Biodiversity and ecosystem function of tropical forest dung beetles under contrasting logging regimes

A key challenge for tropical conservation biologists is to assess how forest management practices affect biodiversity and associated ecosystem functions. Dung beetles (Scarabaeidae: Scarabaeinae) provide an ideal focal guild for such studies. We compared dung beetle assemblages and experimentally assessed rates of dung removal and seed burial in undisturbed forest, low-intensity selectively logged forest under sustainable forest management, and high-intensity logged forest, not under sustainable management in Malaysian Borneo. In total, 7923 individuals from 39 species of dung beetle were collected. There were no significant differences in abundance, biomass or diversity across sites. Species richness was significantly lower in the high-intensity logged sites. Beta diversity was low: of 35 species recorded in undisturbed forest sites, 32 were also captured in low-intensity sites and 29 in high-intensity sites. The environmental and vegetation variables measured had little influence on species composition. An average of 63–99% of dung was removed over 24 h. Mean dung and seed removal were significantly lower in the high-intensity logged sites. Dung removal rates were significantly and positively correlated with dung beetle species richness, but not with dung beetle biomass or abundance. However, the biomass of large-bodied, nocturnal dung beetles was positively correlated with dung removal. In contrast to previous studies, dung beetle biomass and abundance were not correlated with species richness, indicative of density compensation. Overall, dung beetle communities and associated ecosystem functions were robust to low-intensity but not high-intensity selective logging. These differences may be related to changes in the abundance and biomass of particular dung beetle species or guilds rather than community-wide measures of abundance and biomass, highlighting the need to move beyond simplistic biodiversity-ecosystem functioning correlations to understand the functional consequences of habitat modification in high-diversity ecosystems.     

Supply of tree-holes limits nest density of cavity-nesting birds in primary and logged subtropical Atlantic forest

Tree cavities are proposed to limit populations and structure communities of cavity-nesting birds. Although the greatest diversity of cavity-nesting birds is found in tropical and subtropical moist forests, little is known about how tropical logging affects the abundance of cavities or cavity-nesting birds. We compared the abundance of cavities and cavity nests between primary and selectively-logged subtropical moist Atlantic forest in Argentina, and conducted the first before-after controlled nest-box addition experiment to determine whether nest sites limit the breeding density of cavity-nesting birds in tropical or subtropical moist forest. Visual inspection of 86 cavities identified through ground-surveys revealed that only 19% were suitable for nesting birds, suggesting that cavity abundance may be overestimated in the literature on tropical forests. Suitable cavities were found in fewer than 1% of 1156 trees <60 cm dbh but 20% of 20 trees >100 cm dbh. Logged forest had half the basal area of primary forest, one third the density of large trees, nine times fewer cavities suitable for nesting birds, and 17 times fewer active nests. When we added nest boxes, nesting density increased on treatment plots but not on control plots in both logged and primary forest, suggesting that cavity supply can limit nest density even in relatively undisturbed forest. This is the first experiment to show how reduced cavity supply in logged tropical forest can limit breeding density of cavity-nesting birds. International initiatives such as forest certification should promote tropical timber management strategies that conserve large live cavity-bearing trees.     

Effects of selective logging on vascular epiphyte diversity in a moist lowland forest of Eastern Himalaya, India

Effect of selective logging on vascular epiphyte diversity was investigated in a moist lowland forest of Eastern Himalaya. Three epiphytic groups viz. orchids, pteridophytes and non-orchid angiosperm epiphytes were specifically studied in closed, selectively logged and in unlogged forests with treefall gaps. Logging reduced the structural complexity of the forests and altered their microclimate. With logging, there was a general decline in richness and abundance of epiphytes except orchids. The abundance and species composition of pteridophytes and non-orchid angiosperm epiphytes were related to microclimate and substrate features while their richness were correlated only with canopy cover. In contrast, orchid species composition was related to forest structure. A combination of management strategies is required for conservation of all epiphyte groups. A mosaic of logged and unlogged forest patches with undisturbed forests in proximity would maintain the diversity of pteridophytes and other angiosperms. However, for full representation of orchids, it is necessary to maintain the structural diversity of the tree forms.     

