Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The potential contribution of vegetation ecology to biodiversity research

The potential contribution of vegetation ecology to biodiversity research Austin, M. P. 1999. The potential contribution of vegetation ecology to biodiversity research. - Ecography 22: 465-484. The contribution of vegetation ecology to the study of biodiversity depends on better communication between the different research paradigms in ecology. Recent developments in vegetation theory and associated statistical modelling techniques are reviewed for their relevance to biodiversity. Species composition and collective properties such as species richness vary as a continuum in a multi-dimensional environmental space; a concept which needs to be incorporated into biodiversity studies. Different kinds of environmental gradients can be recognised and species responses to them vary. Species response curves of eucalypts to an environmental gradient of mean annual temperature have been shown to exhibit a particular pattern of skewed response curves. Generalised linear modelling (GLM) and generalised additive modelling (GAM) techniques are important tools for biodiversity studies. They have successfully distinguished the contribution of environmental (climatic) and spatial (history and species dispersal ability) variables in determining forest tree composition in New Zealand. Species richness studies are examined at global, regional and local scales. At all scales, direct and resource environmental gradients need to be incorporated into the analysis rather than indirect gradients e.g. latitude which have no direct physiological influence on biota. Evidence indicates that species richness at the regional scale is sensitive to environment, confounding current studies on local/regional species richness relationships. Plant community experiments require designs based on environmental gradients rather than dependent biological properties such as productivity or species richness to avoid confounding the biotic components. Neglect of climatic and other environmental gradients and the concentration on the collective properties of species assemblages has limited recent biodiversity studies. Conservation evaluation could benefit from greater use of the continuum concepts and statistical modelling techniques of vegetation ecology. The future development of ecology will depend on testing the different assumptions of competing research paradigms and a more inclusive synthesis of ecological theory. M . P. Austin (m.austin@dwe.csiro.au),CSIRO Wildlfe and Ecology, GPO Box 284, Canberra A C T 2601, Australia. As our global society faces the problem of productive sustainability of our biosphere, it becomes essential to determine the role of biodiversity. Plant communities are the primary autotrophic component of terrestrial biodiversity. Yet, vegetation ecologists, as opposed to plant ecologists, have not been seen to be contributing to recent studies on the role of plant species richness in determining productivity and stability of ecosystem This is an invited Minireview on the occasion of the 50th anniversary of the Nordic Ecological Society Oikos. Copyright 0 ECOGRAPHY 1999 ISSN 0906-7590 Printed in Ireland - all rights reserved ECOGRAPHY 1 2 5 (1999) function (Kareiva 1994, Lawton 1994, Tilman and Downing 1994, Johnston et al. 1996). Plant community ecology can be defined as the study of the distribution and behaviour of multispecies assemblages of plants in time and space. Vegetation ecology is used here to apply to plant community studies where observational analysis is the dominant method of study. This field has tended to be neglected by ecologists in recent years, yet it has much to offer the study of biodiversity. The term biodiversity as used here can be applied to the total, or to a defined subset, of the biota of an ecosystem, landscape or continent. The question is, why has vegetation ecology been ignored and does it have anything to offer? The existence of disparate, multitudinous research paradigms (sensu Kuhn 1970) in ecology is one factor limiting the use of concepts and methods developed in vegetation ecology. In each paradigm, evidence is frequently sought to support an idea rather than to disprove an hypothesis while methods are selected for their compatibility with the paradigm rather than their suitability for answering the question at hand. Often, new concepts, evidence and suitable methods exist in another ecological paradigm, which could contribute to a better synthesis than could be achieved by either paradigm alone. In this review I present a particular personal paradigm for vegetation ecology and argue that it has potential to elucidate several broader issues in the study of biodiversity. I shall try to examine the problem of communication between paradigms with reference to species richness studies while considering whether vegetation ecology does have anything to contribute to the study of biodiversity. Three issues are important to such a review: 1) The relevance of vegetation ecology’s own paradigm assumptions regarding theory, methodology and experimentation to current questions regarding biodiversity. 2) The need to simultaneously test hypotheses from different paradigms in order to achieve a new synthesis. 3) The use of vegetation survey and analysis techniques in the conservation of biodiversity. Some recent developments in continuum theory and associated statistical modelling methods are reviewed for their potential contribution to biodiversity studies. Species richness studies are briefly examined at global, regional and local scales with the relevance of results from vegetation ecology emphasised. The relationship between recent experimental community studies and observational studies is discussed with suggestions for alternative experimental designs. A similar potential for integration of ideas is shown for conservation studies. The need to incorporate both environmental space and geographical space into any synthesis of community ecology paradigms is then discussed. Continuum and environmental space The continuum concept (Whittaker 1956, Curtis 1959) has been central to vegetation ecology for forty years and is usually presented in general ecology texts (e.g. Huston 1994, Krebs 1994, Shugart 1998). Yet many recent texts and reviews on species diversity and spatial patterns barely mention the concept (Ricklefs and 466 Schluter 1993, Brown 1995, Rosenzweig 1995, Crawley 1997, Tilman and Kareiva 1997, Lawton 1999). The concept states that species populations are distributed along environmental gradients such that vegetation varies continuously and does not form distinct coevolved communities of species (Whittaker 1975). This concept is usually linked with Gleason’s individualistic concept that all species are unique and independent in their distribution patterns along gradients. However, there is no necessary association between the two concepts (Austin 1985). The weakness of the continuum concept is that it remains strictly phenomenological, describing patterns without explicit processes for generating the patterns beyond a token reference to competition. However, any process-based ecological theory should be capable of predicting (explaining) the species patterns which have been demonstrated to occur along environmental gradients on several continents (Whittaker 1956, Curtis 1959, Ellenberg 1988, Austin et al. 1994). Early work by Bond (1957), Beals (1960) and Sabo (1980) showed the applicability of the continuum concept to bird community composition. A reanalysis of Rotenberry and Wiens’ (1980) work on bird distribution along a structural gradient of vegetation has shown that the results support a continuum interpretation contrary to their conclusion (Austin 1999). Recently animal ecologists’ concerns with broad-scale patterns of species, whether at the community or biogeographical level, have emphasised ranges and geographical distance (Brown 1995, Rosenzweig 1995, Gaston and Blackburn 1999, Lawton 1999). Two features of the animal ecologists’ paradigm may account for the lack of interest in the continuum concept. First the attention paid to a collective property of the biota, species richness rather than compositional changes and secondly, the assumption that climate plays a secondary role in determining faunal patterns. Lawton (1982) argues in a study concerning bracken insect faunas that “legitimate worries about climate. . . are arguments about ‘second order’ effects. . . they may modify, but are not primarily responsible for, major differences in faunal richness between geographical regions” (see also Srivastava 1999). In comparison, recent statistical models developed for analysis of tree species distributions (Austin et al. 1990, Yee and Mitchell 1991, Leathwick and Mitchell 1992) have shown that climatic variables such as mean annual temperature, and mean annual rainfall are important predictors of regional distributions. These models examined the behaviour of biotic responses in a multivariable environmental space rather than as a univariate test of the importance of a single predictor variable (Austin et al. 1984, Margules et al. 1987). The influence of a variable may often be hidden by a second variable and multiple regression may be needed to recognise the true significance of a variable. The limitations imposed by single variable analysis and linear ECOGRAPHY 2 2 2 (1999) regression in some ecological paradigms are discussed in Austin (1999). Simultaneous analysis of climatic and geographical predictors of fauna as done by vegetation ecologists is needed to test the secondary importance of climate when testing geographical hypotheses. There are other aspects of the paradigm used by vegetation ecologists which may have led to its neglect by others. The lack of clearly defined testable hypotheses is a major one (Keddy 1987, 1992, 1994) and a second is the pre-occupation of many vegetation ecologists with ordination methodology. Recent conceptual developments The continuum concept has been elaborated and testable propositions have been put forward. Austin (1987) and Shipley and Keddy (1987) have tested several hypotheses from earlier work. Shipley and Keddy (1987) attempted to distinguish between the individualistic and community concepts by investigating the coincidence of species limits along a watertable gradient. They concluded that some limits were clustered contradicting the individualistic concept, but upper and lower limits did not co-occur contradicting the community concept. Propositions that the individual species response curves approximated a Gaussian (bell-shaped) curve along environmental gradients and that the modes of major species tend to have a uniform distribution along a gradient were tested (Austin 1987). In the eucalypt forest studied, positive-skewed curves were characteristic of the major canopy species and there was no evidence that species modes were uniformly distributed along a mean annual temperature gradient. Austin and Smith (1989) then put forward a set of testable phenomenological propositions regarding species and vegetation responses along environmental gradients. The ideas of Ellenberg on species responses to environmental gradients were used to describe the variety of possible response shapes (Mueller-Dombois and Ellenberg 1974, Austin 1999). Shapes can range from the conventional Gaussian curve through skewed curves to bimodal responses. The usual niche theory model used by animal ecologists is a special case of the more general theory of Ellenberg (Austin 1999). Austin and Smith (1989) also argued that three types of environmental gradient could be recognised (see Austin 1980): 1) Indirect gradients or complex gradients (Whittaker 1978) are those where the variable has no physiological relationship with biological growth and where variables having a direct physiological relationship are usually confounded. Latitude, altitude and ocean depth are examples. Altitude has no direct impact on organisms. It is highly correlated with both rainfall and temperature. The nature of the correlation is different for each variable and such relationships are location specific. 2) Direct gradients are those where the enviECOGRAPHY 2 2 5 (1999) ronmental variable e.g. temperature has a direct effect on growth but is not consumed. 