Functional diversity indices have been used to differentiate the relative contribution of stochastic and deterministic processes that modulate the assemblage of communities; however, knowledge regarding the relative contribution of assemblage mechanisms in forest communities is scarce. We analysed the assembly mechanisms driving forest assemblages along a topographic gradient at two spatial scales (1000 m2 and 3000 m2) for three different forest types from subtropical mountain forests (pine-oak, mixed pine-broadleaf and broadleaf forest) in western Mexico, using null models of multi-trait indices. The forest structure differed along the topographic gradient. Upper slopes were dominated by Pinus douglasiana with an importance value index (IVI) of 0.8, while 10 and 13 tree species were required in the middle and lower slopes, respectively, to reach the same IVI. The results support the idea that the subtropical montane forest of western Mexico is a mosaic of communities, when analysed at a scale of 1000 m2 the forest assembly was mainly explained by stochastic processes, while analysis at the scale of 3000 m2 showed that functional convergence of species were the main mechanisms of the assemblage of the pine-oak forest communities due to an abiotic stressful environment.

When one or several classes are much less prevalent than another class (unbalanced data), class error rates and variable importances of the machine learning algorithm random forest can be biased, particularly when sample sizes are smaller, imbalance levels higher, and effect sizes of important variables smaller. Using simulated data varying in size, imbalance level, number of true variables, their effect sizes, and the strength of multicollinearity between covariates, we evaluated how eight versions of random forest ranked and selected true variables out of a large number of covariates despite class imbalance. The version that calculated variable importance based on the area under the curve (AUC) was least adversely affected by class imbalance. For the same number of true variables, effect sizes, and multicollinearity between covariates, the AUC variable importance ranked true variables still highly at the lower sample sizes and higher imbalance levels at which the other seven versions no longer achieved high ranks for true variables. Conversely, using the Hellinger distance to split trees or downsampling the majority class already ranked true variables lower and more variably at the larger sample sizes and lower imbalance levels at which the other algorithms still ranked true variables highly. In variable selection, a higher proportion of true variables were identified when covariates were ranked by AUC importances and the proportion increased further when the AUC was used as the criterion in forward variable selection. In three case studies, known species–habitat relationships and their spatial scales were identified despite unbalanced data.

Sagebrush control has historically involved mechanical and chemical (2,4-D) treatments designed for total control and which detrimentally affect nontarget plant species, resulting in decreases in native wildlife abundance and diversity. Tebuthiuron, a potentially selective herbicide with thinning capabilities, was applied at various rates to big sagebrush plots near Ten Sleep, WY in 1979 and Hyatteville, WY in 1983. Plant and small mammal communities were evaluated in 1992 and 1993. Big sagebrush cover in untreated areas was 31 ± 1% in 1992 and 34 ± 1% in 1993 at Ten Sleep, and 44 ± 4% at Hyatteville (α = 0.10). Big sagebrush control increased as application rate increased. Productivity of herbaceous species (graminoids, forbs, or both) tended to be greatest at about 11 to 17% big sagebrush cover. In general, plant community diversity tended to be greatest where sagebrush was thinned to this level. Small mammal community diversity was least at both sites where big sagebrush cover was less than 5%, and was strongly correlated with plant community diversity at Ten Sleep (R2W = 0.99; 4 degrees of freedom). In addition, greater abundance of or better habitat quality for endemic, stenotypic (habitat specialist) species was associated with about 15% big sagebrush cover. This suggests (albeit retroductively) that biodiversity at both local (alpha) and global (epsilon) scales might be conserved by thinning big sagebrush with tebuthiuron.

We revisited the role of dissimilarity of host assemblages in shaping dissimilarity of flea assemblages using a non-linear approach. Generalized dissimilarity models (GDMs) were applied using data from regional surveys of fleas parasitic on small mammals in four biogeographical realms. We compared (1) model fit, (2) the relative effects of host compositional and phylogenetic turnover and geographic distance on flea compositional and phylogenetic turnover, and (3) the rate of flea turnover along gradients of host turnover and geographic distance with those from earlier application of a linear approach. GDMs outperformed linear models in explaining variation in flea species turnover and host dissimilarity was the best predictor of flea dissimilarity, irrespective of scale. The shape of the relationships between flea compositional turnovers along host compositional turnover was similar in all realms, whereas turnover along geographic distance differed among realms. In contrast, the rate of flea phylogenetic turnover along gradients of host phylogenetic turnover differed among realms, whereas flea phylogenetic turnover did not depend on geographic distance in any realm. We demonstrated that a non-linear approach (a) explained spatial variation in parasite community composition better than and (b) revealed patterns that were obscured by earlier linear analyses.