A comparison of orang-utan density in a logged and unlogged forest on Sumatra

Several studies have shown that there is a strong decline in orang-utan densities shortly after logging. Nevertheless, there is little information on whether orang-utan densities return to their pre-logging values when logged forest is left to recover. This study investigates the orang-utan density in a 22-year-old selectively logged forest and compares it with the orang-utan density in a nearby ecologically similar primary forest. The results show that the orang-utan density did not differ significantly between primary forest and the selectively logged forest. Since we found no difference in fruit availability between the selectively logged and primary forest, we suggest that the selectively logged forest regenerated sufficiently well to sustain pre-logging levels of orang-utans. This study confirms previous studies that suggest fruit availability is the best ecological predictor of orang-utan densities and found a positive overall correlation between orang-utan density and fruit availability. The food attraction hypothesis, which explains local fluctuations in orang-utan density as a result of variation in fruit availability, was not supported.     

Comparison of breeding bird and vegetation communities in primary and secondary forests of Great Smoky Mountains National Park

We compared breeding bird communities and vegetation characteristics at paired point locations in primary (undisturbed) and mature secondary forest (70-100 years old) sites in Great Smoky Mountains National Park, USA to understand how sites logged prior to creation of the park compare to undisturbed sites following 70 years of protection from human disturbance. We found that bird and vegetation communities are currently similar, but retain some differences in species composition. Rank abundance curves for primary and secondary forest bird communities showed very similar patterns of species dominance. Species composition was also similar on the two sites which shared 24 of the 25 most frequently recorded species. Nonetheless, comparisons of density estimates derived from distance sampling showed three bird species were more abundant on primary forest sites and that one bird species was significantly more abundant on secondary forest sites. Notably, comparisons based on raw counts (unadjusted for potential differences in detectability) produced somewhat different results. Analyses of vegetation samples for the paired sites also showed relative similarity, but with some differences between primary and secondary forests. Primary forest sites had more large trees (trees greater than 50 cm diameter at breast height) and late successional species. Primary forest sites had a denser tall shrub layer while secondary forest sites had a denser canopy layer. Nonetheless, tree species richness, basal area of live trees and number of standing snags did not differ between primary and secondary forest sites. Results indicate that breeding bird communities on sites within the park that were logged commercially 70 years ago are currently quite similar to bird communities on sites with no history of human disturbance. Similarities between the bird communities on previously disturbed and undisturbed sites in Great Smoky Mountains National Park may exceed those on more fragmented landscapes because large patches of primary forest, adjacent to commercially logged sites, remained in the park when it was established in 1935. These patches of primary forest may have served as source areas for commercially logged sites.     

Traditional forest management has limited impact on plant diversity and composition in a tropical seasonal rainforest in SW China

In order to determine the impacts of different traditional forest management types on plant diversity of the seasonal tropical rainforests and infer effective conservation strategies, four types of forests with different management histories were studied in Nabanhe National Nature Reserve (NNNR), Xishuangbanna, China. They were: old-growth forest (non-timber product collection allowed), understorey planted old-growth forests, old secondary forests (∼200-years after slash and burn), and young secondary forest (15–50-years after slash and burn). Although human activities affected tree diversity and composition of the forests in NNNR, the forest regeneration potential of the different management systems were good. Even the young secondary forests, that showed the lowest Fisher’s alpha diversity at the plot level, had similar diversity levels to old-growth forest when all plots were combined. Number of red list tree species, timber species, and edible plant species in young secondary forests was as high as those of old-growth forests, and higher than old secondary forests. Additionally, there were a number of vulnerable and endangered species that were more common in the secondary than old-growth forests, indicating the high conservation value of secondary forests. Understorey plantation in old-growth forest, however, impaired regeneration of the climax species. The beneficial effects of traditional forest use depend strongly on its small scale and its close proximity to undisturbed forest, which serves as a species source during secondary forest regeneration. Unfortunately, traditional forest use is now under serious threat by expanding large-scale monoculture rubber plantations.     

