Morphological Variation of Strychnos spinosa Lam. Morphotypes: A Case Study at Bonamanzi Game Reserve, KwaZulu-Natal, South Africa

: Strychnos spinosa Lam. of the Loganiaceae family is associated with versatility, poverty eradication, and rural economic development. However, the morphological diversity of S . spinosa is not well documented. This limits efforts toward its improvement and commercial exploitation. This study aimed to characterize the variability, vegetative and reproductive traits, and heritability of S . spinosa morphotypes at Bonamanzi Game Reserve. The majority of the morphotypes had green, rough, round immature fruits with dark green, elongated, open fully developed leaves. Fruits varied between roundish and pyriform shape as well as green and purple colour. Positive correlation was mainly seen between fruit and seed traits. Principal component analysis indicated fruit and seed traits as major discriminating factors for morphotypes, followed by leaf size and seed thickness. A dendrogram primarily grouped morphotypes according to fruit texture. Genotypic variance and genotypic coefﬁcient of variation were higher than environmental variance and environmental coefﬁ-cient variation in reproductive traits of S . spinosa . A rough pericarp texture and purple tinge on the immature leaves of some morphotypes was recorded for the ﬁrst time. Differences in morphological features across S . spinosa morphotypes indicate a high level of diversity that could be utilized by breeders to generate new cultivars. This ﬁrst report on variability and heritability among S . spinosa morphotypes forms the basis of available germplasm essential for future breeding programs.


Introduction
Strychnos spinosa Lam. of the Loganiaceae family is endemic to Africa, where in South Africa it grows in the Eastern Cape, Limpopo, KwaZulu-Natal, and Mpumalanga provinces [1].In KwaZulu-Natal, S. spinosa trees are abundant and ranked as the most important indigenous fruit tree species with potential for the growth of rural economies and the eradication of poverty [2].
Prolonged droughts have increased because of global climate change [3].Drought and lack of available water have historically been the primary factors that restrict crop productivity [4].However, S. spinosa is generally found in drought-prone areas and semiarid regions, where the tree remains productive even when water is unavailable [5].During the dry season, it plays a crucial role in ensuring food security [6].The expected growth of the global population to more than 9 billion by 2050 provides a significant challenge on how to develop stress-resistant/tolerant crops that are more competitive and can thrive in marginal soils to assure food production [7].
The conservation status of S. spinosa is categorized as "least concern", as its distribution and abundance possess a low risk of extinction [1].Strychnos spinosa can grow up to 10 m tall, with a trunk that is sometimes fluted, up to 25 cm in diameter [8].It has a bark which is almost corky, flaking, and greyish with young reddish branches [9].The specific name "spinosa" refers to the morphological characteristic of the paired spines on nodes of the branches, and this characteristic also sets it apart from the similar tree Strychnos madagaransies.The leaves are opposite and decussate, obovate with the broadest part near the apex, entire and sometimes wavy margin, and light-to-dark-green above but paler beneath [10].Strychnos spinosa fruits are described as indehiscent and sphericalshaped, with a green hard pericarp which becomes yellow as the fruit matures and resembles an orange [11,12]; hence the common English name "green or orange monkey orange", as it is also frequently consumed by monkeys in the wild [13].Strychnos spinosa is valued for its wide range of uses, such as food, medicine, economic and recreational purposes [14].
Morphological markers are among the markers used to assess genetic diversity within a species [15].Even though the species is highly regarded for its versatility and is essential for the development of rural economies and the eradication of poverty [2], knowledge on the morphological diversity of S. spinosa is not well documented.This limits efforts toward improvement and commercial exploitation of the plant [15].Knowledge on species morphological diversity will provide the scientific basis for breeding programs aimed at improving the plant for agricultural production.This is the first step towards the evaluation of genetic diversity and to develop strategies for the conservation and preservation of genetic resources [16].Therefore, the aim of this study was to characterize variability among S. spinosa morphotypes using morphological traits.

