Terrestrial ecosystems originate when particular plant species attain dominance at specific locations under specific environmental regimes. Ecosystems terminate, gradually or abruptly, when the dominant species or functional types are replaced by others, usually owing to environmental change or severe and irreversible disturbance. Assessing whether current ecosystems are sustainable in the face of future environmental change can be aided by examining the range of environmental variation those ecosystems have experienced in the past, and by determining the environmental conditions under which those ecosystems arose. The range of environmental variation depends on the time scale at which it is assessed. A narrow time span (e.g. 200–300 years) may underestimate the range of variation within which an ecosystem is sustainable, and it may also underestimate the risk of major transformation or disruption of that ecosystem by environmental change. Longer time spans (e.g. 1000–2000 years) increase the range of variation, by encompassing a larger sample of natural variability as well as non-stationary variability in the earth system. Most modern ecosystems disappear when the time span is expanded to 10000–15 000 years owing to secular changes in earth's climate system. Paleo-ecological records can pinpoint the time of origination of specific ecosystems, and paleo-environmental records can reveal the specific environmental changes that led to development of those ecosystems and the range of environmental variation under which those ecosystems have maintained themselves in the past. This information can help identify critical environmental thresholds beyond which specific modern ecosystems can no longer be sustained.

Questions: 1. Are trees in a Bornean tropical rain forest associated with a particular habitat? 2. Does the strength of habitat association with the species-specific optimal habitat increase with tree size?

Location: A 52-ha plot in a mixed dipterocarp forest in a heterogeneous landscape at the Lambir Hills National Park, Sarawak, East Malaysia.

Methods: Ten species from the Sterculiaceae were chosen as representative of all species in the plot, on the assumption that competition among closely related species is more stringent than that among more distantly related taxa. Their habitat associations were tested using data from a 52-ha plot by a torus-translation test.

Results: The torus-translation test showed that eight out of the ten species examined had significant association with at least one habitat. We could not find negative species-habitat associations for rare species, probably due to their small sample sizes. Among four species small trees were less strongly associated with habitat than large trees, implying competitive exclusion of trees in suboptimal habitats. The other four species showed the opposite pattern, possibly owing to the smaller sample size of large trees. A habitat had a maximum of three species with which it was significantly positively associated.

Conclusions: For a species to survive in population equilibrium in a landscape, habitats in which ‘source’ subpopulations can be sustained without subsidy from adjacent habitats are essential. Competition is most severe among related species whose source subpopulations share the same habitat. On the evidence of source subpopulations identified by positive species-habitat association, species-habitat association reduces the number of confamilial competitors. Our results therefore indicate that edaphic niche specialization contributes to coexistence of species of Sterculiaceae in the plot, consistent with the expectations of equilibrium hypotheses.

Nomenclature: Ashton (1980).

Questions: How do species diversity, frequency and composition in tussocks differ from those in similar sized plots outside tussocks? Does the extent of the differences depend on community types or environmental conditions?

Location: A sub-alpine grassland in the Swiss National Park.

Methods: In each of the two communities (short grass and tall graminoid) differing in species composition, grazing intensity and soil nutrient availability, relevés were made in 40 pairs of small circular plots, with one plot located inside a randomly selected Carex sempervirens tussock and the other outside.

Results: We found 92 vascular species, of which 46 had a frequency higher than 5%. Species richness (S), pooled cover, Shannon's diversity (H) and cumulative species number (CS) were higher outside than inside the C. sempervirens tussocks, but evenness (J) was lower. S, H and CS differed more in the tall graminoid community than in the short grass community. However, dissimilarity between the paired relevés inside and outside tussocks did not differ between the two communities. Of the 46 most frequent species, 12 were statistically more and only one less frequent outside than inside the tussocks. Vegetation inside and outside tussocks could be clearly distinguished in the ordination space.

Conclusion: Vegetation inside C. sempervirens tussocks is different from that in the surrounding area and represents an impoverished but homogenized version of the surrounding vegetation. Although tussocks of C. sempervirens were systematically avoided by grazers, there is little evidence that tussocks facilitate the species growing inside them.

Abbreviations: CS = Cumulative species number; H = Shannon's diversity index; J = Evenness; PCO = Principal Coordinate Analysis; S = Species richness; SNP = Swiss National Park.

