Variation in the response of eastern and southern Africa provenances of Faidherbia albida (Delile A. Chev) seedlings to water supply: A greenhouse experiment.

Rural communities value Faidherbia albida in farming systems and pastoralism. Faidherbia albida provides products such as medicine, fodder, fuel, wood, food and services such as shade, soil fertility and nutrient cycling. Excessive browsing by animals, branch lopping and pod harvesting, have critically reduced the natural regeneration in some areas which exposes it to challenges due to dependence upon natural regeneration. The objective of this research was to evaluate response of Faidherbia albida provenances from eastern (Taveta Wangingombe) and southern Africa (Lupaso, Kuiseb Manapools) to different watering regimes to aid in selection of provenances for domestication. The observed difference in growth was analyzed to determine whether they are genetic or environmentally induced. Genotype  [Formula: see text]  interaction were significant at ([Formula: see text]0.001, p[Formula: see text]0.05) in seedling height, diameter and leaf numbers. Seedling height (r=0.94 p=0.001) recorded the highest correlation coefficient among all the growth variables analyzed. The growth variation was greater for seedling height than that of diameter and leaf numbers (h[Formula: see text]=0.97). Hierarchical cluster analysis grouped the provenances into three clusters with cluster iii consisting of Taveta, Kuiseb and Lupaso while cluster ii and i composed of Wangingombe and Manapools respectively. Manapools recorded the highest genetic distance from Taveta, Kuiseb and Lupaso at 84.55 units. Wangingombe and Manapools are closely related genetically at a distance of 7.32. The maximum inter-cluster distance between cluster i and iii indicated wider genetic diversity between the provenances in these clusters and selection should be from this clusters for hybridization program to achieve novel breeds.

Introduction

Tree domestication and improvement programs must understand patterns of variation in tree species in order to effectively select, manage and conserve their genetic resources. Pattern of natural variation in adaptively important traits is essential in development of tree improvement and conservation strategies for native hardwood species (Weber et al., 2008). This is particularly urgent for Faidherbia albida (del.) A. Chev (Bastide and Diallo, 1996, Billand and De Framond, 1993, Dangasuk et al., 1997, Roupsard et al., 1998, Sniezko and Stewart, 1989, Wanyancha et al., 1994) and other species that are over extracted (Arnold, 2004, Bowen et al., 1977, Dawson and Powell, 1999, Diallo et al., 2000, Ofori and Cobbinah, 2007, Ofori et al., 2007, Tchoundjeu et al., 1997). Ræbild et al. (2003) report the findings of FAO and the Danida Forest Seed Centre who analyzed several trials to identify provenances from Africa and other continents for reforestation. The number of provenance trials for exotic species outnumber that of native hardwoods as reported by (Appiah, 2003; Diallo et al., 2000; El Amin and Luukkanen, 2006; Loha et al., 2006; Ofori and Cobbinah, 2007; Raddad, 2007; Weber et al., 2008; Wolde-Mieskel and Sinclair, 2000). The situation is worrying because of the changing pattern of climate necessitating the need to focus research on native better adapted provenances. The adaptive capacity that is required to mitigate the effects of climate change will be achieved among others by selection of species that make up a farming system (Afreen et al., 2011). While there has been significant effort to select and breed drought tolerant crop species, there is need to also consider trees when exploring species for consideration in climate smart agricultural systems. This is because trees provide multiple important benefits to farmers and their farming systems (De Leeuw et al., 2014). First, trees positively influence microclimatic and edaphic conditions that are relevant to the production of crop species; second trees provide many goods such as fruits and energy that are relevant to farmers.

Of particular interest for consideration to include in climate smart agriculture are those tree species that have a wide climatological niche. These species are interesting because the adaptation of trees to climate change is dependent on response to the present temperature and rainfall conditions (Austin, 1992, Langlet, 1971). Faidherbia albida is such a species; it is distributed throughout the African continent spanning a wide range of environmental conditions (Barnes and Fagg, 2003). It grows well under deep sandy-clay soils, rocky, heavy and cracking clays and remarkable gradient of 50–1800 mm average annual rainfall across which the species occurs. This broad distribution with respect to rainfall is due to the fact that Faidherbia is a groundwater dependent species (Roupsard et al., 1999).

