Timing the bloom season: a novel approach to evaluating reproductive phenology across distinct regional flora

Context

Just as the timing of the vegetative growing season affects a host of ecological processes, the seasonality of floral availability impacts ecological processes from nectar availability and allergen production to competition for pollinator attention. However, no existing methodology is capable of evaluating multi-species bloom phenology in a standardized fashion across multiple ecosystems or compositionally distinct local flora. Thus, the manner in which the onset of the bloom season (during which the majority of species flower) differs along climate gradients and among distinct local flora remains largely unknown.

Objectives

This study evaluates differences in the timing of the bloom season throughout the western United States, and the relationship of the bloom season to the vegetative growing season and to local climate conditions.

Methods

This study estimated the season during which all but the earliest and latest 5 % of local species flower (the bloom season) using digital herbarium records. Bloom season timing was compared to land surface phenology, SI-x phenoclimate metrics, and PRISM climate normals.

Results

Local differences in mean temperature of the coldest month explained 76 % of observed variation in bloom season onset. Variation in land surface phenology explained 50 % of observed variation, while SI-x Bloom estimates explained 64 % of observed variation in bloom season onset.

Conclusions

These results confirm that bloom season phenology is distinct from the vegetative growing season, and that local temperature is a good predictor of bloom season onset. This work represents a new modality for studying multi-taxa flowering phenology at landscape and regional scales.

     

The use of native vegetation as a proxy for habitat may overestimate habitat availability in fragmented landscapes

Context

Native vegetation is often used as a proxy for habitat to estimate habitat availability in landscapes. This approach may lead to incorrect estimates of the impacts of habitat loss and fragmentation on species, which have not been thoroughly quantified so far.

Objectives

We quantified to what extent the loss of native vegetation reflect actual habitat loss by native species in landscapes. We tested the hypothesis that habitat availability declines at greater rates than native vegetation and thus is overestimated when it is quantified on the basis of native vegetation.

Methods

Using simulations, we quantified how the loss of native vegetation in artificial and real landscapes affects habitat availability for species with different habitat requirements. We contrasted a generalist species, which uses all native vegetation, with 10 habitat-specialist species classified into three categories (interior, patchy and riparian species).

Results

Habitat availability generally declined at greater rates than native vegetation for all specialist species. This pattern was apparent for different specialist species in a broad range of landscape types. Interior species always lost habitat availability more rapidly than the generalist species. Most riparian species lost habitat availability more rapidly than the generalist species. Responses of patchy species were more complex, depending on their dispersal abilities and landscape structure.

Conclusions

Habitat availability is likely to be overestimated when native vegetation is used as proxy for habitat, because habitat availability will generally decline at greater rates than native vegetation. Therefore, a species-centered approach should be adopted when estimating habitat availability in landscapes.

     

Drivers and trends in landscape patterns of stand-replacing fire in forests of the US Northern Rocky Mountains (1984–2010)

Context

Resilience in fire-prone forests is strongly affected by landscape burn-severity patterns, in part by governing propagule availability around stand-replacing patches in which all or most vegetation is killed. However, little is known about drivers of landscape patterns of stand-replacing fire, or whether such patterns are changing during an era of increased wildfire activity.

Objectives

(a) Identify key direct/indirect drivers of landscape patterns of stand-replacing fire (e.g., size, shape of patches), (b) test for temporal trends in these patterns, and (c) anticipate thresholds beyond which landscape patterns of burn severity may change fundamentally.

Methods

We applied structural equation modeling to satellite burn-severity maps of fires in the US Northern Rocky Mountains (1984–2010) to test for direct and indirect (via influence on fire size and proportion stand-replacing) effects of climate/weather, vegetation, and topography on landscape patterns of stand-replacing fire. We also tested for temporal trends in landscape patterns.

Results

Landscape patterns of stand-replacing fire were strongly controlled by fire size and proportion stand-replacing, which were, in turn, controlled by climate/weather and vegetation/topography, respectively. From 1984 to 2010, the proportion of stand-replacing fire within burn perimeters increased from 0.22 to 0.27. Trends for other landscape metrics were not significant, but may respond to further increases proportion stand-replacing fire.

Conclusions

Fires from 1984 to 2010 exhibited tremendous heterogeneity in landscape patterns of stand-replacing fire, likely promoting resilience in burned areas. If trends continue on the current trajectory, however, fires may produce larger and simpler shaped patches of stand-replacing fire with more burned area far from seed sources.

