Community ecology of small mammals in the selous ecosystem, Tanzania

Thumbnail Image



Journal Title

Journal ISSN

Volume Title


Sokoine University of Agriculture


Historically, the protected areas in East Africa and possibly the rest of Africa were for conservation of large mammals. This inheritance approach has led to inadequate attention to other components of the ecosystems including small mammals which play a fundamental role in maintaining ecosystem health. This has resulted in inadequate information on their population dynamics in the tropics and miombo areas. Considering the current anthropogenic development trajectories, various studies have highlighted that, small mammals and other wildlife species are faced with a number of threats which include those related to management practices, economic development, and climate change. For example, as part of a wildlife habitat management tool, prescribed burning is practiced as a natural ecological phenomenon in miombo woodland that highly influences vegetation structure and composition with the aim of providing nourishing pasture. This practice has been reported to affect small mammal distribution and abundance. On the other hand, various studies have reported that climate change is posing an ever-increasing threat to wildlife management around the world making it a primary concern and driver of change. Climate change is also linked with increased zoonotic prevalence, especially with increased contact between humans and small mammals such as plague, hantaviruses pulmonary syndrome, and leptospirosis. Therefore, the current study aimed at investigating the community ecology of small mammals in the Selous ecosystem, Tanzania, as a basis for effective and holistic ecosystem management. The studies to assess the community ecology of small mammals in the Selous ecosystem, Tanzania, were carried out for two consecutive years from July 2018 to June 2020. The ecosystem constitutes a globally important example of vegetation types that is between Somali-Masai and Zambezian regional centers of endemism. The area possesses diverse flora with an estimated total of over 2 000 species with an exceptionally high variety of habitats that are ideal for small mammals. Specifically, the studies were conducted in the northern part of the Selous ecosystem which covers parts of Nyerere National Park (NNP) and Selous Game Reserve (SGR). The study aimed at addressing four key specific objectives; (i) establishing the diversity and distribution of small mammals (ii) determining the effects of prescribed burning on rodents‟ ecology (iii) estimating the relative home ranges and species – habitat association of small mammal species and, (iv) assessing the current and potential future distribution of small mammals in the Selous ecosystem. To address the specific objectives, various approaches were used; (i) for abundance, diversity, distribution, and home ranges estimations, four main habitats namely closed woodland, forest, seasonal riverine forest, and perennial riverine forest/thicket were selected; (ii) For effects of prescribed burning on rodents‟ ecology, prescribed burning was set twice during this study, on August of 2018 and 2019 and two grids in each area (burnt and unburnt) were used. Information of interest from this study include abundance, richness, diversity, age structure, sex ratio, breeding patterns, recruitment, and survival proportions; (iii) For current and potential future distribution of small mammals in the ecosystem, a total of 143 sites across nine selected habitats were used in different areas of the Selous ecosystem. In all approaches, Capture Mark Recapture (CMR) using medium-sized (LFA, 7.5 x 9 x 23 cm) Sherman‟s traps (H.B. Sherman Inc.) and random placement of Havahart traps (60 x 15 x 170 cm) were used. Trapping for estimating the current and potential future distribution of small mammals, two trapping nights were used from December 2018 to November 2019. In each selected site, a grid of 70 m x 70 m was established with seven parallel lines at 10 m intervals from each line and between traps, and in total, each grid had 49 stations. Species abundance was estimated using Minimum Number Alive (MNA) for the most captured species from the CMR data history. Shapiro_test function was used to test for normality of data in R software version 3.4. Kruskal Wallis, Wilcoxon, and Friedman's Chi-square tests were used to detect variation between and within habitats and seasons. Also, a non-normal generalized linear model with Poisson error distribution was performed by utilizing the “glm” function with link = log argument to compare mean differences of abundance between areas across seasons. In addition, robust regression by utilizing “rlm” function from MASS and “sfsmisc” packages in R software version 3.4 was used to assess the trend of abundance on monthly basis. To establish the small mammal species – habitat associations, a principle component was used. Ordination technique - Non-metrics Multi-dimensional Scaling (NMDS) using the Sorensen distance measure in vegan package using metaMDS function in R version 3.4. Age groups were compared using Wilcoxon.test for M. natalensis between burnt and unburnt areas. A generalized linear model with non-normal Poisson error distribution was used to compare mean differences between age groups between burnt and unburnt areas. Recruitment