Conversion of a tropical forest into agroforest alters the fine root-related carbon flux to the soil

Large areas of remaining tropical forests are affected by anthropogenic disturbances of various intensities. These disturbances alter the structure of the forest ecosystem and consequently its carbon budget. We analysed the role of fine root dynamics in the soil carbon budget of tropical moist forests in South-east Asia along a gradient of increasing disturbance intensity. Fine root production, fine root turnover, and the associated carbon fluxes from the fine root system to the soil were estimated with three different approaches in five stands ranging from an old growth forest with negligible anthropogenic disturbance to a cacao agroforestry system with planted shade trees. Annual fine root production and mortality in three natural forest sites with increasing canopy openness decreased continuously with increasing forest disturbance, with a reduction of more than 45% between the undisturbed forest and the forest with large timber extraction. Cacao agroforestry stands had higher fine root production and mortality rates than forest with large timber extraction but less than undisturbed forest. The amount of carbon annually transferred to the soil carbon pool through fine root mortality was highest in the undisturbed forest and generally decreased with increasing forest use intensity. However, root-related C flux was also relatively high in the plantation with planted shading trees. In contrast, the relative importance of C transfer from root death in the total above- and below-ground C input to the soil increased with increasing forest use intensity and was even similar to the C input via leaf litter fall in the more intensively managed agroforest. We conclude that moderate to heavy disturbance in South-east Asian tropical moist forests has a profound impact on fine root turnover and the related carbon transfer to the soil.     

Peat swamp forest avifauna of Central Kalimantan, Indonesia: Effects of habitat loss and degradation

Peat swamp forest is a unique wetland ecosystem covering extensive areas in Southeast Asia that has received relatively little scientific attention and is now being lost at a rapid pace. This study examines the effects of anthropogenic degradation on bird communities in disturbed peat swamp forest habitats – namely, intact logged forest, a degraded forest fragment, and non-forest regrowth – in Central Kalimantan, Indonesia. Results show that species richness was significantly higher and species composition significantly different in intact logged forest in comparison to the degraded forest fragment and non-forested area. Nectarivore and tree foliage-gleaning insectivore abundance declined outside the intact forest, while the regrowth was dominated by the yellow vented bulbul, an open country insectivore–frugivore. Surveys reveal that large intact tracts of logged peat swamp forest can harbour threatened and near threatened bird species (36% of records) and thus play a role in their conservation, especially for a few habitat specialists. Given the extent of unmanaged degraded peatlands and continuing pressure to development them, urgent conservation actions are needed to rehabilitate and protect this ecosystem.     

Bat activity on riparian zones and upper slopes in Australian timber production forests and the effectiveness of riparian buffers

Unlogged buffers are used to ameliorate impacts to riparian areas in timber production forests. One function of these buffers is to protect the biodiversity of riparian areas. We measured bat activity in buffered streams with ultrasonic detectors across four different stream orders in logged, regrowth and mature forests (60 sites). Bat activity, foraging rates and species richness were similar in buffered streams surrounded by logged, regrowth and mature forests, suggesting that riparian areas effectively provide habitat for foraging and commuting bats in selectively logged forests. Vespadelus pumilus was the only species that responded to logging history, with decreased activity in mature forests. We found higher activity on larger rather than smaller order streams, a pattern also not affected by logging history. Bat activity along paired forest trail flyways on upper slopes (60 sites) was measured simultaneously with riparian flyway activity (for a total of 120 sites) to determine the importance of riparian areas relative to other available flyways. Activity was higher on upper slopes than on small streams, but similar to levels on larger streams. Total foraging activity was similar between riparian zones and upper slopes. Upper slopes contained higher species diversity, with Chalinolobus gouldii, Miniopterus schreibersii, Mormopterus norfolkensis, Scotorepens sp., Vespadelus pumilus and, to a lesser extent, Vespadelus darlingtoni detected more often than along streams. Other species (Rhinolophus megaphyllus, Nyctophilus spp. and Miniopterus australis) were not affected by topography. Estimates of total vegetation cover and, in particular, rainforest cover, were negatively associated with bat activity, highlighting the need for management of forest ‘clutter’ in regrowth forests for a suite of bat species. Streams and forest trails provide areas of lower clutter, which assist in maintaining high species diversity in regrowth forests. Our results support the use of riparian buffers, and point to the need for greater recognition of tracks on upper slopes as important habitat.     

