The African baobab (Adansonia digitata) is a multi-purpose tree that is important among African communities as it provides food and a range of raw materials. Its fruits provide essential nutrients and are sold to generate income. As baobab fruits are important to the livelihoods of many people, it is important to understand the causes of variation in fruit production in order to maximise use and for conservation purposes. Many studies have examined fruit production to understand the causes of variation in fruit yields. In Venda, variation in baobab fruit yield has been recorded for 10 years, thus classifying individual trees as either poor producers or producers (Venter and Witkowski, 2011). Poor producers are adult trees producing less than five fruits each year and some not producing at all. While adult trees producing more than five fruits each year are producers. Causes of this variation in fruit production have not been identified. Among other factors, the observed variation in fruit production could be related to differences in ploidy-level among baobab trees. Importantly, few or no studies to our knowledge have been carried out to confirm if difference in fruit production among baobab trees is related to a difference in ploidy-level. The well-known and widespread mainland African baobab, Adansonia digitata, is known to be a tetraploid (doubled number of chromosomes). Recently, a difference in ploidy-level has been revealed. A new diploid species, Adansonia kilima, has been identified in Africa (Pettigrew et al., 2012). Morphological characteristics (floral, pollen, and stomatal size and density), ploidy and molecular phylogenetics suggest the presence of a new species. This new species has been found to overlap the well-known and widespread tetraploid A. digitata’s distribution in Venda. Consequently, the presence of a diploid species that reproduces with a tetraploid species could lead to variation in fruit production in baobab trees. The objectives of this study were to assess if there is any difference in ploidy-level between the poor producers and producers baobab trees in Venda using flow cytometry, assess if stomatal density and size correlate to differences in ploidy-level, and to use microsatellites to estimate levels of gene flow between these baobab trees. Morphological results showed that stomatal size and density are not significantly different between poor producer and producer trees and these features may not be true indicators of difference in ploidy-level for baobabs. Therefore, it is unlikely that difference in ploidy-level is causing variation in fruit production. However, gene flow results showed that there is high mean genetic heterozygosity and low population differentiation expressed in all populations. This suggests that inbreeding is not responsible for the variation in fruit production between poor producer and producer trees. Low population differentiation observed among the populations indicates a large number of common alleles are shared among the populations. Therefore, the high gene flow observed among the populations suggests that poor producer and producer trees are sharing alleles, and what is causing the differences in fruit production remains unclear.
Keywords: African baobab, flow cytometry, fruit producers, gene flow, ploidy-level, poor producers
List of Figures
Figure 2.1. A stomatal opening with two guard cells around it of a baobab
leaf photographed using Nikon Imaging Software elements
connected to an Olympus light microscope at 200X magnification.
(Photo: R. Tivakudze) ………………………………………………………………………….15
Figure 3.1. Comparison of (mean ?? S.D) of stomatal counts between
poor producer (N = 14) and producer (N = 14) baobab fruit
trees. Results from a t-test showed no significant
differences (P = 0.16, ?? ‘ 0.05)………………………………………………………………19
Figure 3.2. Comparison of (mean ?? S.D) of stomatal length between
poor producer (N = 14) and producer (N = 14) baobab fruit
trees. Results from a t-test showed no significant
difference (P = 0.79, ?? ‘ 0.05)……………………………………………………………….20
Figure 3.3. Comparison of (mean ?? S.D) of stomatal area between poor
producer (N = 14) and producer (N = 14) baobab fruit trees.
Results from a t-test showed no significant
differences (P = 0.23, ?? ‘ 0.05)………………………………………………………………20
Figure 3.4. Flow cytometry analysis of relative fluorescence intensity of
baobab nuclei alone………………………………………………………………………………26
Figure 3.5. Flow cytometry analysis of relative fluorescence intensity
of maize (standard) nuclei alone…………………………………………………………….27
Figure 3.6. Flow cytometry analysis of relative fluorescence intensity
of baobab (P2) and maize (P3) nuclei. ……………………………………………………27
List of Tables
Table 2.1. Microsatellite loci list for Adansonia digitata with their base pair size ranges…………………………………………………………………………………………..16
Table 3.5. Summary of the number of alleles found per locus among the
producer and poor producer baobab trees………………………………………………..22
Table 3.6. Summary of the number of alleles found per locus among the
four populations, A, B, C and Q (see text for details)………………………………..23
Table 3.7. Summary of multilocus average heterozygosity (HE) and inbreeding
coefficient (FI) for all samples and between the poor producer and
producer baobab trees……………………………………………………………………………23
Table 3.8. Summary of multilocus average heterozygosity (HE) and inbreeding
coefficient (FI) for all samples and also in the four geographical
locations (see text for details)…………………………………………………………………24
Table 3.9. Global F-statistics for measuring differentiation among poor
producer and producer trees…………………………………………………………………..25
Table 3.10. Global F-statistics for measuring differentiation among the
four geographic locations, A, B, C and Q (see text for details)…………………..25
Table 3.11. Estimation of relative DNA content of baobab samples using flow
cytometry. The Prunus hybrid cultivar ‘Marianne’ was used as
Table 3.12. Nested ANOVA of stomatal density comparing poor producer
and producer baobab trees. No significant differences between
groups but significantly different among the individual trees were found…….41
Table 3.13. Nested ANOVA of stomatal length comparing poor producer and
producer baobab trees. No significant differences between groups
but significantly different among the individual trees were found………………41
Table 3.14. Nested ANOVA of stomatal area comparing poor producer
and producer baobab trees. No significant differences between
groups but significantly different among the individual trees were found…….42
1.1 Literature Review
Indigenous fruit trees
Indigenous fruit trees have many uses and form an important part of the livelihoods of many African communities (Gouwakinnou et al., 2011; Shackleton, 2002). For instance, their importance is due to their nutritional value, medicinal uses, timber uses, social, and economic value (Akinnifesi et al., 2006). Some of the important fruit trees include the African plum (Prunus africana Hook.f; Kalkman), marula (Sclerocarya birrea (A. Rich; Hochst.), baobab (Adansonia digitata L.), tamarind (Tamarindus indica L.), wild mango (Irvingia gabonensis (Aubry-Lecomte ex O’Rorke; Baill.), wild loquat (Uapaca kirkiana Mull. Arg.), monkey orange (Strychnos spinosa Lam.) and ber (Ziziphus mauritiana Lam.) (Shackleton et al., 2000; Akinnifesi et al., 2006; Jama et al., 2007; Wickens and Lowe, 2008). Each part of many fruit trees can be used for a number of purposes. For instance, trunks and branches provide shade in homes and can be used to make wood carvings and firewood. Further, leaves may be used as relish or for extracts of some medicines. Bark and sap can be used to produce utensils, ropes, and glues. Fruit pulp is often used to make juices, wine, and jam, all of which contribute to the diet of African communities (FAO, 1996). Seeds from some fruits yield oil that is used in industry to make varnishes, paints and by pharmaceutical companies to produce facial creams (i.e., EcoProducts Baobab Oil; SCUC, 2006). Most importantly, fruits can be harvested and sold locally and internationally to generate income to meet livelihood needs (Leakey et al., 2005; Vedeld et al., 2007). Consequently, many communities value the fruit trees around them.
Fruit tree usage often depends on what products are most needed by communities, and as a result, different communities prefer certain tree species to others (Poulton and Poole, 2001; Garrity, 2006; Wickens and Lowe, 2008). For example, if trees supply leaves used as relish, trees producing a lot of leaves may be preferred over those that do not produce many leaves. For trees harvested for use as fire wood, species that do not burn out quickly and do not produce too much smoke are preferable (Tietemam, 1991). In some trees where the leaves are harvested and cooked as relish, tree species that produce leaves that are regarded as good-tasting are often harvested (Dhillion and Gustad, 2004). On the other hand, if fruits are required for eating, trees that produce fruits with high nutritional value or are sweet may be preferable to those that do not produce sweet fruits (Babicz-Zieli??ska and Zag??rska, 1998). Since fruit trees are harvested for a variety of purposes to meet the needs of communities, local people play a central role in sustainably harvesting trees around them and conserving these natural resources (Agrawal and Gibson, 1999).
Even though fruit trees have many uses among African communities, they play a major role in food supply among African communities. During periods of droughts and poor crop harvests, food becomes scarce and hunger becomes prominent (Akinnifesi et al., 2006). When such food shortages occur, fruit trees become vital in meeting the dietary requirements of people because they provide essential nutrients. Some fruits have been recorded to have high contents of vitamins, phosphorus, calcium, as well as other essential minerals, and can provide nutrition during food shortages (Akinnifesi et al., 2004). For example, baobab fruit pulp is known to contain more than 10 times as much vitamin C on a mass basis as orange (Sidibe and Williams, 2002). For these reasons, fruit trees are an important part of many rural communities.
Fruit production studies
Studies that have focused on fruit production have suggested several potential reasons behind the variation in fruit production in a number of different tree species. Given the importance of fruit trees as a food source, fruit characteristics such as fruit yield (Shackleton, 2002), size, and taste have been well studied, often in order to maximize fruit production. Furthermore, these traits are also often used criteria to determine which fruit tree species or individuals are preferable. Identifying causes behind fruit variation is necessary to build guidelines for sustainable harvesting and ensure trees will be available for future community use. Consequently, much work has examined potential drivers behind variation in fruit yield in a number of fruit trees. Rainfall has been shown to affect fruit production in many tree species (Stephenson, 1981; Udovic, 1981). For example, rainfall received immediately after pollination has been shown to wash away pollen grains, thus resulting in low fruit set and ultimately low fruit production in both almond (Orteda et al., 2004) and loquat trees in Jordan (Freihat et al., 2008). Further, Shackleton (2002) found that rainfall differences could explain the variation in fruit production between two fruiting seasons in Sclerocarya birrea (marula) in South Africa.