Study Area, Naming and Selection of Morphotypes
A study on variation in morphology among S. spinosa morphotypes was conducted at Bonamanzi Game Park, with an abundance of naturally growing plants of this species.Bonamanzi is located in north-eastern KwaZulu-Natal, South Africa (28 • 03 47.17 S, 32 • 18 06.79E) [17] (Figure 1).The collection site falls within the savannah biome classified as Western Maputaland Clay Bushveld No SV120.According to Köppen-Geiger classification [18], Bonamanzi is comprised of a warm temperate, fully humid and warm summer (Cfb).The vegetation type consists of mixed but mainly compound-leaved, short (5-10 m) woodland and wooded grassland.It has a mean annual rainfall range of 440-1200 mm and temperature ranges from 11-30.4 • C [19].
One hundred and twenty-six trees (three trees per morphotype) of S. spinosa at fruiting stages were chosen in a 4000 ha region using a stratified sampling strategy based on morphotypes.The definition of leaf and fruit qualitative characteristics was determined through preliminary research that was conducted throughout the growing season (July 2018-April 2019).Vegetative and reproductive data was collected in season 1 (July 2019-April 2020) and in season 2 (July 2020-April 2021).
The names of morphotypes were coined based on the colour, texture, and shape of the immature fruits, the colour of recently sprouted but open leaves (young leaves), as well as the colour, shape, and form of fully developed leaves (Table 1).The distinct variation in leaf size within a morphotype was categorised into small (1) and big (2) leaves, based on a ratio of 1 small:1.5 big for length and width categories.The colour of immature fruits was either green or purple (Figure 2a,b).Immature fruit texture was classified into smooth, smooth and corrugated, rough, rough and corrugated, very rough, or very rough and corrugated (Figure 2c-g).The shape of immature fruits was either roundish or pyriform (Figure 2h,i).The colour of young leaves was either pure green or had purple tinge, whereas the fully developed leaves varied between green and dark green (Figure 3a,b).Further, the leaf shape was either elongated or roundish, whereas the leaf form varied between folded and open (Figure 3c).Leaf margins were characterized into either wavy or not wavy.

Tree and Leaf Characteristics
Canopy radius (m), tree height (m), and stem diameter (cm) were measured using measuring tape, a telescopic height rod, and diameter tape, respectively.Canopy sizes were categorised as small (≤5 m), medium (>5-7 m) and large (>7 m).Plant height was divided into three sizes: short (≤4 m), medium (>4-6 m) and tall (>6 m) trees.Stem diameter was classified as thin (≤40 cm), medium (>40-70 cm) and thick (>70 cm).Leaf size at maturity was measured using a ruler.Leaf length (mm) was measured from the lamina tip to the intersection of the lamina and petiole along the lamina midrib.Leaf width (mm) was measured from tip to tip at the widest lamina points.Short (≤40 mm), medium (>40-60 mm) and long (>60 mm) leaves; as well as narrow (≤20 mm), medium (>20-30 mm) and broad (>30 mm) leaves, were used to classify leaf length and width, respectively.
Chlorophyll content was recorded using a SPAD-502 m (Konica-Minolta, Tokyo, Japan).Twelve SPAD measurements were taken per leaf on either side of the midvein, at approximately 5 mm from the periphery, and an average of absolute SPAD values was recorded.Measurements were made on the most recent fully expanded leaf closest to the growing tip, which ranged from leaf three to six depending on the leaf production and expansion rate [20,21].The chlorophyll concentration of the leaves was divided into two distinct groups, namely, green leaves had a lower concentration of chlorophyll (<55) and dark green leaves had a higher concentration of chlorophyll (≥55).One hundred and twenty-six trees (three trees per morphotype) ing stages were chosen in a 4000 ha region using a stratified samplin morphotypes.The definition of leaf and fruit qualitative characterist through preliminary research that was conducted throughout the g

Fruit and Seed Characteristics
Number of fruits per tree was obtained by direct counting at harvest and was categorized as few (≤30), moderate (>40-60), and numerous (>70).This was followed by the harvesting of a total of twelve fruits per tree, where three fruits were sourced from four different directions, when they started to acquire yellow patches on the pericarp (Figure 4).Mass (g) for individual fruits was determined using a balance (Mettler Toledo PM 200).Fruit mass was categorized into light (≤250 g), medium (>250-350 g), and heavy (>350 g).Fruit length (mm) and width (mm) were measured using Vernier callipers.Fruit sizes were divided into small (≤80 mm), medium (>80-90 mm), and big (>90 mm) (Figure 4).Fruit yield per tree (kg/tree) was also calculated using the number of fruits per tree and total fruit mass and categorized into low (≤5 kg/tree), intermediate (>5-10 kg/tree), and high (>10 kg/tree).Fruits were cracked open, and the pericarp thickness was also measured using Vernier callipers (mm) and defined as thin (≤3 mm), medium (>3-4 mm), and thick (>4 mm) pericarp.The quantity of seeds in each fruit was manually counted and classified as few (≤50), moderate (>50-80), and numerous (>80) seeds per fruit.Vernier callipers were used to measure the length (mm), width (mm), and thickness (mm) of 10 seeds from each fruit.Seed length was defined as short (≤20 mm) and long (>20 mm), seed width narrow (≤15 mm) and broad (>15 mm), and seed thickness as thin (≤4.5 mm) and thick (>4.5 mm).Total seed mass (g) per fruit was measured using a balance (Mettler Toledo PM 200) and was then characterized as light (≤40 g), moderate (>40-60 g), and heavy (>60 g) seed per fruit.