3) Resource gradients (e.g. nitrogen) are those where the resource is consumed by plants. It is critical that the three types be distinguished. Huston (1994) discusses the different types of response expected for the different types of gradients. Studies of the relationship of species richness to latitude (complex gradient) require a different analytical framework to those concerned with a resource gradient (Huston 1994, 1999, Rosenzweig 1995). Niche theory for animals is usually based on resource gradients (Giller 1984). Direct gradients (termed regulator gradients by Huston 1994) are often climatic and may be more varied in the type of species response. Fauna studies often do not make clear distinctions between these types of gradients (Brown 1995, Rosenzweig 1995, Lawton 1999). If gradients are not equivalent then comparative analysis is not possible. When the propositions of Austin and Smith (1989) were compared with the work of other researchers, significant gaps in the way plant ecologists analyse species’ responses were revealed (Austin and Smith 1989). Different types of environmental gradients were not recognised. Possible changes in the response shape of fundamental and realised niches were not considered (Austin 1999). Various assumptions were made regarding the collective properties of vegetation, species richness, dominance and standing crop (Table 1). Plant ecologists still have a long way to go before their paradigms achieve agreement. The propositions of Austin and Smith (1989) hopefully provide a framework within which comparisons and debate can be conducted. An example of the type of hypothesis testing which can now be undertaken, thanks to the adoption of new statistical modelling techniques and the use of large data sets (discussed in the next section), concerns the position, shape and skewness of species response curves. Species limits towards the extremes of an environmental gradient are determined by physiological tolerances while competition controls both the limits towards the centre of the gradient and the shape of the species response curve (Fig. 1, Austin 1990). An analysis of nine species of eucalypt in relation to seven environmental predictors examined this hypothesis (Austin et al. 1994). A database of 8377 plots of presence/absence data from south-east New South Wales, Australia were analysed using Generalised Linear Modelling (GLM, McCullagh and Nelder 1989) and p-function response curves. Eight of the nine species showed the postulated pattern along a mean annual temperature gradient (Fig. 1). Austin and Gaywood (1994) subsequently found the pattern applied to 22 out of 24 eucalypt species. A pattern of species packing of this kind along an environmental gradient implies that there are assembly rules associated with the continuum concept (Keddy 1992). Smith and Huston (1989) 467 provide supporting circumstantial evidence based on a mechanistic simulation model which assumes that no species can be simultaneously adapted to extreme environmental conditions and mesic highly productive conditions (cf. Grime 1979). Neglect of the responses of individual species, in preference to collective properties such as species richness, is likely to mean failure to detect assembly rules. However, this is a single result from a vegetation type where almost the entire temperature gradient is dominated by a single genus Eucalyptus. The generality of the pattern needs to be investigated. Three features of these recent developments in vegetation theory need to be incorporated into the study of biodiversity. Firstly, the basic concept that species vary continuously in occurrence and abundance in a multidimensional environmental space. Failure to incorporate this into analyses is likely to obscure rather than clarify results. Secondly, environmental gradients are of different kinds and species responses to them will differ. Thirdly, the types of species niche responses postulated by Ellenberg are more appropriate for investigation than the Gaussian curve model. Methodology Lack of interest by many ecologists in plant community ecology and its theoretical concepts has been due to the preoccupation of some vegetation ecologists with methodology. Strong criticism of the exploratory multivariate approach has been made. Crawley (1997, p. 510) quotes May (1985) as saying “the wilderness of meticulous classification and ordination of plant communities, in which plant ecology has wandered for so long, began in pursuit of answers to questions but then became an activity simply for its own sake”. This criticism is not without substance though see van der Maarel(l989) and Kent and Coker (1992) for a spirited response. Ordination methodology continues to be an issue. Consider the papers presented at a symposium in Uppsala in 1985 on “Theory and models in vegetation science” by ter Braak (1987) (see also ter Braak 1986) on canonical correspondence analysis (CCA) (see also Jongman et al. 1987), and those by Faith et al. (1987) and Minchin (1987) (see also Minchin 1989) comparing correspondence analysis (CA) and multidimensional scaling (MDS). The two paradigms, one using canonical correspondence analysis based methods (ter Braak 1986) and the other, new forms of multidimensional scaling (Belbin 1991) continue to exist today. The basic ecological assumptions have still not been examined in depth, though as the above papers make clear the methods are very sensitive to the assumptions. The controversies continue (Wartenberg et al. 1987, Peet et al. 1988, James and McCulloch 1990), and continue (Belbin 1991, Tausch et al. 1995, Oksanen and Minchin 468 ECOGRAPHY 22.5 (1999) Fig. 1. The upper figure shows the hypothesised pattern of species response curves along an environmental gradient with possible explanatory processes (Austin 1990, reproduced with permission of Academic Press). The lower figure shows the results of fitting generalised linear models with a p-function for temperature to data from nine eucalypt species (Austin et al. 1994 reproduced with permission of Opulus Press). Note the pattern of skewness of the species response curves about the intermediate mean annual temperature of 11.5"C. Increasing skewness + Increasing role of competition t + Increasing role of physiological tolerance + 8 K ? 3 r .-. - 0.0 a 0.5 1997, Okland 1999). Failure to evaluate the issues in the vegetation literature has led some ecologists in other disciplines to continue to use linear models such as principal components analysis (PCA). A technique shown to be inadequate in the early 1970's (Austin 1985, 1999). Much of the controversy concerns the choice of ordination method. Buried under the methodology lies an ecological controversy which revolves around the nature of the non-monotonic species response shape and the robustness of the methods to differences in the response shape and species packing along gradients. The issue posed by May (1985) reECOGRAPHY 22.5 (1999) mains but the development of more detailed continuum propositions and more rigorous direct gradient techniques may lead to the development of a new paradigm for addressing the conceptual questions. The application of powerful new regression techniques, Generalised Linear Models (GLM) (Austin and Cunningham 1981, McCullagh and Nelder 1989, Nicholls 1989, Austin et al. 1990), and Generalised Additive Models (GAM) (Hastie and Tibshirani 1990, Yee and Mitchell 1991, Leathwick 1995, Austin and Meyers 1996) to vegetation analysis has expanded the opportunities to examine issues of vegetation organ469 isation (Austin et al. 1994). These methods relax many and the deeper deposits close to the eruption centre on of the constraints of traditional regression analysis. forest composition was important. The deep pumice Factors and variables can easily be combined as predic- deposits were interpreted as having favoured the curtors. Error functions other than normal are simply rent dominance of those sites by rapidly dispersing accommodated. Use of GAM means that the functional conifer species as opposed to the slower dispersing form of the predictive relationship does not have to be Nothofagus species. A statistical description (model) of predefined and a smoothing function is fitted (Yee and a complex regional pattern of species distributions was Mitchell 1991). Recognition that large sets of data obtained incorporating both environmental controls suitable for this type of statistical modelling could be and historical factors. The results support those of collated from vegetation surveys led to a number of Austin and colleagues (Austin et al. 1984, 1990) for studies of species response to environmental variables forests in south-east Australia that climate variables are (Austin et al. 1984, Margules et al. 1987, Leathwick and important predictors of regional species distributions Mitchell 1992). Presence/absence data from plots of and that the species response shapes are often complex specified size and location form a minimum data set for skewed surfaces. this type of analysis. Given the location, estimates of Leathwick (1995) extended his analysis to most of the the environmental predictors can be obtained from indigenous forest of New Zealand using ca 16000 plots geographic information systems (GIS). These methods of 0.4 ha and GAM. The results confirm the advantages have the potential to transform the regional analysis of of GAM (Yee and Mitchell 1991). The high residual vegetation, testing continuum concepts, the importance deviances associated with the models for the Nothofaof climate within regions, and the adequacy of environ- gus species provided support for the view that their mental models which ignore history and geographic distribution may be limited by volcanic destruction and barriers. As regression methods, GLM and GAM suffer slow rates of reinvasion. from all the well-known statistical limitations e.g. Following the approaches of Legendre (1993), Smith multi-collinearity, inadequacies of stepwise regression (1994), Borcard and Legendre (1994), Leathwick et al. procedures and that it is correlation not causation (1996) extended the GAM modelling approach incorpo(James and McCulloch 1990). These methods provide a rating a water balance model, spatial autocorrelation, statement of the observable pattern of relationships and Nothofagus competition (cf. Austin and Cunningbetween a species and the environment. They also ham 1981) and used it to evaluate the response of the provide a relative measure of the importance of differ- forests to global warming. Leathwick (1998) examined ent components of the environment e.g. climate versus the extent of spatial autocorrelation in the residuals history, and a statistical test of the interpretation. after fitting climatic and other environmental predictors Hopefully they will also contribute to resolving the to consider whether existing forests were in equilibrium ordination and classification issues in vegetation with the current environment. These regression models confirmed the relative importance of spatial autocorreecology. The statistical modelling work of Leathwick on the lation for Nothofagus species as compared with conifers distribution of forest trees in New Zealand provides a and broad-leaved angiosperm trees. The differences beclear example of the power of these methods (Leath- tween predictions of the environmental model with and wick and Mitchell 1992, Leathwick 1995, 1998, Leath- without spatial effects highlighted the areas of the well wick et al. 1996, 1998). New Zealand has a varied tree known “beech gaps” in New Zealand. These gaps have flora, large climatic gradients, and a turbulent history similar climates to adjacent areas with Nothofagus and of glaciation, volcanism and human invasion. The dif- the lack of invasion is ascribed to poor dispersal and/or ferential regional processes of disturbance, climate the requirement for ectomycorrhiza. For a given envichange, local elimination due to volcanic eruptions and ronment, the full range of densities of Nothofagus can subsequent migrations are all present together with be found including natural absences. Leathwick and local and regional control of forest composition due to Austin (unpubl.) have therefore been able to examine present climate, topography and lithology (Leathwick the influence of the major dominant Nothofagus on the and Mitchell 1992). Leathwick and Mitchell (1992) realised environmental niche of other tree species. investigated the distribution of forest trees in an area of There are significant improvements in the statistical ca 20000 km2 in the central North Island of New models of species when the competitive effects of Zealand using presence/absence species data from 2172 Nothofagus species are included. plots of 0.4 ha and GLM. They concluded that the The studies of Leathwick and colleagues demonstrate primary predictors of the eleven species were mean an observational methodology which can detect the annual temperature and solar radiation. Of secondary relative importance of environmental variables, historic importance were mean annual rainfall, topography and processes and biotic effects such as competition. There drainage. The depth of the Taupo pumice deposits was is now a developing literature on statistical modelling of also a significant predictor for several species. The species responses (Franklin 1995) with a variety of influence of the large Taupo Pumice eruption (130 AD) applications (Margules and Stein 1989, Lenihan 1993, 470 ECOGRAPHY 2 2 5 (1999) Beerling et al. 1995, Brzeziecki et al. 1995, Austin and Meyers 1996, Bio et al. 1998, Franklin 1998, Guisan et al. 1998, Lehmann 1998). It is not without controversy (Austin and Nicholls 1997, Oksanen 1997) and the need for performance evaluation (Austin et al. 1995). Statistical models of this kind can indicate the relative importance of variables and test assumptions about whether climate is a secondary factor or not (cf. Lawton 1982, Harrison et al. 1992, Srivastava 1999). The adoption of these types of methods under the stimulation of the continuum concept provides a paradigm of immediate relevance to the study of other forms of biodiversity and particularly current concern for a collective property such as species richness. Species richness Patterns of species richness (number of species per unit area) are a major focus of biodiversity studies of many diverse taxa (Huston 1994, Rosenzweig 1995). The current issues arise at three distinct scales. l) What determines the observed latitudinal gradients in species richness at global or continental scale? There is a voluminous literature of which Rosenzweig (1995) provides one recent summary. 2) What is the relationship between local and regional species richness? Interest in this topic has developed more recently but there is an expanding literature (Ricklefs 1987, Cornell and Lawton 1992, Ricklefs and Schluter 1993, Cornell and Karlson 1997, Zobel 1997). 3) At the local scale, the current question is the role of species richness in determining ecosystem function, particularly its influence on productivity following the early work of Al-Mufti et al. (1977) and Grime (1979). There is now a varied observational and experimental literature (Naeem et al. 1994, 1995, 1996, Tilman et al. 1996, 1997, Huston 1997, Hodgson et al. 1998, Lawton et a]. 