3.2. Significance of cultivation status, genetic group and location on growth response traits
In the linear mixed model analysis, the effects of factors (i.e. , cultivation status, genetic group and location) varied across growth response traits (Table 3). The cultivation status did not have significant effects on RGRA, TNL and TLDW (p >0.05), but did significantly affect SLA and RL (Table 3). On average, wild genotypes had the highest SLA (244 cm-2 g-1) but the difference was only significant with the feral and not with the cultivated genotypes (Appendix Table A. 3.). For RL, cultivated genotypes had a significantly higher average value (0.0087 cm3cm-2) than wild and feral genotypes and there were no significant differences between wild and feral RL values (Appendix Table A.3.). There were no significant interaction effects between cultivation status and treatment for any of the selected traits, except for TL indicating that only for TL, the treatment effect differed across cultivation status. Under ample-water conditions, cultivation status had no significant effects on TL while under restricted-water conditions wild genotypes, had a significantly lower TLthan feral and cultivated genotypes whose TL’s were not significantly affected by water availability. In the restricted-water treatment, wild genotypes had the lowest average TLwhich was 24.4 % lower than the highest average TLobserved in feral genotypes (Appendix Table A.3.). These findings suggest that in terms of TL, wild genotypes might be more sensitive to low water availability than non-wild genotypes.
Genetic groups significantly differed in their TL, TLDW and RL but not in the other three traits (Table 3). Plants from the genetic group SC had the highest mean TL which was 60.9 % higher than the lowest TL observed in genetic group Kibale (Appendix Table A.3.). The effect of the genetic group on TLDW was similar to TL with genetic group SC having 67.6 % higher mean TLDW than genetic group Kibale which had the lowest TLDW (Appendix Table A.3.). For RL, Zoka had the highest value which was 43.9 % higher than the lowest RL observed in the genetic group Kibale (Appendix Table A.3.). Interaction effects between genetic groups and treatment were only observed in RGRA, implying that the magnitude of the response in this trait to the water treatment differed across genetic groups. The RGRA of genetic groups: Budongo, SC and Zoka were significantly reduced due to restricted-water supply but not that of genetic groups Itwara and Kibale (Fig. 1; Appendix Table A.3.).
Overall, location as a factor had stronger effects on growth response traits to ample and restricted-water supply than the two other factors, cultivation status and genetic groups
(Table 3). Location had significant main effects and interaction effects on all traits except on TNL, TLDW and SLA (Table 3). This implies that the growth response values significantly differed depending on the location from which the genotypes were collected. For example, for TL, i.e. the response trait with the strongest location effects (Table 3), location Malabigambo had the highest average TL which was 73.7 % higher than the lowest TL observed in Kibale. Drought had no significant effects on the TL of genotypes collected from Zoka, Itwara, Kibale, Kituza and Kawanda, while it significantly reduced TL of genotypes collected from Malabigambo, Kalangala and Mabira (Appendix Table A.4.). In absolute terms, under ample-water conditions, Malabigambo had a significantly higher TL (7263 (± 153) cm2) than all other locations while Kibale’s TL (1413 (± 38) cm2), was significantly lower than TL’s at all locations except Zoka (Appendix Table A.4.). Similarly, under restricted-water conditions; Malabigambo had the highest TL (3711 (± 62) cm2) compared to all other locations whereas Kibale had the lowest TL(1469 (± 42) cm2) which was significantly lower than TL of all other locations except Zoka (Appendix Table A. 4.).