Data Analysis
Data were subjected to ANOVA using the GenStat 15th edition.Means were separated using Tukey's Honest Significant Difference (HSD) at 5% significant level.Correlations, biplots and Principal Component Analysis (PCA) were implemented to determine multi-character variation.Cluster Analysis (CA) was applied to record variation among morphotypes.The clusters were applied to fifteen variables using Ward's method of linking based on Euclidean distance.a. Estimation of variance components The phenotypic, genotypic and environmental variances and coefficient of variation were calculated according to the formula described by Burton and Devane [22], and cited by Singh [23], as follows: The quantity of seeds in each fruit was manually counted and classified as few (≤50), moderate (>50-80), and numerous (>80) seeds per fruit.Vernier callipers were used to measure the length (mm), width (mm), and thickness (mm) of 10 seeds from each fruit.Seed length was defined as short (≤20 mm) and long (>20 mm), seed width narrow (≤15 mm) and broad (>15 mm), and seed thickness as thin (≤4.5 mm) and thick (>4.5 mm).Total seed mass (g) per fruit was measured using a balance (Mettler Toledo PM 200) and was then characterized as light (≤40 g), moderate (>40-60 g), and heavy (>60 g) seed per fruit.

Data Analysis
Data were subjected to ANOVA using the GenStat 15th edition.Means were separated using Tukey's Honest Significant Difference (HSD) at 5% significant level.Correlations, biplots and Principal Component Analysis (PCA) were implemented to determine multicharacter variation.Cluster Analysis (CA) was applied to record variation among morpho-types.The clusters were applied to fifteen variables using Ward's method of linking based on Euclidean distance.

a. Estimation of variance components
The phenotypic, genotypic and environmental variances and coefficient of variation were calculated according to the formula described by Burton and Devane [22], and cited by Singh [23], as follows: Phenotypic variance (δ 2 p): where: MSG: Square due to genotype MSE: Square of error (environmental variance) R: where δ 2 p: Phenotypic variation δ 2 g: Genotypic variation X: Grand mean of the character studied Estimation of heritability in broad sense: Broad-sense heritability (H 2 ) expressed as the percentage of the ratio of the genotypic variance (δ 2 g) to the phenotypic variance (δ 2 p), according to Allard [24], was calculated with the following formula: Heritability levels were categorised as low (≤40%), moderate (>40-60%), high (>60-80%), or very high (>80%), according to Singh [25] with modifications.Genetic advance (GA) was estimated as per the formula given by Allard [24] and cited by Meena [26].

Fruit and Leaf Attributes Used to Name Strychnos spinosa Morphotypes
A total of 42 morphotypes of S. spinosa were defined based on fruit and leaf characteristics (Table 1).Green was the dominant colour of fruits, exhibited by 83%, while only 17% of morphotypes had purple fruit colour.The six distinct types of fruit texture varied from rough (55%), rough and corrugated (19%), smooth (11%), very rough (10%), smooth and corrugated (2%), and corrugated (2%), in descending order.Many (64%) morphotypes had round fruits, but a minority (36%) had pyriform shape.