1998, Thompson and Hodgson 1999). At each scale, recent results in vegetation ecology both in terms of concept and methodology can contribute to a more comprehensive analysis of the questions. Global species richness Numerous hypotheses have been advanced to account for the observed gradient of species richness from the equator to the poles, including area (Rosenzweig 1995) and energy (Currie 1991). These two hypotheses provide examples of the type of approach which can be adopted and of the problems which arise. The work of Currie and colleagues (Currie and Paquin 1987, Currie 1991, Francis and Currie 1998) has established relationships between tree species richness and actual evapotranspiration (AET) and between vertebrate richness and potential evapotranspiration (PET) for North ECOGRAPHY 2 2 5 (1999) America. The data are based on species ranges recorded as presence in geographical grid cells of 2.5" lat. x 2.5" long. south of 50"N, and 5" lat. x 2.5" long. north of 50"N. Currie and Paquin (1987) found a nonlinear regression relationship between tree species richness and AET with an r2 of 0.76. An additional predictor which explained a further 8% of the variance was elevation variation within the gridcell. Currie (1991) extended the analysis to various vertebrate taxa including amphibians, reptiles and mammals. Potential evapotranspiration explained at least 79% of the variance in species richness of these taxa. Other variables which were found to be important in a multiple regression analysis after taking account of PET included elevation, AET, and spatial variation in elevation and in PET. There was a very strong curvilinear relationship between PET and latitude (Currie 1991: Fig. 8a). An energy measure, a potentially direct gradient accounted for as much if not more than the indirect gradient, latitude. Environmental heterogeneity within the gridcells accounted for additional variability in species richness for both trees and vertebrates. Adams and Woodward (1989) following the approach of Currie and Paquin (1987) examined the patterns of tree species richness in Europe, North America and East Asia. They tested a modelled estimate of net primary production rather than AET and found a very strong non-linear relationship between tree species richness and primary production in each continent. They concluded that glacial history could not explain richness patterns and that differences could be explained by current climate and its impact on productivity. These results and those of Currie and Paquin (1987) and Currie (1991) were criticised by Latham and Ricklefs (1993). They indicate that the data of Adams and Woodward (1989) were not representative of east Asia. They assembled a more extensive data set, finding that though there was a relationship between species richness and AET, mean species richness was 1.9 times higher in Asia than eastern North America. Using a second data set of species richness for approximately one hectare plots, they found after taking account of differences between five regions that AET did not explain any significant additional vari- . ance. The small plot size data set was used as it was argued that large ecologically heterogeneous areas were not suitable for testing the energy-diversity hypothesis. Francis and Currie (1998) responded that for the large area data set the most appropriate model was one with area, AET and region. AET and region were strongly collinear in the small plot data set. As a consequence, either interpretation of species richness as a function of AET or region was equally plausible and could not be distinguished statistically. Kerr and Packer (1997) reanalysed the mammal data of Currie (1991), breaking the data into two sections. Below 1000 mm PET, there is a linear relationship 47 1 between mammal species richness and PET; above 1000 mm species richness and PET are independent. This is consistent with the curvilinear relationship found by Currie. The data above 1000 mm PET were then examined in relation to topographic and PET heterogeneity, and other predictors selected to test other hypotheses concerning species richness. Kerr and Packer (1997) rejected climatic favourability, climatic stability and glacial history ,as explanatory hypotheses on the basis of their regression analyses. Above 1000 mm PET, a significant regression model was found explaining 76.7% of the variance with the terms topographic heterogeneity, PET variability and coastal location. The authors state the result “contradicts previous studies of large scale richness patterns that dismissed the importance of habitat heterogeneity”. The analysis uses PET, a direct gradient sensu Austin (1980), plus a measure of its within gridcell variability and topographic heterogeneity, a measure of variability of a complex gradient, altitude. Potential evapotranspiration is strongly correlated with altitude. Variation in altitude will be associated with variation in PET. If the relationship between species richness and PET holds, cells with high variation in PET will have higher species richness than those with low variance. Two possibilities arise from this consideration of the environmental data. Firstly, only one environmental variable, PET may be associated with richness but through two variables, directly through PET and indirectly through topographic heterogeneity. Secondly, the confounding of PET and its variability above 1000 mm means we cannot be sure of the shape of the relationship above 1000 mm PET. Is there another variable associated with topographic heterogeneity apart from PET that is correlated with species richness? This brief review of the energy-diversity hypothesis raises many issues regarding the analysis of observational data sets particularly those based on large area enumerations of species distributions. Data quality is often unknown and variable. Data also need to be representative of the full range of conditions, for example the criticism of Adams and Woodward (1989) by Latham and Ricklefs (1993) regarding the absence of Asian moist-temperate forests. The use of large area gridcells or polygons must result in massive environmental heterogeneity within some gridcells. It is then necessary to incorporate some measure of heterogeneity for efficient analysis (Currie 1991, Kerr and Packer 1997). Ecological interpretation of such measures can prove difficult as indicated for Kerr and Packer’s (1997) result. Is it environmental heterogeneity in PET or some other variable or both? For patterns of species richness to have a biotic explanation species must interact (Latham and Ricklefs 1993, Huston 1999). For this reason the advantages of large area polygons over small plots distributed over large areas needs to be considered 472 carefully. The analysis of Francis and Currie (1998) emphasises the need to include multiple predictors in any regression model. The problem of multi-collinearity is then clearly apparent as with region and AET in Francis and Currie’s analysis. This is not a statistical problem but an ecological one: are the variables representing a complex indirect gradient or a functional potentially causal relationship? In contrast, Rosenzweig (1995) does not distinguish between latitude, elevation or mean July temperature in terms of the nature of their likely influence. He also asserts that “latitudinal gradients must arise because the tropics cover more area than any other zone. Their greater area stimulates speciation and inhibits extinction.” (Rosenzweig 1995 p. 284). No attempt is made by Rosenzweig (1 995) to examine more than one variable at a time. Yet both his area hypothesis and the energy hypothesis of Currie (1991) could be complementary. Neither PET or area is a pure regulator or resource gradient. There is a complex correlative path between these variables and any causal process determining species richness. A description of species richness in relation to these two variables with estimates of the correlation between PET and area for different continents might yield fresh insight into global richness patterns. For example, using the Currie data set the following model might be fitted: SR = f(PET, H,,, E, He, Area, C) (1) where SR is species richness; PET potential evapotranspiration; HFt a measure of PET heterogeneity within the grid cell; E another environmental variable; Area the area of the latitudinal zone for the continent, and C a correction factor for different sized grid cells. Statistical modelling at this scale needs to incorporate both alternative hypotheses and the influence of data heterogeneity. All ecological paradigms using observational analysis face similar problems. A concensus on the appropriate formulation of hypotheses and suitable methodology has yet to be reached. The approach used by Leathwick (1998) for analysis of forest patterns in New Zealand can be applied to species richness (Leathwick et al. 1998) and indicates a potential direction for development. Regional species richness The current interest in the relationship between local and regional species richness (Ricklefs 1987, Cornell and Lawton 1992, Ricklefs and Schluter 1993, Cornell and Karlson 1997, Zobel 1997) raises similar issues. Species richness is assumed to be a result of local biotic interactions such as competition and predation, and ECOGRAPHY 22:5 (1999) regional or historic processes such as dispersal and speciation; no intra-regional environmental control is envisaged. Local communities may be saturated with species if competition and predation are the dominant processes, or unsaturated if intra-regional dispersal and speciation are determining local richness. These alternatives are distinguished by examining the plot of local richness against regional richness; saturated regions should show a hyperbolic relationship and unsaturated a linear relationship (Cornell and Lawton 1992, Fig. 2) This comes as a surprise to a vegetation ecologist familiar with the complexity of species niches in relation to the numerous environmental gradients within a region. Do any species reach an environmental limit within a region? The results of Harrison et al. (1992) are instructive on this topic. They tested whether high beta diversity (turnover in species composition) along transects of 50 x 50 km squares was associated with poor powers of dispersal. They did this for a wide range of invertebrates, birds, mosses and vascular plants. Both p and a diversity were tested in relation to distance along transects but climate was strongly related to both the north-south and east-west transects. The authors conclude “Our results demonstrate that turnover with distance may be a relatively minor component of regional diversity, especially in the presence of strong environmental gradients,. . . Gradient-driven patterns such as we see in Britain may be characteristic of many temperate biotas at comparable scales.” This result is consistent with what is expected from continuum concept propositions and the statistical models of Austin (1987), Margules et al. (1987), Pausas (1994), Pausas and Carreras (1999, Austin et al. (1996), and Leathwick et al. (1998). These statistical models demonstrate that species richness within a region shows complex patterns of relationships with both climatic and local environmental gradients (see below). Cornell and Lawton (1992) provide a theoretical perspective on the topic of whether “local ecological communities” are ever saturated with species. They refer to local and regional processes in the following terms “ . . . predation, parasitism, competition, and abiotic fluctuation or disturbance are played out within local arenas, whereas long-distance dispersal, speciation, wide-spread extinction, and fluctuation in species’ distributions take place across broad geographic regions.” There is no mention of climatic gradients within or between regions possibly controlling species distribution. A few studies have incorporated an environmental component e.g. depth and habitat type (Cornell and Karlson 1996). The properties of the local and regional floras responding to climatic gradients across regions and to substrate heterogeneities (e.g. peat vs chalk) within regions need to be considered in this type of analysis. Fauna also respond to climate and lithology. The methodology used in this paradigm, particularly the statistical aspects, has received little attention ECOGRAPHY 2 2 5 (1999) (Cresswell et al. 1995, Srivastava 1999). The conceptual framework is represented in Fig. 2a (Cornell and Lawton 1992). An example analysis (Hawkins and Compton 1992) is presented in Fig. 2b where the maximal possible line is shown together with the quadratic curve which was fitted to the data. The effect of fitting a dependent regression (i.e. y = f(x + y)) is mentioned by Cresswell et al. (1995) but not considered in any of the examples of local versus regional relationships presented in Cornell and Karlson (1997). Of necessity, the regression line is constrained below the 45” line (Fig. 2). Several studies (Cornell and Karlson 1997) fit a quadratic regression to demonstrate possible saturation of the local species richness. A hyperbolic function as shown in Fig. 2a cannot be represented by a quadratic function which must decline. A linear function will give a better fit to the saturation curve drawn in Fig. 2a than a quadratic. Null models are used by very few authors (Partel et al. 1996, Caley and Schluter 1997), yet they may have significant implications. Note that in Fig. 2b, the data fill almost all possible combinations of local richness. One null model states that all possible values of local richness are equally probable (Partel et al. 1996). The expected value is then half the maximal local richness or half the regional richness. An alternative null model based on a binomial expansion of the possible combinations of all species, singly, in pairs etc. gives the same result (Nicholls pers. comm.). The expected relationship is then a straight line with a slope of 22.5” (Fig. 2c). Partel et al. (1996) used such a null model to test the strength of the correlation between the actual species pool and the regional species pool and between species richness. They concluded that the relationships tested differed significantly from that expected under the null model. The null model of Caley and Schluter (1997) assumes the local individuals are randomly sampled from a regional pool of species whose abundances have a canonical log-normal distribution, a further complicating assumption. Srivastava (1999) criticises Partel et al. (1996) for making between habitat comparisons (e.g. grassland with forest) rather than within a habitat where interpretation is less ambiguous. In spite of this, the work (Partel et al. 1996) represents a major step forward in the use of data. They defined the regional and actual species pools for the community on the basis of phytosociological information. The regional flora capable of growing in the community was determined from the ecological amplitude of each species recorded as indicator values in Ellenberg et al. (1992). The actual species pool was recorded from community descriptions and actual species lists from surveys, then compared with local small-scale richness in 1 m2-quadrats. Comprehensive data of this kind are only available where intensive vegetation studies have been done. Few animal taxa or communities could be analysed in this way. It would be interesting to have a similar analysis done within phytosociological associations. 