Twelve Acacia senegal provenances were collected from two major ecological zones (Sudan and Sahel) with strong evidence of the species in Yobe state, Nigeria in 2006 (Table 1). Collection was specifically made at Yusufari, Jajimaji Bulkati and Usur (Sahelian ecology) and Damaturu, Ngaburwa, Fune and Gujba (Sudan ecology). The selection criteria were based on the provenance trial recommendations of Burley and Wood [7] , where provenance nomenclature was standardized using locational details with a minimum interval of 100 meters apart to avoid inclusion of likely half-sibs in the collection. Healthiest pod in the best stage of maturity was selected from 25 trees in each provenance because of heterogeneous nature of the population.

The provenance trial was established at the experimental field of the Rubber Research Institute of Nigeria, Gum Arabic Sub-Station, Gashua, located at latitude 12˚46'N, longitude 11˚00'E and altitude 360 m above sea level. Seedlings of the twelve (12) Acacia senegal provenances were nurtured in the nursery for four months before transplanting to the field in August, 2008 in a Randomized Complete Block Design (RBCD) with three replications. Each provenance consisted of a square plot with sixteen trees (4 × 4) row spaced at 3 m × 3 m [8] .

Each provenance was exploited for gum yield at four years after establishment using tapping knife to make an incision of 3 cm × 15 cm [9] . Number of incision was determined by counting the number of tapping cuts. Bark thickness (cm) and stem diameter (cm) were determined by using Venier caliper. Gum yield (kg/ha) was determined by harvesting the gum and shade drying to 15% moisture content before weighing on a digital weighing balance (Hecky High Precision). Analysis of morphological traits and gum yield were carried out for three years (2011 to 2013). Each year number of tree incisions, bark thickness and stem diameter were taken at the commencement of tapping while number of gum bearing incisions and gum yield were taken monthly for three month.

Data collected were subjected to analysis of variance; F-test was used to test for significant differences among the measured traits [10] . Provenance means were compared using Duncan’s Multiple Range Test [11] . The data were analyzed using the general linear model procedure of statistics analysis system software version 9.0 [12] . The mean squares were used to estimate phenotypic variance () and genotypic variances () [13] . The methods of Burton and Devane were used in calculating the genotypic and phenotypic coefficient of variations [14] , as: GCV(%) = × 100/grand mean and PCV (%) = × 100/grand mean. () = genetic variance () + environment variance (σ²e) by Acquaah (2007). The recommendation of Nechif and co-workers were followed in the determination of Relative difference (RD %) between PCV (%) and GCV (%) = [100 (PCV ?

GCV)/PCV [15] . Broad sense heritability and genetic advance were

calculated by using the methods adopted by Acquaah and Falconer & Mackay respectively [16] [17] .

Provenances 11 and 12 though statistically similar, had superior growth traits compared to the rest of the provenances most times except gum yield averaged over 3 years tapping season as presented in Table 2. The highest gum yield was achieved with P9 while the lowest was recorded by P6 and P3. Provenance three (P3) often resulted in significantly lower morphological and yield traits. The superior growth put up by provenances of Gujba origin is not surprising as variability in growth among provenance population of Acacia senegal has been very well documented [18] . Also, the higher gum yield recorded with P9 is in order because Fune vegetation type is reminiscent of sahelian agro-ecological zone. The specie has been reported to put strong growth and yield performance under hash environmental conditions [19] .

The analysis of variance result is shown in Table 3. The provenances showed significant differences for all traits measured in the study, indicating the presence of an adequate variability among the provenances for selection and identification of superior genotypes. Year effects were also significant for all traits except bark thickness probably due to yearly differences in weather variables.

As shown earlier in Table 2, the twelve A. senegal provenances showed wide range of variability in most of the traits studied. Tree incisions, number of gum bearing incisions and yield were relatively low. This investigation revealed considerable amount of variations for traits like barks thickness and stem diameter (Table 4). Such wide variations indicate prospects for improvement in these traits. Provenance variations in A. senegal growth traits which include plant height, canopy spread and stem diameter were earlier reported [20] .

The phenotypic variance () was slightly higher than the genotypic variance () for all traits except bark thickness that gave the same estimate. However, phenotypic coefficient of variation (PCV) was higher in proportion than genotypic coefficient of variation (GCV) for all the traits suggesting the presence of environmental influence in the expression of these traits. GCV and PCV values greater than 20% are regarded as high, whereas values less than 10% are considered to be low [21] . Also, values between 10 and 20% are considered to be moderate. High GCV and PCV values were recorded for number of tree incision, stem diameter and gum yield while GCV was moderate and PCV high for number of gum bearing incision. These offer a wide range of

Means followed by the same letter within a column of a treatment group are not significantly different statistically at 5% level of probability using DMRT. TI: tree incision, BT: bark thickness; SD: stem diameter; NGBI: number of gum bearing incision; GY: gum yield.

^*, ^**: Significant at P < 0.05, P < 0.01, respectively (F-test): TI: tree incision, BT: bark thickness; SD: stem diameter; NGBI: number of gum bearing incision; GY: gum yield.

TI: tree incision, BT: bark thickness; SD: stem diameter; NGBI: number of gum bearing incision; GY: gum yield.

opportunities to carry out selection for the traits. High GCV and PCV recorded for number of tree incision, stem diameter and gum yield signifying that these traits are vital for gum Arabic improvement. RD estimates of GCV in relation to respective PCV were determined to judge the extent of genetic variation in these traits. These however, suggest that the observed variations in number of gum bearing incision (63.40%) and gum yield (81.17%) were mostly due to environmental factors.

The estimated heritability ranges from 18.17% for gum yield (kg/ha) to 77.36% for bark thickness. Heritability estimates were earlier classified as low (0 to 30%), moderate (30% to 60%) and high (above 60%). Considering this benchmark, heritability was high for bark thickness and stem diameter indicating a satisfactory possibility of genetic improvement for these traits through selection. High heritability values for bark thickness and stem diameter have been reported [22] . Low estimation of heritability was observed for gum yield(kg/ha) whereas the number of tree incision and number of gum bearing incision showed moderate heritability values which indicated that total variability was due to genetic and environment effects. This result is similar to previous study which observed that heritability estimates for traits of most forest tree species ranged from low to moderate and largely due to dominance, epistatic and environmental effects [23] .

The genetic advances (GA) at 10% selection intensity for 5 traits are presented in Table 2. Estimate of GA for gum yield was 22.31, indicating that whenever 10% selection is made among high yielding genotypes as parents, mean yields of progenies could be improved by 22.31 kg/ha for the first cycle; meaning that the mean genotypic value of the new population for gum yield could be improved from 95.47 to 117.78 kg/ha. Bark thickness and number of gum bearing incision expressed low genetic advance. Thus high heritability coupled with moderate genetic advance observed for stem diameter and number of tree incision indicated that the traits were mainly controlled by additive genes and that direct selection for this trait could be effective whereas moderate or high heritability coupled with low genetic advance expressed by the remaining traits signified that the traits were largely under non additive gene control (dominance and/or epistasis) and environmental.

This study revealed that there was sufficient genetic variability in A. senegal provenances available in the germ- plasm that could be used for improvement in gum yield and other component attributes of the species through direct selection and hybridization.