Total Chlorophyll Content
Trees with dark green leaves contained significantly more chlorophyll than those with green leaves (Table 2).Morphotype GvRR-dGEO had the greatest total chlorophyll content in season 1 (67.04SPAD value) and season 2 (68.51 SPAD value), whereas GvRR-GRO had the lowest chlorophyll content in the first season (44.21 SPAD value) and the second season The majority (64%) of morphotype bore fruits till maturity in both seasons, whereas only morphotypes GRP-dGEF, GRP-dGEO2, GRP-dGRO2, GRR-dGEO2, GRR-GEO1, GRR-GEF2, GRxCR-dGEO2, GRxCR-dGRO, GSR-GEF2, GSR-GRO, GSxCR-dGRF, GvRR-dGEO, PRR-dGEF2, PRxCP-dGEO2 and PRxCP-GEO produced fruits till maturity in only one of the two seasons (Figure 5).This variation was observed as the number of fruits fluctuated in successive seasons.Fruit bearing fluctuated significantly in all morphotypes, except for GSR-dGRF and PRR-dGRO1, as these morphotypes retained a significantly high amount of fruits, which amounted to above 60, during both seasons.The highest number of fruits was produced by morphotype PRR-dGRO1 (85) in season 1 and morphotype GRP-GEO2 (112) in season 2, while morphotypes GRR-GEF1 (3) and GRxCR-dGRO (1) produced the least number of fruits in the first and second season, respectively.
Amongst the whole population, morphotypes which produced numerous fruits were only 10% in season 1 and 24% in season 2; and those which produced a moderate amount of fruits were 5% in season 1 and 19% in season 2. However, more morphotypes with the least fruit production were seen in season 1 (60%) than season 2 (48%).
GRP-GEO1 (38.45 kg/tree) and PRR-dGRO (33.33 kg/tree) had the highest fruit yield, while GRR-GEF1 (1.67 kg/tree) and GRxCR-dGEF (1 kg/tree) had the lowest fruit yield in season 1 and season 2, respectively (Figure 6).Fruit yield for GSR-dGRF and PRR-dGRO1 was consistently the highest across seasons.Fruit production varied seasonally within the indicated categories.In season 1, 24% of morphotypes had an intermediate output, but this fell to 14% in season 2. Furthermore, there were only 14% of high-yielding morphotypes in season 1, but these trees were able to yield more (43%) in season 2. However, this variation was not present among the low-yielding trees, which remained constant throughout the different seasons.

Fruit Size
Morphotypes GSR-dGRF and GSR-GEF2 produced the heaviest fruits in season 1 (464 g) and season 2 (556 g), respectively (Table 3).The lightest fruits were produced by morphotype GRxCP-dGEF in season 1 (245 g) and GRxCR-dGEF in season 2 (182 g).There was seasonal variation among the morphotypes in fruit mass.These fluctuations were evident in trees with fruits that had a light mass in season 1 (2%) and increased in season 2 (14%); fruits that were medium in season 1 (69%) declined in fruit mass in season 2 (43%); heavy fruits in season 1 (29%) also increased in season 2 (43%).Fruits with an average mass outnumbered others in the population in season 1.However, a comparable proportion of medium and heavy fruits were produced in season 2.
Most morphotypes with a rough-textured pericarp had a brown and thick pulp, where a minority had a white-coloured pulp.On the contrary, morphotypes with a smoothtextured pericarp had a light brown watery pulp.The heaviest pulp mass was obtained from morphotypes GSR-dGRF in season 1 (121.5 g) and GRxCP-dGEF in season 2 (144.5 g).However, morphotypes GRxCP-dGEF in season 1 (56.0 g), and PRR-dGEF 2 in season 2 (45.4 g), produced the lightest pulp.The mass of the pulp constituted 26 % of the total fruit mass.

Seed Size
Fruits with an average number of seeds per fruit constituted the majority of the population, followed by fruits with numerous seeds and fruits with few seeds, in both seasons (Table 4).Morphotypes GRR-GEF1 and PCR-dGRF produced the greatest number of seeds in the first (105) and second (100) seasons, whereas morphotypes GRP-dGRO1 and GRxCR-dGEF produced the least number of seeds in season 1 (50) and in season 2 (37).
The majority of trees had fruits that produced seeds with an average mass, and the minority produced seeds with a light mass, in all seasons (Table 4).However, heavy seed mass did not change seasonally.In season 1, morphotypes GRR-GEF1 (93.8 g) in the first season, and PCR-dGRF (78.4 g) in the second season, had the heaviest seeds per fruit, whereas morphotypes GRxCR-dGEF (26.8 g) and GRP-dGRO1 (38.6 g) had the lightest seeds per fruit in season 1 and season 2, respectively.Seed mass accounted for approximately 20% of the total seed mass.
In season 2, morphotypes GSxCR-dGRF (6.2 mm) and GRxCP-GEF (6.2 mm) were trees with fruits with the thickest seeds, whereas morphotypes GSR-GEO (3.4 mm) and GvRxCR-GEF (3.9 mm) had the thinnest seeds (Table 4).Fruits with the thickest seeds also produced the biggest seed-sized fruits in length and width.The thinnest seeds corresponded to the smallest seeds which had the shortest length and the narrowest seeds.