Aside from its distribution across diverse habitats, Faidherbia albida has a unique reverse phenology of shedding leaves during the rainy season allowing it to grow among field crops without overshadowing them during the wet season and provides shade during dry season (Roupsard et al., 1999). Falling leaf mulch promotes higher microbial activities in the soil, thus improving the soil structure, stability and permeability under the canopy. Increase in yield from the crop grown below the trees has attributed to increase fertility due to nitrogen fixation, dung from the stocks browsing and fallen leaves and pods (Dangasuk et al., 1997). In addition, Faidherbia albida has remarkable capacity for recycling nutrient from underground to the surface due to its very deep root system (Okore et al., 2007).

Beside the benefits mentioned above, Faidherbia albida is appreciated by herdsmen and farmers in arid and semi-arid regions of Africa (Okore et al., 2007). The leaves and pods are palatable to livestock like cattle, goats and sheep. Pastoralist lops the branches to provide dry season browse for their stock (Barnes and Fagg, 2003). Ground pods are highly recommended cattle feeds for milk production (Bwire et al., 2004). The wood of Faidherbia albida is used for construction of dugout canoes, boats, paddles, kitchen utensils, art objects, troughs and fencing (Mokgolodi et al., 2011). The wood ash is used for soap and as a depilatory and bark used as fish poison in Botswana (Barnes and Fagg, 2003). The bark has a high concentration of active component that treats kidney pain and mental illness (Okore et al., 2007). The crushed tree bark homogenized in water is used to treat diarrhea in human (Wondimu et al., 2007). In Nigeria leaf and fruit decoction help to cure leprosy while bark infusions are taken to treat fever, coughs and aid in child birth (Oluwakanyinsola et al., 2010). Seeds of Faidherbia albida are eaten by humans as famine foods although seed requires a long preparation to remove toxins (Barnes and Fagg, 2003).

Because it has many uses, there is intensive extraction pressure on Faidherbia albida in African dry lands. Excessive browsing by animals, branch lopping and pod harvesting, have critically reduced the natural regeneration in some areas which exposes it to challenges due to the fact that it is entirely dependent upon natural regeneration (Wahl and Bland, 2013). Wild animals such as elephants and giraffes have been in many cases identified as the cause of population decline in F. albida leading to this population deficit through low regeneration. Despite adequate seed production, natural regeneration by seed may be limited because of heavy seed predation and high seedling mortality (Turnbull et al., 2008). This, together with the fact that few communities protect and manage natural regeneration, has dramatically reduced the abundance of Faidherbia albida in many areas. In addition, farmers and pastoralists state that many trees are dying due to increasingly hotter, drier conditions in the dry lands and the relatively slow growth during the first few years after planting (Okore et al., 2007). The situation is worsened by little systematic research on genetic variation in growth and survival of native hardwood species in Africa. FAO initiated, and the Danida Forest Seed Centre analyzed several trials to identify some superior provenances from Africa and other continents for reforestation in arid and semi-arid zones in Africa (Ræbild et al., 2003). Numerous provenance trials of several exotic species have been established and evaluated (Andrew et al., 2004, Khasa et al., 1995, Langat and Kariuki, 2004, Pedersen et al., 2007, Ræbild et al., 2003). In contrast, there are relatively few publication of genetic variation in growth, survival and other commercially or adaptively important traits of native African hardwoods (Diallo et al., 2000, El Amin and Luukkanen, 2006, Raddad, 2007, Roupsard et al., 1998).

This paper therefore, evaluated response of five provenances of F. albida to different watering regimes and analyses the observed difference in growth to determine whether the difference is genetic or environmentally induced to aid in selection of suitable provenances for domestication in different environments.

Materials and methods

Study site

The study was conducted in the greenhouse at the World Agroforestry Centre, (ICRAF) tree nursery at latitude 1°33′S longitude 37°14′E in Nairobi.