     

Spatio-temporal variation in foodscapes modifies deer browsing impact on vegetation

Context

Ungulate browsers often alter plant composition and reduce diversity in forests worldwide, yet our ability to predict browse impact on vegetation remains equivocal. Theory suggests, however, that ungulate distribution and foraging impacts are shaped by scale-dependent decisions based on variation in habitat composition and structure encountered within their home range.

Objective

Examine how variation in habitat composition at landscape (259 ha) scales modulates browse impact on vegetation at local scales.

Methods

We measured vegetation richness and abundance in plots with and without white-tailed deer (Odocoileus virginianus) at 23 northern hardwood forest sites distributed across a 6500 km² area in Pennsylvania, USA. Experimental sites were embedded within landscapes with varying levels of habitat composition and deer densities.

Results

Browsing reduced vegetation richness and cover by as much as 53 and 70%, respectively; however, we found browse impact was modulated by variation in the relative abundance of managed habitats that alter forage availability. Specifically, relative to fenced areas, browse impact weakened and ultimately disappeared as the proportion of forage-rich habitats (e.g., recent harvests) increased to ≥20%. Conversely, vegetation grew increasingly depauperate as landscapes contained greater proportions of forage-poor habitats (i.e., older harvests), particularly when browsed.

Conclusions

Our results underscore how management actions that alter forage availability to ungulates throughout the landscape (i.e. the foodscape) can shape forest-ungulate interactions and suggest a new paradigm whereby managers evaluate and undertake actions at the appropriate spatio-temporal scales to proactively limit the deleterious impact of browsing on plant biodiversity.

     

Integrating models across temporal and spatial scales to simulate landscape patterns and dynamics in mountain pasture-woodlands

Context

Pasture-woodlands are semi-natural landscapes that result from the combined influences of climate, management, and intrinsic vegetation dynamics. These landscapes are sensitive to future changes in land use and climate, but our ability to predict the impact on ecosystem service provisioning is limited due to the disparate scales in time and space that govern their dynamics.

Objectives

To develop a process-based model to simulate pasture-woodland landscapes and the provisioning of ecosystem services (i.e., livestock forage, woody biomass and landscape heterogeneity).

Methods

We modified a dynamic forest landscape model to simulate pasture-woodland landscapes in Switzerland. This involved including an annual herbaceous layer, selective grazing from cattle, and interactions between grazing and tree recruitment. Results were evaluated within a particular pasture, and then the model was used to simulate regional vegetation patterns and livestock suitability for a ~198,000 ha landscape in the Jura Vaudois region.

Results

The proportion of vegetation cover types at the pasture level (i.e., open, semi-open and closed forests) was well represented, but the spatial distribution of trees was only broadly similar. The entire Jura Vaudois region was simulated to be highly suitable for livestock, with only a small proportion being unsuitable due to steep slopes and high tree cover. High and low elevation pastures were equally suitable for livestock, as lower forage production at higher elevations was compensated by reduced tree cover.

Conclusions

The modified model is valuable for assessing landscape to regional patterns in vegetation and livestock, and offers a platform to evaluate how climate and management impact ecosystem services.

     

Spatial heterogeneity of subsurface soil texture drives landscape-scale patterns of woody patches in a subtropical savanna

Context

In the Rio Grande Plains of southern Texas, subtropical savanna vegetation is characterized by a two-phase pattern consisting of discrete woody patches embedded within a C₄ grassland matrix. Prior trench transect studies have suggested that, on upland portions of the landscape, large woody patches (groves) occur on non-argillic inclusions, while small woody patches (clusters) are dispersed among herbaceous vegetation where the argillic horizon is present.

Objective

To test whether spatial heterogeneity of subsurface soil texture drives the landscape-scale pattern of woody patches in this subtropical savanna.

Methods

Landscape-scale spatial patterns of soil texture were quantified by taking spatially-specific soil samples to a depth of 1.2 m in a 160 m × 100 m plot. Kriged maps of soil texture were developed, and the locations of non-argillic inclusions were mapped.

Results

Visual comparison of kriged maps of soil texture to a high resolution aerial photograph of the study area revealed that groves were present exclusively where the non-argillic inclusions were present. This clear visual relationship was further supported by positive correlations between soil sand concentration in the lower soil layers and total fine root biomass which mapped the locations of groves.

