The identification of non-random species composition patterns predicted by assembly rules is a central theme in community ecology. Based on life history characteristics, species composition patterns of rodents and shrews should be consistent with predictions from nestedness rather than competition hypotheses. This study investigated the seasonal changes in rodent and shrew assemblages in eleven savanna vegetation types in a protected reserve in South Africa. Rodents and shrews were sampled between 2009 and 2010 at Phinda Private Game Reserve (PPGR), KwaZulu-Natal. Sample-based rarefaction curves showed that rodent and shrew abundance and richness varied among seasons and vegetation types. Species richness estimators indicated that inventories for rodents (80%) and shrews (100%) were fairly complete. Null-model analyses found no evidence that species co-occurrence patterns in the reserve were non-random with respect to predictions from Diamond’s Assembly rules, niche limitation hypothesis and nestedness hypothesis. I also investigated seasonal changes in species richness and abundance of rodent and shrew assemblages on cattle, pineapple and former cattle farms surrounding PPGR, and used cluster analyses to compare the species composition of rodents and shrews at farm and PPGR study sites. Small mammal assemblages exhibited a heterogeneous distribution and species composition patterns changed between seasons. Sample-based rarefaction curves showed that rodent and shrew abundance and richness varied among seasons and study sites. Species richness estimators indicated that inventories for the rodents (91%) and shrews (100%) on the farms were essentially complete. Rodent and shrew species composition patterns did not group study sites according to land use, nor could species composition patterns be explained by vegetation characteristics. My results suggest that complex biotic and abiotic processes other than competition, nestedness, land use and vegetation characteristics operate at different spatial and temporal scales to structure rodent and shrew assemblages. / Thesis (M.Sc.)-University of KwaZulu-Natal, Westville, 2013.

In recent years several regions in South Africa including KwaZulu-Natal (KZN) have experienced a significant shift in their farming practice, moving away from conventional commercial farming to the apparently more lucrative private game farming. This is clearly evident in the northern parts of KZN, where most Private Game Reserves (PGRs) occur in semi-arid areas, which are poorly suited to agriculture or livestock farming. These conversions of land cover or change in land use (i.e. from conventional commercial farming to private game farming) is occurring fairly swiftly even though the extent (area of change), impacts and implications of this trend have not been established. Very limited scientific research has been carried out on the environmental effects, impacts and implications of these land cover/use changes related to PGRs. In an attempt to document these changes as well as the environmental implications, three PGRs i.e. Kube Yini, Thanda and Phinda and the Mkhuze Game Reserve within northern KZN were selected for in-depth study. Mkhuze, which is owned and managed by the state, was selected as the control for this study, as it has remained relatively undisturbed for a long period of time. A comparison between the PGRs and Mkhuze was carried out to compare the land cover/use changes within the study area since the establishment of the PGRs. Remote sensing software was used to achieve the aim and objectives of this study. Using satellite images from 1990 and 2007, a change detection analysis was performed to determine the land cover/use changes that occurred within the study area during the period 1990-2007. Sixty-four land cover/use classes were generated from the analysis. These sixty-four classes were reclassified into five broad land cover/use classes which were identified as (1) water and riverine vegetation, (2) sand forest, forest and rock faces, (3) woodlands, (4) old fields and disturbed areas, (5) and grasslands. The results illustrated that during the period 1990 to 2007, the land cover/use of the entire study area changed significantly, with the overall land cover/use changes illustrating a decrease in the water bodies and riverine vegetation (6 percent to 3 percent); sand forest, forest and rock faces (19 percent to 16 percent); woodlands (26 percent to 25 percent); and old fields and disturbed areas (26 percent to 25 percent) whilst an increase in the grasslands (23 percent to 31 percent) was evident. Within each individual PGR, significant land cover/use changes related to private game farming were evident. In Kube Yini PGR, a decrease in the water bodies and riverine vegetation (4 percent to 1 percent); old fields and disturbed areas (23 percent to 11 percent); and grasslands (6 percent to 5 percent) were observed whilst an increase in the sand forest, forest and rock faces (31 percent to 47 percent); and grasslands (6 percent to 5 percent) land covers were seen. There has been no change in the woodlands cover (i.e. remains 36 percent). Old fields and disturbed areas have decreased due to the clearing programme that was implemented by Kube Yini, which has also contributed to the increased sand forest, forest and rock faces; and grassland