Seed sources

Seeds of the five provenances of F. albida under study were obtained from the World Agroforestry Centre (ICRAF) genebank. The geographical range of the seeds varied from 3°24′S, 37°42′E to 23°34′S, 15°02′E and altitude of 360 to 1700 m above the sea level. Taveta and Wangingombe represented eastern Africa provenances while Lupaso, Kuiseb and Manapools represented southern Africa provenances. The seed sources are presented in (Fig. 1).

Fig. 1

Eastern and southern Africa countries where the provenances under study were collected (Koech et al., 2014).

Seed germination

One hundred seeds per provenances were selected for germination. The seeds were nicked at the distal end near the microphyle using a nicking caliper and soaked in water for 24 h before sowing (Moser, 2006). The seeds were then germinated in sterilized sand in germination trays measuring at a uniform sowing depth (Zhang and Maun, 1990). The sand was sterilized using the oven method at 70 °C for 48 min. Sand was then saturated with water before sowing the seeds. The germination trays containing the seeds were then kept under a temperature range of 25–30 °C and monitored for four week before transplanting the seedlings to potting media.

Potting and establishment of seedlings

After 30 days of seed germination, 72 healthy seedlings of the same root sizes were transplanted to 400 cm3 pots where forest soil, manure and sand in the ratio of 3:1:1 was used as potting media. The seedlings were grown in a greenhouse with full daylight and controlled temperature within a range of 25–30 °C. The air humidity was maintained at 25%–70% relative humidity and vapor pressure deficit (1.0 kPa) with a proper mixing of the air.

Randomized complete block design with provenances, watering regimes and blocks was used to study the provenance trial. Seedlings of different provenances were randomized within the block. Seedlings were supplied with the same amount of water for two months to allow seedling establishment before initiating the watering regimes. After two months, watering regime was initiated. The watering regimes were based on the field capacity. In the abundantly-watered treatment, 18 pots of each provenance were watered to field capacity with provision of quantities of water equal to that is lost through transpiration; soil water content of the other 18 pots of each provenance was always kept at 25%, 50% and 75% of field capacity throughout the experiment. Evaporation from the soil surface was prevented by enclosing the pots in plastic bags tied to the stems of the plants. The amount of water supplied daily was determined by weighting ten individuals randomly chosen from each watering regime provenance combination. In addition, 10 pots without seedlings were used to monitor evaporative water loss from the soil surface throughout each treatment.

Statistical analysis

The following general linear model was established for all the variables: where; is the value of the variable for the th-seedling from the th-family within th-provenance, located in the th-block in the th-treatment, is overall mean, is the effect of a covariate (seedling height at two months), is the effect of the th-treatment (100%, 75% 50% or 25%), is the effect of the th-provenance (1–5), is the effect of the th-block (1–3) and is the experimental error. The interaction terms and were also included in the model. A significant watering treatment by provenance interaction for a particular variable would indicate significant differences in the phenotypic plasticity between provenances for that trait. Pearson correlation coefficient for seedling growth was determine to establish the best time for selection based on seedling growth traits.

Phenotypic correlation coefficients were computed to examine the degree of association among the height, diameter and leaf numbers of F. albida seedling traits. Multivariate analysis of variance (MANOVA) was conducted to reveal the patterns of variation of quantitative traits studied. Means of each quantitative character were standardized before subjecting it to the principal component analysis (PCA) as was suggested by Reddy et al. (2009). The standardized data of quantitative traits was then used as an input for the PCA biplot loading and cluster analysis. An agglomerative, hierarchical cluster classification technique with Average linkage strategy was performed. Mead et al. (2002) indicated that the measures of similarity and dissimilarity were derived by calculating the Euclidean distance between pairs of objects. The Euclidean measure of distance was used for computing genetic distance among the populations (Mohammadi and Prasanna, 2003). Average linkage treats the distance between two clusters as the average distance between all pairs of items where one member of a pair belongs to each cluster (Kahraman et al., 2011). Heritability was calculated for each trait based on the formula: across all the watering regimes and during different period of growth. All the data was analyzed using GenStat Discovery, version number 12.1.3338, 12th Edition Program (Chope et al., 2012).