Conclusions

Subsurface non-argillic inclusions may favor the establishment and persistence of groves by enabling root penetration deeper into the profile, providing greater access to water and nutrients that are less accessible on those portions of the landscape where the argillic horizon is present, thereby regulating the distribution of grove vegetation and structuring the evolution of this landscape.

     

Getting to the root of the matter: landscape implications of plant-fungal interactions for tree migration in Alaska

Context

Forecasting the expansion of forest into Alaska tundra is critical to predicting regional ecosystem services, including climate feedbacks such as carbon storage. Controls over seedling establishment govern forest development and migration potential. Ectomycorrhizal fungi (EMF), obligate symbionts of all Alaskan tree species, are particularly important to seedling establishment, yet their significance to landscape vegetation change is largely unknown.

Objective

We used ALFRESCO, a landscape model of wildfire and vegetation dynamics, to explore whether EMF inoculum potential influences patterns of tundra afforestation and associated flammability.

Methods

Using two downscaled CMIP3 general circulation models (ECHAM5 and CCCMA) and a mid-range emissions scenario (A1B) at a 1 km² resolution, we compared simulated tundra afforestation rates and flammability from four parameterizations of EMF effects on seedling establishment and growth from 2000 to 2100.

Results

Modeling predicted an 8.8–18.2 % increase in forest cover from 2000 to 2100. Simulations that explicitly represented landscape variability in EMF inoculum potential showed a reduced percent change afforestation of up to a 2.8 % due to low inoculum potential limiting seedling growth. This reduction limited fuel availability and thus, cumulative area burned. Regardless of inclusion of EMF effects in simulations, landscape flammability was lower for simulations driven by the wetter and cooler CCCMA model than the warmer and drier ECHAM5 model, while tundra afforestation was greater.

Conclusions

Results suggest abiotic factors are the primary driver of tree migration. Simulations including EMF effects, a biotic factor, yielded more conservative estimates of land cover change across Alaska that better-matched empirical estimates from the previous century.

     

Holding the line: three decades of prescribed fires halt but do not reverse woody encroachment in grasslands

Context

Encroachment of woody vegetation represents a significant global threat to biodiversity in grasslands, but practices used to reverse encroachment are rarely evaluated comprehensively. Several factors may drive encroachment, such as land use history, alteration of disturbance regimes, and local environment, but their relative importance is poorly understood. Another complicating factor is that encroachment may proceed via positive feedbacks that result in thresholds, beyond which its reversal is difficult.

Objectives

We ask what impact reintroducing frequent fire has on encroachment relative to the influences of landscape context and historical vegetation. We investigate whether woody cover frequency distributions suggest that feedbacks reinforce encroachment after a threshold of woody cover is surpassed.

Methods

We analyze aerial photos in glade grasslands in Missouri, USA, to assess encroachment patterns over a 75-year period. Fire was excluded from this landscape for the first 45 years, and then reintroduced at varying frequencies in the last 30 years.

Results

Woody vegetation cover increased sevenfold from 1939 to 2014 overall. After the reintroduction of prescribed fire, woody cover stayed approximately constant in burned glades, but continued increasing in unburned glades. Woody cover followed bimodal frequency distributions in burned areas. Fire-tolerant vegetation tended to encroach near historically wooded areas, while fire-sensitive vegetation responded more to fire history.

Conclusions

Altered disturbance regimes, in addition to numerous recognized drivers, can cause ecosystem state changes associated with losses to biodiversity. Conducting management early in the encroachment process and restoring grasslands at broad landscape scales may help counteract local feedbacks that promote encroachment.

     

Living on the edge: utilising lidar data to assess the importance of vegetation structure for avian diversity in fragmented woodlands and their edges

Context

In agricultural landscapes, small woodland patches can be important wildlife refuges. Their value in maintaining biodiversity may, however, be compromised by isolation, and so knowledge about the role of habitat structure is vital to understand the drivers of diversity. This study examined how avian diversity and abundance were related to habitat structure in four small woods in an agricultural landscape in eastern England.

Objectives

The aims were to examine the edge effect on bird diversity and abundance, and the contributory role of vegetation structure. Specifically: what is the role of vegetation structure on edge effects, and which edge structures support the greatest bird diversity?

Methods

Annual breeding bird census data for 28 species were combined with airborne lidar data in linear mixed models fitted separately at (i) the whole wood level, and (ii) for the woodland edges only.

Results

Despite relatively small woodland areas (4.9–9.4 ha), bird diversity increased significantly towards the edges, being driven in part by vegetation structure. At the whole woods level, diversity was positively associated with increased vegetation above 0.5 m and especially with increasing vegetation density in the understorey layer, which was more abundant at the woodland edges. Diversity along the edges was largely driven by the density of vegetation below 4 m.