Principal Component Analysis (PCA)
The first five informative principal components (PC1-4) were responsible for 66.033% cumulative variability, with each principal component having an eigenvalue greater than 1.0 (Table 6).The first principal component (PC1), with 31.475% of the total variation, was positively associated with mass, length, and width of fruits, number of seeds per fruit, and seed mass.The second principal component (PC2), with 14.270% of the total variability, was positively correlated with leaf length, leaf width and seed width.The third principal component (PC3), responsible for 12.261% of the total variability, was positively associated with number and yield of fruits per tree, but negatively with pericarp thickness.Plant height correlated positively with the fourth principal component (PC4), responsible for 8.027% of total variability.

Scatter Plot Analysis
In the scatter plot, all traits correlated positively with either PC1, PC2 or both (Figure 7a).A positive correlation was determined for fruit width, number of seeds per fruit, fruit length, seed mass and stem diameter in PC1 only.PC2 was positively defined by fruit number, chlorophyll, yield, plant height, leaf length and leaf width.Both PC1 and PC2 were positively defined by seed thickness, pericarp thickness, seed width, seed length, canopy radius and fruit mass.
The highest (2.1) genetic advance was recorded for fruit yield per tree, fruit length, fruit width, pericarp thickness, seed length, seed width and seed thickness, while the lowest (0.9) was in chlorophyll content.The following traits, in descending order, had high genetic advance values (more than one): seed mass, number of seeds per fruit, fruit mass, leaf length and leaf width.

Identification and Tree Characteristics of Strychnos spinosa Morphotypes
Variation between morphotypes, as detected by quantitative descriptors, revealed a great diversity in fruit and leaf traits of Strychnos spinosa (Table 1).The description of the diversity among S. spinosa based on the morphological descriptors allowed for the formation of 42 morphotypes.However, the majority of the morphotypes had green, rough, round fruits with dark green, elongated, open leaves.Plant characterization is the process of documenting and compiling data on significant traits that define variants within species, and therefore enable simple and rapid identification across phenotypes [27].Morphological characterization is an essential method used to identify desired qualities, and these characteristics may help breeders involved in genetic improvement initiatives for agricultural production to select genotypes based on phenotypes [28].Additionally, when a species has been recognized and its phenotype characterized, its variability and life cycle, as well as its ecological niche and role within communities and ecosystems, should also be known in order to fully understand the species [29].
In this study, S. spinosa fruit shapes were observed to vary between roundish and pyriform, and the fruit colour to be green or purple colour at immaturity.A purple tinge on the immature leaves of some morphotypes was also noted.In some trees, fruits may start with a purple colour, which eventually fades to green between 8 to 35 weeks after flowering.However, the exact stage at which the shape changes to roundish and the purple colour begins to fade into green is unknown.As a result, several studies have exclusively reported on green-coloured fruits with round shapes [1,10,11,13].Changes in pericarp colour and texture are governed by the combination of genetic and hormonal variables throughout fruit development [30].During ripening, the cell wall structure deteriorates, resulting in the structure's weakening, and this causes textural changes, although these may vary between species [31].Purple peel formation at an immature stage of the Minhou wild banana (Musa itinerans Cheesman) was a result of higher anthocyanin gene expression and anthocyanin metabolism [32].
Studies that have examined the pericarp texture of S. spinosa have only reported a smooth pericarp texture [1,33,34].The present study has provided evidence for variation in the texture of the pericarp, which included smooth and corrugated, rough, rough and corrugated, very rough, and very rough and corrugated pericarp.A granular fruit pericarp with rough texture has previously been reported for S. cocculoides fruit [12].This shared phenotypic characteristics between S. spinosa and S. cocculoides require investigation to assess of possible hybridization.
The canopy diameter ranges of 3.10 to 8.50 m obtained in this study (Table 2) are within the range of 1.75-9.70m that was recorded for S. spinosa in Umhlabuyalingana, KwaZulu-Natal [35].The canopy of a fruit tree is the primary area for respiration and photosynthesis, as well as the first component to come into contact with light and the environment [36].It is an important plant component that influences fruit tree development and productivity [37].
The tallest trees in this study were 6.9 m, while the shortest trees measured 2.5 m (Table 2).Similarly, plant height ranged from 2.49 to 5.89 m for S. spinosa trees in Umhlabuyalingana, KwaZulu-Natal [35], but the tree height range could be shorter, as noted in Benin where the height varied between 1.97 and 4.11 m [38].This all suggests that S. spinosa is generally not a tall tree.
Many indigenous tree species cannot reproduce until they reach a specific height [39].The trees selected in this study were already at fruit-bearing stage.Thus, S. spinosa trees are capable of bearing fruits at very short heights; less than 3 m.However, an advantage of taller trees could be that they are able to collect and retain nutrients and carbohydrates more effectively and subsequently increase reproductivity [40].
A range of S. spinosa stem diameter from 3.20-10.60cm (Table 2) was somehow comparable to 2.87-26.11cm range among plants in Umhlabuyalingana, KwaZulu-Natal [35], but lower than that of 10.41-29.40cm among plants in Benin [38].Stems of woody plants can be used for storage of food, which can be used in reproduction.Hence, the thickness of the stem can be an important parameter that influences the reproductive capacity of trees, especially those with a biannual fruit-bearing habit such as S. spinosa.There is a need to investigate the role of stem diameter in relation to fruit production.Stems undergo radial symmetric growth in one direction, which contributes to the increase in stem diameter [41].Strychnos spinosa stem is fissured, and the degree of fissuring increases with the age of the plant [10].Variation in stem diameter across trees or within the same species is influenced by factors such as age, internal regulatory mechanisms and external environmental variation throughout the growth period time [42].

