Carbon flux and diversity of nematodes and termites in Cameroon forest soils

Theoretically, there are three principal ways in which ecosystem processes might respond to reductions in species richness. These theories are reviewed, and then considered in the context of a study of the diversity of soil nematodes and termites in near-primary forest sites at Mbalmayo, Cameroon, and the contribution made by these two taxa to carbon fluxes (CO₂ and CH₄) from the forest floor. Nematode abundances average 2.04 × 10⁶ m^-2, and termites between 2933 and 6957 m^-2. The site is the most species-rich yet investigated for both groups anywhere in the world, so that a very large number of species contribute to carbon fluxes. We speculate about how much redundancy might be built into the functioning of both assemblages, and point out the enormous difficulties of resolving such questions, and of producing such detailed species-inventories.     

The effect of termite biomass and anthropogenic disturbance on the CH4 budgets of tropical forests in Cameroon and Borneo

The exchange of CH₄ between tropical forests and the atmosphere was determined by simultaneously measuring the net CH₄ flux at the soil surface and assessing the flux contribution from soil-feeding termite biomass, both within the soil profile and in mounds. In Cameroon the flux of CH₄ ranged from a net emission of 40.7 ng m^–2 s^–1 to a net CH₄ oxidation of –53.0 ng m^–2 s^–1. Soil-inhabiting termite biomass was significantly correlated with CH₄ flux. Termite mounds emitted up to 2000 ng s^–1 mound^–1. Termite-derived CH₄ emission reduced the soil sink strength by up to 28%. Disturbance also had a strong effect on the soil sink strength, with the average rate of CH₄ oxidation, at – 17.5 ng m^–2 s^–1, being significantly smaller (≈ 36%) at the secondary forest site than the –27.2 ng m^–2 s^–1, observed at the primary forest site. CH₄ budgets calculated for each site indicated that both forests were net sinks for CH₄ at – 6.1 kg ha^–1 y^–1 in the near-primary forest and – 3.1 kg ha^–1 y^–1 in the secondary forest. In Borneo, three forest sites representing a disturbance gradient were examined. CH₄ oxidation rates ranged from 0 to – 32.1 ng m^–2s^–1 and a significant correlation between the net flux and termite biomass was observed only in an undisturbed primary forest, although the biomass was insufficient to cause net emission of CH₄. Rates of CH₄ oxidation were not significantly different across the disturbance gradient but were, however, larger in the primary forest (averaging – 15.4 ng m^–2 s^–1) than in an old-growth secondary forest (–13.9 ng m^–2s^–1) and a young secondary re-growth (– 10.8 ng m^–2s^–1). CH₄ flux from termite mounds ranged from net oxidation in an abandoned mound to a maximum emission of 468 ng s^–1 mound^–1. CH₄ budgets calculated for each site indicated that CH₄ flux from termite mounds had an insignificant effect on the budget of CH₄ at the regional scale at all three forest sites. Annual oxidation rates were – 4.8, – 4.2 and – 3.4 kg ha^–1 y^–1 in the primary, secondary and young secondary forests, respectively.     

A pilot analysis of gut contents in termites from the Mbalmayo Forest Reserve,Cameroon

Abstract.

• 1 Gut content analysis was carried out on eight species of higher termites representing the four included subfamilies and notional soil-feeding and wood-feeding forms. A lower termite species feeding on wood was also examined.
• 2 Pooled homogenates equivalent in volume to twenty hindguts of Thoracotermes macrothorax were diluted, stained and scored present/absent in 500 haemacytometer fields for ten content item categories: macerated organic material, lignified tissue, plant tissue fragments, fungal mycelium, arthropod parts, plant roots, safranin +ve, gentian violet +ve, humus and silica. For interspecific comparisons, the occurrence of each category was expressed as a proportion of the total haemacytometer cells examined.
• 3 A hierarchical classification of the species, based on gut contents, was prepared using a two-way indicator species analysis, and suggested the following rank order of species along a hypothetical humification gradient (soil to sound wood): Thoracotermes macrothorax, Astalotermes quietus (both soil-feeders), Termes hospes, Amalotermes phaecocephalus, Pseudacanthotermes militaris, Microtermes congoensis, Nasutitermes lujae, Microcerotermes parvus, Schedorhinotermes putorius (all notional wood-feeders). Arthropod parts, silica and humus were identified as indicator factors.
• 4 It is proposed that wood-feeding forms can be subdivided into a group consuming some silica and humus (five species: humified wood-feeders) and a second group of sound wood-feeders (two species).
• 5 Pianka's equations for diet breadth and diet overlap were also applied to the data. These identified Microtermes congoensis, Schedorhinotermes putorius and Pseudacanthotermes militaris as the most specialized feeders, i.e. that they can extract their nutrients from the least heterogenous substrates.