473 Maximum possible Type I Type II Regional richness Maximum possible I - ii I I I Regional richness l5 Maximum possible Mean when all possible local richness values are equi-probable Regional richness Fig. 2. (a) shows the proposed relationships between local and regional species richness from Cornell and Lawton (1992 reproduced with permission of Blackwell Science). Type I is unsaturated and Type I1 is for a fully saturated local fauna. (b) shows results for fig wasps from Cornell and Karlson (1997) based on data from Hawkins and Compton (1992 reproduced with permission). The fitted curve is a quadratic. (c) shows the line expected under a particular null model where all values of local richness are equally likely. ECOGRAPHY 2 2 5 (1999) Environmental predictors of regional patterns of plant species richness have been reviewed by Pausas and Austin (unpubl.). Margules et al. (1987) showed that eucalypt species richness for a region of ca 40000 km2 in south-eastern Australia could be modelled using climatic variables, mean annual temperature, mean annual precipitation and their interaction, plus a relative measure of solar radiation. Austin et al. (1996) in a more extensive study using eight predictors, both regional climatic variables and local topographic factors, showed that various measures of species richness (total tree species, eucalypt species, rainforest species and richness of Eucalyptus subgenera) could all be described by curvilinear GLM models based on 7208 plots. The models for species richness of the eucalypt subgenera Monocalyptus and Symphomyrtus raise significant issues of interpretation (Austin et al. 1996, Austin 1998). The models suggest that there is an optimum environment where Monocalyptus species richness is maximal. Species richness for Symphomyrtus is highest in other environments and the maxima are complimentary with reference to climatic variables. The processes generating optimal environments for the species richness of a subgenus with associated displacement of another are obscure (Austin 1998). The predicted patterns are however consistent with independent circumstantial evidence (Noble 1989). Pausas (1994) and Pausas and Carreras (1995) showed that different growth forms in pine forests in the Pyrenees had predictable patterns of species richness in relation to environmental variables. A model has also been developed for arboreal marsupial species richness (Pausas et al. 1995). The model contains climatic predictors as well as specifically faunal predictors such as food quality and potential presence of hollows for nesting. These papers and others cited in Pausas and Austin (unpubl.) support the contention that species richness patterns within regions vary strongly with environment. This is in marked contrast with the regional richness control hypothesis adopted by animal ecologists (Cornell and Karlson 1997, Gaston and Blackburn 1999, Lawton 1999). Srivastava (1999) comments in discussing regional effects on herbivore insect communities on bracken Pteridium aquilinum in South Africa " . . . two species appear to have range limits within the region under study". Perhaps bracken herbivore communities have a continuum type of organisation and show individualistic responses to climate. The two paradigms being contrasted here differ in several aspects. Some vegetation ecologists use plot data assumed to be homogeneous, GLM or GAM statistical models and fit descriptive correlative models with environmental variables. Some other ecologists working at the same organisational level use various units (geographic gridcells, ponds, communities or a single fig fruit, Cornell and Karlson 1997), classical linear often uni-variate regression but with a specific ECOGRAPHY 2 2 5 (1999) hypothesis (Austin 1999). Srivastava (1999) in discussing the regional richness approach accepts ponds as a suitable local unit. Ponds are highly environmentally diverse internally. An acid pond will have a different flora from an alkaline one and may also differ in size, depth and hydrology. Use of heterogeneous units will obscure more than it reveals. A model similar to eq. (1) might be used to distinguish between the hypotheses but agreement on units would be necessary first. The real question is not which paradigm is right, it is what is the relative importance of different processes in determining the composition and collective properties of vegetation and faunal assemblages? Local species richness While regional and global patterns of species richness are actively researched, it remains true that little is known about the processes determining local richness patterns in vegetation. Al-Mufti et al. (1977) and Grime (1979) discovered the humped-back curve of species richness in response to a gradient in productivity/fertility. This has assumed importance in recent years (Huston 1993, 1994, Tilman and Pacala 1993) and generated controversy concerning its application to biodiversity conservation and sustainable agriculture (Huston 1993, Margules and Gaston 1994). The postulated mechanism to account for the pattern is that under high productivity, superior competitors suppress other species while under low productivity few species can tolerate the extreme conditions. It is only under intermediate conditions that higher numbers of species can maintain themselves in a community (Grime 1979). This has led to a number of developments (Grime et al. 1988, Tilman and Pacala 1993, Huston 1994). Tilman and Pacala (1993) outline a number of explanations for the unimodal pattern of species richness including changes in nutrient competition, soil spatial heterogeneity, light competition and increasing plant size along the productivity gradient. They also present a number of examples of unimodal graphs where productivity is equated with biomass, drainage, biomass plus litter, soil PO,, a normalised phosphorus plus potassium index and a climatic moisture index. It is assumed that a productivity gradient has an identical effect on species richness regardless of the type of gradient which may show an increase in biomass as the variable increases. This has been challenged by Austin and Gaywood (1994). They point out that a universal pattern of that kind cannot be distinguished from a more restricted response to one environmental gradient conditional on a second, see Fig. 3. The ring response of species richness to biomass equivalent to hump-back response occurs when maximal biomass is assumed to occur under intermediate conditions for two environmental gradients. This pattern (Fig. 3b) cannot be 475 distinguished from Fig. 3d unless the observations or the experiments are conducted over the full two-dimensional environmental space. Such studies do not seem to have been done. Goldberg and Novoplansky (1997) review field competition experiments along gradients with respect to competitive intensity and the pulsed nature of resources. They distinguish between resources which may be accumulated ‘by plants (nutrients) and those less likely to accumulate (water). Water may act as a resource under low rates of supply and as a toxic direct gradient sensu Austin (1980) when creating waterlogged conditions. Goldberg and Novoplansky (1997) expect a “greater difference in competition intensity between xeric and mesic environments than between infertile and fertile environments. However, the few field experiments along water gradients show highly variable results that are not consistent with this prediction.” Their hypothesis implies that these gradients should be different. The inconsistent results indicate that the gradients or the experimental designs for the different gradients differ sufficiently to warrant further study. If competitive intensity differs, then species richness along gradients might be expected to differ too. Goldberg and Miller (1990) investigated the impact of increased watering and nutrients (nitrogen, phosphorus and potassium) on a weed community. The addition of water led to a greater increase in biomass than did addition of nitrogen. but species diversity declined only with nitrogen addition. Light levels were similar in both treatments, so the difference was not explained by differential canopy closure and competition for light. Mortality was consistently higher in the nitrogen treatments. Observational analyses of natural communities and field experiments, both using a two-dimensional gradient design, are needed to resolve this issue (see next section). Because studies of the humped-back curve have focussed on a collective property of vegetation, we are left with another problem. What we do not know is whether the humped-curve reflects the co-occurrence of species maxima (realised niche optima) or the zone of overlap of species limits. Experimenta1 community ecology The dominant research paradigm in plant ecology in the 1960’s and 1970’s was that of the experimental study of plant competition using pair-wise species comparisons (Harper 1977). The accepted methodology was the de Wit replacement series (de Wit 1960). Annual weeds were the preferred experimental material. Diffuse competition between perennial species in multispecies communities was not commonly examined. If environmental conditions were studied, analysis of variance was the standard statistical procedure. Tests were made of the significance of two or three levels of a factor. The experimental testing of hypotheses derived from multivariate vegetation analysis was rare, though see Goldsmith (1973a, b). The use of resource gradients and estimation of species response curves was not done. Since the late 1970’s, a more truly community-level experimental ecology has developed where perennials, multispecies mixtures and environmental gradients are common place (Grime 1979, Tilman 1982, 1988, Grime et al. 1988, Keddy 1989). A wide variety of experimental designs and methods are now used. Various schools of experimental ecology have developed associated with the names of Grime, Tilman and Keddy. Most recently debate between these schools and others has centred upon the relationship between species richness and ecosystem function (Tilman and Downing 1994, Naeem et al. 1994, 1995, 1996, Johnston et al. 1996, Tilman 1996, Tilman et al. 1996, Garnier et al. 1997, Huston 1997, Hodgson et al. 1998, Lawton et al. 1998, Thompson and Hodgson 1999). A large and growing literature has developed on the topic (Lawton et al. 1998) which will not be reviewed in detail here. Some observations pertinent to the thesis being developed here of the relevance of vegetation ecology to biodiversity studies will be made. Huston (1997) drew attention to a difficulty with the Tilman experiment (Tilman and Downing 1994, Tilman 1996). Tilman and Downing claimed that increasing species richness made plant ECOGRAPHY 2 2 5 (1999) @‘* Envimnmenm!gradient. €1 Fig. 3. Possible biomass and species richness patterns in a two-dimensional environmental space (dark shading indicates maximum values). (a) Biomass with a central maximum: (b) consequent species richness response with a maximum at intermediate biomass values: (c) projection of environmental space onto a biomass ( = productivity) gradient, where all radial gradients (AX to AY and AX to AZ) are projected onto a single biomass gradient AB: (d) an alternative pattern of species richness: if the pattern were projected onto the biomass gradient AB in (c), it would be indistinguishable from the pattern in (b). Reproduced with permission of Opulus Press. communities more resistant to the effects of drought, see also Tilman (1996). Huston (1997) points out that the level of species richness was a result of a nitrogen treatment and the response to drought could be interpreted as an interaction between rainfall and the additional biomass produced on the high nitrogen/low richness plots. In the experiments by Naeem et al. (1996) and Tilman et al. (1996), a selection probability effect ensures that species mixtures with higher numbers of species will contain the larger more productive species. Comparison of the mean biomass for 1, 2, 4, 8, etc. species mixtures must result in increasing biomass with increasing species richness (Huston 1997). The concentration on biological variables to the exclusion of the environment led to problems of interpreting the results. The choice of analysis in the Naeem et al. (1996) experiment led to results where their Fig. 2 shows that maximum biomass in certain monocultures was greater than in any of the 8 or 16 species mixtures. If biomass production increases with species richness, how does this relate to the hump-backed curve of species richness with biomass? Subsequent studies (Tilman et al. 1997, Lawton et al. 998) have focused on the influence of functionally ifferent types of plants e.g. grasses and legumes. The behaviour of individual species in relation to these richness gradients has not usually been examined, though see Naeem et al. (1996) and Tilman et al. 7). It must be of some concern that the relations between ecosystem functions and species richness do not have an r2 > 0.22, or for functional richness and composition > 0.36 (Tilman and Downing 1994, Tilman et al. 1996, 1997). Some unknown factors are exercising a greater influence on ecosystem production and function than is diversity. If the species response patterns of Ellenberg (Mueller-Dombois and Ellenberg 1974, Austin 1999) as part of a vegetation continuum (Austin and Smith 1989) are correct, then all biotic properties are functions of environment. The ultimate role of environmental gradients in determining the biological composition and functioning of ecosystems has yet to be investigated. An independent approach to the study of plant species competition was developed by Ellenberg (1953, 1954) in Germany concentrating on competition along gradients. An experiment with six grass species along a water-table gradient was carried out. When grown together in mixture the species showed marked shifts in their ecological (realised niche) compared to their physiological (fundamental niche) maxima in monoculture (Ellenberg 1953). Each species came to dominate a different segment of the water-table gradient. Subsequent work extended the approach to a nutrient gradient, with different levels of species richness (five and ten grass species), and developed a statistical analysis procedure for the experimental design (Austin and Austin 1980, Austin 1982). Species response under multispecies ECOGRAPHY 22:5 (1999) competition was shown to be a function of the particular species present in the mixture and the position along the nutrient gradient. Maximal biomass varied with the number of species depending on the position along the gradient (Austin and Austin 1980). Other studies have been carried out with thistles (Austin et al. 1985) and grass/legume mixtures