Effective conservation of global species diversity requires a clear understanding of spatial scales that support overall diversity across broad scales. Abandonment of semi-natural grasslands has increased their fragmentation and decreased their areal extent. We quantified diversity patterns of plant communities in Japan across hierarchical scales to facilitate the development of an effective nationwide strategy for conserving species diversity in remnant semi-natural grasslands. We applied additive partitioning of plant species diversity, using a nested hierarchical design at three spatial scales (quadrat, grassland, and western and eastern regions of Japan) for three groups of plant species (all species, grassland species and national Red Listed species). We consistently found lower proportions of among-quadrats diversity, and higher proportions of among-grasslands diversity and between-regions diversity in the overall diversity of the entire species complement than would be expected by chance. The high contribution of among-grasslands diversity to overall diversity suggests that each grassland had a unique species content. The second-ranking contributor to overall diversity differed between grassland species and Red Listed species: the second-ranking contributor for grassland species was diversity at the among-quadrats scale but the second-ranking contributor for all species and for Red Listed species was diversity at the between-regions scale. Thus, effective conservation of diversity of the entire species complement in remnant semi-natural grasslands requires preservation of beta diversity in individual grasslands. Our findings highlight the importance of strengthening local preservation and restoration activities within each grassland, and of nationwide strategies for conserving Red Listed species in remnant semi-natural grassland communities.

Plant invasions can have large effects on ecosystem services. Some plant invaders were introduced specifically to restore key services to ecosystems, and other invaders are having unintended, detrimental effects on services, such as the quantity and quality of water delivered, flood control, erosion control, and food production. Many ecosystem services are difficult to measure directly, and although there are extensive studies on plant invaders and ecosystem processes, a number of challenges prevent us from confidently extrapolating those processes as proxies for services. To extrapolate local, short-term measures of processes to ecosystem services, we must: (1) determine which processes are the key contributors to a service, (2) assess how multiple processes interact to provide a given service, (3) determine how vegetation types and species affect those processes, and (4) explicitly assess how ecosystem services and their controls vary over space and time, including reliance of ecosystem services on “hot spots” and “hot moments” and a minimum size of a vegetation type in the landscape. A given invader can have positive effects on some services and negative effects on others. It is important to consider that, in some systems, shifting environmental conditions may no longer support native species and that invasive species may be critical contributors to the resilience of ecosystem services.

Examining how variability in population abundance and distribution is allotted among different spatial scales can inform of processes that are likely to generate that variability. Results of studies dealing with scale issues in marine benthic communities suggest that variability is concentrated at small spatial scales (from tens of centimetres to few metres) and that spatial patterns of variation are consistent across ecosystems characterized by contrasting physical and biotic conditions, but this has not been formally tested. Here we quantified the variability in the distribution of intertidal rocky shore communities at a range of spatial scales, from tens of centimetres to thousands of kilometres, both in the NE Atlantic and the Mediterranean, and tested whether the observed patterns differed between the two basins. We focused on canopy-forming macroalgae and associated understorey assemblages in the low intertidal, and on the distribution of Patella limpets at mid intertidal levels. Our results highlight that patterns of spatial variation, at each scale investigated, were consistent between the Atlantic and the Mediterranean, suggesting that similar ecological processes operate in these regions. In contrast with former studies, variability in canopy cover, species richness and limpet abundance was equally distributed among spatial scales, possibly reflecting the fingerprint of multiple processes. Variability in community structure of low intertidal assemblages, instead, peaked at the largest scale, suggesting that oceanographic processes and climatic gradients may be important. We conclude that formal comparisons of variability across scales nested in contrasting systems are needed, before any generalization on patterns and processes can be made.

Although seed dispersal plays a critical role for plant regeneration, the long-term benefits of frugivores for recruitment success have been poorly explored. We evaluated the relative importance of tree species abundance and of frugivore-related factors for successful plant recruitment on 18 tree species in a lowland Colombian rain forest. We combined census data from four 1-ha plots of trees (>10 cm dbh), saplings (1–5 cm dbh) and seedlings (<1.7 m) with a dataset describing tree–frugivore interactions. Seedling abundance was higher for large-seeded species dispersed by the spider monkey, while sapling abundance was higher for large-seeded species dispersed by birds. The identity of the dispersal agent and its interaction with seed size explained 20–30% of the total variance in seedling and sapling abundance across scales. Seed size consistently influenced the species abundance of seedlings and saplings across scales, but in opposite ways. These developmental changes suggest that what is beneficial to seedlings is not necessarily beneficial to saplings. Species identity explained 10–50% of the total variance in seedling and sapling abundance among and within 1-ha plots. Overall, our findings suggest that recruitment success is context-dependent as the relative importance of the different variables addressed may shift along spatial and temporal scales.