Leaf Size and Chlorophyll Content
Leaf shape variation among S. spinosa has been noted previously [12].In the current study, variation in leaf characteristics included shape, width, length and leaf margin (Table 2).Strychnos spinosa demonstrated extreme variation in leaf size and shape [12], which are important traits that may express the extent of leaf area, and hence the seasonal integral of light interception, which can directly affect plant yield [43]. Leaf size (length, width and area), together with the chlorophyll content, are among the most influential photosynthetic components of the tree canopy.The shape and size of the leaf affect the distribution of light in the canopy, and this, together with chlorophyll, determines how light is utilized by the plant in photosynthesis.Differences are expected between the open-and closed-leaf morphotype in the use efficiency of incident light in photosynthesis.Likewise, differences are expected in the capture and use efficiency of incident light between morphotypes with different leaf shapes and sizes.A narrow leaf is expected to lessen the effect of leaf overshadowing, and hence allow a larger photosynthetically active leaf area per canopy size compared with a canopy with large wide leaves, and can directly affect plant yield [43]. Morphotypes with high chlorophyll content are expected to perform more photosynthetically compared to those with lower chlorophyll.This will be more advantageous in deep canopies where light penetration becomes a limiting factor.A wavy leaf edge will help to allow deeper light penetration into the canopy, and as such create gaps through which light filters down to lower leaves.
The total chlorophyll content of S. spinosa morphotypes ranged from 44.21-68.51SPAD value (Table 2).Strychnos spinosa leaves which had a higher chlorophyll content were dark green compared to those which had less chlorophyll content.Similarly, in Nigeria, leaf colour for S. spinosa also ranged from light to dark green above and paler beneath [10].Additionally, leafy vegetables with dark green foliage had a high SPAD value compared to other light green leafy vegetables [44].Chlorophyll is the molecular basis for the function of photosystems and is also a promising tool for ecological prediction [45].Chlorophyll pigment is an important underlying plant trait that varies widely among species [46].

Fruit Number and Yield
In the current study, fruit number ranged from 1-112 fruits per tree (Figure 5), resulting in fruit yield that ranged from 1-38.45 kg/tree (Figure 6).These values are substantially lower than those reported from in a study in Zimbabwe, where a single tree can produce as much as 300-700 fruits, this being approximately equivalent to 40-100 kg/tree [5].These differences show that fruits at Bonamanzi Game Reserve, although fewer, were three times heavier than the ones in Zimbabwe.This could be related to the hotter climate in Zimbabwe [47] than in South Africa, which might promote the formation of numerous but lighter fruits.Despite these adverse climatic conditions, S. spinosa continues to produce and thrive in drought-prone and semi-arid conditions [5].Most morphotypes showed biennial fruit bearing, in which in one season they produced many fruits, but showed less or no fruit production in the following season.This fruiting behaviour in S. spinosa was recently reported [48].It is a widespread phenomenon in many fruit tree species and causes severe labour, marketing and economical problems [49].Some of the problems experienced by the rural communities concerning the cultivation of the underutilized fruit of a crop such as S. spinosa is low yield when compared with other domesticated fruit trees [1].Thus, alternate bearing of some morphotypes in the subsequent season exacerbates the problem of low yields [50].