under ambient and elevated C 0 2 (Navas et al. in press). Microcosms are becoming part of plant community ecology (Fraser and Keddy 1997). The experimental design usually consists of different treatments with a limited number of levels which are tested for whether they have a significant effect on species or vegetation properties. Compare Weiher and Keddy (1995) with seven treatments each at two levels with Austin and Austin (1980) with one treatment and 16 levels. The types of questions and the possible degree of resolution of the response are quite different, see also Keddy et al. (1994). If the responses of species and vegetation to gradients results are complex response curves, these will not be detected with less than five treatment levels. Campbell and Grime (1992) in a novel experimental design examined multispecies responses in relation to two environmental gradients. They constructed continuous gradients of nutrient concentration and disturbance in the same container. This allowed two gradients each with five levels to be studied together with the interactions in one container i.e. 25 combinations. Monocultures for seven species and the seven species mixture were replicated three times. The spatial autocorrelation between residuals for the neighbours was not significant. This experimental design (see also Burke and Grime 1996) offers the opportunity to explore species and plant community responses to an environmental space in a logistically efficient manner. The equivalent large-scale field experiments at the mercy of the weather and with the necessity of weeding are costly and logistically demanding. Gradient experiments with numerous levels and response surface analysis rather than ANOVAs offer advantages. The complex responses of individual species and of collective vegetation properties can then be examined in detail. A synthesis between observational vegetation ecology and experimental plant ecology will then be possible. Conservation studies Vegetation ecology has seen the establishment of large data bases of species by sites. Three distinct types can be recognised: a) phytosociological association tables where the site is a selected relevee of varying size (Westhoff and van der Maarel 1978); b) vegetation survey data where the site is a quadrat or plot of a specific size and the survey design may have been stratified or purposive; c) geographic distribution data 477 where species occurrence is recorded for large contigu- may contain species that are different from those in ous gridcells, and may be based on survey data of an species rich areas and have high conservation value. unspecified type or herbarium specimens. In general, Hotspots may occur for two main reasons. An unusual fauna databases on geographical distribution are re- environment may support endemic species or relict stricted to type (c). The limitations of such data for species from a previous climatic regime. An area may conservation and biodiversity studies are discussed by have high species richness due to steep local environmental gradients. Species richness may result from the Margules and Austin (1994) (see also Austin 1998). Very large phytosociological databases can be devel- artificial combination of species e.g. counting Koalas oped e.g. 35000 samples for the account of British and Grey kangaroos together when one is an arboreal Plant Communities (Rodwell 1991). The conceptual folivore and the other a ground dwelling herbivore. framework of subjectively identifiable associations and There is a strong case for combining the vegetation hierarchical syntaxonomic classification has practical paradigm ecology with conservation evaluation apvalue but remains controversial (Westhoff and van der proaches, while reducing the emphasis on species Maarel 1978, Rodwell 1991). The primary data have richness. immense value for conservation independent of the conceptual framework providing the subjective sampling bias is recognised. Quadrat surveys can also gen- Discussion erate large data bases. Grime et al. (1988) use a data base of > 10000 quadrats for a region in central Eng- The view of plant community ecology presented here, land of 3000 km2. Information on species preferences of a vegetation continuum embedded in environmental can also be built up from such survey data, for example space using statistical models to describe patterns in the indicator values of species for various environmen- species and collective properties such as species richness tal gradients developed by Ellenberg et al. (1992, see or dominance, is a very partial and personal view. A also Grime et al. 1988). The results of analysing these synthesis of vegetation ecology paradigms is needed. At data bases so far have reflected the particular paradigm the present time an increased dialogue between animal adopted by the authors. Australian experience has and plant community ecologists is probably of greater shown that a minimum data set of plots with presence/ priority. Lawton (1999) has provided a rather pessimistic reabsence data collated from various surveys, plot locaview of community ecology. “[It] is a mess, with so tion and a GIS can make a significant contribution to conservation evaluation (Austin 1998). Statistical mod- much contingency that useful generalisations are hard to find.” He is concerned that any laws, rules and elling of such data allows various different approaches mechanisms will be contingent i.e. “only true under to conservation evaluation to be adopted, see for examparticular or stated circumstances.” A more extreme ple Margules and Nicholls (1987), Margules and Stein quote is “the rules are contingent in so many (1989), and Margules and Austin (1991). State agencies ways. . . as to make the search for patterns unworkin New South Wales have adopted these methods in a able.” He is similarly concerned about population dymajor study of forest conservation requirements in the namics and advocates the search for pattern at the State (e.g. Anon. 1994a, b). Several hundreds of thou- macro-scale, that is the study of macroecology (Brown sands of hectares of forest were reserved as a result of 1984, 1995, Brown and Maurer 1989). However, as the the study. Much of the Australian work in this field has reviews above of species richness at various scales have been put together as a manual and package of com- indicated, even global patterns of species richness are puter programs called BIORAP (Margules et al. 1995). contingent on glacial history, continental patterns of It represents the collation of experience in vegetation mountain barriers, and current climate. Most results of ecology, climate estimation, environmental analysis, ecological studies are contingent on environment and and conservation evaluation using complementarity. history. What can be done? Examining and developing the links between paradigms There is a more intimate interdependency between in conservation ecology and mainstream ecology would concept, analytical methods and data than is commonly repay a detailed review. recognised in ecology (Austin 1998, 1999). Lawton There is an extensive literature on conservation eval- (1999) quotes as an example of useful generalisations in uation which is based on a very different paradigm but ecology, a diagram from Whittaker (1975) which shows also using databases and GIs. Vegetation communities the relationship between plant biomes and two climatic are mapped and wildlife habitat models are developed variables, mean annual temperature and mean annual based on the assumption of homogeneous plant com- precipitation. Whittaker’s earlier work on regional vegmunities over large areas (Scott et al. 1993). Yet an- etation patterns using direct gradient analysis is not other conservation focus is on species richness with the mentioned. This work on the vegetation of the Great detection and conservation of species rich “diversity Smokey Mountains (Whittaker 1952, 1956), the hotspots” (Prender@st ef al. 1993). However, this ap- Siskiyou Mountains (Whittaker 1960) and the Santa proach neglects the possibility that species poor areas Catalina Mountains. (Whittaker and Niering 1965) was 478 ECOGRAPHY 2 2 5 (1999) of the pioneering works which established the ,.ontinuurn concept. The results also show the existence of strong within-region patterns. There are general patterns of community organisation at regional scale with continuous change in composition, structure and collectiva properties contingent on environmental gradients. Whittaker (1952) demonstrated that similar patterns oc~uf for individual insect species with no species having identical patterns in a two-dimensional environmental Wace of elevation and topographic moisture. Brown (1984) mentions the work of Whittaker but then proceeds to emphasise geographic ranges rather than environmental gradients. One possible explanation is the difficulty of collecting suitable data for such analysis for animals. Another contributing factor may be the traditional concern with the temporal patterns of faunal abundance at a location rather than spatial distribution patterns. Statistical models capable of establishing quantitatively the existence of these patterns now provide a firm basis for their existence (Austin et al. 1984, 1990, 1994, Margules et al. 1987, Leathwick and Mitchell 1992, Franklin 1995, 1998, Leathwick 1995, 1998). Given the work of Whittaker (1952) and others (e.g. Bond 1957, Beals 1960, Rotenberry and Wiens 1980 and Sabo 1980), it is perhaps time that vegetation ecology paradigms were re-examined and re-applied to faunal data. Lawton (1999) discusses Schoener’s (1986) attempt to develop a contingent theory for community ecology. Schoener defines six primitive organismic axes (body size, generation time, etc.) and six environmental axes (severity of physical factors, spatial fragmentation etc.). These axes create a twelve-dimensional space within which community assembly patterns might be found. Lawton (1999) is sceptical: “whether it is worth the effort to shoe-horn every study into a point in n-dimensional space defined by combinations of at least 12 primitive axes, I have my doubts.” Yet, vegetation properties and species distributions are now being regularly examined in a seven-dimensional environmental space (Austin et al. 1994, Austin and Meyers 1996, Leathwick et al. 1998, see Pausas et al. 1995 for faunal example). The organismic axes can be regarded as equivalent to the functional attributes of plant ecologists (Grime et al. 1988, Ellenberg et al. 1992, Smith et al. 1997, Westoby 1998). These are now a significant concern for those plant ecologists investigating climate change (Smith et al. 1997). The pioneering work of the Sheffield school (Grime and Lloyd 1973, Grime 1979, Grime et al. 1988) in this respect represents a standard against which to judge the importance of functional attribute patterns. Perhaps the vegetation studies in organismic and environmental space need to be reviewed before the topic is dismissed. The expression of any biotic component is contingent on the environment. A first step would therefore be to examine patterns of biotic properties and processes in a relevant ECOGRAPHY 225 (l999) multidimensional environmental space. By relevant I mean not indirect gradients where physiologically influential variables are confounded in a location specific manner. The formulation of Schoener (1986) resembles a simpler formulation of Jenny (1941), the functional factorial approach to the study of soil properties (s): s = f(c1, p, r, 0,t...) where the five principal factors are regional climate (cl), parent material (p), topography (r), biota (0)and time (t). An array of soil properties can then be studied as a function of the environmental factors. Major (1951) suggested it could equally well be applied to vegetation. (In fact it could apply to any other dependent variable in an ecosystem.) Billings (1952) and Crocker (1952) commented on the mathematical limitations, the need to account for interactions of plants and the sheer complexity of the functional relationships which were known to occur (cf. Lawton 1999). Jenny (1941) intended the equation to be a guide. One would study a soil property, say soil nitrogen, as a quantitative function of rainfall while controlling as far as possible for the other factors i.e. Many of the studies quoted in this review do not bother to examine whether environmental variables may influence or confound their analysis. Is climate uniform within the region? Is parent material confounded with climate in the study? Does topographic heterogeneity within geographic pixcels contribute to the residual deviance? Can these experimental results obtained on sandy glacial till be extrapolated to heavy clay soils? This check list could materially improve analysis of community ecology issues. Vegetation ecologists have not adopted this approach with the exception of one neglected study. Perring (1958, 1959, 1960) explicitly adopted Major’s (1951) suggestion to investigate climatic and topographic gradients in chalk grassland. He examined species patterns in relation to aspect and slope gradients stratifed by four climatic areas while controlling for parent material, history and recent disturbance (Perring 1958). It was a graphical analysis without the benefit of statistics or computers. It does clearly show that patterns of species distribution in relation to topographic gradients are contingent on climate (Perring 1959). This contingency is however a regular ordered response to a climatic gradient probably associated with evaporative stress (Perring 1960). The towns nearest the different climatic study areas were York, Cambridge, Dorchester and Rouen. The inferences which can be drawn from this old study are that any study of biotic patterns covering England and Northern France cannot consider that the region is 479 environmentally homogeneous or that species response patterns across the region are insensitive to the local climatic or environmental gradients. This would also apply to a region containing any two of the adjacent areas. The framework incorporates currently important factors such as the regional flora as a potential determinant of the dependent variable. The questions then and today are what is the relative importance of the different factors and does their importance vary in a predictable way with environment or taxa? In other words, are the values of local contingent predictors of biotic patterns predictable from intra-regional