Results

Effect of factors

The block effect was not significant across all the studied variables. The effect of watering regime, provenance and interaction terms of the model were significant for most of the studied variables except for the replication meaning that the provenances reacted differently in response to stress. There were significant differences at (, ) among provenances in seedling height, diameter and leaf numbers (Table 1).

Table 1

Analysis of variance for F. albida seedling traits for six months after initiating the watering regime.

SV	d.f	di1	di2	di3	di4	di5	di6	hm1	hm2	hm3	hm4	hm5	hm6	le1	le2	le3	le4	le6
Rep	2	0.003^ns	0.007^ns	0.003^ns	0.034^ns	0.024^ns	0.017^ns	0.341^ns	0.179^ns	0.179^ns	0.203^ns	0.293^ns	0.97^ns	1.915^ns	0.831^ns	1.214^ns	1.59^ns	1.353^ns
Trt	3	0.05⁎⁎⁎	0.06⁎⁎⁎	1.89⁎⁎⁎	3.33⁎⁎⁎	5.10⁎⁎⁎	11.13⁎⁎⁎	38.22⁎⁎⁎	38.5⁎⁎⁎	212.3⁎⁎⁎	480.9⁎⁎⁎	857.1⁎⁎⁎	896.8⁎⁎⁎	83.86⁎⁎⁎	2643.7⁎⁎⁎	617.1⁎⁎⁎	2.98⁎⁎⁎	112 684.8⁎⁎⁎
prv	4	2.40⁎⁎⁎	0.39⁎⁎⁎	11.30⁎⁎⁎	8.47⁎⁎⁎	7.42⁎⁎⁎	13.81⁎⁎⁎	61.45⁎⁎⁎	481.4⁎⁎⁎	571.9⁎⁎⁎	303.7⁎⁎⁎	334.7⁎⁎⁎	470.9⁎⁎⁎	451.58⁎⁎⁎	1482.1⁎⁎⁎	6537.9⁎⁎⁎	1.59⁎⁎⁎	11 845.7⁎⁎⁎
Trt X Prv	12	0.17⁎⁎⁎	0.05⁎⁎⁎	0.02⁎⁎⁎	0.42⁎⁎⁎	0.76⁎⁎⁎	1.04⁎⁎⁎	9.91⁎⁎⁎	14.3⁎⁎⁎	27.4⁎⁎⁎	86.7⁎⁎⁎	67.9⁎⁎⁎	60.2⁎⁎⁎	146.38⁎⁎⁎	937.3⁎⁎⁎	3561.4⁎⁎⁎	2.24⁎⁎⁎	4367.02⁎⁎⁎
Error	38	0.004	0.006	0.01	0.02	0.004	0.02	0.38	0.27	0.39	0.27	0.30	0.22	0.43	0.29	0.36	4.57	1.15
CV %		1.9	3.3	2.0	2.5	1.0	1.9	1.9	0.8	0.9	0.6	0.6	0.5	1.3	0.4	0.3	0.3	0.3

Variables measured 6 months after initiating the watering regime: height = stem height, Diameter = Basal diameter, Leaf number = all of leaves present during the counting period. di1 = seedling diameter month one, di2 = seedling diameter month two, di3 = seedling diameter month three, di4 = seedling diameter month four, di5 = seedling diameter month five, di6 = seedling diameter month six, hm1 = seedling height month one, hm2 = seedling height month two, hm3 = seedling height month three, hm4 = seedling height month four, hm5 = seedling height month five, hm6 = seedling height month six, le1 = all the number of leaves present during month one, le1 = all the number of leaves present during month one, le2 = all the number of leaves present during two, le3 = all the number of leaves present during month three, le4 = all the number of leaves present during month four, le5 = all the number of leaves present during month five, le6 = all the number of leaves present during month six,: SV stands for source of variation, df = degree of freedom, Source of variation Prv  = provenance, trt = treatment, Rep = replication, Prv x trt  = provenance by treatment interaction, Error = residual error.