Fruit Size
Morphotype GSR-GEF2 produced the largest fruits, measuring 96.0 mm in length and 98.9 mm in width, whereas GRxCR-dGEF produced the smallest fruits, measuring 64.9 mm in length and 67.6 mm in width (Table 3).Strychnos spinosa fruit size ranges from 80-120 mm in diameter [50] and from 60-150 mm in both length and width [11].In general, S. spinosa fruits require approximately 100 days to reach full size at maturity [50].
Fruit mass is important in the composition of fruit quality traits that can also be used to estimate tree productivity [51].Many breeding efforts are being made towards increasing fruit size and mass [52].Fruit mass is a highly variable trait of S. spinosa.In the present study, it varied from 182 to 556 g, which was comparable to the range observed by Omotayo and Anemu [11].Mass selection is one of the earliest strategies of selection employed by farmers; it increases the frequency of favourable genes in a population [4].Fruit pulp constituted of about 26% of the total fruit mass.This small proportion of the pulp is one of the constraints towards its processing into porridge and fermented and unfermented beverages, especially by rural communities [5].The variation of white, light brown and brown colour, as well as the watery and thick texture of the pulp among the studied morphotypes, corresponds with the variation in colour and texture of S. spinosa fruit pulp in Zimbabwe [5].Strychnos spinosa fruits are covered with a very hard pericarp [10].The protective nature of the pericarp on the pulp and its effects on the storability of the fruit are not at present known.However, it might be speculated that thicker pericarps infer longer storability of the fruit.Nevertheless, this needs to be assessed experimentally.Whatever the case, the pericarp must be broken to access the fruit pulp.It is therefore tempting to think that the thicker the pericarp, the harder it is to break open the fruit, whether by hand or by machine.In this study, fruit pericarp thickness ranged from 2.8 mm to 5.3 mm among the morphotypes (Table 3), which suggests that size of the pericarp is not constant, and it appeared that most of the fruit mass was contributed to by the very hard pericarp.

Seed Size
Morphotype GRR-GEF1 (106) produced the greatest number of seeds, whereas GRxCR-dGEF (37) produced the least (Table 4).These were both morphotypes which were green and round, but the morphotype with the most and heaviest seeds was rough-textured, unlike the morphotype with the least and lightest seeds, which had a corrugated skin texture.
For terrestrial plants, the size of seeds is a key fitness-related trait that links germination to seedling establishment and later life stages [53], as fitness of an individual is determined by the size of its offspring and the effectiveness with which its genotype is transmitted to future generations [29].The data obtained in the present study and other studies [10,54] shows that there is a wide range in seed length, as obtained among plants in this study (14.4-22.7 mm), and the range from 22-24 mm reported previously by Azusu and Nwosu [10] and Boon [54].The heaviest (93.75 g) and lightest (26.83 g) seeds per fruit were recorded in morphotypes GRR-GEF1 and GRxCR-dGEF, respectively (Table 4).Seed mass plays an important role in the establishment of the juvenile phase of the life cycle, principally under conditions where resources are scare [55].However, seed mass constituted about 20% of the overall fruit mass, which is almost proportional to the pulp mass.Respondents in Zimbabwe mentioned that the separation of pulp from the numerous seeds is another constraint which results in low juice yield during product processing [5].

Correlation, Principal Component Analysis and Cluster Analysis
Correlation was chiefly based on the fruit and seed traits that were positively associated with each other (Table 5).This probably means that these traits are good determinants of variation among S. spinosa morphotypes and can also be used in plant improvement and breeding programs [56].According to the PCA, the major discriminating factors for morphotypes are primarily based on the fruit and seed traits (PC1), and later on leaf size and seed thickness (PC2).
Morphotypes were primarily grouped according to fruit texture, where cluster I in the dendrogram had fruits with a rough pericarp (Figure 8).In addition, morphotypes in cluster I were also associated based on lighter, shorter, and narrower fruits, with shorter, narrower and lighter seeds (Tables 3 and 4).The morphotypes with a smooth pericarp texture (GSR-dGRF, GSR-GEF1, GSR-GEF2, GSR-GEO, GSR-GRO, and GSxCR-dGRF), in cluster II (Figure 8), also shared traits such as heavier, longer and wider fruits, with longer, broader, and heavier seeds (Tables 3 and 4).This clustering can be attributed to the traits that positively defined the first principal component, namely: mass, length and width of fruits; as well as the number and mass of seeds (Table 6).This grouping of morphotypes, as determined by the traits in PC1, means that the texture of the fruits was also affected by the size, number and mass of fruits.This can be associated with smooth-textured fruits that were bigger and heavier with numerous seeds when compared with rough-textured fruits.Pericarp texture was also a morphological characteristic differentiating two subgroups in the dendrogram for Citrus maxima [57].