Litter decomposition in primary forest and adjacent fire-disturbed forests in the Gran Sabana,southern Venezuela

Litter decomposition and nutrient dynamic were studied in tall primary forest (TF) and in adjacent slightly fire-affected (MF) and strongly fire-affected (LF) forests of the Gran Sabana, southern Venezuela. The aim of the study was to compare the mass and nutrient loss of litter in undisturbed forest and adjacent fire-disturbed forests growing under the same soil conditions. The results showed no significant differences in the dry-mass reduction among TF, MF, and LF after 1-year of decomposition. At the end of the decomposition period, the mass loss was 31% in TF, 24% in MF, and 25% in LF. With few exceptions, the initial nutrient content of the litter did not show significant differences among the studied forests. The initial litter was poor in nutrients, especially in P, with C/N ratio extremely high, particularly in LF. Both residual P content and C/N ratio were the most effective predictors of dry-mass loss. The general trend in the three studied forests was net N and to less extend P immobilization and release of K, Mg, and Ca. We concluded that during a 1-year period, the decomposition process was similar in undisturbed and adjacent fire-disturbed forests in the Gran Sabana and that the low litter decomposition seems to be mainly controlled by the low chemical quality of the decomposing litter.     

Impacts of selective logging on canopy and ground assemblages of tropical forest butterflies: Implications for sampling

Commercial selective logging is a major cause of habitat disturbance in Southeast Asian rainforests, yet despite much research there is little consensus on impacts of disturbance on biodiversity. There is also little consensus on the most appropriate methods for sampling tropical species, making it difficult to draw general conclusions from published studies. For example, many studies have used butterflies to assess impacts of selective logging but sampling has usually been conducted at ground level and the canopy fauna has often been ignored. In this study, we investigate the importance of sampling in the canopy by using fruit-baited traps to investigate impacts of selective logging on Nymphalid butterflies in primary forest and forest selectively logged 15 years previously in Sabah (Malaysian Borneo). Analysing combined data from canopy (≈25 m above ground level) and ground-level traps showed significantly lower diversity in selectively logged forest. However, this difference was not observed when data from only ground traps were analysed. Primary forest supported a butterfly assemblage comprising species with more restricted geographical ranges, and thus higher conservation value, compared with selectively logged forest. This result was observed regardless of whether or not canopy data were included in this analysis. We conclude that sampling in the canopy is critical when producing species inventories, but of little importance when determining the impacts of selective logging on restricted-range species.     

Harvesting of non-timber forest products and implications for conservation in two montane forests of East Africa

Plant species-level research that comprises inventories, impact studies and monitoring is necessary if plant resources are to be harvested sustainably by human populations living adjacent to protected areas in sub-Saharan Africa. This research assessed the extraction of plant products from two montane forest ecosystems, Uzungwa Scarp Forest Reserve (USFR) and Bwindi Impenetrable National Park (BINP), East Africa. In USFR, data from vegetation sampling and interviews with local people were used to understand the ecological and socio-economic aspects of non-timber forest product (NTFP) harvesting. The densities of eight commonly harvested tree species, most of which were used as building poles, were approximately 2.4-4.5 times lower in disturbed versus undisturbed habitats across all four sites in USFR. Interviews with 91 women and 54 men suggested that most species were harvested for medicinal uses (57 species) and building purposes (50 species). In BINP, the liana Loeseneriella apocynoides (Apocynaceae), is harvested for basketry weaving. Evidence suggested that the liana was harvested in both protected and unprotected areas of BINP. Data collected suggested a negative impact on this species in an unprotected versus a protected area, with stem diameters larger than 1 cm significantly more abundant in the protected area. This study reveals that harvesting of NTFPs occurs even in these two protected forest areas, and that over-exploitation not only threatens species of high-demand, but could also alter forest structure and composition. Management practices that encourage the monitoring of sustainable harvesting levels of species and promote alternative plants for the same uses should be considered as part of conservation strategies.     

Bird community responses to reduced-impact logging in a certified forestry concession in lowland Bolivia

We studied bird community composition and abundance within the logged and unlogged forest areas of a certified forestry concession in lowland Bolivia. The logged forest was harvested using reduced-impact logging techniques between one and four years previously. We used canonical correspondence analysis to describe the relationship between selected environmental variables and bird species abundance data, and the Indicator Value procedure to test for associations between bird species and the logged and unlogged habitats. Approximately one-third of birds were restricted to either the logged or unlogged areas, with 20% of all species only encountered in, or significantly more abundant in, the unlogged areas of the concession. The majority of birds found in significantly higher abundance in the unlogged areas of the concession were associated with forest habitats dominated by large trees, or a high diversity of trees, providing dense canopy cover and deep leaf litter, with an understorey dominated by ferns. Over 40% of bird species that were significantly associated with the unlogged areas of the concession are of conservation concern. In contrast, the majority of birds associated with the logged areas of the concession are known to be relatively resilient to human disturbance. The majority of species which exhibited significant lower abundances in the logged areas of the concession belonged to insectivorous or frugivorous feeding guilds. We discuss whether current management practices within this certified concession are sustainable and how our results can be used to guide future research and inform better practice.     