Question: How accurately can a suite of suggested functional traits predict plant species response to succession from semi-open woodland to closed deciduous canopy forest?

Location: Southeastern Sweden.

Methods: Abundance of 46 field-layer plant species in a temperate deciduous forest, measured as frequency of occupied plots, was estimated in 1961, 1970 and 2003. Abundance change over time across species was tested for correlations with functional traits and literature information on habitat preference.

Results: Increase in abundance was positively correlated with specific leaf area (SLA), weakly negatively correlated with seed mass and not significantly correlated with plant height or start, peak and length of the flowering period. Change in abundance was correlated with the Ellenberg light indicator value, whereas no correlations were found with Ellenberg values for nitrogen, calcium and moisture, or forest preference according to the literature.

Conclusions: SLA was a better predictor of how field layer plants responded to succession from semi-open woodland to closed canopy forest than empirically-derived measures of habitat preference. The same holds for SLA in relation to seed size, indicating that interactions in the established life-cycle phase are more important than the recruitment phase for species response to succession.

Abbreviations L = Light; N = Nitrogen; SLA = Specific Leaf Area.

Nomenclature: Mossberg & Stenberg (2003).

Questions: What is the variability of succession over a large geographical area? What is the relative importance of (1) local site factors and (2) landscape factors in determining spontaneous vegetation succession?

Location: Various regions of the Czech Republic, Central Europe. The regions represent two categories characterized by agrarian lowlands, with a relatively warm and dry climate, and predominant woodland uplands with a relatively cold and wet climate.

Methods: Gravel-sand pits ranged in age from 1–75 years since abandonment. Three types of sites were distinguished: dry, wet and hydric in shallow flooded sites. Vegetation relevés were recorded with species cover (%) visually estimated using the space-for-time substitution approach. Local site factors, such as water table and soil characteristics, and landscape characteristics, namely climatic parameters, presence of nearby (semi-) natural plant communities and main land cover categories in the wider surroundings, were evaluated.

Results: Ordination analyses showed that water table was the most important local site factor influencing the course of spontaneous vegetation succession. Succession was further significantly influenced by soil texture, pH, macroclimate, the presence of some nearby (semi-) natural communities and some land cover categories in the wider surroundings. Spontaneous vegetation succession led to the formation of either shrubby grassland, deciduous woodland, Alnus and Salix carrs, and tall sedge or reed and Typha beds in later stages depending predominantly on the site moisture conditions.

Conclusions: Although the water table was the most influential on the course of vegetation succession, the landscape factors together explained more vegetation variability (44%) than local site factors (23%).

Nomenclature: Kubát et al. (2002).

Question: This paper studies the establishment and performance of Cortaderia selloana (Pampas grass), an alien South American ornamental species that is invading many parts of the world. We asked whether (1) early successional stages were the most susceptible to C. selloana invasion; (2) soil microdisturbances increased invasion at any point of succession, and (3) C. selloana invasion of later successional stages was modulated by vegetation type.

Location: Delta del Llobregat (Catalonia, NE Spain).

Methods: We monitored survival and growth of transplanted C. selloana seedlings in disturbed and non-disturbed plots throughout a successional gradient with an age range of < 1 to > 10 years in different vegetation types and within the area of influence of coexisting species with similar growth form.

Results: Although seedling survival was extremely low in all treatments, our results revealed that early successional stages were not the most easily invaded since we found no significant differences in the percentage survival of C. selloana along the successional gradient. However, survival and seedling biomass were enhanced by soil disturbance at any seral stage. This result suggested that inhibition ruled C. selloana invasion. Invasibility neither depended on the invaded vegetation type nor on the co-existing species with similar growth form. Finally, C. selloana invasion was not enhanced by decreasing competition with Phragmites australis, a native coexisting species because survival rates after a year were not significantly different. However, Phragmites increased C. selloana leaf length probably due to shading.

Conclusions: C. selloana recruitment appears to be positively affected by soil disturbance but it is independent of successional stage or vegetation type.

Questions: To what extent can spatial structure and its causes be determined in a highly disturbed environment? What are the main determinants of pattern and are these species-specific? How much do spatial patterns change over generations?