gradients? Chapin et al. (1996) have recently focussed on this conceptual framework in the context of the sustainability of ecosystems. They use the Jenny equation as the framework within which interactions within the ecosystem between the dependent variables, vegetation and soils and their feedback to processes such as soil erosion and nutrient cycling can be considered. The weakness of the vegetation ecology paradigm presented here and of some other vegetation ecology studies is the absence of any allowance for spatial pattern. The book edited by Tilman and Kareiva (1997) illustrates the potential importance of the spatial arrangement of organisms and their environment. The key questions are again, what is the relative importance of spatial patterns and how does that importance change with environment? The proximal causes of ecological patterns and temporal behaviour are highly contingent as maintained by Lawton (1999). There is another question to be asked, are there any patterns in the importance of different proximal causes as a function of the distal environmental causes? A synthesis of the questions posed by animal ecologists with the environmental framework and statistical methods of vegetation ecologists may yield a less pessimistic outcome than that implied by Lawton (1999). Gaston and Blackburn (1999) present a defensive review of the role and importance of macroecology in ecology. They contrast observational analysis with reductionist experimental analysis. This has been an ongoing debate throughout the history of ecology (McIntosh 1985). The increasing importance of plant community experiments (Campbell and Grime 1992, Tilman and Downing 1994, Naeem et al. 1996, Tilman 1996) and their informed criticism (Huston 1997, Grime 1998, Hodgson et al. 1998) is leading to better experiments (Tilman et al. 1997), and a greater interface between observational plant ecology and experimentation. A more valid criticism of macroecology is the failure to adopt a sufficiently critical conceptual framework and analytical methodology when addressing large scale ecological patterns. Three strategies of analysis may be adopted. A collective biodiversity property may be correlated with a single predictor. A significant result can be regarded as a robust result having been obtained against a background of numerous other sources of variation. If the predictor is a complex indirect variable e.g. latitude, then a biological explanation of the relationship may still be elusive. A second strategy is multiple regression with well-defined direct or resource gradients as potential predictors, testing a particular hypothesis. The analysis of Currie (1991) testing the energy hypothesis for species richness is a clear example of this approach. The difficulty of distinguishing between regional and energy hypotheses due to collinearity problems remains (Francis and Currie 1998). The potential complementarity of hypotheses such as the correlation of terrestrial area with latitude (Rosenzweig 1995) and the energy relationship of Currie (1991) could also be addressed. A third as yet unrealised strategy would be to use a functional factorial approach similar to that of Jenny (1941) to both design the collection of data and to stratify the data for statistical analysis in order to test multiple hypotheses. This would require that the species data are such that the species actually co-occur and interact at a local scale (Huston 1999). It is only in this way that regional or global scale patterns can have ecological meaning. The environmental gradients selected would also need to be consistent with the biology of the taxa studied. One reason for the potential contribution of vegetation ecology to addressing these issues is that plants are easy to sample and measure; they do not run away. Large stratified surveys and the resulting data bases can be established relatively cheaply. Plot data at a scale relevant to species or individual interactions can be collected and within plot environmental heterogeneity relatively easily allowed for. To collect equivalent unambiguous data for fauna will be much more difficult and costly. Although the continuum concept seems to be generally accepted by vegetation ecologists its potential for organising questions about the community level of ecology has yet to be realised. If the continuum concept (Austin and Smith 1989) has any validity, then characteristics of species’ responses and patterns of collective properties such as species richness, dominance, standing biomass and assembly rules should be detectable in appropriate environmental spaces. Ecological processes are contingent on environment, space and history. This is true whether we study local, regional or global patterns. The research issues are how to determine the relative importance of the different distal causes and to determine how contingent the processes are on environment etc. This is no easy task given the high degree of correlation which exists among potential predictors. The concepts and methods of vegetation ecology have a role to play if better communication between the different paradigms can be established. ECOGRAPHY 2 2 5 (1999) Austin, M. P. et al. 1994. Determining species response functions to an environmental gradient by mealis of a P-func. tion. J . Veg. Sci, 5: 21 j-228. Austin. M. P. et al. 1995. Modelling of landscape patterns and processes using biological data. Subproject 5: simulated data case study. Rep. ERIN. CSIRO Wildlife and Ecol.. Canberra. Beak. E. W. 1960. Forest bird communities in the Apostle References Islands of Wisconsin. - Wilson Bull. 72: 156-181. Beerling. D. J.. Huntley, B. and Bailey. J. P. 1995. Climate and Adams, J. M. and Woodward, F. I. 1989. Patterns in tree the distribution of Fullopiu , j q w ? i a i : use of an introduced species richness as a test of the glacial extinction hypothespecies to test the predictive capacity of response surfaces. sis. - Nature 339: 699-701. - J. Veg. Sci. 6: 269-282. Anon. 1994a. Fauna of north-east NSW forests. North-east Belbin, L. 1991. Semi-strong hybrid scaling. a new ordination forests biodiversity study report No. 3. Int. Rep., NSW algorithm. - J. Veg. Sci. 2: 491-496. National Parks and Wildlife Serv.. Sydney. Billings, W. D. 1952. The environmental complex in relation Anon. 1994b. Flora of north-east NSW forest. North-east to plant growth and distribution. Quart. Rev. Biol. 27: forests biodiversity study report No. 4. Int. Rep., NSW 251 -265. National Parks and Wildlife Serv.. Sydney. Bio, A. M. F., Alkemade, R. and Barendregt, A. 1998. DeterAl-Mufti, M. M. et al. 1977. A quantitative analysis of shoot mining alternative models for vegetation response analysis: phenology and dominance in herbaceous vegetation. J . a non-parametric approach. J. Veg. Sci. 9: 5-16. EcoI. 65: 759-791. Austin, M. P. 1980. Searching for a model for use in vegeta- Bond, R. R. 1957. Ecological distribution of breeding birds in the upland forests of southern Wisconsin. - Ecol. Monogr. tion analysis. - Vegetatio 42: l 1-21. 27: 351-384. Austin. M. P. 1982. Use of relative physiological performance value in the prediction of performance in multispecies Borcard, D. and Legendre, P. 1994. Environmental control and spatial structure in ecological communities: an exammixtures from monoculture performance. - J. Ecol. 70: ple using oribatid mites (Acari, Oribatei). - Environ. Ecol. 559-570. Stat. 1: 37-61. Austin, M . P. 1985. Continuum concept, ordination methods, Brown, J. H. 1984. On the relationship between abundance and niche theory. Annu. Rev. Ecol. Syst. 16: 39-61. Austin, M. P. 1987. Models for the analysis of species response and distribution of species. - Am. Nat. 124: 255-279. to environmental gradients. Vegetatio 69: 35-45. Brown, J. H. 1995. Macroecology. Chicago Univ. Press. Austin, M. P. 1990. Community theory and competition in Brown, J. H. and Maurer, B. A. 1989. Macroecology: the vegetation. - In: Grace, J. B. and T'ilman, D. (eds), division of food and space among species on continents. Perspectives on plant competition. Academic Press, pp. Science 243: 1145-1150. 215-238. Brzeziecki, B., Kienast, F. and Wildi, 0. 1995. Modelling Austin, M. P. 1998. An ecological perspective on bidiversity potential impacts of climate change on the spatial distribuinvestigations: examples from Australian eucalypt forests. tion of zonal forest communities in Switzerland. - J. Veg. - Ann. Missouri Bot. Gard. 85: 2-17. Sci. 6: 257-268. Austin, M . P. 1999. A silent clash of paradigms: some incon- Burke, M. J. W. and Grime, J. P. 1996. An experimental study sistencies in community ecology. - Oikos 86: 170-178. of plant community invasibility. - Ecology 77: 776-790. Austin, M . P. and Austin, B. 0. 1980. Behaviour of experi- Caley, M. J. and Schluter, D. 1997. The relationship between mental olant communities alone a nutrient gradient. - J. local and regional diversity. Ecology 78: 70-80. Ecol. 68': 891-918. Campbell, B. D. and Grime, J. P. 1992. An experimental test Austin, M. P. and Cunnineham. R. B. 1981. Observational of plant strategy theory. - Ecology 73: 15-29. analysis of environmenh gradients. Proc. Ecol. SOC. Chapin, F. S. 111, Torn, M. S. and Tateno, M . 1996. Principles Aust. 11: 109-119. of ecosystem sustainability. - Am. Nat. 148: 1016-1037. Austin, M. P. and Gaywood, M. J. 1994. Current problems of Cornell, H. V. and Lawton, J. H. 1992. Species interactions, environmental gradients and species response curves in local and regional processes, and limits to the richness of relation to continuum theory. - J. Veg. Sci. 5: 473-482. ecological communities: a theoretical perspective. J. Austin, M . P. and Meyers, J. A. 1996. Current approaches to Anim. Ecol. 61: 1-12. modelling the environmental niche of eucalypts: implications for management of forest biodiversity. - For. Ecol. Cornell, H. V. and Karlson, R. H. 1996. Species richness of reef-building corals determined by regional and local proManage. 85: 95-106. cesses. J. Anim. Ecol. 65: 233-241. Austin, M. P. and Nicholls, A. 0. 1997. To fix or not to fix the species limits, that is the ecological question: response to Cornell, H. V. and Karlson, R. H. 1997. Local and regional processes as controls of species richness. - In: Tilman, D. Jari Oksanen. - J. Veg. Sci. 8: 743-748. and Kareiva, P. (eds), Spatial ecology: the role of space in Austin, M. P. and Smith, T. M. 1989. A new model for the population dynamics and interspecific interactions. Princecontinuum concept. Vegetatio 83: 35-47. ton Univ. Press, pp. 250-268. Austin, M. P., Cunningham, R. B. and Fleming, P. M. 1984. New approaches to direct gradient analysis using environ- Crawley, M. J. 1997. The structure of plant communities. In: Crawley, M. J. (ed.) Plant ecology, 2nd ed. Blackwell, mental scalars and statistical curve-fittine mocedures. pp. 475-531. Vegetatio 55: 11-27. Austin. M. P.. Nicholls. A. 0. and Mareules. C. R . 1990. Cresswell, J. E., Vidal-Martinez, V. M. and Crichton, N . J. 1995. The investigation of saturation in species richness of Measurement of the realised qualitative iiche: environmencommunities: some comments on methodology. - Oikos tal niches of five Eucalyptus species. - Ecol. Monogr. 60: 72: 301-304. I61 177. Austin, M. P., Pausas, J. G. and Nicholls, A. 0. 1996. Patterns Crocker, R. L. 1952. Soil genesis and the pedogenic factors. Quart. Rev. Biol. 27: 139-168. of tree species richness in relation to environment in southeastern New South Wales, Australia. Aust. J. Ecol. 21: Currie, D. J. 1991. Energy and large-scale patterns of animaland plant-species richness. Am. Nat. 137: 27-49. 154- 164. Austin, M . P. et al. 1985. Relative growth of six thistles along Currie, D. J. and Paquin, V. 1987. Large-scale biogeographical patterns of species richness of trees. - Nature 329: 326a nutrient gradient with multispecies competition. - J. Ecol. 73: 667-684. 327. ~~ ~ ~ k ~ o w l e d ~ c n l eI nthank A . B. Wellington. J . R. W. Reid. t.~ D, M. Sprdtt and A . J. Kearns for useful discussions and on the manuscript. D. Grice for assistance with the figures and preparation of the manuscript and the Divisional library staff for assistance with references. ECOGRAPHY 22 5 (1999) Curtis, J. T. 1959. The vegetation of Wisconsin: an ordination of plant communities. - Univ. Wisconsin Press. de Wit, C. T. 1960. On competition. - Versl. Landbouwkonderz. 66: 1-82. Ellenberg, H. 1953. Physiologisches und okologisches Verhalten derselben Pflanzenarten. - Ber. Deutsch Bot. Ges. 65: 351-362. Ellenberg, H. 1954. Uber einige Fortschritte der kausalen Vegetationskunde. - Vegetatio 5/6: 199-21 1. Ellenbere. H. 1988. Vegetation ecologv of central EuroDe. Camcridge Univ. Press. Ellenberg. H. et al. 1992. Zeigerwerte von Pflanzen in Mitteleuropa: - Scr. Geobot. 18: Faith, D. P., Minchin, P. R. and Belbin, L. 1987. Compositional dissimilarity as a robust measure of ecological distance. - Vegetatio 69: 57-68. Francis, A. P. and Currie, D. J. 1998. Global patterns of tree species richness in moist forests: another look. - Oikos 81: 598-602. Franklin, J. 1995. Predictive vegetation mapping: geographic modelling of biospatial patterns in relation to environmental gradients. - Prog. Physical Geogr. 19: 474-499. Franklin, J. 1998. Predicting the distribution of shrub species in southern California from climate and terrain-derived variables. - J. Veg. Sci. 9: 733-748. Fraser, L. H. and Keddy, P. 1997. The role of experimental microcosms in ecological research. - Trends Ecol. Evol. 12: 46-49. Garnier, E. et al. 1997. A problem for biodiversity-productivity studies: how to compare the productivity of multispecific plant mixtures to that of monocultures? - Acta Oecol. 18: 657-670. Gaston, K. J. and Blackburn, T. M. 1999. A critique for macroecology. - Oikos 84: 353-368. Giller, P. S. 1984. Community structure and the niche. Chapman and Hall. Goldberg, D. E. and Miller, T. E. 1990. Effects of different resource additions on species diversity in an annual plant community. - Ecology 71: 213-225. Goldberg, D. E. and Novoplansky, A. 1997. On the relative importance of competition in unproductive environments. - J. Ecol. 85: 409-418. Goldsmith, F. B. 1973a. The vegetation of exposed sea cliffs at South Stack. Anglesev. I. The multivariate aDDrOaCh. - J. ._ Ecol. 61: 787-8fi. Goldsmith. F. B. 1973b. The vegetation of exDosed sea cliffs at South Stack, Anglesey. 