Abstract
A study aimed at assessing the structure of rodent and shrew assemblages inhabiting a degradation gradient while considering rainfall patterns, was conducted in one of few remaining lowland tropical forests in Eastern Africa. We collected a unique dataset of rodents and shrews, representing 24 species (19 rodents, 5 shrews). The most abundant species alternated in dominance as species abundance significantly fluctuated across the study period following a degradation gradient (F2,33 = 5.68, p = 0.007). While only generalist species were observed near the degraded forest edge, habitat specialists such asDeomys ferrugineus, Malacomys longipes and Scutisorex congicus , were observed in the primary forest interior suggesting a significant (X2 = 1165.329, P<0.001) association between species and their associated habitats and habitat attributes. There was also an observed correlation between rainfall patterns and species abundance. Capturing more species in adjacent fallows and along the degraded forest edge suggests that many species are able to live in degraded habitats that offer a variety of food resources. The continued pressure on forest resources, however, may lead to changes in habitat structure. This, coupled with the dependence of forest ecological functions on rainfall, which is typically not the case, may ultimately cause the local extinction of highly specialized but less adaptable species.
Key words: Habitat degradation, Rainfall patterns, Habitat association, Rodents, Shrews

1.0 INTRODUCTION

Deforestation in Uganda has been severe over the years due to the continued dependence on natural resources for food, energy, medicine and the growing human settlements close to protected areas. Forests in Uganda have been severely degraded and their diversity compromised. Obua et al. (2010) noted an estimated loss of 86% of Uganda’s tropical moist forest especially on private land, while National Forestry Authority reported an 88% and 2 % loss of tropical well stocked forest on private land and protected areas, respectively, between 1990 and 2015 (NFA 2016). This is mainly as a result of the increasing human population that has overtime encroached on forested areas for both settlement and farming (Mulugo et al. 2019).
Few countries of equivalent size have as rich and diverse rodent and shrew fauna as Uganda. Apart from many species of rats and mice there is a great variety of other forms including gerbils, squirrels, flying squirrels, porcupines, cane rats, mole rats, and dormice (Delany 1975). The main factors causing this high diversity are the geographical position at the margin of several major biogeographical areas (Linder et al. 2012) and the wide range of altitudes and the complex array of vegetation which support a characteristic rodent fauna (Delany 1975).
A variety of micro habitats with varying characteristics can be found in tropical forests, and rodents and shrews are two of the most species-rich groups of mammals that live there (Carleton & Musser 2005). Because of their diversity and abundance, they are some of the most significant players in the ecological dynamics of forest ecosystems (Angelici & Luiselli 2005). They are essential to the transport of nutrients and materials through the ecosystem and to the food web (Nicolas et al. 2009). In disturbed ecosystems, rodents play important roles in succession as seed dispersers (Nicolas et al. 2009). In contrast, shrews prey on insects and smaller vertebrates (Nicolas et al. 2009) and the population dynamics of their prey are regulated by their voracious feeding habits. Shrews and rodents are both eaten by larger vertebrates like birds and snakes. They are economically important as crop pests and medically important, with many zoonotic heamoparasites such as borrelia, trypanosomes, bacilli, plasmodia and coccobacilli (Katakweba et al. 2012), and can also be employed in ecological evaluation for conservation decision-making.
One of the elements that shapes and influences small mammal groups in tropical forests is habitat structure (Tews et al. 2004). In a Brazilian rain forest, Pardini et al. (2005) discovered that the forest structure affected both the overall abundance of species and the abundance of particular species on an individual basis. In Africa, where the majority of these natural areas are threatened by anthropogenic activities, there has generally been a dearth of ecological study on rodents and shrews in all habitats (Nicolas et al. 2009, Obua et al. 2010). Even though there is a lot of forest damage in Africa, these activities’ immediate and long-term repercussions are rarely examined (Malcom & Ray 2000). Few studies have specifically looked at how habitat degradation affects small mammal groups over time in Africa (e.g. Struhsaker 1997, Malcolm & Ray 2000). Small mammal communities’ reactions, whether individual or collective, provide evidence of the effects of disruptions. Some of the most significant human-mediated disruptions to forest ecosystems are expanding agriculture, selective harvesting for lumber, and charcoal burning, among others (Malcolm & Ray 2000, Baranga 2007, Obua et al. 2010). Depending on the extent, severity, and kind of the disturbance, as well as the context of the landscape, human-induced habitat alterations have both direct and indirect consequences on small mammal groups (Malcolm & Ray 2000).