Location: Wimmera region of southern Australia.

Methods: Broad-leaved weeds were counted in 225 000 contiguous 20-cm square quadrats. A substantial number of these quadrats were recorded again after two and four years. An hierarchical ‘adaptive analysis’ approach was used to select spatial analytical methods to examine specific aspects of pattern and variation in pattern from year to year.

Results: Patterns varied among species and included both dense and sparse patches surrounded by areas of zero density, diffuse gradations of density and clear anisotropy. Patterns in Erodium botrys and Oxalis pes-caprae persisted over years, whereas patterns in Arctotheca calendula were less pronounced and varied over time. Edaphic factors appeared to have only a minor influence over the spatial distribution of the weed community as a whole. In Oxalis pes-caprae, whose patches were hypothesized to have been shaped by cultivation, there was no spread in four years, despite further tillage. Outlying plants of O. pes-caprae failed to establish new patches, even in the year of greatest population increase. Little evidence of localised recruitment events was found.

Conclusions: Despite repeated annual disturbances by natural and anthropogenic mechanisms, clear and interpretable spatial structure develops in annual weeds over a range of spatial resolutions. Adaptive analysis is a useful approach to the characterization of such patterns.

Abbreviations: EC_e = Electrical conductivity; EC_a = Apparent electrical conductivity; DTM = Digital Terrain Model.

Nomenclature: Walsh & Entwisle (1994–1999).

Question: The formation of large woody debris (LWD) piles during floods has significant impacts on riparian succession through pioneering plants often establishing in association with wood. We assess the importance of LWD for seed regeneration of riparian plants after a century-scale flood disturbance in a semi-arid environment.

Location: The Sabie River within Kruger National Park in the semi-arid northeast of South Africa.

Methods: Our approach was to quantify the riparian soil seed bank, to record the frequency of establishment of riparian plants in woody debris piles, and to conduct experimental out-plantings of common riparian trees in plots with and without LWD.

Results: We found the abundance and diversity of seedlings were higher in soils taken from wood piles than from open reference areas, and most seedlings were herbaceous species. Surveys indicated that numbers of seedlings recorded within woody debris were significantly greater than in open reference areas or within established vegetation. Seedling establishment in various cover-types also varied for different riparian tree species. Experimental out-planting of seedlings of two riparian tree species (Philenoptera violacea and Combretum erythrophyllum) revealed that, after 433 days, planted seedlings survived only in woody debris piles.

Conclusion: LWD formed after a large flood creates heterogeneous patches that may influence post-disturbance regeneration of riparian vegetation by providing a variety of environmental niches for seedlings establishment. We suspect that higher seedling survival in LWD is due to increased moisture (particularly in the dry season) and nutrients, and protection from seasonal flooding and herbivory.

Abbreviations: KNP = Kruger National Park; LWD = Large woody debris.

Questions: How do gap abundance and the spatial pattern of trees and snags change throughout stand development in Picea mariana forests? Does spatial pattern differ among site types and structural components of a forest?

Location: Boreal forests dominated by Picea mariana, northern Quebec and Ontario, Canada.

Methods: Data on the abundance, characteristics and spatial location of trees, snags and gaps were collected along 200 m transects at 91 sites along a chronosequence. Spatial analyses included 3TLQV, NLV and autocorrelation analysis. Non-parametric analyses were used to analyse trends with time and differences among structural components and site types.

Results: Gaps became more abundant, numerous and more evenly distributed with time. At distances of 1–4 m, tree cover, sapling density and snag density became more heterogeneous with time. Tree cover appeared to be more uniform for the 10–33 m interval, although this was not significant. Patch size and variance at 1 m were greater for overstorey than for understorey tree cover. Snags were less spatially variable than trees at 1 m, but more so at intermediate distances (4 - 8 m). Few significant differences were found among site types.

Conclusions: During stand development in P. mariana forest, gaps formed by tree mortality are filled in slowly due to poor regeneration and growth, leading to greater gap abundance and clumping of trees and snags at fine scales. At broader scales, patchy regeneration is followed by homogenization of forest stands as trees become smaller with low productivity due to paludification.

Abbreviations: 3TLQV = Three-term local quadrat variance; NLV = New local variance.