Conclusions

The results demonstrate that bird diversity was maximised by a diverse vegetation structure across the wood and especially a dense understorey along the edge. These findings can assist bird conservation by guiding habitat management of remaining woodland patches.

     

The influence of matrix quality on species richness in remnant forest

Context

Habitat destruction is the leading threat to terrestrial biodiversity, isolating remnant habitat in a matrix of modified vegetation.

Objectives

Our goal was to determine how species richness in several broad taxonomic groups from remnant forest was influenced by matrix quality, which we characterized by comparing plant biomass in forest and the surrounding matrix.

Methods

We coupled data on species-area relationships (SARs) in forest remnants from 45 previously published studies with an index of matrix quality calculated using new estimates of plant biomass derived from satellite imagery.

Results

The effect size of SARs was greatest in landscapes with low matrix quality and little forest cover. SARs were generally stronger for volant than for non-volant species. For the terrestrial taxa included in our analysis, matrix quality decreased as the proportion of water, ice, or urbanization in a landscape increased.

Conclusions

We clearly demonstrate that matrix quality plays a major role in determining patterns of species richness in remnant forest. A key implication of our work is that activities that increase matrix quality, such as active and passive habitat restoration, may be important conservation measure for maintaining and restoring biodiversity in modified landscapes.

     

Landscape history,time lags and drivers of change: urban natural grassland remnants in Potchefstroom,South Africa

Context

The history of the landscape directly affects biotic assemblages, resulting in time lags in species response to disturbances. In highly fragmented environments, this phenomenon often causes extinction debts. However, few studies have been carried out in urban settings.

Objectives

To determine if there are time lags in the response of temperate natural grasslands to urbanization. Does it differ for indigenous species and for species indicative of disturbance and between woody and open grasslands? Do these time lags change over time? What are the potential landscape factors driving these changes? What are the corresponding vegetation changes?

Methods

In 1995 and 2012 vegetation sampling was carried out in 43 urban grassland sites. We calculated six urbanization and landscape measures in a 500 m buffer area surrounding each site for 1938, 1961, 1970, 1994, 1999, 2006, and 2010. We used generalized linear models and model selection to determine which time period best predicted the contemporary species richness patterns.

Results

Woody grasslands showed time lags of 20–40 years. Contemporary open grassland communities were, generally, associated with more contemporary landscapes. Altitude and road network density of natural areas were the most frequent predictors of species richness. The importance of the predictors changed between the different models. Species richness, specifically, indigenous herbaceous species, declined from 1995 to 2012.

Conclusions

The history of urbanization affects contemporary urban vegetation assemblages. This indicates potential extinction debts, which have important consequences for biodiversity conservation planning and sustainable future scenarios.

     

Grain-dependent functional responses in habitat selection

Context

Spatial scale is an important consideration for understanding how animals select habitat, and multi-scalar designs in resource selection studies have become increasingly common. Despite this, examination of functional responses in habitat selection at multiple scales is rare. The perceptual range of an animal changes as a function of vegetation association, suggesting that use, selection and functional responses may all be habitat- and scale-dependent.

Objectives

Our objective was to determine how varying grain size affects our interpretation of functional response in habitat selection and to elucidate scalar and landscape effects on habitat selection.

Methods

We quantified the functional response of GPS-collared, female white-tailed deer (Odocoileus virginianus, n = 18) in Riding Mountain National Park, Canada, to different habitat types. Functional responses were quantified at multiple spatial scales by regressing proportion of habitat used against proportion of habitat available at different buffer radii (ranging from 75–1000 m radius) surrounding used (telemetry) locations and available points within the individual’s seasonal home range. We examined how functional responses changed as a function of grain by plotting grain size against the slope of the functional response.

Results

We detected functional responses in most habitat types. As expected, functional responses tended to converge towards 1 (use proportional to availability) at large buffer sizes; however, the relationship between scale and functional response was typically non-linear and depended on habitat type.

Conclusions

We conclude that a multi-scalar approach to modelling animal functional responses in habitat selection is important for understanding patterns in animal behaviour and resource use.