In addition to rainfall and other environmental factors, such as soil type and land form, have been found to affect fruit production in marula trees in north-central Namibia (Botelle et al., 2002). Additionally, Botelle et al. (2002) noted that trees with larger trunk sizes yielded significantly more fruits than the trees with smaller trunks. In Mexican guava trees, variation in fruit yield has been associated with soil conditions such as soil fertility and soil acidity, diseases, and other environmental conditions (Delgado et al., 2007). Alternatively, other factors may contribute to variation in fruit yield. For instance, variation in fruit yield may be due to the number of flowers and premature death of young developing fruits (Stephenson, 1981) or reduced pollinator activity (Freihat et al., 2008). It has also been suggested that fruit yield may be affected by damage on trees due to the harvesting of leaves and bark (Dhillion and Gustad, 2004). Clearly, variation in fruit yield is of considerable interest, yet a conclusion regarding potential reasons for observed variation has not been found.
Perhaps one of the best studied fruit trees is the iconic baobab tree (Adansonia digitata) due to its importance among African communities. A better understanding of fruit production is necessary since baobab fruits are important in the livelihoods of many people, particularly those in Venda where the trees are economically important (Venter and Witkowski 2013a). In an effort to maximise use of the fruit trees around them, local people often observe and note certain characteristics that affect how they benefit from them (Assogbadjo et al., 2009). Local people often look at characteristics of leaves, bark, and fruits, and often note variation among fruit trees. Through these observations, local people collect information about trees that is useful from a conservation and scientific point of view. For example, variation in fruit yield has been observed in baobabs in Benin and South Africa (Assogbadjo et al., 2008; Venter and Witkowski, 2011). The locals observed that some baobab trees never produce any fruits, while others produced fruits (Assogbadjo et al., 2008), thus identifying poor-fruiting trees as ‘male’ and fruiting trees as ‘female,’ however, the baobabs are bisexual (Sidibe and Williams, 2002; Assogbadjo et al., 2008) with both male and female parts in the same flower. Even though local people made these critical observations in distinguishing between these trees (Assogbadjo et al., 2008), they had no scientific explanation as to what caused some baobabs to produce fruits and some to fail to produce any fruits despite producing flowers.
Similarly, observations of baobab fruit in Mali and Sudan note variation in fruit yield, size, and nutritional values (De Smedt et al., 2011; Gebauer and Luedeling, 2013), as well as which trees produce tasty fruits. In these populations, fruit yield was negatively influenced by the degree to which people harvested fresh leaves for cooking, which in turn resulted in the number of fruits per adult tree declining (Dhillion and Gustad 2004).
The observations made by local people have been corroborated by findings of Venter and Witkowski (2011). In that study, fruit production was found to vary between baobab trees in Northern Venda. Approximately 41 % of adult trees consistently produced fewer than five fruits per year, and were then classified as ‘poor producers’. Other trees in the same study area consistently produced more than five fruits (and usually much more than five) each year, and were thus classified as ‘producers’. Interestingly, the ‘poor producer’ trees also produced flowers; however, few of these flowers produce fruits (S. Venter, 2013, pers. comm.). Venter and Witkowski (2011) suggested that environmental conditions may not be causing differences in the observed variation in fruit yields because the poor producer and producer trees were often found growing next to each other and most likely to be sharing same environmental conditions. Further, fruit production in these baobab trees also varied between years (Venter and Witkowski 2011). Venter and Witkowski (2011) also found tree size and land-use type not influencing trees to be poor producers or producers. Moreover, in the same study rainfall received did not correspond to the fruit production in the same season. Therefore, the reasons behind some trees being poor producers and producers remain unclear.
Causes of variation in fruit yield
Although many ecological causes have been explored, relatively few genetic causes have been examined. One possibility is that inbreeding may result in reduced fruit production for some individuals. Inbreeding leads to reduced fitness (inbreeding depression) of offspring for certain traits, such as germination rate, competitive ability, growth rate, pollen quantity, number of ovules, and amount of seed (Jain, 1976; Silvertown, 2001; Keller and Waller, 2002; Frankham et al., 2003). However, Baum (1995) conducted self-pollination trials on Madagascan baobab trees (Adansonia grandidieri, A. rubrostipa, A. madagascariensis and A. gregorii) and found that there was no inhibition of pollen tube growth in the style, which suggests that these species may be self-compatible. Thus, if baobabs can pollinate their own flowers, inbreeding depression could potentially cause the observed variation between poor producer and producer trees. In a similar trial, Baum, (1995) further examined A. gibbosa and found that there was a delayed abortion after approximately one month of self pollinated and non-pollinated flowers. As a result, the likelihood of baobab inbreeding and causing some baobab trees to be poor producers and producers is unresolved.
Other studies suggest that the mainland African baobab may be self-incompatible (unable to self pollinate). For example, Rao (1954) noted that it is common to have sterile A. digitata, observing that fruits generally develop well with tender and juicy walls, but become hard after a while, resulting in the seeds failing to develop. These data suggest that A. digitata may be self-incompatible (Wickens and Lowe, 2008). Further, Assogbadjo et al. (2008) suggested that baobab trees in Benin that did not produce any fruits have been influenced by either genetic inbreeding among particular baobab trees or some incompatibility within the reproduction system of baobab trees that did not produce fruits. In addition, A. digitata exhibits considerable morphological variations across its range. Assogbadjo et al. (2009) went on to study the genetic differentiation among eight different morphotypes observed within baobab populations in Benin. The different phenotypes were recognised through a morphological classification system which local farmers used identifying trees with desired or undesired combinations of traits. Amplified fragment length polymorphism (AFLP) marker information was used, but found no genetic variation among the morphotypes (Assogbadjo et al., 2009), which suggests that the eight different baobab phenotypes studied in Benin are genetically similar.
Another possible reason for the noted variation in fruit yield between poor producer and producer trees in Venda may be differences in ploidy-level in the genus Adansonia. Adansonia digitata is tetraploid (four sets of chromosomes) and is found only on mainland Africa, whereas Adansonia species found in either Madagascar or Australia are diploid (two sets of chromosomes, like most organisms; Baum, 1995; Wickens, 1982). Recently, work has suggested that there is a possibility that a diploid progenitor exists in mainland Africa (Pettigrew et al., 2012). This new diploid species, Adansonia kilima Pettigrew, Bell, Bhagwandin, Grinan, Jillani, Meyer, Wabuyele and Vickers, sp. nov, may have subtle morphological (floral and pollen characteristics, and stomatal length and density) and distribution differences (moderate elevation of about 650- 1500 m) from the widespread A. digitata, though both species are said to overlap in northern South Africa in the Venda region (Pettigrew et al., 2012). Consequently, the presence of A. kilima may represent a possible explanation for the observed variation in fruit production in baobab trees that occur in northern South Africa.
Polyploidy (whole genome duplication) has long been reported in plants (Levin, 1983; Stebbins, 1971) and is associated with enhanced vigour, altered morphology, increased sterility, higher pest or disease tolerance, and restoration of hybrid fertility. In addition, it can influence reproductive compatibility and fertility (Stebbins, 1971). Ramsey and Schemske (2002) highlighted that infertility in polyploids is complex and may be due to meiotic aberrations, and physiological effects of polyploidy, or genetic factors. Incidental effects of polyploidy may result in increased differences in the way information from genes is used in synthesis of functional genes, which reduces the number or viability of gametes produced and may also affect the growth and development of organisms (Ramsey and Schemske, 2002). Meiotic aberrations have been shown to be the most general factor affecting fertility in polyploids due to the high incidence of unpaired chromosomes and non-homologous chromosome pairing during meiosis (Stebbins, 1971; Ramsey and Schemske, 2002). Furthermore, reproduction between tetraploid (A. digitata) and diploid (A. kilima) baobab trees may have produced triploid offspring, which often result in infertility as suggested above. Therefore, infertility may be caused by lack of homologous pairing resulting from the production of unbalanced, unviable, and semi-sterile gametes (Ramsey and Schemske, 2002) and result variation fruit production between poor producer and producer trees in Venda.
Polyploidy often affects plant morphology, with the most direct and universal effect being an increase in cell size (Baum et al., 1998; Stebbins, 1971). Interestingly, within the baobab distribution, there is evidence indicating the existence of a number of forms differing in fruit size and shape, habit, vigour and leaf morphology (Pakenham, 2004; Pettigrew et al., 2012; Munthali et al., 2013). Many varieties have been described and may be a result of morphological and genetic diversity observed within the African baobab population (Pettigrew et al., 2012). For instance, Sanchez et al. (2010) studied leaf morphology (e.g., leaf length and thickness, and stomatal density and size on the leaf surfaces) of baobab trees in Benin from different agro-climatic zones and found significant variation in leaf size and stomatal characteristics. The authors linked the observed differences in leaf morphologies to the environment and inherent drought tolerance of baobabs. An alternative explanation might be that there is a difference in ploidy-level that lead to the observed differences in leaf morphologies. The number and density of stomata can also be influenced by the ploidy-level of the plant. Diploid plants tend to possess leaves with greater stomatal densities and with stomata that are smaller in size (aperture) than in tetraploid plants (Stebbins, 1971). Interestingly, Pettigrew et al. (2012) found that Adansonia kilima (diploid) leaves have smaller stomatal apertures (mean length of 26.1 ??m) and higher stomatal densities (5 per 100 ??m2) than the tetraploid A. digitata. Adansonia digitata leaves were found to have bigger stomatal apertures (38.1 ??m) and lower stomatal density (1.6 per 100 ??m2).