Variables significantly different at 0.001 level of significance.

Correlation among growth traits

The age to age correlation values with all possible combinations of variable height, diameter and leaf number growth from the first month to the sixth month of F. albida seedling were summarized in (Table 2). The correlations between seedling diameter at month six (d6 and either d1, d2, d4 or d5) revealed increasing significant correlation () with diameter month six recording the highest correlation coefficient . Seedling height month six (hm6 ) recorded the highest correlation coefficient among all the growth variables analyzed. The correlations between leaf number with seedling diameter and height was weak as it ranged from to 0.73.

Table 2

Age–age correlations between seedling traits of F. albida provenances for six months of seedling growth under the watering regimes: (d1–d6) stands for diameter from month one to month six; hm1–hm6 stands for height from month one to month six; le1–le6 stands for leaf numbers from month one to month six.

Char	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
di1	–
di2	0.68⁎⁎	–
di3	0.79⁎⁎	0.77⁎⁎	–
di4	0.80⁎⁎	0.67⁎⁎	0.91⁎⁎	–
di5	0.54⁎⁎	0.63⁎⁎	0.81⁎⁎	0.79⁎⁎	–
di6	0.56⁎⁎	0.70⁎⁎	0.90⁎⁎	0.84⁎⁎	0.92⁎⁎	–
hm1	0.37⁎⁎	0.20^ns	0.18^ns	0.16^ns	0.14^ns	0.02	–
hm2	0.78⁎⁎	0.69⁎⁎	0.82⁎⁎	0.79⁎⁎	0.67⁎⁎	0.69⁎⁎	0.46⁎⁎	–
hm3	0.81⁎⁎	0.67⁎⁎	0.85⁎⁎	0.90⁎⁎	0.81⁎⁎	0.79⁎⁎	0.37⁎⁎	0.89⁎⁎	–
hm4	0.57⁎⁎	0.57⁎⁎	0.74⁎⁎	0.78⁎⁎	0.79⁎⁎	0.81⁎⁎	0.15^ns	0.52⁎⁎	0.77⁎⁎	–
hm5	0.49⁎⁎	0.62⁎⁎	0.75⁎⁎	0.78⁎⁎	0.78⁎⁎	0.82⁎⁎	0.08^ns	0.60⁎⁎	0.80⁎⁎	0.92⁎⁎	–
hm6	0.44⁎⁎	0.60⁎⁎	0.73⁎⁎	0.70⁎⁎	0.78⁎⁎	0.83⁎⁎	0.09^ns	0.65⁎⁎	0.81⁎⁎	0.83⁎⁎	0.94⁎⁎	–
le1	0.16^ns	−0.17^ns	−0.14^ns	−0.07^ns	−0.19^ns	−0.25^ns	−0.06^ns	−0.31⁎	−0.14^ns	0.01^ns	−0.14^ns	−0.31⁎	–
le2	0.66⁎⁎	0.27⁎	0.26⁎	0.41⁎⁎	0.13^ns	0.07^ns	0.31⁎	0.45⁎⁎	0.39⁎⁎	0.14^ns	0.02^ns	−0.05^ns	0.23^ns	–
le3	0.36⁎⁎	0.19^ns	0.03^ns	0.20^ns	0.12^ns	0.03^ns	0.10^ns	−0.03^ns	0.14^ns	0.31⁎	0.11^ns	−0.08^ns	0.59^ns	0.55^ns	–
le4	0.15^ns	0.08^ns	0.10^ns	0.15^ns	0.17^ns	0.07^ns	0.21^ns	0.10^ns	0.22^ns	0.17^ns	0.19^ns	0.13^ns	0.20⁎⁎	0.13⁎⁎	0.17^ns	–
le6	0.20^ns	0.37⁎⁎	0.51⁎⁎	0.53⁎⁎	0.66⁎⁎	0.73⁎⁎	−0.25⁎	0.27⁎	0.45⁎⁎	0.73⁎⁎	0.71⁎⁎	0.65⁎⁎	−0.07^ns	−0.04^ns	0.21^ns	0.09^ns	–