Questions: 1. How are the long-term dynamics of the root hemiparasite Rhinanthus angustifolius related to vegetation structure, grassland management and climate? 2. Does R. angustifolius have a long-term impact on standing crop and community composition?

Location: A formerly fertilized grassland, part of a larger brook valley system in the nature reserve ‘Drentsche Aa’, near Groningen, The Netherlands.

Methods: Vegetation development has been monitored since 1973 in 54 permanent plots in nine management regimes without fertilizer application.

Results: 1. The hemiparasite established when standing crop was less than 600 g.m⁻² and performed best under annual haymaking using machinery. Since its appearance, the population fluctuated stochastically, with two peaks. Coinciding collapses in six adjacent grasslands and comparison with an integrated climatic index suggest that the population collapses are induced by spring drought. 2. We did not find a relationship between total standing crop and R. angustifolius cover. Cover of grasses was negatively related to the abundance of the hemiparasite in the same and the previous year. Forb cover tended to increase with the parasite.

Conclusions: R. angustifolius shows stochastic population fluctuations, mainly determined by spring drought, to which this species is probably highly vulnerable because of its parasitic and annual life style without a persistent seed bank. The hemiparasite also shows long-term relationships with grass cover (negative) and forb cover (positive), but it seems to have no lasting impact on standing crop.

Abbreviations: AC = Autocorrelation; CC = Cross-correlation coefficient; Loef S = Study site within Drentsche Aa nature reserve.

Nomenclature: Zuidhoff et al. (1996).

Questions: What is the mechanism of bog ground layer colonization post-fire? Is species colonization stochastic or does facilitation occur?

Location: Boreal bog peatland near Crow Lake, Alberta, Canada.

Methods: Diaspore-addition treatments were applied in 2003 to autoclaved peat samples from high and low microtopographic positions within a recently burned bog. Colonization was assessed within the plots in 2005 and compared to control plots to determine treatment success and patterns of colonization.

Results: A significant degree of ground layer colonization was found two years after fire, with Polytrichum strictum dominating the site. Colonization was greater in low (wet) plots, although only P. strictum and Sphagnum angustifolium had significant colonization. No effect of diaspore addition was observed and Sphagnum was only found in conjunction with P. strictum.

Conclusions: Environmental conditions and species life history strategy are more important than diaspore availability for post-fire colonization. True mosses (e.g. P. strictum) appear to facilitate Sphagnum colonization.

Nomenclature: Vitt et al. 1988.

Question: What are the composition, conservation status, and structural and environmental characteristics of eight mature tropical forest plant communities that occur along an elevational gradient.

Location: Northeastern Puerto Rico.

Methods: We quantified the species composition, diversity, conservation status, and ecological attributes of eight mature tropical forest plant communities in replicated plots located to sample representative components of important forest types occurring along an elevational gradient. A suite of environmental and vegetation characteristics were sampled at each plot and summarized to characterize communities and analyse trends along the elevational gradient.

Results: The set of communities included 374 species; 92% were native, 14% endemic, and 4% critical elements (locally endangered) to the island. All communities, occurring within a wide range of patch sizes and degree of conservation protection, showed a high percentage of native species (> 89% per plot). The lowland moist forest communities, occurring within a matrix of urbanization, agriculture, and disturbance, had the highest degree of invasion by exotics. Community descriptions were nested within a variety of hierarchies to facilitate extrapolation of community characteristics to larger ecosystem units. Basal area, above-ground biomass, canopy heights, and mean species richness peaked at mid elevations.

Conclusions: It is significant that all of these forest communities continue to be dominated by native species while existing in a matrix of human and natural disturbance, species invasion, and forest regeneration from widespread agriculture. The lowland moist and dry forest types represent a minority of the protected forested areas in Puerto Rico, serve as unique genetic reservoirs, and should be protected.

Abbreviations: AB = above-ground; BA = Basal area; DCA = Detrended Correspondence Analysis; CE = Critical element; I = Introduced; MAP = Mean annual precipitation; MAT = Mean annual temperature; N = Native.

Nomenclature: Liogier & Martorrell (1999).

Question: What is the relative importance of the initial seed bank and subsequent seed dispersal for floristic composition of bank vegetation two years after creation of a newly-cut reach of a river channel?