Reduced-impact logging and temporal activity of understorey bats in lowland Amazonia

Because global timber demands continue to threaten tropical rain forests, identification of sustainable-use forest management protocols that meet human needs while preserving biodiversity is critical. Reduced-impact logging (RIL) protocols are increasingly common in the tropics and may be a viable option for sustainable forest use; however, few studies have documented faunal responses to RIL. Moreover, evaluations of abundance or diversity may not be sufficient for a comprehensive understanding of faunal responses to human activities, especially in the short-term. We assessed the effects of RIL on the temporal activity patterns of abundant bats in lowland Amazonian rain forest. More specifically, we characterized temporal patterns of activity, overlap of temporal activity, and dispersion of activity modes for seven abundant bat species and for four common bat guilds in RIL forest and in undisturbed forest in Tapajós National Forest, Pará, Brazil. Temporal activity of aerial insectivores, nectarivores, and gleaning animalivores did not change in response to RIL. In contrast, three of five species of frugivores and frugivores as a group changed their patterns of temporal activity in response to RIL. RIL had a greater effect on temporal activity of frugivores that foraged in the understorey than on frugivores that foraged in multiple forest strata. Overlap of temporal activity of frugivores was greater than expected by chance in undisturbed forest, but was random in RIL forest. Changes in activity may be a response to a combination of increased predation risk and reduced distances of early evening flights between day roosts and feeding areas in RIL forest compared to undisturbed forest.     

Orangutan population density, forest structure and fruit availability in hand-logged and unlogged peat swamp forests in West Kalimantan, Indonesia

We investigated the population density of Bornean orangutans Pongo pygmaeus pygmaeus and aspects of habitat quality in a selectively hand-logged peat swamp forest in West Kalimantan, Indonesia, and in a comparable unlogged forest nearby. We conducted orangutan nest surveys, measured different parameters of forest structure, recorded monthly changes in fruit availability, and noted the sex and the stage of maturity of orangutans encountered. Nest density, an index of orangutan population density, was 21% lower in the logged area. The forest, logged 2 years previously, had fewer large food trees and a greater number of canopy gaps. We discuss these differences in relation to the lower orangutan nest density in the logged forest. Significantly fewer adult orangutans were observed in the logged study area. We hypothesize that fully adult orangutans, particularly females, are the most severely affected by hand-logging.     

The impacts of logging on landslide activity at Clayoquot Sound, British Columbia

The objective of this study was to investigate the impacts of logging on landslide activity in Clayoquot Sound on the west coast of Vancouver Island, British Columbia. A total of 1004 landslides were documented in order to test the hypothesis that areas affected by logging activities show different density, frequency and magnitude characteristics of landsliding than areas unaffected by logging. The frequency of landslides in logged terrain was found to be nine times higher than in undisturbed forest. An exponential increase in landslide frequency within the area logged was observed on a large watershed scale. Failures in logged terrain occur on gentler slopes than in natural terrain, partly because fewer slopes steeper than 40° have been logged. Debris slides and debris flows are the most frequently occurring mass movements, initiating mostly from road fill failures and from within cutblocks. Most landslides initiated on SE-facing slopes, with the highest landslide frequency near the coast and diminishing exponentially further inland. This pattern of activity most likely reflects the greater impact of severe storms arriving from the Pacific with winds dominating from the southeast. Concave and straight slopes are most susceptible to landslide initiation. A preliminary examination of geologic control on the frequency of landslides suggests that the Quatsino Formation, the Island Plutonic Suite and the Sicker Group are significantly above the average rate. Further work is needed to identify and quantify factors confounding the trends and correlations identified in this overview study. Results obtained in this study will be used to implement changes in forest management.     

The vanishing Brazilian Atlantic forest

The Brazilian Atlantic forest constitutes a unique biome characterised by a high species diversity and a high level of endemism. A long history of land clearing for crops, pastures, timber and firewood has resulted in over two centuries of widespread destruction. Today, less than 6·8% of the total area of the Atlantic forest of Minas Gerais has any form of forest cover and forests in primaeval state are not far from becoming extinct. The remaining forest patches are small, isolated and unprotected. In order to preserve some of the original diversity of the Atlantic forest, urgent efforts must be channelled toward protecting the remaining forest fragments.     

