Introduction

The Eastern Arc Mountains (EAMs) of Africa are one of the most biologically diverse areas in the world (Mittermeier et al. 2011). Unfortunately, over 70% of its forests are degraded due to illegal timber harvest, forest fires, expansion for agriculture, mining and biological invasions (Burgess et al. 2000, Newmark 2002, Hulme et al. 2013). As a result, ≥ 800 vascular plants and ≥ 96 vertebrate species occurring nowhere else in the world may be at risk of extinction (Myers et al. 2000, Burgess et al. 2007, Newmark & McNeally 2018). The protected forests in the EAMs provide important habitat for rare flora and fauna. However, forest degradation, particularly caused by invasive plants, reduces the capacity of these forests to support biodiversity conservation (Hall et al. 2009, Kilawe et al. 2022).

Maesopsis eminii Engl. is a lowland rainforest tree species commonly known as an umbrella tree (Fig. 1). The tree species is native to Africa (Binggeli & Hamilton 1993, Mbuya et al. 1994, Epila et al. 2017) but has been successfully introduced in Australia, Latin America, and Asia continents (CABI 2016). In Tanzania, the species' natural range is limited to the rain forest of Bukoba and Geita Districts (Binggeli 1989, Geddes 1998). Soon after its introduction in East Usambara, Tanzania, M. eminii started to spread into the natural forest (Binggeli et al. 1993). The most recent data show that the species accounted for more than 15% of large trees in the sub-montane part of the forest (Geddes 1998, Frontier Tanzania 2001) and 50% of trees in agroforestry systems (Hall et al. 2010). The spread of M. eminii in East Usambara is fostered by effective dispersals; trumpeter hornbill (Bycanistes buccinators), silvery-cheeked hornbill (Bycanistes brevis), fruit bats (Eidolon helvum) and blue monkey (Cercopithecus mitis) (Binggeli 1989, Cordeiro et al. 2004, Epila et al. 2017); absence of natural predators; rapid vegetative growth; prolific seed production; high seed germination rate; tolerance to drought, pest and stress (Binggeli 1989).

Generally, there is insufficient information regarding the impact of invasive plants on mammals, particularly rodents. It has been found that invasive plants can negatively affect animal communication (Harvey & Fortuna 2012, Stewart et al. 2021), limit food supply (Malo et al. 2013), or expose animals to predators. Studies conducted in Europe and North America found that invasive grasses, such as Ammophila arenaria and Bromus tectorum, as well as invasive shrubs like Rhododendron ponticum and Rhamnus cathartica, reduced the quality of mammal forage (Johnson & De León 2015) and the density, abundance, and richness of rodents (Smith et al. 2017, Guiden & Orrock 2019, Kluever et al. 2019). However, not all invasive plants negatively impact small mammals; some positive effects of invasive plants have also been reported. For example, invasive shrubs R. ponticum and R. cathartica have been found to increase the abundance of rodents (Malo et al. 2013, Guiden & Orrock 2019) and invasive shrubs, Lonicera maackii has been found to reduce rodents' risks against predators (Mattos & Errock 2010).

Fig. 1.

Maesopsis eminii trees (white stems) in Amani Forest Nature Reserve.

Fig. 2.

A map of the study area showing the position of East Usambara Mountain in Tanzania and a land cover map of Amani Forest Nature Reserve.

There is a conflicting account of the impacts of M. eminii on residence plant communities. Some scholars argue that the species impoverishes the understory scrub and herb vegetation of the forest's ecosystem, elevates soil pH and accelerates soil erosion (Binggeli 1989, Macfadyen 1989, Geddes 1998, Mugasha et al. 2000, Mwendwa et al. 2019). On the other hand, other studies found that M. eminii did not affect native tree species in any way (Hall 1995, Holmes 1995, Nero & Mohamed 2005) and had a high mortality rate in the original area of introduction (Kilawe et al. 2018). Generally, there is insufficient information regarding the impact of M. eminii on biodiversity, particularly rodents; therefore, further investigation is required.

This study aimed to determine the relationship between M. eminii invasions and rodent community structure in Amani Forest Nature Reserve (AFNR) to have a sound basis for decision-making about managing the invasive tree and the forest ecosystem. We hypothesised that rodent abundance and diversity would be higher in uninvaded than invaded forest sites.

Table 1.

The number of rodents (measured as the minimum-number-known-alive) and abundance of vegetation in invaded and uninvaded forest sites of the Amani Forest Nature Reserve.