Location: River Cole, West Midlands, United Kingdom.

Methods: We took bank and bed sediment samples from a 0.5-km reach of a new river channel cut into intact flood-plain. After river diversion, seed samples deposited on artificial turf mats placed on the river banks and flood-plain edge were taken in summer and winter 2002 and 2003. Seed rain samples from funnel traps were taken during summer 2002 and 2003. We undertook greenhouse germination trials to assess viable seed species within these samples. In summer 2004, we surveyed river bank vegetation.

Agglomerative cluster analysis was used to investigate floristic similarity between seed bank, seed rain, seed deposition samples and final bank vegetation cover. DCA was used to explore contrasts between the samples and to assess whether these reflected interpretable environmental gradients.

Results: Seed rain samples contained a small subset of species in the summer depositional samples. 38 species were found within the final vegetation, the seed bank, and at least one of the four sets of depositional samples; a further 30 species not present in the seed-bank samples were present in at least one of the four sets of depositional samples and the final vegetation. Floristic composition of the vegetation was most similar to the depositional samples from winter 2002 and 2003 and summer 2003. DCA axis 1 reflected a time sequence from seed-bank samples through depositional samples to the final vegetation.

Conclusions: Newly cut river banks were colonized rapidly. Seed remobilization and hydrochorous transport from the upstream catchment are important for colonization. Species richness was highest in samples deposited during winter when high river flows can remobilize and transport viable seeds from upstream. This process would also have enhanced the species richness of seed production along the banks during the second summer (2003).

Nomenclature: Stace (1999).

Question: How do interactions between rocky landscape features and fire regime influence vegetation dynamics? Location: Continental Eastern USA.

Methods: We measured vegetation, disturbance and site characteristics in 40 pairs of rocky and non-rocky plots: 20 in recently burned stands, and 20 in stands with no evidence of recent fire (‘unburned’ stands). Two-way analysis of variance (ANOVA) was used to assess the main and interaction effects of fire and rock cover on plant community composition. Results: In burned stands, rock cover had a strong influence on vegetation. Non-rocky ‘matrix’ forests were dominated by Quercus, and had abundant ground cover and advance regeneration of early and mid-successional tree species. Burned rocky patches supported greater density of fire-sensitive species such as Acer rubrum, Sassafras albidum and Nyssa sylvatica and had little advance regeneration or ground cover. Quercus had fewer fire scars and catfaces (open, basal wounds) on rocky patches, suggesting that rocky features mitigate fire severity. In unburned stands, differences between rocky and non-rocky patches were less distinct, with both patch types having sparse ground cover, little tree regeneration, and high understorey densities of relatively shade tolerant A. rubrum, N. sylvatica and Betula lenta.

Conclusion: Under a sustained fire regime, heterogeneity in rock cover created a mosaic where fire-adapted species such as Quercus dominate the landscape, but where fire-sensitive species persisted in isolated pockets of lower fire severity. Without fire, species and landscape richness may decline as early-mid successional species are replaced by more shade tolerant competitors.

Nomenclature: Rhoads & Block (2000).

Because of their isolation and geographical position, and in contrast to the multi-species tree canopies of tropical rain forests on the continents, the Hawaiian Islands have only two native dominant canopy species in their rain forests, Acacia koa and Metrosideros polymorpha. The wetter forest ecosystems are dominated by only the latter. In 1905, a dieback of lowland tropical Metrosideros rain forest was observed over a 35 km stretch on the lower east slope of Haleakalā Mountain on Maui Island. This was dubbed ‘The Maui Forest Trouble.’ Although the synchronous decline of so many trees was initially believed to be caused by an epidemic disease, a decade of research yielded no pathogen. The conclusion was that the Hawaiian flora consisted primarily of colonizer species that were unable to continue growing on aging soils. Although this made ecological sense at that time, it was a rather limited and thereby unfortunate conclusion. Further research has shown that the Maui Forest Trouble was a ‘bog-formation dieback’, a process of vegetation dynamics not only related to soil aging but more broadly to geomorphic aging and fundamental landscape change. This process is clearly a marginal-site syndrome, but a natural process of profound consequence for biological conservation. This will be further explained as a paradigm for vegetation ecology.