and apparent survival were computed from the CMR data history. On all occasions, the assumption was that there was an ability to detect a new and old individual through the marks applied. A generalized linear model (GLM) with logistic regression was used to compare recruitment and survival proportions data by using a built-in function called “family = binomial” in R software version 3.4 to explore the relationship between burnt and unburnt areas. For monthly trend variations, a robust regression model using “rlm” function from MASS and “sfsmisc” packages was used to assess monthly trends of capture proportions for recruitment and survival. Sex ratio and breeding patterns were compared using Wilcox.test between areas. A generalized linear model with binomial error distribution was used to assess the trends whereby proportions were treated as the response variables, while season and areas (burnt and unburnt) were used as explanatory variables. Relative home ranges were estimated from small mammals‟ relocation points from CMR history data. Relative home ranges were estimated using a Minimum Convex Polygon (MCP) using “adehabitatHR” package in R software version 3.4 using “mcp.area” function at 100% of all points. Home range overlaps were estimated using the “adehabitatHR” package by utilizing the “Kernel_overlap” function. Kruskal Wallis.test, one and two-way ANOVA were used to detect variation in home range across sex, habitat, and season. To assess the current and future distribution of small mammals in the Selous ecosystem, species occurrence (presence data) data and habitat parameters including elevation and location points were recorded from the field to enable mapping of current and projected suitable habitats. Additional data were obtained from the WorldClim dataset for bioclimatic data downloaded in raster (ASCII) format. Additional topographical data were obtained from Digital Elevation Models (DEM). Maxent algorithms were used to estimate the current and future distribution of small mammals in the Selous ecosystem. Models performance was evaluated using Area Under the Receiver Operating Characteristics Curve (AUC). QGIS v.3.10 was used to generate the distribution maps. The suitable habitat categories ranged between 0 and 1, with five classes values; <0.1 unsuitable habitat, 0.1 – 0.2 least suitable, 0.2 – 04 moderately suitable, 0.4 – 0.6 suitable habitat and > 0.6 highly suitable habitat. The results of this study showed that, abundance of Acomys ngurui was statistically significantly different between the four main habitats ( 2 = 12, df = 3, p = 0.007) and across seasons ( 2 = 6, df = 2, p = 0.049). Overall, forest habitat recorded the highest species diversity (H‟ = 2.065) and the lowest (H‟ = 1.506) diversity was recorded in perennial riverine forest/thicket. Dry seasons recorded the highest species diversity (H‟ = 1.65) and wet seasons the least (H’ = 1.445). On the effects of prescribed burning on rodents ecology, there were statistically significant different (W = 0.892, p = 0.0004) in abundance of A. ngurui between the burnt and unburnt areas and through seasons (Deviance = 2.644, df = 1, p = 0.021). Yet, for Mastomys natalensis, although its abundance increased after the prescribed burning events, the differences between the burnt and unburnt areas were not statistically significant (W = 344.5, p = 0.226), but significant difference between seasons (Deviance = 3.606, df = 1, p = 0.05). Species richness was higher in burnt areas than in unburnt areas suggesting burnt areas promote more species colonozitation compared to unburnt. Seasonality showed effects on richness with more species (5) appearing in the wet season in burnt areas than in unburnt areas. On the other hand, the highest species diversity (H‟ = 1.551) was recorded in burnt areas than in unburnt (H‟ = 0.759). On seasons, higher diversity in both wet and dry seasons was recorded in the burnt area (H‟ = 0.679 and 0.556 respectively) than in unburnt (H‟= 0.419 and 0.382 respectively) suggesting prescribed burning promotes higher diversity. On age groups, no statistically significant differences (Deviance = 3.507, df = 1 p = 0.061) between burnt and unburnt areas and across seasons (Deviance = 0.301, df = 1, p = 0.583) were recorded. On the other hand, the probability for both A. ngurui and M. natalensis survival (E = -0.481±11.233, z = -0.428, p = 0.669 and E = 0.377±7.044, z = 0.054, p = 0.957 respectively) and recruitment (E = 7.316±13.876, z = 0.528, p = 0.598 and E = 1.023±11.598, z = 0.088, p = 0.93 respectively) were not statistically significant different between burnt and unburnt areas. This was the same for breeding patterns and sex ratio suggesting that prescribed burning does not promote breeding and sex ratio variations in the Selous ecosystem. Overall, the results showed that prescribed burning as a management tool has effects on the abundance, richness, and diversity of rodents and not on specific demographic characteristics. On the home range, A. ngurui recorded the largest mean home range (1 087.58 m 2 ) than L. rosalia (831.55 m 2 ) and M. natalensis (166 m 2 ). Home range overlap was not statistically significant different between the two habitats but it was statistically significant difference across seasons in both habitats ( 2 = 33.5, df = 1, p = 0.000 and  2 = 46, df = 1, p = 0.000 respectively). Overall, most small mammals were