Variables measured 6 months after initiating the watering regime: height = stem height, Diameter = Basal diameter, Leaf number = all of leaves present during the counting period. di1 = seedling diameter month one, di2 = seedling diameter month two, di3 = seedling diameter month three, di4 = seedling diameter month four, di5 = seedling diameter month five, di6 = seedling diameter month six, hm1 = seedling height month one, hm2 = seedling height month two, hm3 = seedling height month three, hm4 = seedling height month four, hm5 = seedling height month five, hm6 = seedling height month six, le1 = all the number of leaves present during month one, le1 = all the number of leaves present during month one, le2 = all the number of leaves present during two, le3 = all the number of leaves present during month three, le4 = all the number of leaves present during month four, le5 = all the number of leaves present during month five, le6 = all the number of leaves present during month six.

significance at 5% probability level.

significance at 1% probability level.

Principal component analysis

Principal component analysis (PCA) of the present study of height, diameter and leaf numbers showed PC1 contributed to an appreciable variance (63.18%) followed by PC2 (22.49%) with PC3 (13.05%) Table 3. The difference of eigenvalue between PC 1 (11.37) and PC 3 (2.35) was high. Contribution weight of each variable to the Principal components (PC) based on correlations. The percentage of the total variance explained by PC1, PC2 and PC3 is also provided. PCA results showing the factor coordinates of the variables (×) and provenances () on the plane defined by the three Principal Components. A varimax rotation of the axes was performed. The ellipses denote the three different groups than can be defined according to the position of provenances on the plane (Fig. 2).

Table 3

The inter- and intra-cluster distances of F. albida provenances under study.

Provenance	1	2	3	4	5
Taveta	1
Lupaso	62.19	1
Kuiseb	62.19	47.23	1
Manapools	84.55	84.55	84.55	1
Wangingombe	71.23	71.23	71.23	84.55	1

Fig. 2

Principal component (PC) ordination of height, diameter and leaf number growth of seedlings of F. albida from eastern and southern Africa. The–axis of ordination plot shows PC1 (principal component 1) with eigenvalue 11.37% and 63.18% of total variance of seedling growth–axis of ordination plot shows PC 2 (principal component 2) with eigenvalue 4.05% and 22.49% of total variance of seedling growth. Manapools, Wangingombe, Lupaso, Kuiseb and Taveta are the provenances under study.

On the basis of hierarchical cluster analysis, 5 provenances were grouped in to three clusters (Fig. 3). The maximum number of 3 provenances (Taveta, Kuiseb and Lupaso) was included in cluster 3 while cluster 1 and 2 composed of one provenance each (Manapools and Wangingombe respectively). The cluster pattern proved that geographical diversity need not necessarily be related to genetic diversity.

Fig. 3

Cluster analysis of F. albida seedling based on seedling height, diameter and leaf number among eastern and southern Africa provenances.

Manapools recorded the highest distance genetic distance from Taveta, Kuiseb and Lupaso and a value of 84.55 (Table 4). Wangingombe and Manapools are closely related genetically at a distance of 7.32 (Table 4). The intra-cluster distances ranged from 47.23 to 84.55 with maximum value in cluster 3 followed by cluster 2 and cluster 1. Minimum intra-cluster distance was found in cluster 1. The highest inter-cluster distance was found between cluster 1 and 3(37.32) followed by cluster 1 and 2 (24). The minimum inter-cluster distance was observed between cluster 3 and 2 (7.32).

Table 4

Phenotypic variance (), genetic variance (), heritability () for seedling diameter, height leaf number of five provenances of F. albida under different watering regimes (1 = 250 ml twice per week, 2 = 500 ml twice per week, 3 = 750 ml twice per week 4 = 1000 ml twice per week) for six months of measurement after two months of seedling establishment.