Genetic Parameters
The higher genotypic variance and genotypic coefficient of variation than environmental variance and environmental coefficient variation in reproductive traits of S. spinosa (Table 7) indicates that phenotypic variation of these traits is primarily caused by genetic variation [58].However, because the phenotypic coefficient of variation is higher than genotypic coefficient of variation in these traits, it indicates a minor environmental influence on the expression of the genes in the phenotypic display [59].On the contrary, the phenotypic changes in leaf chlorophyll content were chiefly due to environmental rather than genetic variation.Nonetheless, the genotypic and environmental variation had similar effects on the phenotypic variation in leaf length and width.In this study, variation was present among morphotypes that were found within the same area and experiencing the same climatic condition.On the contrary, morphological variation among S. spinosa in Benin was a result of exposure to different climatic conditions [38].The low heritability and genetic advance in vegetative traits (Table 7) probably indicates that these traits are predominantly governed by non-additive gene action, and direct selection may not be possible because most of the variation is attributed to the environmental effects [60].However, the high heritability and genetic advance in reproductive traits such as fruit yield per tree, fruit length and width, pericarp thickness, and length, width and thickness of seeds (Table 7), suggests the presence of additive genes, and these characteristics are expected to respond to selection with greater efficiency [61].

Conclusions
Differences in morphological features across S. spinosa morphotypes indicate a high level of diversity that could be utilized by breeders to generate new cultivars.The importance of correlation studies in breeding programs stems from the fact that they enable breeders to comprehend the inter-relationships between morphological traits and use the results for selection during the breeding process.Breeding of horticultural crops is important, as fruits are the one of the best foods adapted to provide vitamins that can combat "hidden hunger" in addition to actual hunger.Morphotype GSR-dGRF was not affected by seasonal variation and had a reasonable and balanced fruit production in both seasons, and the rough texture of the morphotype related to a sweeter fruit taste; this can be the starting point for a breeding program for S. spinosa.This is the first study to report purple, pyriform, and rough fruits, as well as a purple tint on the juvenile leaves.However, further phenological research is still required to determine at which stage of plant growth these characteristics are replaced by green, round, smooth fruits with green leaves, and when the purple tint in leaves is replaced by green.Additionally, nutritional and sensory evaluation studies pertaining to these characteristics are still required.These qualitative characteristics are essential for analyzing genetic diversity among S. spinosa morphotypes and could be utilized to choose appropriate morphotypes for crop-improvement breeding programs.Furthermore, all the reproductive traits assessed showed high heritability and genetic advance, which are expected to respond to selection with greater efficiency.The foundation of plant breeding is diversity.There cannot be breeding for a characteristic if there are no variants of that trait.Therefore, it is essential to conserve a collection of genotypes (germplasm) that are indicative of the variety within a species.

Figure 1 .
Figure 1.Map showing the location of Bonamanzi Game Park, under the B KwaZulu-Natal, South Africa.

Figure 1 .
Figure 1.Map showing the location of Bonamanzi Game Park, under the Big Five Municipality, KwaZulu-Natal, South Africa.

Figure 6 .
Figure 6.Fruit yield per tree among S. spinosa morphotypes.Different superscript letter(s) on the bars indicate significant differences between morphotypes and seasons as analyzed simultaneously according to Tukey's Honest Significant Difference (p < 0.05), where initial alphabets are in small cases and then continue to capital letters after "z".The two superscript letters indicate the range of significance among morphotypes across the two seasons.

Figure 6 .
Figure 6.Fruit yield per tree among S. spinosa morphotypes.Different superscript letter(s) on the bars indicate significant differences between morphotypes and seasons as analyzed simultaneously according to Tukey's Honest Significant Difference (p < 0.05), where initial alphabets are in small cases and then continue to capital letters after "z".The two superscript letters indicate the range of significance among morphotypes across the two seasons.

Figure 7 .
Figure 7. Scatter plots based on the first two principal components (PCs) for morphological traits (a) and morphotypes (b) of Strychnos spinosa.Description for morphotypes is in Table1.

Figure 7 .
Figure 7. Scatter plots based on the first two principal components (PCs) for morphological traits (a) and morphotypes (b) of Strychnos spinosa.Description for morphotypes is inTable1.

Table 1 .
Fruit and leaf attributes used to name Strychnos spinosa morphotypes.

Table 3 .
Variation in fruit characteristics among S. spinosa morphotypes.

Table 6 .
Principal components of traits for Strychnos spinosa morphotypes.

Table 7 .
Genetic parameters for morphological traits of Strychnos spinosa morphotypes.