associated with seasonal riverine forest which is characterized by high canopy and shrub cover while most murid species were associated with closed woodland with leaf litter and dead logs. The results of the current and future distribution of small mammals in the Selous ecosystem indicate that most of the current highly suitable habitats will be affected and small mammal species will be concentrated in a few areas within and outside the ecosystem. This study has provided important information for updating the current small mammal distribution maps as most of the small mammal species are reported for the first time from this area with molecular confirmation. Further, Grammomys selousi is being reported for the first time in the northern part of Rufiji River and South of Ruaha River. Considering the finding from this study, small mammal species abundance, diversity, and distribution are largely influenced by habitat types and seasonal variations at the community level and at individual species such as A. ngurui in the Selous ecosystem. This study has demonstrated that, overall, prescribed burning as a management tool favors the high abundance, richness, and diversity of small mammals. While for individual species, prescribed burning has shown varying effects both positive and negative. Acomys ngurui and M. natalesis have shown variations in the effect of prescribed burning on various demographic characteristics, suggesting that, although they share the same habitat, prescribed burning might affect them differently. The study has also shown a varied effect of habitats and seasons on the abundance of individual species suggesting that they are not uniformly affected. In relatively stable habitats with infrequent large herds of herbivores and prescribed burning, abundance has shown to be higher for A. ngurui while for L. rosalia and M. natalensis their abundance was higher in areas frequented by prescribed burning and large herbivores. In addition, this study indicated that large home ranges were recorded in the seasonal riverine forest for A. ngurui and L. rosalia which was relatively stable with infrequent disturbances while for M. natalensis it was different. Small mammal species have shown a high overlap within closed woodland suggesting shared resources with possible separation of time, i. e. crepuscular (L. rosalia) and nocturnal (A. ngurui and M. natalensis) and feeding behaviors with others being generalist such as M. natalensis and omnivorous (A. ngurui). Furthermore, most small mammals were associated with seasonal riverine forest than closed woodland suggesting that areas with limited activities of large mammals and management practices support more species while areas with varied degree of disturbances necessitate an adaptation to be able to survive. The study has further established the expected climate change effects, especially from the „business as usual scenario‟ (RCP 8.5) which is expected to shift the small mammal distribution in the Selous ecosystem both positive and negative based on specific species. The model suggests that most of the current highly suitable range will be affected and species will be forced to concentrate in a few areas within the ecosystem. It is important however to consider that, although the model predicted the observed distribution, there is a chance that this will depend on the species' capacity in adapting to the changing environment and the time expected for these changes to occur. All the species studied suggest being able to adapt to a dynamic habitat and can tolerate even the disturbed area. To address the gaps identified in this study, the following recommendations are made to the managers and policymakers of the protected areas; The information from this study call for an update of various management plans such as General Management Plans (GMP), Prescribed Burning Plan, Infrastructure Development/Management Plans and others to include the distribution of small mammals in the parks/reserves within the Selous ecosystem. Also, conservation managers are advised to maintain the current cycle of prescribed burning since any alteration might affect rodent species' population dynamics. Although no statistically significant variation in most demographic characteristics was vivid, at assemblage levels such as abundance, richness, and diversity, the effects were obvious. Furthermore, areas with low protection status such as Kilombero Game Controlled Area (KGCA) should be upgraded to Game Reserve/National Park standard to protect the important areas for species including small mammals' future strongholds. In addition, the management of different protected areas in the ecosystem are advised to include climate change as a key component in the development of General Management Plans (GMP). This may include a specific chapter to address climate change impacts including intensified patrol to reduce/avoid encroachment of any form and extension services related to the conservation of natural resources in the surrounding villages. Considering the importance of small mammals to the ecosystems, additional studies on their population dynamics are recommended to cover a wider part of the ecosystem including threatened, endangered, and/or rare/endemic species such as Grammomys selouis.




Ecology community, Small mammals, Ecosystems components, Wildlife habitat management tool, Selous ecosystem, Tanzania