Material and Methods

Data analysis

The number of rodents was computed as the minimum-number-known-alive (MNKA) according to Krebs (1966) from the number of individual rodents captured from April 2020 through June 2021. Rodent species diversity was computed according to Shannon-Wiener diversity (H') (Shannon & Wiener 1949) using the ‘vegan’ R package (R Core Team 2022). Statistical comparison of the Shannon diversity index of the two rodent sampling areas (invaded and uninvaded forest sites) was done by Hutcheson’s t-test (Hutcheson 1970). The Hutcheson t-test was conducted using the ‘ecolTest’ R package (R Core Team 2022).

To determine if M. eminii invasion resulted in changes in species community composition in the forest, we compared species counts in the invaded and uninvaded forest sites using the multivariate extension of generalised linear models ‘manyglm’ function (Zuur et al. 2010, Warton et al. 2012) in the ‘mvabund' package (Wang et al. 2012, 2017) in the R software (R Core Team 2022). This model-based approach allows for hypothesis testing, identifying species that expressed an invasion effect, predicting the abundance of each rodent species but without confounding location and dispersion effects due to the misspecification of the mean-variance relationship (Warton et al. 2012). The effect of invasion on the assemblage was evaluated using the ‘anova.manyglm’ function with the P-value calculated from 999 bootstraps via PIT-trap resampling. The option in the ‘anova.manyglm' function was used to assess whether the assemblages in the invaded forest site differed from the uninvaded forest site. We fitted a multivariate generalised linear model with a negative binomial distribution (O’Hara & Kotze 2010) with invasion categories (invaded/uninvaded) as the predictor variable and abundance data as the response variable for mvabund analyses. During the analysis, both numerous rodents (e.g. M. delectorum, Beamys hindei and Lophuromys kilonzoi) and those with occurrence at low abundance (Aethomys chrysophilus and Grammomys dolichurus) species were included. Hence, the model for the abundance of rodents of species ‘j' found at site'i' (Y_ij) is negative binomial (NB): Y_ij ∼ NB (µ_jkl, φ_j) (equation 1) where site ‘i' is in transect ‘k’ and it received invasion ‘l’. The over-dispersion parameter ‘φ_j’ is constant across sites but can vary across species, and the mean of ‘Y_ij’ is ‘µ_jkl’a log-linear function of transect and invasion: log (µ_jkl) = intercept_j + transect_jk + invasion_jl + transect × invasion (equation 2). The model is based on the key assumption that ‘Y_ij’ are independent across forest sites. The appropriateness of the model assumptions was checked by directly plotting the mean-variance relationship (Wang et al. 2012). Data analysis followed data exploration protocols, including visualising outliers, normality tests, and independence between response and variables, as described by Zuur et al. (2010).

Table 2.

Species diversity in invaded and uninvaded forest sites of Amani Forest Nature Reserve.

Table 3.

Univariate tests for rodent species abundance between invaded and uninvaded forest patches.

Fig. 3.

A plot of the rodent abundance in invaded and uninvaded forest sites where: Gram_doli = Grammomys dolichurus, Aeth_chry = Aethomys chrysophilus, Loph_kilo = Lophuromys kilonzoi, Beam_hind = Beamys hindei and Mont_dele = Montemys delectorum.

Results

Discussion

The results of this study reveal that rodents occurred in higher abundance and diversity in uninvaded than invaded forest sites as hypothesised. The invasive tree may be affecting various aspects of the rodent's life. For example, the sites invaded by M. eminii were more open and with more herbaceous cover than uninvaded, potentially affecting rodents' movement, exposure to predators, nesting areas, and food availability. It has been found that invasive plants can limit food supply, expose the animals to predators (Malo et al. 2013) or affect animal communication (Harvey & Fortuna 2012, Stewart et al. 2021). Information is transmitted from plants to animals as either signals and/or incidental cues that animals detect (Schaefer & Ruxton 2011). Hence, invasive plants can interfere with animal communication by altering the habitat or emitting signals and cues, generating an array of effects that can change animal behaviour (Stewart et al. 2021).

Montemys delectorum was the most common of five rodent species recorded in this study, implying that AFNR could be a suitable habitat for this species. It is important to note that M. delectorum inhabits burrows between the roots of large forest trees and under large logs forests (Kingdon 2015). Hence, it is commonly found in forests with high tree density, such as natural montane forests (Bryja et al. 2013, Cassola 2016). Our findings are consistent with Carleton & Stanley (2012) and Ralaizafisoloarivony et al. (2014), who found M. delectorum to be the most common rodent species in the forests of the East and West Usambara Mountains. Montemys delectorum was also discovered dominating the forests of the Ukaguru Mountains (Ademola et al. 2021) and Uluguru Mountains (Chidodo et al. 2020), implying that the species is stable and abundant in the Eastern Arc Mountains' montane forests. Rodent species are abundant in montane forests because of high humidity and soil moisture maintained by a close canopy cover and high tree density.