Given the potential variation in ploidy-level, or genome size, between the two presumed baobab species in mainland Africa, poor fruit production in baobabs may be related to infertility due to differences in ploidy-level. As a result, this study aims to 1) determine if there are ploidy-level differences among the baobab trees sampled in northern Venda and on two islands off the coast of Mozambique and 2) determine if the observed variation in fruit production among the trees in Venda is linked to ploidy-level. The Mozambican trees sampled include trees also classed as poor producer and producer trees. Therefore, there was need to include them in this study.
Use of molecular data
Prior to the advancement of molecular (DNA-based) data, genetic variation, kinship, and phylogenies were estimated using comparisons of phenotypic data from physiology, morphology, and behaviour observed in organisms (Avise, 2004; Conner and Hartl, 2004). Now, however, molecular approaches are widely used in population genetics to examine gene flow among individuals (Avise, 2004) and also to determine ploidy-level. Some of these molecular approaches include microsatellites, flow cytometry, and AFLP. Microsatellites are useful molecular markers to estimate gene flow from both parents due to its co-dominant nature. Microsatellites are stretches of short mono-, tri-, or tetra-repeats of DNA sequences of variable lengths that are distributed throughout the eukaryotic nuclear genome and are found in both coding and non-coding regions (Conner and Hartl, 2004). Genetic knowledge helps to better understand viability in the near future and long-term evolutionary potential of species in view of environmental changes that may occur (Munthali et al., 2013).
Using molecular data, the causes behind the observed variation between poor producer and producer baobabs may be addressed. Molecular data can assist to better understand genetic variation in the poor producer and producer baobabs in Venda. Therefore, this project aimed to investigate whether fruit production variation in poor producer and producer baobab trees is linked to possible differences in ploidy-level among trees in the Venda region of the Limpopo Province, South Africa. In addition, I also aimed to estimate gene flow and test for potential inbreeding among the producer and poor producer trees.
Larsen et al. (2009) suggest that gene flow studies provide an insight into dispersal processes that shape the genetic structure, particularly of baobabs. The co-dominant nature of microsatellites and the wide dispersal across eukaryotic genomes (Koreth et al., 1996; Avise, 2000) makes them useful markers for the study of local gene flow and population structure by determining levels of genetic variation. Spatial genetic structuring in tree species has been shown to be influenced by many biological forces such as gene flow through seed and pollen dispersal, tree density, fragmentation, colonization history, isolation into small numbers, differential mortality, and micro-environmental selection (Kyndt et al., 2009). This same genetic structuring could be evident in the producer and poor producer baobab trees.
Molecular studies have been done in baobabs trees in previous years from West Africa in order to assess genetic variation (Assogbadjo et al., 2009; Kyndt et al., 2009; Larsen et al., 2009), but few studies have used microsatellites. Most of these studies have been carried out in Benin, Ghana, Burkina Faso and Senegal (Assogbadjo et al., 2009; Kyndt et al., 2009). These studies generally showed high levels of genetic variation and that genetic diversity varies between baobab populations in different climatic regions. The authors suggest that observed patterns of genetic variation are influenced by many factors such as seed and pollen dispersal, colonisation history, fragmentation, and micro-environmental selection (Heywood, 1991; Kyndt et al., 2009), which may affect the genetic structure in tree species (Kyndt et al., 2009). Recently, microsatellite primers developed by Larsen et al. (2009) have been used in Malawi to establish genetic differentiation and diversity in baobabs (Munthali et al., 2013). In my study, nine polymorphic microsatellite loci were used to assess gene flow between poor producer and producer baobab trees from Venda.
The African baobab (Adansonia digitata L., Malvaceae) is an iconic tree (Venter and Witkowski, 2010) with multiple traditional uses across different African communities (Pakenham, 2004; Sidibe and Williams, 2002; Wickens and Lowe, 2008). For instance, it is a great source of food because it is a good source of vitamin C and phosphorus (SCUC, 2006). The pulp is mixed with water to make a refreshing drink and is also used as an ingredient in baking. The seeds of baobab fruits are roasted and ground to produce coffee (SCUC, 2006). Twigs, flowers, seeds, leaves and fruits are all used as common ingredients in traditional dishes for rural people (Sanchez et al., 2010). Furthermore, tender young baobab leaves in particular are used as vegetables; they can also be dried and cooked later as they are a good source of vitamin A and calcium (SCUC, 2006).
The economic value of the baobab is derived not only from its value as a food source, but also as an important raw material for a variety of uses. The seeds are crushed to extract oil that is used as an ingredient in the international cosmetic industry (Venter and Witkowski 2013a) and are burnt to ashes for use as soap. Empty seed pods are curved to make cups, fishing floats, and snuff boxes (Pakenham, 2004). Further, the pulp in the fruits contains sterols, saponins and triterpenes, which are used in medicine as they have pain killing (analgesic) and temperature reducing (antipyretic) effects (Pakenham, 2004; SCUC, 2006). The baobab bark is used for fibre to make ropes, fishing lines, nets, bark clothes, baskets and strong harnessing ropes (Pakenham, 2004). All of these products that are obtained from the baobab tree contribute to income and help to alleviate poverty, improve livelihoods and allows participation of marginalized people in a growing cash economy (SCUC, 2006; Venter and Witkowski, 2013a). In addition to industrial uses, huge, hollow African baobab trees have been used for other purposes, such as providing shelter, storage of water, as well as prisons or burial sites. Some are used as religious meeting places, stables, storage rooms, watchtowers, and as restaurants or pubs (Pakenham, 2004; SCUC, 2006; Pettigrew et al., 2012).
Given that baobabs are important for the livelihoods of African communities (Sidibe and Williams, 2002; Venter and Witkowski, 2011), many studies have focused on this iconic tree. One particular area of interest is the variation in fruit production observed between individual trees. This variation has been observed by local people in Benin who use baobab products, and as a result of this variation, they viewed trees that produce fruits in very low numbers as ‘male’ trees, and high fruit producing trees as ‘females’ (Assogbadjo et al., 2008). This pattern is also evident in South Africa in the Venda region, where poorly fruiting trees were named ‘poor producers’ and those producing many fruits ‘producers’ (Venter and Witkowski 2011). Despite a number of studies on variation in fruit production across many tree species, the causes behind these large differences observed among baobabs remain unresolved.
There are many factors that may cause variation in fruit production. Some of the factors suggested to be causing variation in fruit production include adverse conditions such as high or low temperature and low water availability, poor soil fertility, soil salinity and unfavourable soil pH (Stephenson, 1981; Botelle et al., 2002), predation and damage (Dhillion and Gustad, 2004; Venter and Witkowski, 2010). Additionally, variation in fruit production may be caused by limited activities of pollinator agents (Zimmerman and Aide, 1989).
In poor producer and producer baobab trees, causes of variation in fruit yield remain unclear. A possible reason could be linked to the new species recently identified described by Pettigrew et al. (2012), viz., Adansonia kilima using mainly morphological features (floral, pollen, and stomatal size and density) to reveal the presence of this second mainland African baobab, A. kilima. This new species has been noted to be diploid (having two sets of chromosomes) as compared to the widely spread tetraploid (four sets of chromosomes) A. digitata. Polyploidy is known to cause cell size increase (Stebbins, 1971), due to increased DNA content subsequently affecting morphology; Increased DNA content could be one of the reasons why there is variation in fruit production between poor producer and producer baobabs. Further, mating between diploid A. kilima and tetraploid A. digitata could contribute to variation in fruit production among individuals. If mating occurs between diploid and tetraploid baobab trees, the offspring may be infertile triploids, due to unbalanced gametes (Ramsey and Schemske, 2002); this may be causing the differences in fruit production observed. This study aimed to investigate the causes of this variation in fruit production between poor producer and producer trees, and specifically, to test if fruit production is linked to difference in ploidy-level. The study also aimed to examine and compare morphological features (stomatal density and size) of the poor producer and producer trees. Given that Pettigrew et al. (2012) found differences in stomatal size and density between A. digitata and A. kilima, I tested whether stomatal size and density differed between producers and poor producers and whether this correspond with difference in ploidy. Another aim was to examine gene flow between the poor producer and producer trees using nine microsatellite loci and 30 individual trees across four populations in Venda, South Africa and one population from Mozambique.
2.1.2 Objectives of the study
1) To quantify stomatal density and measure stomatal size on the abaxial surface of baobab leaves and correlate these with any differences in ploidy.
2) To use flow cytometry to determine if there is variation in ploidy-level among the mainland African baobab trees in three populations in Venda, South Africa and one population from Mozambique and to correlate any differences with leaf morphology, notably stomatal features.
3) To examine gene flow between producers and poor producers using microsatellite loci.
1) Is stomatal density and size linked to a difference in ploidy-level? And does this match the differences reported by Pettigrew et al. (2012) between A. digitata and A.kilima?
2) Is there difference in ploidy-level between poor producer and producer trees in Venda, South Africa?
3) Is difference in ploidy-level correlated with baobab fruit trees being poor producers or producers?
4) Is there gene flow between producer and poor producer baobab trees in the Venda region of South Africa?
2.2 Materials and methods
2.2.1 Study species
The genus Adansonia of subfamily Bombacoideae in the Malvaceae has eight species (Baum and Oginuma, 1994; Wickens and Lowe, 2008). Two species are endemic to specific regions, A. digitata, is thought to be the only mainland African species that occupies the drier parts of the African continent and A. gregorii F. Muell., is confined to western Australia. The other six species are endemic to Madagascar (Gebauer and Luedeling, 2013; Pettigrew et al., 2012; Wickens and Lowe, 2008). Adansonia digitata is the only species that is tetraploid, unlike the diploid species found in Madagascar and Australia (Pettigrew et al., 2012; Wickens and Lowe, 2008). Recent work by Pettigrew et al. (2012) suggests the presence of a new diploid species (Adansonia kilima), the type of which is in Africa; in southern Africa A. kilima was been near Tshirolwe, in Venda, South Africa. Pettigrew et al. (2012 reported that A. kilima grows in restricted elevations (between 650-1500 m), opposed to the widespread A. digitata usually growing at elevations below 800 m. Surprisingly, this potentially new species went unnoticed despite many years of research on the genus Adansonia (Pettigrew et al., 2012).