     

The spatial pattern of soil-dwelling termites in primary and logged forest in Sabah, Malaysia

Abstract.  1. Primary and logged lowland dipterocarp forest sites were sampled for subterranean termites using soil pits located on a grid system in order to detect any patchiness in their distribution.
2. A spatial pattern in termite distributions was observed in the primary and logged sites, but the response differed between soil-feeding and non-soil-feeding termites.
3. Spatial analysis showed that soil-feeding termites were homogeneously distributed in the primary forest but significantly aggregated in the logged forest. This pattern was reversed for non-soil-feeding termites and may result from differences in resource provisioning between the two sites.
4. Gaps in termite distribution comprised a greater area than patches for both feeding groups and sites, but gaps dominated the logged site.
5. A significant association between soil-feeding and non-soil-feeding termite distributions occurred at both sites. This arose from an association between patches in the primary forest and between gaps in the logged forest.
6. Termite spatial pattern was optimally observed at a minimum extent of 64 m and lag of 2 m.
7. The spatially explicit SADIE (Spatial Analysis by Distances IndicEs) analyses were more successful than (non-spatially explicit) multivariate analysis (Canonical Correspondence Analysis) at detecting associations between termite spatial distributions and that of other biotic and abiotic variables.     

Methane consumption in two temperate forest soils

Forest soils are thought to be an important sink for atmospheric methane. To evaluate methane consumption,¹⁴C-labeled methane was added to the headspace of intact soil cores collected from a mixed mesophytic forest and from a red spruce forest located in the central Appalachian Mountains. Both soils consumed the added methane at initially high rates that decreased as the methane mixing ratio of the air decreased. The mixed mesophytic forest soil consumed an average of 2 mg CH₄ m^–2 d^–1 versus 1 mg CH, m^–2 d^–1 for the spruce forest soil. The addition of acetylene to the headspace completely suppressed methane consumption by the soils, suggesting that an aerobic methane-consuming microorganism mediated the process. At both forest sites, methane mixing ratios in soil air spaces were greater than that in the air overlying the soil surface, indicating that these soils had the ability to produce methane. Models of methane emission from forest soils to the atmosphere must represent methane flux as the balance between production and consumption of methane, which are controlled by very different factors     

Iron oxidation stimulates organic matter decomposition in humid tropical forest soils

Humid tropical forests have the fastest rates of organic matter decomposition globally, which often coincide with fluctuating oxygen (O₂) availability in surface soils. Microbial iron (Fe) reduction generates reduced iron [Fe(II)] under anaerobic conditions, which oxidizes to Fe(III) under subsequent aerobic conditions. We demonstrate that Fe (II) oxidation stimulates organic matter decomposition via two mechanisms: (i) organic matter oxidation, likely driven by reactive oxygen species; and (ii) increased dissolved organic carbon (DOC) availability, likely driven by acidification. Phenol oxidative activity increased linearly with Fe(II) concentrations (P < 0.0001, pseudo R² = 0.79) in soils sampled within and among five tropical forest sites. A similar pattern occurred in the absence of soil, suggesting an abiotic driver of this reaction. No phenol oxidative activity occurred in soils under anaerobic conditions, implying the importance of oxidants such as O₂ or hydrogen peroxide (H₂O₂) in addition to Fe(II). Reactions between Fe(II) and H₂O₂ generate hydroxyl radical, a strong nonselective oxidant of organic compounds. We found increasing consumption of H₂O₂ as soil Fe(II) concentrations increased, suggesting that reactive oxygen species produced by Fe(II) oxidation explained variation in phenol oxidative activity among samples. Amending soils with Fe(II) at field concentrations stimulated short‐term C mineralization by up to 270%, likely via a second mechanism. Oxidation of Fe(II) drove a decrease in pH and a monotonic increase in DOC; a decline of two pH units doubled DOC, likely stimulating microbial respiration. We obtained similar results by manipulating soil acidity independently of Fe(II), implying that Fe(II) oxidation affected C substrate availability via pH fluctuations, in addition to producing reactive oxygen species. Iron oxidation coupled to organic matter decomposition contributes to rapid rates of C cycling across humid tropical forests in spite of periodic O₂ limitation, and may help explain the rapid turnover of complex C molecules in these soils.     