land covers. At Thanda PGR, water bodies and riverine vegetation remained the same overall, while an increase in the grasslands cover (19 percent to 49 percent) was clearly evident. A decrease in the sand forest, forest and rock faces (13 percent to 4 percent); woodlands (29 percent to 18 percent); and old fields and disturbed areas (38 percent to 28 percent) was also noted. This was due to Thanda‟s Management Plan, whereby management had cleared the old fields and disturbed areas to accommodate the increased tourist accommodation, increased game species as well as the increased space that is required for these species. Furthermore over the recent years, Thanda has been practicing bush clearing and thinning of vegetation, which has resulted in the creation of a grassland landscape. A grassland landscape was required for viewing the game species, which resulted in a decreased sand forest, forest and rock faces; woodlands; and old fields and disturbed areas. In Phinda too, the water bodies and riverine vegetation remained the same (i.e. 1 percent). The sand forest, forest and rock faces (32 percent to 19 percent); and woodlands (43 percent to 41 percent) decreased whilst an increased old fields and disturbed areas (15 percent to 26 percent); and grasslands (9 percent to 13 percent) were observed. Sand forests are sensitive and decrease in extent rather easily as a result of disturbances experienced in the area. The decrease in sand forest, forest and rock faces during this period may be due to the increased wildlife and especially elephant population experienced on this reserve. The woodlands land cover decreased in size as a result of clearing to accommodate tourist infrastructure as well as creating a grassland landscape i.e. open space and clear viewing for the game species. The changes in Phinda‟s land cover resulted can be explained by the different management practices that were practised on the reserve. The results from the PGRs were compared against Mkhuze so as to determine whether private game faming was in fact changing the landscape of the study area. During this period 1990 to 2007, Mkhuze experienced only slight changes within it‟s land cover/use classes: water and riverine vegetation (2 percent to 1 percent); sand forest, forest and rock faces (16 percent to 15 percent); and grasslands (29 percent to 26 percent) decreased whilst an increase in the woodlands (27 percent to 30 percent); and old fields and disturbed areas (26 percent to 28 percent) were noted. Since Mkhuze is managed by the state with its priority being conservation first and not increased monetary gain, competition between species and habitants occurred with the least amount of human influence. It is most likely that the water and riverine vegetation; and grasslands decreased due to the drought experienced in the area between 2001-2006, which may have led to the increased old fields and disturbed areas. In addition woodlands may have increased due to its resistance to harsh conditions. Also contributing to the increased old fields and disturbed areas could be the increased wildlife numbers experienced in the reserve. The slight drop in the sand forest, forest and rock faces land cover most likely resulted due to natural competition between the vegetation species as well as management realising the significance of the sand forest, thus assisting in the preservation of the sand forest, forest and rock faces cover. An accuracy assessment was carried out to determine the accuracy of the image classification, and a 75 % accuracy was achieved for the overall analysis. The results illustrate that during the period 1990-2007, the study area‟s land cover changed significantly and that this was mostly like due to the changes within the land cover of the individual PGRs. Climatic data for this period was also taken into consideration to see whether climate variations had occurred and whether this had impacted the change in land cover/use of the study area. The results from the PGRs were compared against the state reserve results as well as the climatic data. The results suggest that the majority of the land cover/use changes within the study area had occurred largely due to private game farming rather than natural processes. The climatic variables such as temperature of the area remained fairly unchanged whilst humidity remained fairly high. With the exception of the drought experienced in the area between the period 2001-2006, it is most likely that the changes in the area are the result of the change to private game farming. Private game farming is having a significant effect on the landscape of the study area and this is largely due to the different management practices practised on each individual PGR. It is recommended that PGRs be monitored and regulated consistently in order to prevent the degradation of the environment and its ecosystem. The study exhibits the value and usefulness of satellite imagery in detecting land cover/use changes related to private game farming. Landsat 5 TM imagery was the most suitable and accurate in identifying and mapping broad scale land cover/use changes related to private game farming. If a more detailed level of investigation is required, the use of higher resolution imagery (e.g. Spot/Landsat 7) is advised. / Thesis (M.Sc.)-University of KwaZulu-Natal, Westville, 2010.