Effects of tree harvesting, forest floor removal, and compaction on soil microbial biomass, microbial respiration, and N availability in a boreal aspen forest in British Columbia

The effects of timber harvesting and the resultant soil disturbances (compaction and forest floor removal) on relative soil water content, microbial biomass C and N contents (C_mic and N_mic), microbial biomass C:N ratio (C_mic-to-N_mic), microbial respiration, metabolic quotient (qCO₂), and available N content in the forest floor and the uppermost mineral soil (0-3 cm) were assessed in a long-term soil productivity (LTSP) site and adjacent mature forest stands in northeastern British Columbia (Canada). A combination of principal component analysis and redundancy analysis was used to test the effects of stem-only harvest, whole tree harvest plus forest floor removal, and soil compaction on the studied variables. Those properties in the forest floor were not affected by timber harvesting or soil compaction. In the mineral soil, compaction increased soil total C and N contents, relative water content, and N_mic by 45%, 40%, 34% and 72%, respectively, and decreased C_mic-to-N_mic ratio by 29%. However, these parameters were not affected by stem only harvesting or whole tree harvesting plus forest floor removal, contrasting the reduction of white spruce and aspen growth following forest floor removal and soil compaction reported in an earlier study. Those results suggest that at the study site the short-term effects of timber harvesting, forest floor removal, and soil compaction are rather complex and that microbial populations might not be affected by the perturbations in the same way as trees, at least not in the short term.     

Ground beetles (Carabidae) and edge effect in oak-hornbeam forest and grassland transects

We evaluated ground beetle diversity in relation to forest edge between an oak-hornbeam forest and adjacent herbaceous grassland. To test our hypothesis that the diversity of ground beetles was higher in the forest edge than the interior, pitfall trap samples were taken along two forest-grassland transects in northern Hungary. The diversity of ground beetles was significantly higher at the forest edge and in the grassland than in the forest interior. Ground beetle assemblages in the forest interior, forest edge and grassland could be separated from each other by ordination. Indicator species analysis detected five groups of species: habitat generalists, grassland-associated species, forest generalists, forest specialists, and edge-associated species. Rank correlation indicated leaf litter, herb, canopy cover, and prey abundance as the most important factors influencing carabid diversity. The high diversity of the forest edge resulted from the presence of edge-associated species and of species characteristic of adjacent habitats. Forest edges seem to play an important role in maintaining diversity. Serving as source habitats, edges also contribute to the recolonisation by ground beetles after habitat destruction or other disturbance in the adjacent habitats.     

Effects of cropping and tree density on earthworm community composition and densities in central Cameroon

Earthworms can have positive effects upon crop growth in the tropics. If soils are to be managed sustainably, then more attention should be paid to the effects of cultivation and cropping practices upon earthworms. When forest vegetation is cleared, slashed, burned and land is tilled and cultivated, earthworm abundance, diversity and activity are reduced. Conversely, retaining trees in agroecosystems may maintain earthworm populations during the cropping phase.Here, we assessed the impact on earthworm species diversity and densities of crop cultivation in the understorey of timber plantations thinned to two tree densities and compared these with uncropped, undisturbed timber plantation controls. The plots were reassessed after two and a half years of fallow to see whether populations had recovered. The experiment was in central Cameroon.Seventeen earthworm species were recorded from Eudrilidae subfamilies Eudrilinae and Pareudrilinae, Ocnerodrilidae and Acanthodrilidae, most of which were endemics. This included two new species from two new genera from the sub-family Pareudrilinae, one new species from one new genus of Ocnerodrilidae, two new species of Dichogaster and one new species of Legonodrilus. Ten species were epigeic, six were endogeic and one was anecic.Generally, earthworm densities were lower in cropped plots than in the undisturbed plantation control. The most abundant species was a Legonodrilus sp. nov. with average densities of 49 individuals m⁻² in the crop phase and 80 ind. m⁻² in the fallow phase. By the fallow phase, densities in the low tree density (120 ind. m⁻²) were higher than in the high density (40 ind. m⁻²). The densities of the epigeic Acanthodrilidae were significantly reduced to 7 ind. m⁻² in the cropped plots compared with 42 ind. m⁻² in the control plots. The effects of cropping were thus species-specific and more work is required to identify which of these endemics are the ecosystem engineers in the system.