ratio, NBV: Number of vegetative branches, NBF: Number of fruiting branches. FF:

Table 4 .
Groups classification matrix on the basis of characters of phenology and morphology.

Table 5 .
Discrimination power of different variables.