Positive plant interactions have strong effects on plant diversity at several spatial scales, expanding species distribution under stressful conditions. We evaluated the joint effect of climate and grazing on the nurse effect of Croton wagneri, by monitoring several community attributes at two spatial scales: microhabitat and plant community. Two very close locations that only differed in grazing intensity were surveyed in an Ecuadorian dry scrub ecosystem. At each location, two 30 × 30-m plots were established at four altitudinal levels (1500, 2630, 1959 and 2100 m asl) and 40 microsites were surveyed in each plot. Croton wagneri acted as community hubs, increasing species richness and plant cover at both scales. Beneath nurses mean richness and cover values were 3.4 and 21.9%, and in open areas 2.3 and 4.5%, respectively. Magnitude of nurse effect was dependent on climate and grazing conditions. In ungrazed locations, cover increased and diversity reduced with altitude, while grazed locations showed the opposite trend. In ungrazed plots the interactions shifted from positive to negative with altitude, in grazed locations interactions remained positive. We conclude that the nurse effect is a key mechanism regulating community properties not only at microsite but also at the entire community scale.

We tested the hypothesis that compositional and/or phylogenetic dissimilarity of host assemblages affect compositional and/or phylogenetic dissimilarity of parasite assemblages, to different extents depending on scale, using regional surveys of fleas parasitic on small mammals from 4 biogeographical realms. Using phylogenetic community dissimilarity metric, we calculated the compositional and phylogenetic dissimilarity components between all pairs of host and parasite communities within realms and hemispheres. We then quantified the effect of compositional or phylogenetic dissimilarity in host regional assemblages, and geographical distance between assemblages, on the compositional or phylogenetic dissimilarity of flea regional assemblages within a realm, respectively. The compositional dissimilarity in host assemblages strongly affected compositional dissimilarity in flea assemblages within all realms and within both hemispheres. However, the effect of phylogenetic dissimilarity of host assemblages on that of flea assemblages was mostly confined to the Neotropics and Nearctic, but was detected in both the Old and New World at the higher scale, possibly because of phylogenetic heterogeneity in flea and host faunas between realms. The clearer effect of the compositional rather than the phylogenetic component of host community dissimilarity on flea community dissimilarity suggests important roles for host switching and ecological fitting during the assembly history of flea communities.

Litter-feeding termites influence key aspects of the structure and functioning of semi-arid ecosystems around the world by altering nutrient and material fluxes, affecting primary production, foodweb dynamics and modifying vegetation composition. Understanding these complex effects depends on quantifying spatial heterogeneity in termite foraging activities, yet such information is scarce for semi-arid savannas. Here, the amount of litter that was removed from 800 litterbags in eight plots (100 litterbags per plot) was measured in Hluhluwe–iMfolozi Park (HiP) South Africa. These data were used to quantify variation in litter removal at two spatial scales: the local scale (within 450-m2 plots) and the landscape scale (among sites separated by 8–25 km). Subsequently, we attempted to understand the possible determinants of termites’ foraging patterns by testing various ecological correlates, such as plant biomass and bare ground at small scales and rainfall and fences that excluded large mammalian herbivores at larger scales. No strong predictors for heterogeneity in termite foraging intensity were found at the local scale. At the landscape scale termite consumption depended on an interaction between rainfall and the presence of large mammalian herbivores: litter removal by termites was greater in the presence of large herbivores at the drier sites but lower in the presence of large herbivores at the wetter sites. The effect of herbivores on termite foraging intensity may indicate a switch between termites and large herbivore facilitation and competition across a productivity gradient. In general, litter removal decreased with increasing mean annual rainfall, which is in contrast to current understanding of termite consumption across rainfall and productivity gradients. These results generate novel insights into termite ecology and interactions among consumers of vastly different body sizes across spatial scales.

We investigated spatial patterns of stony corals, molluscs and echinoderms among six habitat types within the lagoons of six atolls in the Tuamotu Archipelago (French Polynesia). Percent cover (stony corals only), abundance and richness were recorded within habitat types delineated by specific geomorphological features. Among six habitat types defined a priori, four encompassed distinct assemblages of corals and molluscs: (1) pinnacles possessed maximum richness, abundance and coverage, (2) passes showed relatively high abundance and coverage, (3) lagoon floors were marked by the commonness of a few coral genera, and minimum mollusc abundance and richness, and (4) inner reef flat stations, independent of their location in front of spillways, motu or rim zones, had low cover, abundance, and richness. While habitat associations were similar for stony corals and molluscs, echinoderms were preferentially found on the inner reef flat facing the rim. Lowest echinoderm abundances and species richness were recorded on pinnacles and in the vicinity of the pass. Such small-scale spatial heterogeneity appears to be one of the distinctive characteristics of coral communities in French Polynesian reefs. These results, along with those of the TYPATOLL program, demonstrate the importance of both local and regional factors in determining the diversity and structure of coral assemblages. Habitat fidelity was weak: many species/genera were poorly represented among habitats and restricted to a few atolls, while others were present across all lagoons and habitats. Therefore, marine protected areas design based solely on representativeness will require the protecting the vast majority of lagoon habitats.

