| dc.description.abstract | The understanding of factors driving plant invasions in African Savannas which is pivotal
for their management is currently poor. Therefore, this study tested several hypotheses that link human disturbance to species invasions. Firstly, human disturbance facilitates the invasion of introduced plant species into savanna ecosystem. Secondly, species in human disturbed environments, should exhibit traits that are associated with quick regeneration giving advantage to invasive species. The third hypothesis tested was that, human factors when used in conjunction with natural factors should produce habitat prediction models with high accuracy for invasive species. The data for testing the first two hypotheses were collected from 227 sampling plots in three sites in Zimbabwe; Kyle Game reserve and the adjacent Mutirikwi communal lands, Tuli and Mazunga safari areas and Bubiana conservancy and the surrounding Shake and Gwanda communal lands. Vegetation data, including species richness, abundance and frequency of occurrence of introduced plant species were collected. We tested whether plant species richness differed significantly between the human dominated landscape and the adjacent protected areas. In addition, the Z-test was run to determine whether the proportion of introduced exotic plant species was significantly higher in the human-disturbed landscape. The X 2 test was performed to check whether introduced exotic species were significantly associated with human disturbance. To test the third hypothesis, predicting the preferred location on invasive Xanthium strumarium,
105 Presence and absence plots were selected from Gonarezhou National Park and the
adjacent Malipati communal lands. Ten modeling techniques were employed within the
BIOMOD2 environment; thereafter building an ensemble model. For the first two
hypotheses, our results indicate that the proportion of introduced exotic plant species was significantly higher in the communal land compared to the adjacent protected site (Z ≥ 2.89, P < 0.05). Introduced exotic plant species were dominated by the ruderal strategy (annual lifespan and forb growth form). They were also significantly associated with human disturbance at two of the three sites assessed (X 2 ≥ 3.4, P < 0.05). Results also indicate that combining human and natural factors as predictors provided good models for predicting the suitable habitat of X. strumarium Machine learning based techniques performed well (ROC 0.8) at predicting the occurrence of X. Strumarium, although the ensemble prediction outperformed all the individual models (ROC 0.9). X. strumarium was predicted to occur in areas that receive direct radiation from the sun, thus, aspect was identified as a critical factor in habitat selection for X. strumarium. Conclusions drawn from this multi-site study suggest that human disturbance may favor the success of introduced species. | en_ZW |