Furthermore, the abundance of M. delectorum and B. hindei was higher in the uninvaded forest site than in the invaded forest site, implying that plant invasion may have had a negative effect on these species' abundance because these are habitat specialists adapted to forested areas with close canopies and vegetation cover. Maesopsis eminii invasion in AFNR was reported to impoverish understory shrub and herb vegetation and alternated canopy structure (Mwendwa et al. 2019), which are known to be suitable environment characteristics for both M. delectorum and B. hindei. It has been described that deep leaf litter of closed forests provides suitable foraging and nesting areas for M. delectorum (Happold 2013, Monadjem et al. 2015, Thomas et al. 2022) while B. hindei have been reported to be common in areas with dense shrub layers and closed canopies (FitzGibbon et al. 1995). Clusella-Trullas & Garcia (2017) reported that invasive plants have a negative impact on the abundance of native fauna because they directly or indirectly affect food resources for animal communities. Although low species abundance is typical in plant-invaded sites, increased animal abundance in invaded sites has also been reported (Schlaepfer et al. 2010). More L. kilonzoi were found in invaded forest sites than in uninvaded forest sites suggesting that invasion had partial or no effect on the abundance of it in invaded forest sites. This finding is likely due to its high ability to tolerate habitat disturbances, as Ralaizafisoloarivony et al. (2014) pointed out. The observed low and increased abundance of rodents in invaded sites is because the impacts of invasive species vary depending on the characteristics of the resident species, communities and ecosystems (Vilà et al. 2011). As a result, the overall pattern of how invasive plants affect animal abundance is unclear (Hayes & Holzmueller 2012, Schirmel et al. 2015). However, species abundance provides valuable data to indicate changes in animal species between invaded and uninvaded areas (Clusella-Trullas & Garcia 2017).

The finding of this study shows that the occurrence of M. eminii was negatively associated with the abundance of rodent species, specifically B. hindei. The invasion of M. eminii opens the once closed forest and reduces the density and area covered by native tree species changing vegetation structure, canopy cover, ground cover and moist soil (Musila & Leonhartsberger 2006), which are major habitat requirements for B. hindei (FitzGibbon et al. 1995). There are limited studies for comparison, but Stewart et al. (2021) reported that plant invasion can influence plant-animal communication, such as animal behaviour. Thus, the influence of plants on animals remains a signal that may be identified from information conveyed from plants to animals (Schaefer & Ruxton 2011) and is a key pathway via which invasive plants can affect native animals (Harvey & Fortuna 2012). A favourable link between an invasive plant M. eminii and the abundance of rodents in AFNR is most likely owing to changes in forest structure, such as the reduction of understory vegetation, which exposes the rodents to predators. Invasive plants limit the establishment of native food plants, limiting the local food supply for native small rodents (Malo et al. 2013). Other species, such as M. delectorum, L. kilonzoi, G. dolichurus and A. chrysophilus, showed no significant association with M. eminii invasion, implying that the effects of invasive plants vary by species. Kluever et al. (2019) observed that invasive plants can have different effects on the same or different groups of rodents. Although this study did not confirm the association between M. eminii invasion and L. kilonzoi, Dando (2019) and Makundi (2009) reported that L. kilonzoi is currently threatened by the replacement of forested areas by exotic tree plantations.

This study's results further reveal that rodent diversity was higher in uninvaded than invaded forest sites, as hypothesised. We speculate that changes in vegetation cover caused by invasion have possibly reduced niche availability for diverse rodent species. This idea is supported by Freeman et al. (2014), who discovered that changes in vegetation cover due to plant invasion reduce rodent niche availability, resulting in low rodent diversity in invaded sites. Maesopsis eminii has significantly outcompeted native plants in AFNR, changing the vegetation cover of the invaded area (Hall 2010). Low rodent species diversity in plant-invaded sites, particularly invasive shrubs and grasses, has also been reported by Hejda et al. (2009), Ostoja & Schupp (2009) and Freeman et al. (2014) in some regions of Central Europe and USA. Our findings contradict those of Packer et al. (2016), who found no significant difference between Rubus fruticosus invaded and uninvaded sites in South Australia. Thus, the impacts of invasion vary among different levels of ecological complexity; hence, a need to explore the link between invasion and different ecosystems.

Conclusions

This study presented evidence that the invasion of AFNR by M. eminii is associated with a decline in the abundance and diversity of rodent species. If left unmanaged, M. eminii could lead to a loss of habitat for rodents and other native species, threatening the capacity of the forest to support biodiversity. We recommend preventing the spread of M. eminii into the uninvaded parts of the reserve and implementing control measures in invaded sites. The future introduction of the plant outside its ecological range should be avoided or carefully implemented.