Evaluation of methods for estimating soil carbon dioxide efflux across a gradient of forest disturbance

Better understanding of variation in soil carbon dioxide (CO₂) efflux caused by measurement techniques is needed, especially over gradients of site disturbance, to accurately estimate the global carbon cycle. We present soil CO₂ efflux data from a gradient of disturbance to ponderosa pine (Pinus ponderosa C. Lawson var. scopulorum Engelm.) forests in northern Arizona, USA that were obtained using four different techniques: vented static chambers, a Licor 6400‐09, and soil CO₂ diffusion profiles using two different models (Moldrup, Millington–Quirk) to estimate soil gas diffusivity. We also compared soil CO₂ efflux measured by the Moldrup and Millington–Quirk diffusion profile methods to nighttime total ecosystem respiration (TER) data from an eddy covariance tower. We addressed four questions: (1) Does the use of a given method to measure soil CO₂ efflux bias results across a disturbance gradient? (2) Does the magnitude of difference between observed and modeled estimates of soil CO₂ differ between methods and across sites? (3) What is the spatial variability of each method at each site? (4) Which method is closest to the estimate of TER measured by the eddy covariance tower? Although soil CO₂ efflux varied significantly among methods the differences were consistent among sites. Measured and modeled total growing season fluxes were generally higher for the Licor 6400‐09 and Millington–Quirk diffusion gradient methods compared with static chamber and the Moldrup diffusion gradient methods. A power analysis showed that the larger static chamber was the most efficient method at sampling spatial variation in soil CO₂ efflux. Nighttime measurements of soil CO₂ efflux from the Moldrup diffusion gradient method were most strongly related to nighttime TER assessed with eddy covariance. The use of a single, well‐implemented method to measure soil CO₂ efflux is unlikely to create bias in comparisons across a gradient of forest disturbance.     

Forest succession in Kibale National Park, Uganda: implications for forest restoration and management

Forest succession was studied in four plots in former grasslands at the Ngogo study area in Kibale National Park, Uganda. The plots were located in areas that had been protected from fire for 0.58, 25, 9 and ≈30 years for plots 1, 2, 3 and 4, respectively. Species richness reflected the length of time that the plot had been protected from fire; it was highest in plot 4 and lowest in plot 1. Species density, stem density and basal area were all highest in plot 4 and lowest in plot 1. The species densities of plots 2 and 3 were not different. Similarly, plots 2 and 4 did not differ with regard to stem density or basal area. Animal seed dispersers played a vital role in the colonization of grasslands by forest tree species.     

Disentangling the long‐term effects of disturbance on soil biogeochemistry in a wet tropical forest ecosystem

Climate change is increasing the intensity of severe tropical storms and cyclones (also referred to as hurricanes or typhoons), with major implications for tropical forest structure and function. These changes in disturbance regime are likely to play an important role in regulating ecosystem carbon (C) and nutrient dynamics in tropical and subtropical forests. Canopy opening and debris deposition resulting from severe storms have complex and interacting effects on ecosystem biogeochemistry. Disentangling these complex effects will be critical to better understand the long‐term implications of climate change on ecosystem C and nutrient dynamics. In this study, we used a well‐replicated, long‐term (10 years) canopy and debris manipulation experiment in a wet tropical forest to determine the separate and combined effects of canopy opening and debris deposition on soil C and nutrients throughout the soil profile (1 m). Debris deposition alone resulted in higher soil C and N concentrations, both at the surface (0–10 cm) and at depth (50–80 cm). Concentrations of NaOH‐organic P also increased significantly in the debris deposition only treatment (20–90 cm depth), as did NaOH‐total P (20–50 cm depth). Canopy opening, both with and without debris deposition, significantly increased NaOH‐inorganic P concentrations from 70 to 90 cm depth. Soil iron concentrations were a strong predictor of both C and P patterns throughout the soil profile. Our results demonstrate that both surface‐ and subsoils have the potential to significantly increase C and nutrient storage a decade after the sudden deposition of disturbance‐related organic debris. Our results also show that these effects may be partially offset by rapid decomposition and decreases in litterfall associated with canopy opening. The significant effects of debris deposition on soil C and nutrient concentrations at depth (>50 cm), suggest that deep soils are more dynamic than previously believed, and can serve as sinks of C and nutrients derived from disturbance‐induced pulses of organic matter inputs.     