In this project, I focus on the mainland African baobab tree (Adansonia digitata). The African baobab is a deciduous tree, shedding leaves mostly in the winter dry season and bearing flowers and leaves in summer (Wickens and Lowe, 2008). Baobab trees seldom exceed a height of 25 m. The cylindrical trunk gives rise to thick tapering branches resembling a root system, which is why it has often been referred to as the ‘upside-down tree’ (Gebauer and Luedeling, 2013). Baobab trees can be very long lived, and previous age estimates suggest that the oldest baobab trees are over 2000 years old (Wickens, 1982). Interestingly, baobab seedling establishment in northern Venda has been episodic, possibly only occurring every 100’150 years (Venter and Witkowski, 2013b). Trees are restricted to hot, dry woodland on stony, well drained soils, in frost-free areas that receive low rainfall. In South Africa, baobabs are found mainly in the hot and dry Limpopo Province (Wickens, 1982; Wickens and Lowe, 2008).
2.2.2 Study Area
Young leaf samples were collected on 26 February 2013 from 26 individuals in the Venda region in Limpopo province. Individuals sampled in this study are known as either ‘producers’ or ‘poor producers’ based on a study conducted by Venter and Witkowski (2011) that classified adult trees that produced less than five fruit per year as ‘poor producers’ and those that produced more than five fruits per year as ‘producers’. Samples were also collected from three individuals from the Mozambican Islands, Quilalea and Senco, on 11 March 2013. Leaf samples were immediately placed in filter paper in resealable plastic bags with silica gel to rapidly dry the leaves and preserve the DNA. The sample collection was done based on the location these trees found around the Venda villages, and this aided in naming these trees. Poor producer and producer trees were found growing mixed in the same area in all different locations. The locations were named A (most western locality near Muswodi village), B (most northern locality near Matalu and Tshipise villages), C (most eastern locality near Tshikuyu village) and Q (Mozambican). The distance was approximately 25 km between village A and B, and village B and C, but about 60 km was between village A and C. Included in the sampling was the type of A. kilima found near Tshirolwe in Venda.
2.2.3 Stomatal analysis
To measure stomatal density and size, clear fingernail polish was used to create an impression of the abaxial surface of the leaf epidermis. The clear fingernail polish was applied on the abaxial epidermis of the selected leaf following methods outlined by Saltonstall et al. (2007). Once the clear fingernail polish dried on the leaf surface, the dried layer was peeled off by firmly pressing sellotape at the edge of the dried fingernail polish on the leaf, then carefully pulling it off. This peeled layer was then placed on a glass microscope slide, pressed flat using a cover slip, and observed using a light microscope (Olympus BH-2). Stomata counts were recorded for three random fields of view per peel at 200X magnification. A systematic approach to counting was done by observing a particular field of view by first counting from the top left side going down to the bottom, then taking a slight right turn, then counting going upwards, at the top end a right turn was taken again then counting proceeding going downwards. By so doing, all stomata on a single field of view were counted. A haemocytometer was then used to recount stomata to verify the initial counts. The grids on the haemocytometer allowed demarcation of a particular field of view.
The microscope field of view for a 200X magnification was found with the following formula:
Field of View = ??r2
= 22 (0.8 mm) 2
Field of View area = 2.01 mm2
Therefore, each field of view measured 2.01 mm2, and counts were made for three separate fields of view within one leaf peel. An average was then calculated for the three fields of view to give an average number of stomata per 2.01 mm2. The mean values of stomatal density were compared between producer and poor producers using the independent sample Welch t-test in statistical package R 2.12.1 version (R Development Core Team, 2010).
2.2.4 Guard cell size
The same peels of dried impressions used to count stomatal density were also used for guard cell size measurements. Measurements were done using Nikon Imaging Software elements D3.1 (NIS-elements). This software enables image capture, object measurement, and counting of objects on a screen from a microscope (Figure 1). First, calibration was done using a 2 mm micrometer that was placed under the microscope. A measurement of 0.1 mm was done on the micrometer using the NIS-elements and calibrated to measure in microns (1 mm = 1000 ??m). The 0.1 mm was calibrated by equating it to 100 ??m. After calibration, the dried peels were individually put under a microscope at 200X magnification. Thirty stomata were randomly selected to measure length (L) and width (W). The area of the stomata was calculated using the formula of an ellipse, which best represents the shape of the guard cells: Area = 0.5 ?? (L x W). The independent Welch t-test was also used to compare differences in the mean guard cell lengths between producers and poor producers. The mean area of guard cells was also calculated and differences were compared between producers and poor producers using the Welch two sample t-test.
Figure 2.1. A stomatal opening with two guard cells around it of a baobab leaf photographed using Nikon Imaging Software elements connected to an Olympus light microscope at 200X magnification (Photo: R. Tivakudze).
2.2.5 Gene flow analyses
DNA was extracted using Qiagen DNEasy Plant Mini Kit following manufacturer’s instructions with minor modifications as follows. The volume of the buffers, AP1 and P3, was increased from 400 ??l to 800 ??l and from 130 ??l to 260 ??l, respectively. Previously published microsatellite primers for Adansonia digitata (Larsen et al., 2009) were used to amplify microsatellites to estimate gene flow between producers and poor producers. Optimum polymerase chain reaction (PCR) conditions were set for nine polymorphic markers (Table 2.1) to produce amplification products following Larsen et al. (2009). PCR reactions consisted of a 10 ??l final volume: 1.5 ??l of DNA template, 2 ??l of nuclease free water, 0.5 ??l of Bovine Serum Albumin (BSA), 0.5 ??l each of 10 ??M forward and reverse primers, and 5 ??l Phusion Master Mix (Thermo Scientific; Inqaba Biotech, Pretoria, South Africa). The thermo cycler conditions followed instructions supplied with the Phusion Master Mix but annealing conditions followed Larsen et al. (2009). The PCR conditions were as follows: an initial denaturation step at 98 oC for 10 s, followed by 30 cycles of denaturation cycles for 10 s at 98 oC, annealing at 58 oC for 5 s and extension step at 72 oC for 15 s, and final extension at 72 oC for 1 min, the reactions were held at 20 oC. The PCR products were then visualized on 1 % agarose gel stained with SYBRSafe (BIO RAD). After verification of the presence of a band within the correct size range, successful PCR products were multiplexed and sent to the Central Analytical Facility (CAF) at Stellenbosch University for analysis on an ABI 3130.
Table 2.1. Microsatellite loci list for Adansonia digitata with their base pair size ranges.
Locus name Size range Motif
Ado1 94’124 (AG)
Ado2 262’298 (TC)
Ad04 176’224 (CT)
Ad08 265’301 (GAA)
Ad09 181’211 (AAG)
Ad12 159’187 (AG)
Ad14 169’187 (AC)
Ad17 177’201 (AC)
Ad18 251’271 (TG)
Microsatellites were visualized and recorded using PeakScanner v1 (Applied Biosystems, www.appliedbiosystems.com). PeakScanner was used to determine the size of the alleles found in each sample for the selected microsatellite locus. I calculated genetic allele frequency, heterozygosity, inbreeding coefficient, and kinship coefficient between poor producers and producers using SpaGeDi (Hardy and Vekemans, 2002). Gene flow analysis was done on two subpopulations (producers vs. poor producers) within the baobab samples. I further divided the individual trees into four groups based on the geographic location of the three different populations of the baobab trees in Venda and one population in Mozambique. The groups were A, B, C, and the fourth group (Q) comprised the Mozambican trees as described above. The grouping was done because the trees in each group were found in the same locality and in order to analyse gene flow for the trees in the same location. Nonetheless, gene flow between the three Venda populations may be possible as they are in relatively close proximity compared with the Mozambican population.
2.2.6 Flow cytometry for ploidy-level analyses
Flow cytometry was used to determine the relative DNA content for both producer and poor producer trees. A flow cytometer enables visualization and quantification of moving particles in a suspension (Johnston et al., 1999). The inbuilt programme of the flow cytometer then converts the fluorescence signal obtained from the stained particles into a graph. All cells containing the same relative DNA content contribute to the same peak on the graph. Given the variation in fruit production and the presence of a potential new diploid species, I expected that poor producer and producer baobab trees might have different genomic sizes. Fresh young baobab leaves were collected on 26 and 27 October 2013 in Venda for analysis of their ploidy. In the lab, these fresh leaves were weighed together with a standard, Zea mays to obtain a combined mass of 0.05 mg. Both tissues were co-chopped using new razor blades in a petri dish and stained with 500 ??l of DAPI one-step Cystain kit following the manufacturer’s instruction to release nuclei. After chopping the sample for 45’50 s, 500 ??l of DAPI stain was added and the chopped tissue was incubated for 2 min in the dark on ice to allow DNA staining to take place. After incubation, the mixture was filtered through a 30 ??m mesh filter. Filtration was done to eliminate debris, such as the vacuole, cytoplasm and other soluble substances found in the plant cells, obtained through the rough chopping of the plant tissues. The filtrate was then centrifuged at maximum speed for 30 s. The supernatant was discarded, and the DNA was resuspended using 1000 ??l of DAPI stain. This solution was run through a Fortessa flow cytometer at the University of the Witwatersrand, Johannesburg Medical School.