     

Spatial dynamics of coastal forest bird assemblages: the influence of landscape context,forest type,and structural connectivity

Context

Complex structural connectivity patterns can influence the distribution of animals in coastal landscapes, particularly those with relatively large home ranges, such as birds. To understand the nuanced nature of coastal forest avifauna, where there may be considerable overlap in assemblages of adjacent forest types, the concerted influence of regional landscape context and vegetative structural connectivity at multiple spatial scales warrants investigation.

Objectives

This study determined whether species compositions of coastal forest bird assemblages differ with regional landscape context or with forest type, and if this is influenced by structural connectivity patterns measured at multiple spatial scales.

Methods

Three replicate bird surveys were conducted in four coastal forest types at ten survey locations across two regional landscape contexts in northeast Australia. Structural connectivity patterns of 11 vegetation types were quantified at 3, 6, and 12 km spatial scales surrounding each survey location, and differences in bird species composition were evaluated using multivariate ordination analysis.

Results

Bird assemblages differed between regional landscape contexts and most coastal forest types, although Melaleuca woodland bird assemblages were similar to those of eucalypt woodlands and rainforests. Structural connectivity was primarily correlated with differences in bird species composition between regional landscape contexts, and correlation depended on vegetation type and spatial scale.

Conclusions

Spatial scale, landscape context, and structural connectivity have a combined influence on bird species composition. This suggests that effective management of coastal landscapes requires a holistic strategy that considers the size, shape, and configuration of all vegetative components at multiple spatial scales.

     

How landscape,pollen intake and pollen quality affect colony growth in <Emphasis Type="Italic">Bombus terrestris</Emphasis>

Context

Abundance and diversity of bumblebees have been declining over the past decades. To successfully conserve bumblebee populations, we need to understand how landscape characteristics affect the quantity and quality of floral resources collected by colonies and subsequently colony performance.

Objectives

We therefore investigated how amount and composition of pollen collected by buff-tailed bumblebee Bombus terrestris colonies was affected by the surrounding landscape (i.e. the proportion of forest, urban, semi-natural habitats) and how they were related to colony growth.

Methods

Thirty B. terrestris colonies were placed at grassland sites differing in surrounding landscape. Colonies were established in spring when availability of flowering plants was highest, and their weight gain was monitored for 1 month. We additionally recorded the quantity and compared plant taxonomic composition and nutritional quality (i.e. amino acid composition) of pollen stored.

Results

Bumblebee colonies varied little in the pollen spectra collected despite differences in surrounding landscape composition. They collected on average 80 % of pollen from woody plants, with 34 % belonging to the genus Acer. Early colony growth positively correlated with total amount of woody pollen and protein collected and decreased with increasing proportions of semi-natural habitats and total amino acid concentrations.

Conclusions

Our results suggest that woody plant species represent highly important pollen sources for the generalist forager B. terrestris early in the season. We further show that colony growth of B. terrestris is predominantly affected by the quantity, not quality, of forage, indicating that several abundant plant species flowering throughout the bumblebees’ foraging season may cover the colonies’ nutritional needs.

     

Using remote-sensing data to assess habitat selection of a declining passerine at two spatial scales

Context

Detailed information on habitat needs is integral to identify conservation measures for declining species. However, field data on habitat structure is typically limited in extent. Remote sensing has the potential to overcome these limitations of field-based studies.

Objective

We aimed to assess abiotic and biotic characteristics of territories used by the declining wood warbler (Phylloscopus sibilatrix), a forest-interior migratory passerine, at two spatial scales by evaluating a priori expectations of habitat selection patterns.

Methods

First, territories established by males before pairing, referred to as pre-breeding territories, were compared to pseudo-absence control areas located in the wider forested landscape (first spatial scale, N_territories = 66, N_controls = 66). Second, breeding territories of paired wood warblers were compared to true-absence control areas located immediately close-by in the forest (second spatial scale, N_territories = 78, N_controls = 78). Habitat variables predominantly described forest structure and were mainly based on first and last pulse lidar (light detection and ranging) data.

Results

Occurrence of pre-breeding territories was related to vegetation height, vertical diversity and stratification, canopy cover, inclination and solar radiation. Occurrence of breeding territories was associated to vegetation height, vertical diversity and inclination.

Conclusions

Territory selection at the two spatial scales addressed was governed by similar factors. With respect to conservation, habitat suitability for wood warblers could be retained by maintaining a shifting mosaic of stand ages and structures at large spatial scales. Moreover, leaf-off lidar variables have the potential to contribute to understanding the ecological niche of species in predominantly deciduous forests.