11. gxperimental studies. - J. Ecol. 61: 819-830. Grime, J. P. 1979. Plant strategies and vegetation processes. Wiley. Grime, J. P. 1998. Benefits of plant biodiversity to ecosystems: immediate, filter and founder effects. - J. Ecol. 86: 902910. Grime, J. P. and Lloyd, P. S. 1973. An ecological atlas of grassland plants. - Edward Arnold. Grime, J. P., Hodgson, J. G. and Hunt, R. 1988. Comparative plant ecology: a functional approach to common British species. - Unwin Hyman, London. Guisan, A., Theurillat, J.-P. and Kienast, F. 1998. Predicting the potential distribution of plant species in an alpine environment. - J. Veg. Sci. 9: 65-74. Harper, J. L. 1977. Population biology of plants. - Academic Press. Harrison, S., Ross., S. J. and Lawton, J. H. 1992. Beta diversity on geographic gradients in Britain. - J. Anim. Eco~.61: 151-158. Hastie, T. and Tibshirani, R. 1990. Generalised additive models. - Chapman and Hall. Hawkins, B. A. and Compton, S. G. 1992. African fig wasp communities: undersaturation and latitudinal gradients in species richness. - J. Anim. Ecol. 61: 361-372. Hodgson, J. G. et al. 1998. Does biodiversity determine ecosystem function? The Ecotron experiment reconsidered. - Funct. Ecol. 12: 843-848. I _ Huston, M. A. 1993. Biological diversity, soils and economics. - Science 262: 1676-1680. Huston, M. A. 1994. Biological diversity: the coexistence of species on changing landscapes. - Cambridge Univ. Press. Huston, M. A. 1997. Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. - Oecologia 110: 449-460. Huston, M. A. 1999. Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. - Oikos 86: 393-401. James, F. C. and McCulloch, C. E. 1990. Multivariate analysis in ecology and systematics: panacea or pandora’s box. Annu. Rev. Ecol. Syst. 21: 129-166. Jenny, H. 1941. Factors of soil formation. - McGraw-Hill. Johnston, K. H. et al. 1996. Biodiversity and the productivity and stability of ecosystems. - Trends Ecol. Evol. 11: 372-377. Jongman, R. G. H., ter Braak, C. J. F. and van Tongeren, 0. F. 1987. Data analysis in community and landscape ecology. - Pudoc, Wageningen. Kareiva, P. 1994. Diversity begets productivity. - Nature 368: 686-687. Keddy, P. A. 1987. Beyond reductionism and scholasticism in plant community ecology. - Vegetatio 69: 209-21 1. Keddy, P. A. 1989. Competition. - Chapman and Hall. Keddy, P. A. 1992. Assembly and response rules: two goals for predictive community ecology. - J. Veg. Sci. 3: 157-164. Keddy, P. A. 1994. Applications of the Hertzsprung-Russell star chart to ecology: reflections on the 21st birthday of geographical ecology. - Trends Ecol. Evol. 9: 231-234. Keddy, P. A,, Twolan-Strutt, L. and Wisheu, I. C. 1994. Competitive effect and response rankings in 20 wetland plants: are they consistent across three environments? - J. Ecol. 82: 635-643. Kent, M. and Coker, P. 1992. Vegetation description and analysis: a practical approach. - Belhaven Press, London. Kerr, J. T. and Packer, L. 1997. Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. - Nature 385: 252-254. Krebs C. J. 1994. Ecology: the experimental analysis of distribution and abundance, 4th ed. - Harper and Row. Kuhn, T. S. 1970. The structure of scientific revolutions, 2nd ed. - Univ. Chicago Press. Latham, R. E. and Ricklefs, R. E. 1993. Global patterns of tree species richness in moist forests: energy-diversity theory does not account for variation in species richness. Oikos 67: 325-333. Lawton, J. H. 1982. Vacant niches and unsaturated communities: a comparison of bracken herbivores at sites on two continents. - J. Anim. Ecol. 51: 573-595. Lawton, J. H. 1994. What do species do in ecosystems. Oikos 71: 367-374. Lawton, J. H. 1999. Are there general laws in ecology? Oikos 84: 177-192. Lawton, J. H.’et al. 1998. Biodiversity and ecosystem function: getting the Ecotron experiment in its correct context. Funct. Ecol. 12: 848-852. Leathwick, J. R. 1995. Climatic relationships of some New Zealand forest tree species. - J. Veg. Sci. 6: 237-248. Leathwick, J. R. 1998. Are New Zealand’s Nothofagus species in equilibrium with their environment? - J. Veg. Sci. 9: 719-732. Leathwick, J. R. and Mitchell, N. D. 1992. Forest pattern, climate and vulcanism in central North Island, New Zealand. - J. Veg. Sci. 3: 603-616. Leathwick, J. R., Whitehead, D. and McLeod, M. 1996. Predicting changes in the composition of New Zealand’s indigenous forests in response to global warming: a modelling approach. - Environ. Softw. 11: 81-90. Leathwick, J. R., Burns, B. R. and Clarkson, B. D. 1998. Environmental correlates of tree alpha-diversity in New Zealand primary forests. - Ecography 21: 235-246. ECOGRAPHY 225 (1999) Lehmann, A. 1998. GIS modeling of submerged macrophyte distribution using generalized additive models. - Plant Eco]. 139: 113-124. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? - Ecology 74: 1659-1673. Lenihan, J. M. 1993. Ecological response surfaces for North American boreal tree species and their use in forest classification. - J. Veg. Sci. 4: 667-680. Major, J. 1951. A functional, factorial approach to plant ecology. - Ecology 32: 392-412. Margules C. R. and Austin, M. P. (eds). 1991. Nature conservation: cost effective biological surveys and data analysis. - CSIRO, Melbourne, Australia. Margules, C. R. and Austin, M. P. 1994. Biological models for monitoring species decline: the construction and use of data bases. - Philos. Trans. R. SOC.Lond. B Biol. Sci. 344: 69-75. Margules, C. R. and Gaston, K. J. 1994. Biodiversity and agriculture. - Science 265: 457. Margules, C. R. and Nicholls, A. 0. 1987. Assessing the conservation value of remnant habitat ‘islands’: mallee patches on the western Erye Peninsula, South Australia. In: Saunders, D. A. et al. (eds), Nature conservation: the role of remnant vegetation. Surrey Beatty and Sons, pp. 89-102. Margules, C. R. and Stein, J. 1989. Patterns in the distributions of species and the selection of nature reserves: an example from Eucalyptus forests in south-eastern New South Wales. - Biol. Conserv. SO: 219-238. Margules, C. R., Nicholls, A. 0. and Austin, M. P. 1987. Diversity of Eucalyptus species predicted by a multi variables environmental gradient. - Oecologia (Berl.) 71: 229232. Margules, C. R. et al. 1995. BioRap: guidelines for using the BioRap methodology and tools. - CSIRO, Canberra. May, R. M. 1985. Evolutionary ecology and John Maynard Smith. In: Greenwood, P. J., Harvey, P. H. and Slatkin, M. (eds), Essays in honour of John Maynard Smith. Cambridge Univ. Press, pp. 107- 116. McCullagh, P. and Nelder, J. A. 1989. Generalized linear models, 2nd ed. - Chapman and Hall. McIntosh, R. P. 1985. The background of ecology. - Cambridge Univ. Press. Minchin, P. R. 1987. Simulation of multidimensional community patterns: towards a comprehensive model. - Vegetatio 71: 145-156. Minchin, P. 1989. Montane vegetation of the Mt. Field Massif, Tasmania: a test of some hypotheses about properties of community patterns. - Vegetatio 83: 97-1 10. Mueller-Dombois, D. and Ellenberg, H. 1974. Aims and methods of vegetation ecology. - Wiley. Naeem. S. et al. 1994. Declining biodiversity can alter the performance of ecosystems. - Nature 368: 734-737. Naeem. S. et al. 1995. Empirical evidence that declining species diversity may alter the performance of terrestrial ecosystems. - Proc. R. SOC.Lond. B Biol. Sci. 347: 249262. Naeem. S. et al. 1996. Biodiversity and plant productivity in a model assemblage of plant species. - Oikos 76: 259-264. Navas, M.-L. et al. Effect of competition on the responses of grasses and legumes to elevated atmospheric CO, concentration along a nitrogen gradient: differences between isolated plants, monocultures and multi-species mixtures New Phytologist, in press. Nicholls, A. 0. 1989. How to make biological surveys go further with generalized linear models. - Biol. Conserv. 50: 51-76. Noble, I. R. 1989. Ecological traits of Eucalyprus L‘Herit. subgenera Monocalyptus and Symphyomyrtus. - Aust. J. Bot. 37: 207-224. Okland, R. H. 1999. On the variation explained by ordination and constrained ordination axes. - J. Veg. Sci. 10: 131136. Oksanen, J. 1997. Why the beta-function cannot be used to estimate skewness of species responses. J . Veg. Sci. 8: 147- 152. Oksanen, J. and Minchin, P. R. 1997. Instability of ordination results under changes in input data order: explanations and remedies. - J . Veg. Sci. 8: 447-454. Partel, M. et al. 1996. The species pool and its relation to species richness: evidence from Estonian plant communities. - Oikos 75: 11 1- 117. Pausas, J. G. 1994. Species richness patterns in the understory of Pyrenean Pinus syluestris forests. - J. Veg. Sci. 5: 5 17-524. Pausas, J. G. and Carreras, J. 1995. The effect of bedrock type, temperature and moisture on species richness of Pyrenean Scots pine (Pinus syluestris L.) forests. - Vegetatio 116: 85-92. Pausas, J. G., Braithwaite, L. W. and Austin, M. P. 1995. Modelling habitat quality for arboreal marsupials in the south coastal forests of New South Wales, Australia. For. Ecol. Manage. 78: 39-49. Peet, R. K. et al. 1988. Putting things in order: the advantages of detrended correspondence analysis. Am. Nat. 131: 924-934. Perring, F. 1958. A theoretical approach to a study of chalk grassland. - J. Ecol. 46: 665-679. Perring, F. 1959. Topographical gradients of chalk grassland. J. Eco~. 447-481. 47: Perring, F. 1960. Climatic gradients of chalk grassland. - J. Ecol. 48: 415-442. Prendergast, J. R. et al. 1993. Rare species, the coincidence of diversity hotspots and conservation strategies. - Nature 365: 335-337. Ricklefs, R. E. 1987. Community diversity: relative roles of local and regional processes. Science 235: 167-171. Ricklefs, R. E. and Schluter, D. (eds). 1993. Species diversity in ecological communities: historical and geographical perspectives. - Univ. Chicago Press. Rodwell, J. S. 1991. British plant communities. Vol. 1. Woodlands and scrub. Cambridge Univ. Press. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge Univ. Press. Rotenberry, J. T. and Wiens, J. A. 1980. Habitat structure, patchiness, and avian communites in North American steppe vegetation: a multivariate analysis. - Ecology 61: 1228- 1250. Sabo, S . R. 1980. Niche and habitat relations in subalpine bird communities of the white mountains of New Hampshire. Ecol. Monogr. 50: 241-259. Schoener, T. W. 1986. Overview: kinds of ecological communities - ecology becomes pluralistic. - In: Diamond, J. and Case. T. J. (edsl. Communitv ecolow. Haruer and Row. pp. 333-343: Scott, J. M. et al. 1993. Gap analysis: a geographic approach to protection of biological diversity. - Wildlife Monogr. 123 1-41. Shipley, B. and Keddy, P. A. 1987. The individualistic and community-unit concepts as falsifiable hypotheses. - Vegetatio 69: 47-55. Shugart, H. H. 1998. Terrestrial ecosystems in changing environments. - Cambridge Univ. Press. Smith, P. A. 1994. Autocorrelation in logistic regression modelling of species’ distributions. - Global Ecol. Biogeogr. Lett. 4: 47-61. Smith, T. M. and Huston, M. A. 1989. A theory of the spatial and temporal dynamics of plant communities. - Vegetatio 83: 49-69. Smith, T. M., Shugart H. H. and Woodward, F. I. 1997. Plant functional types. - Cambridge Univ. Press. Srivastava, D. S. 1999. Using local-regional plots to test for species saturation: pitfalls and potentials. - J. Anim. Ecol. 68: 1-16. Tausch, R. J. et al. 1995. Patterns of ordination and classification instability resulting from changes in input data order. - J. Veg. Sci. 6: 897-902. I - ” ECOGRAPHY 22:5 (1999) ter Braak, C. J. F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. - Ecology 67: 1167-1179. ter Braak, C. J. F. 1987. The analysis of vegetation-environment relationships by canonical correspondence analysis. - Vegetatio 69: 69-77. Thompson, K. and Hodgson, J. G. 1999. Response to Lawton et al. - Funct. Ecol. 12: 852-853. Tilman, D. 1982. Resource competition and comunity structure. - Princeton Univ. Press. Tilman, D. 1988. Plant strategies and the structure and dynamics of plant communities. - Princeton Univ. Press. Tilman, D. 1996. Biodiversity: population versus ecosystem stability. - Ecology 77: 350-363. Tilman, D. and Downing, J. A. 1994. Biodiversity and stability in grasslands. - Nature 367: 363-365. Tilman, D. and Kareiva, P. 1997. Spatial ecology: the role of space in population dynamics and interspecific interactions. - Princeton Univ. Press. Tilman, D. and Pacala, S. 1993. The maintenance of species richness in plant communities. - In: Ricklefs, R. E. and Schluter, D. (eds), Species diversity in ecological communities: historical and geographical perspectives. Univ. Chicago Press, pp. 13-25. Tilman, D., Wedin, D. and Knops, J. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. - Nature 379: 718-720. Tilman, D. et al. 1997. The influence of functional diversity and composition on ecosystem processes. - Science 277: 1300- 1302. van der Maarel, E. 1989. Theoretical vegetation science on the way. - Vegetatio 83: 1-6. Wartenberg, D., Ferson, S. and Rohlf, F. J. 1987. Putting things in order: a critque of detrended correspondence analysis. - Am. Nat. 129: 434-448. Weiher, E. and Keddy, P. A. 1995. The assembly of experimental wetland plant communities. - Oikos 73: 323-335. Westhoff, V. and van der Maarel, E. 1978. The Braun-Blanquet approach. - In: Whittaker, R. H. (ed.), Classification of plant communities. Junk, The Hague, pp. 287-399. Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. - Plant Soil 199: 213-227. Whittaker, R. H. 1952. A study of summer foliage insect communities in the Great Smokey Mountains. - Ecol. Monogr. 22: 1-44. Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. - Ecol. Monogr. 26: 1-80. Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. - Ecol. Monogr. 30: 279338. Whittaker, R. H. 1975. Communities and ecosystems, 2nd ed. - Macmillian. Whittaker, R. H. 1978. Direct gradient analysis. - In: Whittaker. R. H. (ed.). Ordination of ~ l a n tcommunities. Junk,’ The Hague, pp. 7-51. Whittaker. R. H. and Niering. W. A. 1965. Vegetation of the Santa Catalina Mount&, Arizona: a grdient analysis of the south slope. - Ecology 46: 429-452. Yee, T. W. and Mitchell, N. D. 1991. Generalised additive models in plant ecology. - J. Veg. Sci. 2: 587-602. Zobel, M. 1997. The relative role of species pools in determining plant species richness: an alternative explanation of species coexistence? - Trends Ecol. Evol. 12: 266-269. ECOGRAPHY 2 2 5 (1999) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Ecography Wiley

The potential contribution of vegetation ecology to biodiversity research

Ecography , Volume 22 (5) – Dec 1, 1999

Loading next page...