These procedures were then repeated in November on a flow cytometer at Stellenbosch University Central Analytical Facility (CAF) to confirm relative DNA content. At Stellenbosch CAF a hybrid plum tree cultivar ‘Marianne’ (a hybrid of Prunus munsonian and P. cerasifera) was used as a standard. A two-step Cystain kit was used, with an initial addition of 500 ??l of lyse buffer followed by 80 ??l of DAPI stain, and the other steps were similar to the one-step Cystain kit described above. The DNA C-values (amount of DNA in pico grams) of Marianne was not known and the values for Prunus munsonian and Prunus cerasifera were obtained from Kew Royal Botanic Gardens DNA C-values data base (http://data.kew.org/cvalues). Although values were not available for the two parents, a literature search suggested that P. cerasifera was synonymous to P. domestica (2C DNA content = 0.66pg; Loureiro et al., 2007) and that the other parent, P. munsonian was closely related to P. angustifolia (2C DNA content = 0.61 pg) and were found within a polytomy of the same clade (Baird et al., 1994; Shaw and Small, 2005). I therefore estimated the Marianne genome size by averaging the genome sizes of the close relatives (0.66 pg and 0.61 pg, respectively) of Marianne parents. The average estimated genome size (0.635 pg) was used as the standard in the equation below. Mean genome sizes (picograms) and standard errors for all samples were calculated using the following equation from Saltonstall et al. (2005):
Genome size = (Mean position of baobab peak/mean position of Marianne peak) X 0.63 pg
2.2.7 Statistical analyses
Mean stomatal density was compared between poor producer and producer trees using the independent sample Welch t-test in R version 2.12.1 (R Development Core Team, 2010). The same test was performed on the stomatal density obtained from the haemocytometer counts. Mean guard cell length and width were calculated for 30 randomly selected stomata per leaf sample. Guard cell area was also calculated for each stoma using the formula as mentioned above, obtaining area for the 30 randomly selected stomata. Mean guard cell area of the poor producer and producer trees were compared using the independent sample Welch t-test in R 2.12.1 version (R Development Core Team, 2010). The guard cell length and width were also compared using the same t-tests to test if there were differences between poor producer and producer trees. A nested ANOVA was also conducted to determine if there were differences between individual trees and the two groups (poor producer and producer trees).
3.1.1 Stomatal density and size
The independent Welch t-test was used to compare the stomatal size and density between poor producer and producer trees. Stomatal density did not significantly differ between poor producer and producer trees (t = 1.4642, df = 24.66, P = 0.1558; Figure 3.1). Similarly, the length of the stomata was not significantly different between the poor producer and producer trees (t = ‘0.2713, df = 25.06, P = 0.7884; Figure 3.2). Finally, no significant difference was found in stomatal area between poor producer and producer trees (t = 1.2264, df = 25.214, P = 0.2314; Figure 3.3).
Figure 3.1. Comparison of (mean ?? S.D) of stomatal counts between poor producer (N = 14) and producer (N = 14) baobab fruit trees. Results from a t-test showed no significant differences (P = 0.16, ?? ‘ 0.05).
Figure 3.2. Comparison of (mean ?? S.D) of stomatal length between poor producer (N = 14) and producer (N = 14) baobab fruit trees. Results from a t-test showed no significant difference (P = 0.79, ?? ‘ 0.05).
Figure 3.3. Comparison of (mean ?? S.D) of stomatal area between poor producer (N = 14) and producer (N = 14) baobab fruit trees. Results from a t-test showed no significant differences (P = 0.23, ?? ‘ 0.05).
3.1.2 Nested ANOVA analyses
Result from the nested ANOVA did not suggest differences in stomatal density that were calculated for each individual sample nested within the producer group or the poor producer group. Stomatal density was not significantly different between the poor producer and producer trees (F = 2.14, P = 0.55; Table 3.12). However, stomatal density was significantly different among the individual samples (F = 21.48 P < 0.01; Table 3.12). Results from a nested ANOVA analysis for stomatal size showed no significant differences in stomatal length between poor producer and producer trees (F = 0.074, P = 0.78; Table 3.13), but stomatal length was significantly different among individual samples (F = 17.70, P < 0.01; Table 3.13). Stomatal area among poor producer and producer trees was not significantly different (F = 1.50, P = 0.23; Table 3.14), but stomatal area was significantly different among individual samples (F = 22.51, P < 0.01; Table 3.14).
3.1.3 Gene flow analyses
Allele frequencies appeared to vary between poor producer and producer trees (average 10.33 alleles for poor producer trees vs. 11.67 alleles for producer trees; Table 3.5). The most alleles were scored at locus Ad04 (20) and the least number of alleles was scored at locus Ad18 (8 alleles; Table 3.5). Variation in allele size ranged from 94 to 301 base pairs (bp). Total number of alleles and allele size (bp) varied among the populations across all the nine loci (Table 3.5).
Table 3.5. Summary of the number of alleles found per locus among the producer and poor producer baobab trees.
Number of alleles
Loci Producer Poor producer All
All 11.67 10.33 12.67
Ad01 15 11 16
Ad02 12 13 14
Ad04 17 17 20
Ad08 11 9 11
Ad09 9 9 9
Ad12 12 10 14
Ad14 8 7 9
Ad17 13 10 13
Ad18 8 7 8
The average number of alleles across all four populations was 12.67 (Table 3.6). Generally alleles were shared between trees from populations A, B and C. Trees from population Q (from Mozambique) did not share as many alleles with trees in locations A, B and C. Trees in location Q (from Mozambique) showed low gene flow between trees relative to those in locations A, B and C. The average number of alleles in trees from populations A, B, C and Q was 8.78, 10.56, 8 and 4.78 respectively (Table 3.6).
Table 3.6. Summary of the number of alleles found per locus among the four populations, A, B, C and Q (see text for details).
Number of alleles
Loci A B C Q All
All 8.78 10.56 8 4.78 12.67
Ad01 9 14 8 5 17
Ad02 11 12 10 6 14
Ad04 12 15 9 6 19
Ad08 8 10 6 5 11
Ad09 7 8 7 3 9
Ad12 9 12 8 5 14
Ad14 7 6 7 3 9
Ad17 8 10 11 4 13
Ad18 8 8 6 6 8
To test if poor producers and producer baobab trees showed evidence of inbreeding, I calculated an inbreeding coefficient (FI) for each group. Results among all the baobabs sampled suggest that the trees are all outcrossing (mean FI = ‘0.154; Table 3.7). Both the producer trees and poor producer trees are likely outcrossers (producer mean FI = ‘0.147; poor producer mean FI = ‘0.167; Table 3.7). The average heterozygosity (HE; Nei 1978) for all populations was high (HE = 0.856; Table 3.7), indicating genetic diversity is high across the populations. Both the producer and poor producer groups showed high genetic diversity (HE = 0.865, HE = 0846; Table 3.7).
Table 3.7. Summary of multilocus average heterozygosity (HE) and inbreeding coefficient (FI) for all samples and between the poor producer and producer baobab trees.
Multi locus average HE (heterozygosity, Nei, 1978) FI (individual inbreeding coefficient)
Producer 0.865 ‘0.147
Poor producer 0.846 ‘0.167
All populations 0.856 ‘0.154
Expected heterozygosity and inbreeding were also calculated for baobab individuals that were divided into the four populations recognised according to geography: A, B, C (Venda), and Q (Mozambique). Tree samples in all of the populations were found to be outcrossers (Table 3.8). The results showed that average heterozygosity (HE; Nei 1978) for all populations was high (HE = 0.857; Table 3.8).
Table 3.8. Summary of multilocus average heterozygosity (HE) and inbreeding coefficient (FI) for all samples and also in the four geographical locations (see text for details).
Multi locus average HE (heterozygosity, Nei, 1978) FI (individual inbreeding coefficient)
A 0.849 ‘0.166
B 0.854 ‘0.156
C 0.868 ‘0.151
Q 0.794 ‘0.271
All populations 0.857 0.15
In order to clearly understand the population differentiation among the poor producer and producer trees, Global F-statistics were used to incorporate three levels of population structure (within subpopulations (FIS), among subpopulations (FST) and the individual differentiation within the population (FIT). The average FST across all loci showed that there is little population differentiation between poor producer and producer trees (FST = 0.0018; Table 3.9). The average FIS for all loci demonstrated that the individual tree samples are out-crossing (FIS = ‘0.1551), which corroborates the inbreeding estimates. Global F-statistics were also used to fully understand the population differentiation among the four geographic categories, A, B, C and Q. The average FST on all loci is 0.0182, showing that there is little population differentiation among the four locations. The average FIS for all loci was ‘0.1652, demonstrating that the individual tree samples are out-breeding, which corroborates the FI values (Table 3.10).
Table 3.9. Global F-statistics for measuring differentiation among poor producer and producer trees.
Locus FST FIS FIT Pairwise Ds (Nei’s 1978 standard distance)
All 0.0018 ‘0.1551 ‘0.153 ‘0.026
Ad01 0.0009 ‘0.1353 ‘0.1343 ‘0.0383
Ad02 0 ‘0.1334 ‘0.1334 ‘0.0408
Ad04 ‘0.0016 ‘0.0711 ‘0.0728 ‘0.0796
Ad08 ‘0.0028 ‘0.1264 ‘0.1295 ‘0.0561
Ad09 0.0224 ‘0.22 ‘0.1927 0.0625
Ad12 0.0048 ‘0.1541 ‘0.1486 ‘0.0073
Ad14 0.0027 ‘0.2696 ‘0.2661 ‘0.0248
Ad17 ‘0.0032 ‘0.1297 ‘0.1332 ‘0.0622
Ad18 ‘0.057 ‘0.1804 ‘0.1871 ‘0.0616
Table 3.10. Global F-statistics for measuring differentiation among the four geographic locations, A, B, C and Q (see text for details).