A number of surveys of rodents and shrews of Uganda Forest Reserves have been previously conducted. Basuta and Kasenene (1987) reported 14 species from a study in Kibale Forest National Park, Davenport et al. (1996) and Howard et al. (1996) reported on more comprehensive small mammal surveys from Ugandan forests. In the Mabira Central Forest Reserve (CFR), Dickinson & Kityo (1996) conducted earlier surveys of the rodent and shrew population, followed by the recent survey conducted by Waswa et al. (2016) and Ssuuna et al. (2020). These studies assessed small mammals throughout the forest with a variety of objectives; Dickinson & Kityo (1996) focused on the distribution of species, whereas Waswa et al. (2016) examined the composition and organization of rodent communities and Ssuuna et al. (2020) assessed rodent communities in different forest compartments with varying degrees of degradation. However none of these studies has looked at spatial and temporal dynamics of rodent and shrew communities in MCFR along a degradation gradient while considering habitat characteristics and rainfall patterns.
Mabira CFR is uniquely located between metropolitans of Lugazi, Mukono, Jinja, and Kampala city (Fig. 1). As a result, the forest is under a lot of pressure to provide a variety of products, the most important of which are charcoal and timber. The forest is currently being invaded by the invasive paper mulberry, which has taken over all of the forest boundaries as gaps have opened up. Numerous groups of rodents and shrews that depend on forests, particularly highly specialized ones with low degrees of adaptation, may be pushed toward local extinction as the size of natural habitats declines. The situation is made worse by the general lack of studies, species monitoring programs, and/or habitat monitoring programs in the region, which causes biodiversity to decline in the absence of a baseline.
A thorough grasp of rodents and shrew ecological niches is necessary to properly explain their importance to the dynamics of the forest ecosystem. Such data is required to classify specific forest sections’ management regimes and/or forest habitats according to how important they are to the local wildlife. It is hoped that the study highlights the effect of forest habitat deterioration on a micro scale hence stimulate fresh approaches for management of MCFR and other forests in Uganda. Mainly the study focuses on temporal and spatial changes in assemblages of rodents and shrews living along a degradation gradient. The main specific objectives are: (i) to assess the diversity and distribution of rodents and shrews along a habitat degradation gradient; (ii) to evaluate how habitat characteristics affect the occurrence of rodents and shrews along a habitat degradation gradient; (iii) to analyse how rainfall patterns affect the abundance of rodents and shrews along a degradation gradient.

2.0 METHODS

2.1 Study areas

The study was conducted in Mabira (CFR); (Fig. 1) in Uganda Griffin Falls (0°26’14.28”N, 32°57’14.31”E, 1179 m a.s.l.) from August 2018-December 2019. This is the largest forest reserve in Central Uganda (Colwell & Coddington 1994), managed by Uganda National Forestry Authority (NFA) as a Central Forest Reserve (CFR). According to Howard (1991) MCFR is considered secondary regenerating, in which the most dominant vegetation represents sub culmination communities, heavily influenced by man through continued excess of illegal resource use and encroachment. However, some parts are now fully regenerated with tall mature trees especially in the strict nature reserve compartment of the forest (Ministry of water and environment 2010). The MCFR’s vegetation is classified as a semi-deciduous medium altitude forest. Mabira forest continues to exist in central Uganda as an isolated but sizable forest island without any land linkages to the various little pockets of forest in the Lake Victoria basin. Mabira (CFR) has a special position as the last substantial sanctuary for forest biota in this area due to its survival as the only fairly large contiguous forest estate in this area (Kityo 2008).