Effects of forest clear-cutting and soil disturbance on the biology of small forest brooks

As a part of the larger Nurmes research project, we studied the effects of clear-cutting and soil disturbance (ditching, ploughing, mounding) on the biology of small forest brooks. After these forestry activities, incoming light, temperature, and nutrient content of the water increased significantly. Mean algal biomass peaked during the first summer after clear-cutting. During the first year following soil disturbance, algal biomass initially decreased because of turbidity, but began to rise again as suspended solids in the water decreased in the summer months. Species composition changed, too; after clear-cutting, Cryptomonas and Chlamydomonas species increased their densities (48% of the algal biomass) and following soil disturbance, the species numbers and densities of Conjugatophyceae (52% of the algal biomass) were higher than before. Canonical correspondence analysis suggested that increased nutrient levels and acidity were the main factors behind the changes in the flora following clear-cutting. The most important effect of soil disturbance was the increased water temperature. Effects of forest clear-cutting on algal productivity in the brooks remain evident at least for three years and those of soil disturbance for an even longer period. Restricted clear-cutting, with a protective zone left uncut around the brooks, appears to reduce the effects.     

森林生态系统土壤CO2释放随海拔梯度的变化及其影响因子

联合国气候框架公约的签署提升了人们对全球变暖、碳循环的关注。土壤CO2释放作为土壤-大气CO2交换的主要途径之一,成为了各国生态学家研究的重点内容。通过对1800～2155m海拔梯度上森林生态系统土壤CO2释放进行研究,揭示了较小空间尺度上土壤CO2释放的变化规律及其控制机制。在研究区域内,随着海拔梯度的增加,森林土壤CO2释放由(1.94±0.06)μmolm-2s-1逐渐增加至(2.22±0.07)μmolm-2s-1。土壤温度、土壤水分、土壤有机碳(SOC)、全N、全P与土壤CO2释放呈显著正相关(n=14,P<0.05);土壤容重与土壤CO2释放速率呈显著负相关(n=14,P<0.05);土壤pH对土壤CO2释放影响不显著。作为一个复杂的生态学过程,环境因子及其交互作用对土壤CO2释放产生影响,为了减少因子共线性影响,逐步降低因子维数,采用主成分分析(PCA)揭示了土壤温度、土壤水分、SOC、全N、全P、容重6个因子的联合作用,其累积贡献率达到了57%以上;进一步运用逐步回归分析方法,探讨了影响土壤CO2释放沿海拔梯度分布的主导因子,结果表明土壤水分是研究区域森林生态系统土壤CO2释放沿海拔梯度变化的主导因子。     

全球变化对森林土壤甲烷吸收的影响及其机制研究进展

大气CO₂浓度升高、降水格局改变、全球氮沉降增加和土地覆盖变化等全球变化不仅改变了森林土壤理化性质,而且影响了植物的生长和微生物活性,导致森林土壤碳、氮循环发生改变,进而影响土壤CH₄的吸收.本研究综述了森林土壤CH₄吸收的重要性,森林土壤CH₄吸收对大气CO₂浓度升高、降水格局改变、全球氮沉降增加和土地覆盖变化等全球变化的响应差异及驱动机制.大气CO₂浓度升高抑制土壤CH₄吸收;降水减少倾向于促进土壤CH₄吸收;外源氮输入抑制富氮森林土壤CH₄吸收,而对贫氮森林土壤CH₄吸收则表现为促进或不影响;森林转化为草地、农田或人工林会减少土壤CH₄的吸收量,而植树造林则会增加土壤CH₄的吸收量.今后的研究重点是探讨全球变化对森林土壤CH₄吸收产生长期影响和综合效应,并借助分子生物学方法进一步探究土壤CH₄吸收的微生物学机制.     

The effect of human disturbance on tree species composition and demographic structure in Kalinzu Forest Reserve,Uganda