Locus FST FIS FIT Pairwise Ds (Nei’s 1978 standard distance)
All 0.0182 ‘0.1652 ‘0.144 0.0908
Ad01 0.0234 ‘0.1528 ‘0.1258 0.1642
Ad02 0.018 ‘0.1439 ‘0.1233 0.0679
Ad04 0.0069 ‘0.0763 ‘0.0689 0.0337
Ad08 0.038 ‘0.1551 ‘0.1112 0.2909
Ad09 0.0272 ‘0.2244 ‘0.191 0.1602
Ad12 0.0143 ‘0.1671 ‘0.1504 0.0976
Ad14 0.0124 ‘0.2796 ‘0.2637 ‘0.0835
Ad17 0.0384 ‘0.1444 ‘0.1005 0.3351
Ad18 ‘0.0175 ‘0.165 ‘0.1854 ‘0.2538
3.1.4 Flow cytometry
Results obtained from the Fortessa flow cytometer at the University of the Witwatersrand Medical School were inconclusive. When a one-step Cystain kit was used, running the baobab stained DNA material alone, good output peaks were obtained (Figure 3.4). Similarly, when the standard was run alone in the Fortessa, it yielded good peaks with a defined position and size (Figure 3.5). However, when baobab DNA material was stained together with the standard (maize) the results showed unclear peaks, different from the ones obtained by baobab DNA alone, and maize DNA alone, making it difficult to distinguish the two peaks (Figure 3.6). Therefore, I was unable to calculate reliable estimates of DNA content using this approach. The samples were then taken to Stellenbosch CAF to obtain clearer results using the Prunus hybrid cultivar Marianne as a standard and the two-step Cystain kit. Although data obtained from the Stellenbosch CAF analyses suggest variation in genome size (DNA content) between poor producer and producer baobab trees, the genome sizes estimates obtained were also inconclusive. Estimation of relative DNA content of the unclear graphs obtained showed that the producer trees AP4, AP5 and AV1 may be diploids (0.35 pg, 0.44 pg and 0.47 pg respectively; Table 3.11). While the results showed that some of the poor producer (AP3 and BP4) and producer trees (CV5, AV3, AV4 and BF3; Table 3.11) may be tetraploid.
Figure 3.4. Flow cytometry analysis of relative fluorescence intensity of baobab nuclei alone.
Figure 3.5. Flow cytometry analysis of relative fluorescence intensity of maize (standard) nuclei alone.
Figure 3.6. Flow cytometry analysis of relative fluorescence intensity of baobab (P2) and maize (P3) nuclei.
Table 3.11. Estimation of relative DNA content of baobab samples using flow cytometry. The Prunus hybrid cultivar ‘Marianne’ was used as the standard.
Fruiting history Sample Marianne Fluorescence Value CV% Baobab Fluorescence value CV % Baobab DNA estimates (pico grams)
P CV5 36.2 17 64.4 9.7 1.13 4x
PP AP3 35.81 17.3 85.264 12.8 1.51 4x
P AP4 61.022 9 34.09 12.7 0.35 2x
P AP5 41.444 13 28.64 15.7 0.44 2x
P AV1 41.145 13.1 30.741 15.2 0.47 2x
P AV3 36.136 17.6 56.433 19.1 0.99 4x
P AV4 42.465 12.7 69.063 22.5 1.03 4x
P BF3 39.65 16.1 65.437 8.3 1.05 4x
PP BP4 50.818 12.4 84.639 6.2 1.06 4x
P = producer, PP = poor producer
4.1.1Stomatal density and size
Polyploidy is known to influence the cell size of organisms and can affect reproductive function in plants. In addition, the number and density of stomata can change relative to ploidy-level. Diploid plants tend to possess leaves with greater stomatal densities and with stomatal apertures that are smaller in size than tetraploid plants (Stebbins, 1971). In baobab trees, Pettigrew et al. (2012) found that A. kilima (diploid) has significantly smaller stomatal apertures and higher stomatal densities than the tetraploid A. digitata. Tetraploid A. digitata individuals were found to have larger stomatal apertures and lower stomatal density. This study examined the stomatal density on the leaf surface, and the length and area of individual stomata to test for differences between poor producer and producer trees and to correlate the differences with ploidy-level. This study also estimated the ploidy-level of the poor producer and producer trees using flow cytometry. It also examined gene flow between poor producer and producer trees, to test whether these trees are exchanging genes or if fruit variation in these trees is a result of inbreeding.
In this study, stomatal length and area were not significantly different between poor producer and producer trees (Figure 3.2; Figure 3.3). Poor producer and producer trees had mean stomatal lengths of 26.54 ??m and 26.28 ??m, respectively. Similarly, a mean stomatal length of 26.1 ??m was obtained for the type of A. kilima (a diploid species; Pettigrew et al., 2012). Therefore, poor producer and producer baobab trees both have similar stomatal lengths to that of A. kilima as identified by Pettigrew et al. (2012), and poor producer and producer baobab trees in this study cannot be distinguished using stomatal length and size. Surprisingly, the results suggest that baobabs in Venda may all be diploid if only stomatal length is considered. However, it is more likely that stomatal density and stomatal size are not effective indicators of difference in ploidy-level between poor producer and producer baobab trees. A similar study by Saltonstall et al. (2005) showed that stomatal density showed a significant relationship with subspecies and was useful in distinguishing between two subspecies of Phragmites australis, but DNA content was the same for both of the subspecies. They concluded that morphological features in Phragmites australis may not be accurate indicators of difference in ploidy-level. Therefore, the stomatal length and area in poor producer and producer baobabs may not be a true representative of whether the trees are diploid or tetraploid.
In addition, I found no significant difference in mean stomatal density between poor producer and producer trees. This suggests that poor producer and producer trees cannot be distinguished based on stomatal density. The similarity in mean stomatal density may be a result of these trees occurring in the same locality with similar environmental conditions. This is in accordance to Sanchez (2010) in a study on relationship between stomatal characteristics and drought adaptation in Benin and Malawi, baobab leaves in Benin were found having higher stomatal density in high temperature and low rainfall areas, whereas smaller guard cell length was obtained in high temperature and low rainfall areas. This however was not consistent in baobabs in Malawi. Even though the mean stomatal density between poor producer and producer trees was not significantly different, it is difficult to compare the densities from this study and the one by Pettigrew et al. (2012). In this study, the abaxial surface of the leaflets was examined for stomatal length and density, whereas Pettigrew et al. (2012) reportedly studied the adaxial surface. No stomata were observed on the adaxial surface in this study when I examined the leaflets at 200X magnification. This is consistent with findings of Sidibe and Williams (2002), where stomata in baobabs were reported only on the abaxial surface of the leaflets. However, Rao and Ramayya (1981) noted that stomata appear on both the abaxial and adaxial surfaces of the leaflets. Pettigrew et al. (2012) used a different microscope at 600X magnification, which may have aided observing stomata on the adaxial surface of the leaves. Sanchez (2010) observed (at 400 X magnification on a similar Olympus microscope) that stomata appear mainly on the abaxial surface but do occur on the adaxial surface of the medial leaflet where they are restricted to alongside the midvein. The stomata were noted to be were absent from the adaxial lamina surface of the leaflets. It is possible that if the mid-vein area of leaflets of poor producer and producer trees had been viewed in this study at higher magnification, a few stomata might have been observed. However, leaflet impressions were easily peeled starting from the edge of the leaflets, whereas impressions from the mid-vein area of the leaflet were difficult to obtain for measurements in this study and were unfortunately not viewed.
4.1.2 Gene flow
Inbreeding (mating of closely related organisms) may result in inbreeding depression. Inbreeding can also affect certain traits, such as germination rate, competitive ability, growth rate, pollen quantity, number of ovules, and amount of seed produced (; Jain, 1976; Silvertown, 2001; Keller and Waller, 2002; Frankham et al., 2003). This study examined gene flow and inbreeding in poor producer and producer trees. I hypothesized that poor producer trees were more likely to be inbred relative to the producer trees. Further, I hypothesized that gene flow would be mainly occurring among producer trees and that poor producers would not contribute significantly to gene flow. However, results showed that both poor producer and producer trees have high mean heterozygosity (Table 3.7). This suggests that both poor producer and producer trees outcross, which corroborates the calculated inbreeding coefficients. Surprisingly, these results are not consistent with recent work on genetic differentiation and diversity carried out in Malawi. Munthali et al. (2013) found evidence for low genetic diversity among baobab populations in Malawi. However, genetic diversity, obtained from AFLPs of Benin baobab populations varies from high to low across the different climatic regions where the trees are found (Assogbadjo et al., 2009). In West Africa, spatial genetic structuring from AFLP data showed high level of within-population genetic diversity (Kyndt et al. 2009). Collectively, these results may suggest that West African baobabs and Venda baobabs have different levels of gene flow and that population structuring may be more prevalent in East and West Africa relative to southern Africa.
Global F-statistics of population differentiation also suggested little population differentiation between producer and poor producer trees in Venda (FST = 0.0018). The G F-statistics values of less than 0.05 represent little population differentiation. Moderate population differentiation is shown when values are 0.05’0.15; values between 0.151’0.25 represent great differentiation and values above 0.25 represent very great differentiation (Conner and Hartl, 2004). These data also suggest that there is a high level of gene flow between poor producer and producer trees preventing differentiation. Little population differentiation implies that there are many common alleles being shared between poor producer and producer trees, with few rare alleles present. Again, southern African baobab populations differ in genetic structuring from populations in Malawi. For example, Malawian populations appear to be moderately genetically differentiated (Munthali et al. 2013). Compared to populations in Malawi, the poor producer and producer trees in Venda all show much higher levels of gene flow that is preventing population differentiation. Consequently, the baobabs of Venda appear to be a single, cohesive population, unlike baobab populations in other parts of Africa.