     

Elephants,termites and mound thermoregulation in a progressively warmer world

Context

With global change, microclimates become important refuges for temperature-sensitive, range-restricted organisms. In African savannas, woody vegetation on Macrotermes mounds create widely-dispersed microclimates significantly cooler than the surrounding matrix, which buffer against elevated temperatures at the finer scale of mounds, allowing species to persist at the landscape scale. Termite colonies cultivate symbiotic fungi to digest lignin, but the fungi require temperatures between 29 and 32 °C, which termites strive to maintain. Mound-associated vegetation is a hot-spot for elephant herbivory, so removal of woody species cover by elephants could influence mound-associated microclimates, impacting temperature regulation by termites.

Objectives

We explored the interaction between two prominent ecosystem engineers (termites and elephants) to ascertain whether elephant removal of mound woody cover affects (1) external mound-associated microclimate and (2) internal mound temperature.

Methods

We surveyed 44 mounds from three sites in Kruger National Park, South Africa, during an El Niño/Southern Oscillation-induced drought and heatwave, recording whether sub-canopy, external, mound-surface and internal mound temperatures varied with vegetation removal by elephant.

Results

Elephant damage to mound-associated vegetation reduces the fine-scale microclimate effect provided by vegetation on Macrotermes mounds. Despite this, termites were able to regulate internal mound temperatures, whereas internal temperatures of abandoned mounds increased with elevated surface temperatures.

Conclusions

Termites can persist despite loss of mound-associated microclimates, but the loss likely increases energetic costs of mound thermoregulation. Since mound vegetation buffers against drought, loss of widely-dispersed, fine-scale microclimates could increase as megaherbivores remain constrained to protected areas, impacting climate-sensitive organisms and ecosystem function at a range of scales.

     

A network model framework for prioritizing wetland conservation in the Great Plains

Context

Playa wetlands are the primary habitat for numerous wetland-dependent species in the Southern Great Plains of North America. Plant and wildlife populations that inhabit these wetlands are reciprocally linked through the dispersal of individuals, propagules and ultimately genes among local populations.

Objective

To develop and implement a framework using network models for conceptualizing, representing and analyzing potential biological flows among 48,981 spatially discrete playa wetlands in the Southern Great Plains.

Methods

We examined changes in connectivity patterns and assessed the relative importance of wetlands to maintaining these patterns by targeting wetlands for removal based on network centrality metrics weighted by estimates of habitat quality and probability of inundation.

Results

We identified several distinct, broad-scale sub networks and phase transitions among playa wetlands in the Southern Plains. In particular, for organisms that can disperse >2 km a dense and expansive wetland sub network emerges in the Southern High Plains. This network was characterized by localized, densely connected wetland clusters at link distances (h) >2 km but <5 km and was most sensitive to changes in wetland availability (p) and configuration when h = 4 km, and p = 0.2–0.4. It transitioned to a single, large connected wetland system at broader spatial scales even when the proportion of inundated wetland was relatively low (p = 0.2).

Conclusions

Our findings suggest that redundancy in the potential for broad and fine-scale movements insulates this system from damage and facilitates system-wide connectivity among populations with different dispersal capacities.

     

How does forest fragmentation affect tree communities? A critical case study in the biodiversity hotspot of New Caledonia

Context

The biodiversity hotspot for conservation of New Caledonia has facing high levels of forest fragmentation. Remnant forests are critical for biodiversity conservation and can help in understanding how does forest fragmentation affect tree communities.

Objective

Determine the effect of habitat configuration and availability on tree communities.

Methods

We mapped forest in a 60 km² landscape and sampled 93 tree communities in 52 forest fragments following stratified random sampling. At each sampling point, we inventoried all trees with a diameter at breast height ≥10 cm within a radius of 10 m. We then analysed the response of the composition, the structure and the richness of tree communities to the fragment size and isolation, distance from the edge, as well as the topographical position.

Results

Our results showed that the distance from the forest edge was the variable that explained the greatest observed variance in tree assemblages. We observed a decrease in the abundance and richness of animal-dispersed trees as well as a decrease in the abundance of large trees with increasing proximity to forest edges. Near forest edges we found a shift in species composition with a dominance of stress-tolerant pioneer species.

Conclusions

Edge-effects are likely to be the main processes that affect remnant forest tree communities after about a century of forest fragmentation. It results in retrogressive successions at the edges leading to a dominance of stress-tolerant species. The vegetation surrounding fragments should be protected to promote the long process of forest extension and subsequently reduce edge-effects.