Tree abundance and species composition in the mechanically logged, intensively pitsawn and minimally disturbed areas of Kalinzu Forest Reserve were determined. The spatial and diameter size‐class distribution of ten selected tree species representing pioneer, secondary colonizer, understorey, canopy, dominant and endangered species were also assessed. The species were: Musanga leo‐errerae, Trema orientalis, Funtumia africana, Strombosia scheffleri, Oxyansus speciosus, Parinari excelsa, Tabernaemontana holstii, Newtonia buchananii, Lovoa swynnertonii and Entandrophragma angolense. The mean stem density of all trees ≥0.5 m in height was 2809.1 per hectare and 150 species were enumerated in the three forest areas. Most individuals (47.73%) and species (75.0%) were recorded in the minimally disturbed and least in the mechanically logged areas. Of the selected species, F. africana was the most abundant (n = 789) and widely distributed in each of the areas. Entandrophragma angolense was the least abundant (n = 63) and most of its individuals (74.6%) were recorded in the minimally disturbed area. Funtumia africana and S. scheffleri (subcanopy), O. speciosus and T. holstii (understorey), N. buchananii (canopy) and P. excelsa (an upper canopy and a climax species of this forest reserve) had an inverse J‐shaped diameter size‐class structure while pioneer species (M. leo‐errerae and T. orientalis) had a bell‐shaped size‐class structure.     

The effect of forest disturbance on the leaf litter ant fauna in Ghana

In order to observe the effect of forest loss on the leaf litter ant fauna in Ghana, West Africa, samples were taken in primary forest, secondary forest and in cocoa plots. Ants were extracted from the leaf litter by sieving followed by suspension in Winkler bags. The species composition and species richness in the three different habitats were compared and no significant difference was found between them. It was concluded that most primary forest leaf litter ant species continue to survive in parts of the agricultural landscape which has largely replaced their original habitat.     

Dynamics of carbon stocks in soils and detritus across chronosequences of different forest types in the Pacific Northwest, USA

We investigated variation in carbon stock in soils and detritus (forest floor and woody debris) in chronosequences that represent the range of forest types in the US Pacific Northwest. Stands range in age from <13 to >600 years. Soil carbon, to a depth of 100 cm, was highest in coastal Sitka spruce/western hemlock forests (36±10 kg C m^?2) and lowest in semiarid ponderosa pine forests (7±10 kg C m^?2). Forests distributed across the Cascade Mountains had intermediate values between 10 and 25 kg C m^?2. Soil carbon stocks were best described as a linear function of net primary productivity (r²=0.52), annual precipitation (r²=0.51), and a power function of forest floor mean residence time (r²=0.67). The highest rates of soil and detritus carbon turnover were recorded on mesic sites of Douglas‐fir/western hemlock forests in the Cascade Mountains with lower rates in wetter and drier habitats, similar to the pattern of site productivity. The relative contribution of soil and detritus carbon to total ecosystem carbon decreased as a negative exponential function of stand age to a value of ～35% between 150 and 200 years across the forest types. These age‐dependent trends in the portioning of carbon between biomass and necromass were not different among forest types. Model estimates of soil carbon storage based on decomposition of legacy carbon and carbon accumulation following stand‐replacing disturbance showed that soil carbon storage reached an asymptote between 150 and 200 years, which has significant implications to modeling carbon dynamics of the temperate coniferous forests following a stand‐replacing disturbance.     

Methane uptake rates in Japanese forest soils depend on the oxidation ability of topsoil, with a new estimate for global methane uptake in temperate forest

To clarify the reason for the higher CH₄ uptake rate in Japanese forest soils, twenty-seven sites were established for CH₄ flux measurement. The first order rate constant for CH₄ uptake was also determined using soil core incubation at 14 sites. The CH₄ uptake rate had a seasonal fluctuation, high in summer and low in winter, and the rate correlated with soil temperature at 17 sites. The annual CH₄ uptake rates ranged from 2.7 to 24.8 kg CH₄ ha⁻¹ y⁻¹ (the average of these rates was 9.7 or 10.9 kg CH₄ ha⁻¹ y⁻¹, depending on method of calculation), which is somewhat higher than the uptake rates reported in previous literature. The averaged CH₄ uptake rate correlated closely with the CH₄ oxidation rate of the topsoil (0–5 cm) in the study sites. The CH₄ oxidation constant of the topsoil was explained by a multiple regression model using total pore volume of the soil, nitrate content, and C/N ratio (p < 0.05, R ² = 0.684). This result and comparison with literature data suggest that the high CH₄ uptake rate in Japanese forest soils depends on the high porosity probably due to volcanic ash parent materials. According to our review of the literature, the CH₄ uptake rate in temperate forests in Europe is significantly different from that in Asia and North America. A new global CH₄ uptake rate in temperate forests was estimated to be 5.4 Tg y⁻¹ (1 SE is 1.1 Tg y⁻¹) on a continental basis.     