The poor producer and producer baobab trees have shown high levels of gene flow between them and that they are more outbreeding with high heterozygosity, which is a healthy situation for this species. However, since it remains unclear what drives the variation in fruit production observed between poor producer and producer trees, further studies are required to ascertain the causes of this variation in fruit production between poor producer and producer observed in Venda. Possible studies could pollination studies, pollen viability self incompatibility, flowering and fruit set between the poor producer and producer trees. Pollination studies particularly artificial pollination may help to test if trees are really good outcrossers, even though high gene flow was shown between poor producer and producer, but what is not known is the direction of this gene flow.
4.1.3 Flow cytometry
Although estimates of DNA content suggest variation among the individual baobab trees tested, the estimates of DNA content were inconclusive. The peaks obtained were not reliable enough to distinguish between the baobab sample and the maize standard. However, the data do suggest that there is variation present, which needs to be further verified. The DNA estimates obtained showed that both the poor producer and producer trees may either be diploid or tetraploid (Table 3.11). This seems to conflict with expectations that polyploidy results in increased vigour and increased productivity (Stebbins, 1971). In addition, these results appear to suggest that stomatal measurements do not always correlate with genome size estimates. Furthermore, fruit production does not appear to correlate with variation in DNA content either. However, earlier genomic size estimates of A. digitata using Feulgen microdensitometry (Fe) showed a DNA 2C-value of 7.7 pg (Bennet and Leitch, 1997). Relative DNA content of baobabs in the current study is very low with an average size of only 0.89 pg. The current study used flow cytometry, while Bennet and Leitch (1997) used Fe. This may be the reason why the genome sizes obtained are very different, or because of unreliable flow cytometry output. On another note, difficulties faced in using flow cytometry to obtain good and reliable results can be attributed to secondary chemistry taking place. Baobab leaves are known to contain toxicants such as hydrocyanic acid, oxalate, phytic acid, and tannins (Chadare et al., 2009). The possibility remains that these toxins in baobabs could be reacting with the DAPI stain, thus preventing adequate staining of the nuclei for analysis. The staining protocol was altered several times in order to try and obtain clearer results. However, most alterations did not yield sufficient staining for subsequent analysis. Consequently, there is need for an alternative method of DNA content determination or use of other staining protocols.
The results obtained in this study effectively rule out the possibility of inbreeding and reduced gene flow causing the observed variation in fruit yield between poor producer and producer baobab trees. Further, comparisons between stomatal measurements in this study and those of Pettigrew et al. (2012) suggest that either all baobabs in Venda are diploid, or that stomatal measurements are not a reliable measure of ploidy-level in baobabs. Future work should continue to assess ploidy-level using flow cytometry to better explore potential variation in genome size as a driver for variation of fruit production. Due to the economic and nutritional value of baobab trees, producer trees remain the prime target for harvesting by local people.
To fully explore possible differences in ploidy-level, extensive sampling in southern Africa may be required, so that a much bigger area is covered. Moreover, seed germination followed by chromosome counts on root tips may be done to verify if there is any difference in ploidy-level among the baobab trees. In addition, more morphological features may be added to explore if they correlate with difference in ploidy-level, for example using features such as floral traits from the trees, or from voucher specimens collected; e.g. pollen grain diameter, stamen length and stalk diameter, maximum calyx diameter, and staminal corolla width can also be measured and the number of free staminal filaments could also be counted. Furthermore, hand pollinations could be carried out between poor fruit producers and producers to establish the viability of seeds from these crosses. A number of possibilities exist that could cause variation in fruit production, and changes due to polyploidy (potentially in the morphological features described above), which would be beneficial to examine for a broader understanding of what drives differences in fruit production between poor producer and producer baobab trees in Venda.
I wish to express his grateful thanks to Dr. K. L Glennon, Prof E. T. F. Witkowski and Prof G. V. Goodman-Cron for their kind supervision and helpful suggestions. Thanks are also due to Dr. Sarah Venter for the dried leaf samples provided and the fruiting history she kindly shared and for accompanying us on a leaf-collecting trip. I am also grateful for assistance given by M. Goodman driving around Venda and in the field during fresh leaf sample collection. My family and all other colleagues are gratefully acknowledged for providing assistance throughout my studies. This work is based on the research supported in part by the National Research Foundation of South Africa through their Integrated Biodiversity Information Programme (Grant Number 86959).
Agrawal, A., Gibson, C., 1999. Enchantment and Disenchantment: The Role of Community in Natural Resource Conservation. World Development 27(4), 629’649.
Akinnifesi, F.K., Kwesiga, F.R., Mhango, J., Mkonda, A., Chilanga, T., Swai, R., 2004. Domesticating priority miombo indigenous fruit trees as a promising livelihood option for smallholder farmers in southern Africa. Acta Horticulturae 632, 15’30.
Akinnifesi, F.K., Kwesiga, F., Mhango, J., Chilanga, T., Mkonda, A., Kadu, C.A.C., Kadzere, I., Mithofer, D., Saka, J.D.K., Sileshi, G., Ramadhani, T., Dhliwayo, P., 2006. Towards the development of miombo fruit trees as commercial tree crops in southern Africa. Forests, Trees and Livelihoods 16, 103’121.
Assogbadjo, A.E., Kakai, R.G., Chadare, F.J., Thomson, L., Kyndt, T., Sinsin, B., Van Damme, P., 2008. Folk classification, perception, and preferences of baobab products in West Africa: consequences for species conservation and improvement. Economic Botany 62, 74’84.
Assogbadjo, A.E., Kyndt, T., Chadare, F.J., Sinsin, B., Gheysen, G., Eyog-Matig, O., Van Damme, P., 2009. Genetic fingerprinting using AFLP cannot distinguish traditionally classified baobab morphotypes. Agroforestry Systems 75, 157’165.
Avise, J.C., 2000. Phylogeography: the History and Formation of Species. Harvard University Press, Cambridge, MA.
Avise, C.J., 2004. Molecular Markers, Natural History and Evolution (2nd Edition). Sinauer Associates, Inc; USA
Babicz-Zieli??ska, E., Zag??rska, A., 1998. Factors affecting the preferences for vegetables and fruits. Polish Journal of Food and Nutrition Sciences 4, 755’762.
Baird, W.V., Estager, A.S., Wells, J.K., 1994. Estimating nuclear DNA content in Peach and related diploid species using laser flow cytometry and DNA hybridization. Journal of the American Society for Horticultural Science 119, 1312’1316.
Baum, D.A., Oginuma, K., 1994. A review of chromosome numbers in Bombacaceae with new counts for Adansonia. Taxon 43, 11’20.
Baum, D.A., 1995. A systematic revision of Adansonia (Bombacaceae). Annals of Missouri Botanical Gardens 82, 440’470.
Baum, D.A., Small, R.L., Wendel, J.L., 1998. Biogeography and Floral Evolution of Baobabs (Adansonia, Bombacaceae) as Inferred From Multiple Data Sets. Systematic Biology 47(2), 181’207.
Bennet, M.D., Leitch, I.J., 1997. Nuclear DNA Amounts in Angiosperms-583. New Annals of Botany 80, 169’196.
Botelle, A., du Plessis, P., Pate, K., Laamanen, R., 2002. A Survey of Marula Fruit Yields in North-Central Namibia. CRIAA SA-DC, Namibia, 22pp (+Appendices and photographs).
Chadare, F.J., Linnemann, A.R., Hounhouigan, J.D., Nout, M.J.R., Van Boekel, M.A.J.S., 2009. Baobab Food Products: A Review on their Composition and Nutritional Value. Critical Reviews in Food Science and Nutrition, 49, 254’274.
Conner, J.K., Hartl, D.L., 2004. A Primer of Ecological Genetics. Sinauer Associates, Inc; USA
De Smedt, S., Alaerts, K., Kouyate, A.M., Van Damme, P., Potters, G., Samson, R., 2011. Phenotypic variation of baobab (Adansonia digitata L.) fruit traits in Mali. Agroforestry Systems 82, 87’97.
Dhillion, S.S., Gustad, G., 2004. Local management practices influence the viability of the baobab (Adansonia digitata L.) in different land use types, Cinzana, Mali. Agriculture, Ecosystems and Environment 101, 85’103.
FAO, 1996. Domestication and commercialization of non-timber forest products in agroforestry systems. Food and Agriculture Organisation. Rome, Italy
Frankham, R., Ballou, J.D., Briscoe, D.A., 2003. Introduction to Conservation Genetics. Cambridge University Press. Cambridge, UK.
Freihat, N.M., Al-Ghzawi, A., Zaitoun, S., Alqudah, A., 2008. Fruit set and quality of loquats (Eriobotrya japonica) as effected by pollinations under sub-humid Mediterranean. Scientia Horticulturae 117, 58’62.
Garrity, D., Okono, A., Grayson, M., Parrott, S., (Eds), 2006. World Agroforestry into the Future. World Agroforesty Centre; Nairobi.
Gebauer, J., Luedeling, E., 2013. A note on baobab (Adansonia digitata L.) in Kordofan, Sudan. Genetic Resources and Crop Evolution, Springer; Science Business Media Dordrecht.
Gouwakinnou, G.N., Assogbadjo, E.A., Lykke, A.M., Sinsina, B., 2011. Phenotypic variations in fruits and selection potential in Sclerocarya birrea subsp.birrea. Scientia Horticulturae 129, 777’783.
Hardy, O.J., Vekemans, X., 2002. Spagedi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Molecular Ecology Notes 2, 618’620.
Heywood, J.S., 1991. Spatial analysis of genetic variation in plant populations. Annual Review of Ecology and Systematics 22, 335’355.