Knowledge of the ecological and evolutionary causes of dispersal can be crucial in understanding the behaviour of spatially structured populations, and predicting how species respond to environmental change. Despite the focus of much theoretical research, simplistic assumptions regarding the dispersal process are still made. Dispersal is usually regarded as an unconditional process although in many cases fitness gains of dispersal are dependent on environmental factors and individual state. Condition-dependent dispersal strategies will often be superior to unconditional, fixed strategies. In addition, dispersal is often collapsed into a single parameter, despite it being a process composed of three interdependent stages: emigration, inter-patch movement and immigration, each of which may display different condition dependencies. Empirical studies have investigated correlates of these stages, emigration in particular, providing evidence for the prevalence of conditional dispersal strategies. Ill-defined use of the term ‘dispersal’, for movement across many different spatial scales, further hinders making general conclusions and relating movement correlates to consequences at the population level. Logistical difficulties preclude a detailed study of dispersal for many species, however incorporating unrealistic dispersal assumptions in spatial population models may yield inaccurate and costly predictions. Further studies are necessary to explore the importance of incorporating specific condition-dependent dispersal strategies for evolutionary and population dynamic predictions.

Relating patterns of species abundance to the processes that determine them at local and regional spatial scales is one of the central aims of ecological research. To examine the relationship between distribution and growth in tropical rain forest, seasonal and annual changes in tree diameter of two tree species with contrasting distribution patterns, Celtis mildbraedii and Strombosia glaucescens, were monitored across topographic gradients in moist semi-deciduous and moist evergreen forest in Ghana over 2 years, 1997-9. Concurrent measurements were made of soil water availability, and during 1997/8, of rainfall, nutrient availability and photosynthetically active radiation (PAR). The two sites received similar, low, total rainfall during 1997/8, but soil water availability in the dry season at the evergreen site exceeded that at the semi-deciduous forest site. Soil water availability was higher during 1998/9 at both sites. During 1997/8, PAR was similar at the two sites. The evergreen site had soils of lower pH, available P, Mg, Ca and K concentrations than the semi-deciduous forest site. Water availability and the sand content of the soil increased and concentrations of total N and C decreased, from summit to valley positions at both sites. Celtis mildbraedii had significantly faster growth rates than S. glaucescens, and growth of both species was greater at the semi-deciduous than the evergreen forest site during the wet year (1998/9) but not during the dry year (1997/8). Celtis mildbraedii in semi-deciduous forest grew less in valley than summit and slope positions. We conclude that in the absence of a severe dry season, growth is higher in semi-deciduous than evergreen forest, and is related to the higher soil fertility in more seasonal forest. The patterns are consistent with the hypothesis that concentrations of available P and/or cations in the soil limit growth in evergreen forest, while concentrations of N in valley soils limit growth of C. mildbraedii in semi-deciduous forest. There was no evidence that variation in PAR influenced growth rates in these forests. A reduction in growth rate on less fertile soils may be a factor determining the distribution limit of dry forest species, such as C. mildbraedii, in evergreen forest.

The opportunity for selection, I, calculated as the variance in relative fitness, sets an upper limit to the amount of adaptive change that selection may produce. Therefore, it is a potentially valuable, and frequently used, measure of the potential of action of phenotypic selection. Although many different aspects of I calculation and analysis have been explored, the effect of the spatial scale chosen for calculation received little attention, notwithstanding the growing evidence that natural populations are not homogeneous and present a hierarchical spatial structure. The effect of scale on the estimation of I was examined from data collected in two populations of southern elephant seals (Mirounga leonina), an easily observable and strongly polygynous species. A significant effect of spatial scale on three important aspects of I calculation and analysis was found: dependence of I on mean fitness, between population variation of I, and effect of local demography on I.

In a 3-year study, we examined landscape features (aspect, slope, sun exposure, canopy cover, type of ground cover, and nearest water source) that were potentially related to prevalence of infection with Plasmodium mexicanum in fence lizards (Sceloporus occidentalis) within a 4.5 ha study area in northern California, USA. Logistic regression analysis showed that ground cover type was the primary mediator of the probability of P. mexicanum infection. Infected lizards were captured more often in rock and/or leaf litter locations than in grassy ones. In another experiment, the study area was divided into 9 sites (0.07–0.33 ha), and infection prevalence was calculated for each. Three sites with high (>30%) infection prevalence had significantly more rocky outcrops and leaf litter than those with low (<20%) or moderate (20–30%) infection prevalence (N = 3 sites each). We conclude that lizard site selection may influence the probability of exposure to infected vectors and thus the likelihood of P. mexicanum infection. We also demonstrate that studies at different spatial scales may be required to understand fully the relationship between landscape features and parasite distribution.