Influence of atmospheric CO2 enrichment on methane consumption in a temperate forest soil

Rates of atmospheric CH₄ consumption of soils in temperate forest were compared in plots continuously enriched with CO₂ at 200 µL L^?1 above ambient and in control plots exposed to the ambient atmosphere of 360 µL CO₂ L^?1. The purpose was to determine if ecosystem atmospheric CO₂ enrichment would alter soil microbial CH₄ consumption at the forest floor and if the effect of CO₂ would change with time or with environmental conditions. Reduced CH₄ consumption was observed in CO₂‐enriched plots relative to control plots on 46 out of 48 sampling dates, such that CO₂‐enriched plots showed annual reductions in CH₄ consumption of 16% in 1998 and 30% in 1999. No significant differences were observed in soil moisture, temperature, pH, inorganic‐N or rates of N‐mineralization between CO₂‐enriched and control plots, indicating that differences in CH₄ consumption between treatments were likely the result of changes in the composition or size of the CH₄‐oxidizing microbial community. A repeated measures analysis of variance that included soil moisture, soil temperature (from 0 to 30 cm), and time as covariates indicated that the reduction of CH₄ consumption under elevated CO₂ was enhanced at higher soil temperatures. Additionally, the effect of elevated CO₂ on CH₄ consumption increased with time during the two‐year study. Overall, these data suggest that rising atmospheric CO₂ will reduce atmospheric CH₄ consumption in temperate forests and that the effect will be greater in warmer climates. A 30% reduction in atmospheric CH₄ consumption by temperate forest soils in response to rising atmospheric CO₂ will result in a 10% reduction in the sink strength of temperate forest soils in the atmospheric CH₄ budget and a positive feedback to the greenhouse effect.     

Different responses of soil respiration to environmental factors across forest stages in a Southeast Asian forest

Soil respiration (SR) in forests contributes significant carbon dioxide emissions from terrestrial ecosystems and is highly sensitive to environmental changes, including soil temperature, soil moisture, microbial community, surface litter, and vegetation type. Indeed, a small change in SR may have large impacts on the global carbon balance, further influencing feedbacks to climate change. Thus, detailed characterization of SR responses to changes in environmental conditions is needed to accurately estimate carbon dioxide emissions from forest ecosystems. However, data for such analyses are still limited, especially in tropical forests of Southeast Asia where various stages of forest succession exist due to previous land‐use changes. In this study, we measured SR and some environmental factors including soil temperature (ST), soil moisture (SM), and organic matter content (OM) in three successional tropical forests in both wet and dry periods. We also analyzed the relationships between SR and these environmental variables. Results showed that SR was higher in the wet period and in older forests. Although no response of SR to ST was found in younger forest stages, SR of the old‐growth forest significantly responded to ST, plausibly due to the nonuniform forest structure, including gaps, that resulted in a wide range of ST. Across forest stages, SM was the limiting factor for SR in the wet period, whereas SR significantly varied with OM in the dry period. Overall, our results indicated that the responses of SR to environmental factors varied temporally and across forest succession. Nevertheless, these findings are still preliminary and call for detailed investigations on SR and its variations with environmental factors in Southeast Asian tropical forests where patches of successional stages dominate.     

Habitat use by endemic and introduced rodents along a gradient of forest disturbance in Madagascar

We used logistic and Poisson regression models to determine factors of forest and landscape structure that influence the presence and abundance of rodent species in the rain forest of Ranomafana National Park in southeastern Madagascar. Rodents were collected using live-traps along a gradient of human disturbance. All five endemic rodent species (Nesomys rufus, N. audeberti, Eliurus tanala, E. minor and E. webbi) and the introduced rat Rattus rattus were captured in both secondary and primary forests, but the introduced Mus musculus was only trapped in secondary forest. The abundance of R. rattus increased with the level of habitat disturbance, and it was most common in the heavily logged secondary forest. Furthermore, the probability of the presence of R. rattus increased with decreasing distance from forest edge and decreasing canopy cover, while the probability of presence increased with increasing herbaceous cover, altitude and overstory tree height. The species was never observed farther than 500 m away from human habitation or camp-site. N. rufus prefered selectively-logged forest at altitudes above 900 m a.s.l. Its probability of presence increased with increasing canopy cover, herbaceous cover and distance from forest edge, and with decreasing density of fallen logs, overstory tree height and distance from human habitation. N. audeberti prefered heavily-logged areas, while E. tanala was the only species occurring along the entire range of forest disturbance. We suggest that in the Ranomafana National Park the spread of R. rattus is associated with deforestation.