Jain, S.K., 1976. The Evolution of Inbreeding in plants. Annual Review of Ecology and Systematics 7, 469-495.
Jama, B.A., Mohammed, A.M., Mulaya, J., Njui, A.N., 2007. Comparing the ‘Big Five’: A framework for the sustainable management of indigenous fruit trees in the drylands of East and Central Africa. Ecological Indicators 8, 170’179.
Johnston, J.S., Bennett, M.D., Rayburn, A.L., Galbraith, D.W., Price, H.J. 1999. Reference standards for determination of DNA content of pant nuclei. American Journal of Botany 86(5), 609’613.
Keller, L.F., Waller, D.M., 2002. Inbreeding effects in wild populations. Trends in Ecology and Evolution 17(5), 230’241
Kew Royal Botanic gardens. http://data.kew.org/cvalues accessed 05/02/2014
Koreth, J., O’Leary, J.J., McGee, O’D.J., 1996. Microsatellites and PCR genomic analysis. Journal of Pathology 178, 239’248.
Kyndt, T., Assogbadjo, A.E., Hardy, O.J., Kakai, R.G., Sinsin, B., Van Damme, P., Gheysen, G., 2009. Spatial genetic structuring of baobab (Adansonia digitata, Malvaceae) in the traditional agroforestry systems of West Africa. American Journal of botany 96(5), 950’957
Larsen, A.S., Vaillant, A., Verhaegen, D., Kj??r, E.D., 2009. Eighteen SSR-primers for tetraploid Adansonia digitata and its relatives. Conservation Genetic Resources 1, 325’328.
Leakey, R.R.B., Tchoundjeu, Z., Schreckenberg, K., Shackleton, S.E., Shackleton, C.M., 2005. Agroforestry Tree Products (AFTPs): Targeting poverty reduction and enhanced livelihoods. International Journal for Agricultural Sustainability 3, 1’23.
Levin, D.A., 1983. Polyploidy and novelty in flowering plants. The American naturalist 122, 1’25.
Loureiro, J., Rodriguez, E., Dolezel, J., Santos, C., 2007. Two new nuclear isolation buffers for plant DNA flow cytometry: a test with 37 species. Annals of Botany 100, 875’888.
Ortega, E., Dicenta, F., Egea, J., 2007. Rain effect on pollen-stigma adhesion and fertilization in almond. Scientia Horticulture 112, 345’348.
Munthali, C.R.Y., Chirwa, P.W., Akinnifesi, F.K., 2013. Phenotypic variation in fruit and seed morphology of Adansonia digitata L. (baobab) in five selected wild populations in Malawi. Agroforestry Systems 85, 279’290.
Nei, M., 1978. Estimates of average heterozygosity and genetic distance from a small number of individuals. Genetics 89, 583’590
Pakenham, T., 2004. The remarkable baobab. Great Britain: Jonathan Ball Publishers.
PeakScanner v1, 2012. Applied Biosystems, www.appliedbiosystems.com)
Pettigrew, F.R.S.J.D., Bell, K.L., Bhagwandin, A., Grinan, E., Jillani, N., Meyer, J., Wabuyele, E., Vickers, C.E., 2012. Morphology, ploidy and molecular phylogenetics reveal a new diploid species from Africa in the baobab genus Adansonia (Malvaceae: Bombacoideae). Taxon 61, 1240’1250.
Poulton, C., Poole, N., 2001. Poverty and Fruit Tree Research: Issues and Options Paper, DFID Forestry Research Programme, Wye College, Asgford, England, 70p.
R Development Core Team, 2010. R: A language and environment for statistical
computing. R Foundation for Statistical Computing. ISBN 3-900051-07-0,
URL http://www.R-project.org/. Vienna, Austria.
Ramsey, J., Schemske, D.W., 2002. Neopolyploidy in flowering plants. Annual Review of Ecology and Systematics 33, 589’639.
Rao, C.V., 1954. A contribution to the embryology of Bombacaceae. Proceedings of the Indian Academy of Science, Section B 39, 51’75.
Rao, S.R.S., Ramayya, N., 1981. Distribution of stomata and its relation to plant habit in the order Malvales. Indian Journal of Botany 4, 149’156.
Saltonstall, K., Glennon, K., Burnett, A., Hunter, R.B., Hunter, K.L., 2007. Comparison of morphological variation indicative of ploidy-level in Phragmites australis (Poaceae) from Eastern North America. Rhodora 109(940), 415’429.
Sanchez, A.C., 2010. Predicting suitable areas for cultivation and conservation of the baobab tree and investigating superior sources of planting material. University of Southampton, Faculty of engineering and the environment. PhD thesis, 231 pages.
Sanchez, A.C., Haq, N., Assogbadjo, A.E., 2010. Variation in baobab (Adansonia digitata L.) leaf morphology and its relation to drought tolerance. Genetic Resources and Crop Evolution 57, 17’25.
SCUC, 2006. Baobab Manual. Field Manual for Extension Workers and Farmers. University of Southampton. Southampton UK.
Shackleton, C.M., Dzerefos, C.M., Shackleton, S.E., Mathabela, F.R., 2000. The use and trade in indigenous edible fruits in the Bushbuckridge savanna region, South Africa. Ecology of Food and Nutrition 39, 225’245.
Shackleton, C.M., 2002. Growth and fruit production of Sclerocarya birrea in the South African lowveld. Agroforestry Systems 55, 175’180.
Shaw, J., Small, R., 2005. Chloroplast DNA phylogeny and phylogeography of the North American plums (Prunus subgenus Prunus section Prunocerasus, Rosaceae. American Journal of Botany 92(12), 2011’2030.
Sidibe, M., Williams, J.T., 2002. Baobab: Adansonia digitata L. Field manual for extension workers and farmers. International Centre for Underutilised Crops.
Silvertown, J., Charlesworth, D., 2001. Introduction to Plant Population Biology. Fourth Edition. Blackwell Science. Oxford, UK.
Stebbins, G.L., 1971. Chromosomal evolution in higher plants. London. Edward Arnold.
Stephenson, A.G., 1981. Flower and Fruit Abortion: Proximate Causes and Ultimate. Annual Review of Ecology and Systematics 12, 253’279.
Suda, J., Travnicek, P., 2006. Reliable DNA ploidy determination in dehydrated tissues of vascular plants by DAPI flow cytometry-New prospects for plant research. International Society for Analytical Cytology, Cytometry Part A, 69A, 273’280.
Tietemam, T., Ditlhogo, M., Tibones, C., Mathalaza, N., 1991. Characteristics of eight firewood species of Botswana. Biomass and Bioenergy 1(1), pp. 4146.
Udovic, D., Aker, C., 1981. Fruit abortion and the regulation of fruit number in Yucca whipplei. Oecologia 49, 245’248.
Vedeld, P., Angelsen, A., Bojo, J., Sjaastad, E., Berg G.K., 2007. Forest environmental incomes and the rural poor. Forest Policy and Economics 9, 869’879.
Venter, S.M., Witkowski, E.T.F., 2010. Baobab (Adansonia digitata L.) density, size-class distribution and population trends between four land-use types in northern Venda, South Africa. Forest Ecology and Management 259, 294’300.
Venter, S. M., Witkowski, E.T.F., 2011. Baobab (Adansonia digitata L.) fruit production in communal and conservation Land-use types in Southern Africa. Forest Ecology and Management 261, 630’639.
Venter, S. M., Witkowski, E.T.F., 2013a. Fruits of our labour: contribution of commercial baobab (Adansonia digitata L.) fruit harvesting to the livelihoods of marginalized people in northern Venda, South Africa. Agroforestry Systems Systems 87, 159’172.
Venter, S. M., Witkowski, E.T.F., 2013b.Where are the young baobabs? Factors affecting regeneration of Adansonia digitata L. in a communally managed region of southern Africa. Journal of Arid Environments 92, 1’13
Wickens, G.E., 1982. The baobab- Africa’s upside-down tree. Kew Bulletin 37, 173-209.
Wickens, G.E., Lowe, P., 2008. The baobabs: pachycauls of Africa, Madagascar and Australia. Springer, UK.
Wiimer, C.M., 1983. Stomata. London: Longman.
Zimmerman, J.K., Aide, T.M., 1989. Patterns of fruit production in a neotropical orchid: pollinator vs. resource. American Journal of Botany 76(1), 67’73.
Table 3.12. Nested ANOVA of stomatal density comparing poor producer and producer baobab trees. No significant differences between groups but significantly different among the individual trees were found.
Source of variation D.F SS MS F P Variance component in %
Between groups 1 3990.96 3990.96 2.144 0.155 ns 6.95
Among subgroups within groups 26 48402.40 1861,63 21.48 <0.01*** 81.16
Within samples 56 4853.33 86.67 11.89
Total 83 57246.70 100.00
ns- not significant
Table 3.13. Nested ANOVA of stomatal length comparing poor producer and producer baobab trees. No significant differences between groups but significantly different among the individual trees were found.
Source of variation D.F SS MS F P Variance component in %
Between groups 1 14.24 14.24 0.07 0.78 ns 0
Among subgroups within groups 26 5031.20
193.51 17.7 <0.01*** 35.76
Within samples 812 8878.58 10.93 64.24
Total 839 13924.02 100
ns- not significant
Table 3.14. Nested ANOVA of stomatal area comparing poor producer and producer baobab trees. No significant differences between groups but significantly different among the individual trees were found.
Source of variation D.F SS MS F P Variance component in %
Between groups 1 93548.63 93548.63 1.50 0.231 ns 1.55
Among subgroups within groups 26 1617176.21 62199.08 22.51 <0.01*** 41.11
Within samples 812 2243670.49 2763.14 57.34
Total 839 3954395.33 100.00
ns- not significant