2.2 Study design

The study was conducted in the village of Namusa in the MCFR for 12 months between August 2018 and December 2019. Data were intermittently collected (with a one-month break after two months of data collection) along a gradient of habitat degradation that included a primary forest interior, a degraded forest edge, and fallows, gardens, sugarcane plantations, homesteads close to the forest, collectively referred to as adjacent habitats (Figure 1). For every sampling regime, nine transects were set in subjectively selected sites, namely primary forest interior, depleted forest edges and adjacent habitats (Figure 1). Rodents and shrews were trapped using Sherman traps set in transects of 200 m with 20 stations and an inter-station distance of 10 m with a trapping effort of 40 traps per transect (Mulungu et al. 2011). Traps were baited with a mixture of peanut butter, maize flour, ripe bananas and silver fish, and traps laid out for 3 nights. Morphometric measurements were taken from every captured specimen including the total length (TL), tail vertebrae length (TV), hind foot length (Hf), ear length, and weight (Wt). All measurements were recorded in millimeters and weight in grams. All specimens collected were kept as wet specimens in 75% ethanol and kept as vouchers in Makerere zoological museum. Collected specimens were identified to species using morphometric measurements cross references with published identification guides (Brambell 1973, Delany 1975, Thorn & Kerbis 2009, Monadjem et al. 2015) For selected individuals of each morphotype, identifications were confirmed by sequencing of partial mitochondrial cytochrome b gene (i.e. DNA barcoding) from 96% ethanol-preserved samples at the Institute of Vertebrate Biology (IVB) of the Czech Academy of Sciences. Obtained sequences were compared with the sequences in GenBank and further unpublished sequences in the database of IVB. The barcoding protocol (i.e. used primers, PCR conditions and sequencing) is described in Bryja et al. (2014). Microhabitat variables were also subjectively recorded by observation for every transect. These were canopy cover (Delineated by percentage cover where any cover below 40% was considered as no canopy cover), forest undergrowth (measured by low, medium, max), while water source (with aa 500m proximity) and leaf litter (which was measured by presence or absence). Vegetation cover type (garden, fallow and plantation) was used for adjacent habitats. Average monthly rainfall data was obtained from the Sugar Cooperation of Uganda Lugazi (SCOUL) and used to assess how species abundance fluctuates with rainfall over time

2.3 Data analysis

Trapping success (TS) was computed using the formula;
\begin{equation} TS=\left(\frac{N_{i}}{T_{n}}\right)\times 100\nonumber \\ \end{equation}
Where; \(N_{i}=\) number of specimens collected\(,T_{n}=total\ number\ of\ traps\ se\)t.
Shannon-Wiener diversity index was computed from the formula;
H’=−pi ∑ ln pi
where
pi is the proportion of individuals found in species i
ln Natural logarithm
Species abundance was assessed based on the number of individuals recorded for every species.
Correspondence analysis (CA) was performed for the different sampling sites to show the chi-square distance among sites and also assess the significance of association between species and habitats.
To certify adequate sample size, rarefaction curves were developed using Vegan statistical package in R (Oksanen et al. 2016). Where the smoothed averages of the individual curves represent the statistical expectation of species accumulation per sampling site.
Small mammal community composition and association was measured using “multipatt” function in “R” using packages “vegan” and “indicspecies” (Miquel De Caceres 2013). In order to run an indicator species analysis, a vector containing the classification of the sites micro habitat types into groups was done using non-hierarchical cluster analysis with different “r” functions in packages “indicspecies” and “vegan”. In order to determine which species can be used as indicators of different sampling sites; an approach commonly used in ecology is the use of the Indicator Value index (Dufr^ene & Legendre 1997, De Caceres et al. 2010). The approach calculates the “IndVal” (indicator value) index between the species and each site group and then looks for the group corresponding to the highest association value, the statistical significance of this relationship is then tested using a permutation test. All analyses were performed with R statistical software R Core Team (2013). Multiple linear regression was carried out to assess the relationship between species richness and species abundance with rainfall. The assumption is that species richness and abundance follow rainfall patterns. Line graphs were made to examine for any temporal trends in species richness and abundance, while cross correlation graphs were made to seek for relationship patterns between species richness, abundance and rainfall.