Trait		1	2	3	4			1	2	3	4			1	2	3	4
σ^2p	di1	0.59	0.21	0.16	0.23	σ^2g	di1	0.58	0.20	0.15	0.23	h^2	di1	75.44	77.75	81.86	92.66
	di2	0.09	0.03	0.08	0.01		di2	0.09	0.03	0.08	0.01		di2	73.71	77.81	82.97	94.4
	di3	0.87	0.98	0.26	1.01		di3	0.87	0.98	0.26	1.01		di3	72.42	78.64	82.28	95.2
	di4	0.85	1.24	1.08	0.45		di4	0.85	1.24	1.08	0.45		di4	68.59	78.42	81.73	93.2
	di5	0.4	1.39	1.03	0.98		di5	0.41	1.39	1.03	0.91		di5	70.65	78.76	83.04	89.4
	di6	0.38	2.32	2.72	1.21		di6	0.26	1.82	2.24	1.10		di6	68.40	78.53	82.55	91.0
	hm1	0.65	13.1	2.72	14.8		hm1	0.49	10.5	2.29	14.47		hm1	75.38	80.14	84.19	97.1
	hm2	103	48.3	30.8	40.5		hm2	72.1	38.6	25.5	38.19		hm2	69.48	79.92	82.71	94.2
	hm3	95.0	85.0	42.10	44.3		hm3	66.0	69.2	35.9	42.66		hm3	69.46	81.45	85.45	96.1
	hm4	41.9	53.5	59.50	38.1		hm4	28.9	43.0	50.5	36.03		hm4	69.07	80.33	84.93	94.5
	hm5	31.9	35.6	83.02	23.0		hm5	25.4	28.5	71.5	21.36		hm5	79.67	80.17	86.12	92.6
	hm6	50.5	57.2	92.97	59.9		hm6	37.1	47.1	78.7	55.78		hm6	73.47	82.31	84.68	93.0
	le1	64.9	161.0	24.03	85.0		le1	51.5	135	20.9	82.35		le1	79.28	83.77	87.16	96.8
	le2	659	267	80.27	659		le2	518	225	71.6	627.46		le2	78.69	84.15	89.29	95.1
	le3	1693	332	446.16	3990		le3	1353	280	391.0	3736.6		le3	79.97	84.38	87.66	93.6
	le4	578.0	4980	2201.5	927.0		le4	462.0	4164	1924	876.01		le4	79.85	83.61	87.43	94.4
	le5	452.0	4256	3807.5	2593		le5	360.0	3625	3364	2465.9		le5	79.62	85.19	88.37	95.0
	le6	456.0	1890	3951.4	3237		le6	359.0	1601	3427	3007.5		le6	78.78	84.72	86.74	92

Variables measured 6 months after initiating the watering regime: height = stem height, Diameter = Basal diameter, Leaf number = all of leaves present during the counting period. di1 = seedling diameter month one, di2 = seedling diameter month two, di3 = seedling diameter month three, di4 = seedling diameter month four, di5 = seedling diameter month five, di6 = seedling diameter month six, hm1 = seedling height month one, hm2 = seedling height month two, hm3 = seedling height month three, hm4 = seedling height month four, hm5 = seedling height month five, hm6 = seedling height month six, le1 = all the number of leaves present during month one, le1 = all the number of leaves present during month one, le2 = all the number of leaves present during two, le3 = all the number of leaves present during month three, le4 = all the number of leaves present during month four, le5 = all the number of leaves present during month five, le6 = all the number of leaves present during month six.

Discussion

Large differences in seedling height, diameter and leaf numbers of F. albida provenances under different watering regimes was detected in the present study (). This indicates that there is adequate genetic variability for seedling growth in the present material which can be utilized in tree improvement programs to select genetically productive provenances for domestication. As the experiment was conducted under similar environment, the variation in seedling growth among the seedlings was due to the genotype which was evidenced from the provenance by treatment interaction. The growth variation was greater for seedling height than that of diameter and leaf numbers () thus height was identified as the best trait for predicting provenance growth in the greenhouse. Weber et al. (2015) recommended the selection of native species based on seedling height is more effective which supports the result of the current study. These results also agree with the finding of provenance trials of other native hardwoods e.g.  Ky-Dembele et al. (2014), Loha et al. (2006) and Weber et al. (2015). The results confirm the findings of Sniezko and Stewart (1989) and Ibrahim et al. (1997).

Water supply impacted on most of the studied traits suggesting that water deficit reduces carbon gain and slows down ontogenetic development Faidherbia albida, like in many other species. Growth of Faidherbia albida provenances related to rainfall gradients of the seed sources. Provenances from drier zones with lower rainfall had better growth than provenances from regions receiving higher rainfall, this suggest that Faidherbia albida provenances from the drier parts are genetically better adapted compared with populations from the more humid parts of the region due to differences in selection pressures (Langlet, 1971). If the observed clines reflect adaptive variation it would be necessary for tree improvement and conservation programs to collect Faidherbia albida provenances from parts receiving low rainfall for domestication in arid and semi-arid areas. Kuiseb from southern Africa naturally occur in region with dry climate receiving average annual rainfall and thus have a higher drought tolerance and was able to record best growth performance across all the watering regimes. Lupaso, Manapools, Taveta and Wangingombe are naturally distributed in region which receives relatively high rainfall explain its better growth when supplied with more water ( twice per week). This observation is explained by Passioura (1982) who hypothesized that plant species adapted to arid lands employ either a conservative or prodigal water use strategy. In its natural habitat, Kuiseb provenance utilizes a conservative water use strategy while in the experimental condition under study; it employed a prodigal water use strategy hence better growth and more biomass accumulation.

The correlations among seedling height stem diameter and leaf numbers increased from age 3 to 9 months after sowing the seed. The increasing trends of the age–age correlations signify the utilization of any of these traits for selection of provenances for further testing. Nevertheless, involving the three seedling characters, seedling height revealed more positive correlations than that of stem diameter and leaf numbers. This implies that seedling height is a superior quantifier of selection than stem diameter and leaf numbers. The strong correlations between hm5 and hm6 (0.94); di5 and di6 (0.92) recommended that selection of clones 9 month after sowing was reasonable for this species.

The heritability values were higher for most traits under study. Almqvist et al. (2001), Van Wijk et al. (2005) and Westendorp et al. (1997) reported higher heritability value for the growth traits in the climatic chamber experiment reflecting enhanced genetic expression under controlled environmental. These results agree with the findings of Karlsson et al. (2002) and O’Connor et al. (1994). There is however need to test the provenance under the field conditions before making conclusion based on this result.

PCA not only provided a visual demonstration of the relationship but also indicates which characters were the most vital in defining that relationship. In the present study, rainfall of the seed source, height growth, and collar diameter growth were the most important characters measured in PCA which defined drought as an adaptive trait in identification of F. albida seedlings originating from eastern and southern Africa. This result was also supported by cluster analysis. Dendrogram illustrated clusters of seedlings origins with greater similarities of seedling growth. The seedlings from Taveta, Kuiseb and Lupaso were the most distinct from the rest of the seedlings origins because they possessed higher height, stem diameter and leaf numbers. Since the environmental variation was insignificant, the observed growth performances may be due to genetic this observation needs verification by using genetic marker analysis of growth traits. The highest inter-cluster distance was found between clusters i and iii followed by cluster i and ii. The minimum inter-cluster distance was observed between cluster iii and ii indicating wider genetic diversity between the trees in these clusters and selection of parents from such cluster for breeding programs would help to achieve novel breeds. The minimum inter-cluster distance indicated that trees in these clusters indicated close genetic relationship among provenances therefore; selection of provenances from these two clusters is to be avoided.

This study provide useful technical recommendations for selection, tree improvement and conservation programs, however, sustainable management of tree genetic resources requires an enabling social and political environment for it success. This is limited because rural communities are not part of the policies addressing land ownership and uses hence unsustainable conservation and management of tree genetic resources (Yatich et al., 2008). The situation is worsened by the conflicts that exist between related to use and management of natural resources. The government, non-governmental institution, rural communities should therefore address these challenges and create an enabling environment to ensure sustainable use, management and conservation of Faidherbia albida genetic resources.