 
/lp/wiley/the-potential-contribution-of-vegetation-ecology-to-biodiversity-acdPOh3qoY

References (167)

Publisher
Wiley
Copyright
1999 Ecography
ISSN
0906-7590
eISSN
1600-0587
DOI
10.1111/j.1600-0587.1999.tb01276.x
Publisher site
See Article on Publisher Site

Abstract

Austin, M. P. 1999. The potential contribution of vegetation ecology to biodiversity research. - Ecography 22: 465-484. The contribution of vegetation ecology to the study of biodiversity depends on better communication between the different research paradigms in ecology. Recent developments in vegetation theory and associated statistical modelling techniques are reviewed for their relevance to biodiversity. Species composition and collective properties such as species richness vary as a continuum in a multi-dimensional environmental space; a concept which needs to be incorporated into biodiversity studies. Different kinds of environmental gradients can be recognised and species responses to them vary. Species response curves of eucalypts to an environmental gradient of mean annual temperature have been shown to exhibit a particular pattern of skewed response curves. Generalised linear modelling (GLM) and generalised additive modelling (GAM) techniques are important tools for biodiversity studies. They have successfully distinguished the contribution of environmental (climatic) and spatial (history and species dispersal ability) variables in determining forest tree composition in New Zealand. Species richness studies are examined at global, regional and local scales. At all scales, direct and resource environmental gradients need to be incorporated into the analysis rather than indirect gradients e.g. latitude which have no direct physiological influence on biota. Evidence indicates that species richness at the regional scale is sensitive to environment, confounding current studies on local/regional species richness relationships. Plant community experiments require designs based on environmental gradients rather than dependent biological properties such as productivity or species richness to avoid confounding the biotic components. Neglect of climatic and other environmental gradients and the concentration on the collective properties of species assemblages has limited recent biodiversity studies. Conservation evaluation could benefit from greater use of the continuum concepts and statistical modelling techniques of vegetation ecology. The future development of ecology will depend on testing the different assumptions of competing research paradigms and a more inclusive synthesis of ecological theory. M . P. Austin (m.austin@dwe.csiro.au),CSIRO Wildlfe and Ecology, GPO Box 284, Canberra A C T 2601, Australia. As our global society faces the problem of productive sustainability of our biosphere, it becomes essential to determine the role of biodiversity. Plant communities are the primary autotrophic component of terrestrial biodiversity. Yet, vegetation ecologists, as opposed to plant ecologists, have not been seen to be contributing to recent studies on the role of plant species richness in determining productivity and stability of ecosystem This is an invited Minireview on the occasion of the 50th anniversary of the Nordic Ecological Society Oikos. Copyright 0 ECOGRAPHY 1999 ISSN 0906-7590 Printed in Ireland - all rights reserved ECOGRAPHY 1 2 5 (1999) function (Kareiva 1994, Lawton 1994, Tilman and Downing 1994, Johnston et al. 1996). Plant community ecology can be defined as the study of the distribution and behaviour of multispecies assemblages of plants in time and space. Vegetation ecology is used here to apply to plant community studies where observational analysis is the dominant method of study. This field has tended to be neglected by ecologists in recent years, yet it has much to offer the study of biodiversity. The term biodiversity as used here can be applied to the total, or to a defined subset, of the biota of an ecosystem, landscape or continent. The question is, why has vegetation ecology been ignored and does it have anything to offer? The existence of disparate, multitudinous research paradigms (sensu Kuhn 1970) in ecology is one factor limiting the use of concepts and methods developed in vegetation ecology. In each paradigm, evidence is frequently sought to support an idea rather than to disprove an hypothesis while methods are selected for their compatibility with the paradigm rather than their suitability for answering the question at hand. Often, new concepts, evidence and suitable methods exist in another ecological paradigm, which could contribute to a better synthesis than could be achieved by either paradigm alone. In this review I present a particular personal paradigm for vegetation ecology and argue that it has potential to elucidate several broader issues in the study of biodiversity. I shall try to examine the problem of communication between paradigms with reference to species richness studies while considering whether vegetation ecology does have anything to contribute to the study of biodiversity. Three issues are important to such a review: 1) The relevance of vegetation ecology’s own paradigm assumptions regarding theory, methodology and experimentation to current questions regarding biodiversity. 2) The need to simultaneously test hypotheses from different paradigms in order to achieve a new synthesis. 3) The use of vegetation survey and analysis techniques in the conservation of biodiversity. Some recent developments in continuum theory and associated statistical modelling methods are reviewed for their potential contribution to biodiversity studies. Species richness studies are briefly examined at global, regional and local scales with the relevance of results from vegetation ecology emphasised. The relationship between recent experimental community studies and observational studies is discussed with suggestions for alternative experimental designs. A similar potential for integration of ideas is shown for conservation studies. The need to incorporate both environmental space and geographical space into any synthesis of community ecology paradigms is then discussed. Continuum and environmental space The continuum concept (Whittaker 1956, Curtis 1959) has been central to vegetation ecology for forty years and is usually presented in general ecology texts (e.g. Huston 1994, Krebs 1994, Shugart 1998). Yet many recent texts and reviews on species diversity and spatial patterns barely mention the concept (Ricklefs and 466 Schluter 1993, Brown 1995, Rosenzweig 1995, Crawley 1997, Tilman and Kareiva 1997, Lawton 1999). The concept states that species populations are distributed along environmental gradients such that vegetation varies continuously and does not form distinct coevolved communities of species (Whittaker 1975). This concept is usually linked with Gleason’s individualistic concept that all species are unique and independent in their distribution patterns along gradients. However, there is no necessary association between the two concepts (Austin 1985). The weakness of the continuum concept is that it remains strictly phenomenological, describing patterns without explicit processes for generating the patterns beyond a token reference to competition. However, any process-based ecological theory should be capable of predicting (explaining) the species patterns which have been demonstrated to occur along environmental gradients on several continents (Whittaker 1956, Curtis 1959, Ellenberg 1988, Austin et al. 1994). Early work by Bond (1957), Beals (1960) and Sabo (1980) showed the applicability of the continuum concept to bird community composition. A reanalysis of Rotenberry and Wiens’ (1980) work on bird distribution along a structural gradient of vegetation has shown that the results support a continuum interpretation contrary to their conclusion (Austin 1999). Recently animal ecologists’ concerns with broad-scale patterns of species, whether at the community or biogeographical level, have emphasised ranges and geographical distance (Brown 1995, Rosenzweig 1995, Gaston and Blackburn 1999, Lawton 1999). Two features of the animal ecologists’ paradigm may account for the lack of interest in the continuum concept. First the attention paid to a collective property of the biota, species richness rather than compositional changes and secondly, the assumption that climate plays a secondary role in determining faunal patterns. Lawton (1982) argues in a study concerning bracken insect faunas that “legitimate worries about climate. . . are arguments about ‘second order’ effects. . . they may modify, but are not primarily responsible for, major differences in faunal richness between geographical regions” (see also Srivastava 1999). In comparison, recent statistical models developed for analysis of tree species distributions (Austin et al. 1990, Yee and Mitchell 1991, Leathwick and Mitchell 1992) have shown that climatic variables such as mean annual temperature, and mean annual rainfall are important predictors of regional distributions. These models examined the behaviour of biotic responses in a multivariable environmental space rather than as a univariate test of the importance of a single predictor variable (Austin et al. 1984, Margules et al. 1987). The influence of a variable may often be hidden by a second variable and multiple regression may be needed to recognise the true significance of a variable. The limitations imposed by single variable analysis and linear ECOGRAPHY 2 2 2 (1999) regression in some ecological paradigms are discussed in Austin (1999). Simultaneous analysis of climatic and geographical predictors of fauna as done by vegetation ecologists is needed to test the secondary importance of climate when testing geographical hypotheses. There are other aspects of the paradigm used by vegetation ecologists which may have led to its neglect by others. The lack of clearly defined testable hypotheses is a major one (Keddy 1987, 1992, 1994) and a second is the pre-occupation of many vegetation ecologists with ordination methodology. Recent conceptual developments The continuum concept has been elaborated and testable propositions have been put forward. Austin (1987) and Shipley and Keddy (1987) have tested several hypotheses from earlier work. Shipley and Keddy (1987) attempted to distinguish between the individualistic and community concepts by investigating the coincidence of species limits along a watertable gradient. They concluded that some limits were clustered contradicting the individualistic concept, but upper and lower limits did not co-occur contradicting the community concept. Propositions that the individual species response curves approximated a Gaussian (bell-shaped) curve along environmental gradients and that the modes of major species tend to have a uniform distribution along a gradient were tested (Austin 1987). In the eucalypt forest studied, positive-skewed curves were characteristic of the major canopy species and there was no evidence that species modes were uniformly distributed along a mean annual temperature gradient. Austin and Smith (1989) then put forward a set of testable phenomenological propositions regarding species and vegetation responses along environmental gradients. The ideas of Ellenberg on species responses to environmental gradients were used to describe the variety of possible response shapes (Mueller-Dombois and Ellenberg 1974, Austin 1999). Shapes can range from the conventional Gaussian curve through skewed curves to bimodal responses. The usual niche theory model used by animal ecologists is a special case of the more general theory of Ellenberg (Austin 1999). Austin and Smith (1989) also argued that three types of environmental gradient could be recognised (see Austin 1980): 1) Indirect gradients or complex gradients (Whittaker 1978) are those where the variable has no physiological relationship with biological growth and where variables having a direct physiological relationship are usually confounded. Latitude, altitude and ocean depth are examples. Altitude has no direct impact on organisms. It is highly correlated with both rainfall and temperature. The nature of the correlation is different for each variable and such relationships are location specific. 2) Direct gradients are those where the enviECOGRAPHY 2 2 5 (1999) ronmental variable e.g. temperature has a direct effect on growth but is not consumed. 3) Resource gradients (e.g. nitrogen) are those where the resource is consumed by plants. It is critical that the three types be distinguished. Huston (1994) discusses the different types of response expected for the different types of gradients. Studies of the relationship of species richness to latitude (complex gradient) require a different analytical framework to those concerned with a resource gradient (Huston 1994, 1999, Rosenzweig 1995). Niche theory for animals is usually based on resource gradients (Giller 1984). Direct gradients (termed regulator gradients by Huston 1994) are often climatic and may be more varied in the type of species response. Fauna studies often do not make clear distinctions between these types of gradients (Brown 1995, Rosenzweig 1995, Lawton 1999). If gradients are not equivalent then comparative analysis is not possible. When the propositions of Austin and Smith (1989) were compared with the work of other researchers, significant gaps in the way plant ecologists analyse species’ responses were revealed (Austin and Smith 1989). Different types of environmental gradients were not recognised. Possible changes in the response shape of fundamental and realised niches were not considered (Austin 1999). Various assumptions were made regarding the collective properties of vegetation, species richness, dominance and standing crop (Table 1). Plant ecologists still have a long way to go before their paradigms achieve agreement. The propositions of Austin and Smith (1989) hopefully provide a framework within which comparisons and debate can be conducted. An example of the type of hypothesis testing which can now be undertaken, thanks to the adoption of new statistical modelling techniques and the use of large data sets (discussed in the next section), concerns the position, shape and skewness of species response curves. Species limits towards the extremes of an environmental gradient are determined by physiological tolerances while competition controls both the limits towards the centre of the gradient and the shape of the species response curve (Fig. 1, Austin 1990). An analysis of nine species of eucalypt in relation to seven environmental predictors examined this hypothesis (Austin et al. 1994). A database of 8377 plots of presence/absence data from south-east New South Wales, Australia were analysed using Generalised Linear Modelling (GLM, McCullagh and Nelder 1989) and p-function response curves. Eight of the nine species showed the postulated pattern along a mean annual temperature gradient (Fig. 1). Austin and Gaywood (1994) subsequently found the pattern applied to 22 out of 24 eucalypt species. A pattern of species packing of this kind along an environmental gradient implies that there are assembly rules associated with the continuum concept (Keddy 1992). Smith and Huston (1989) 467 provide supporting circumstantial evidence based on a mechanistic simulation model which assumes that no species can be simultaneously adapted to extreme environmental conditions and mesic highly productive conditions (cf. Grime 1979). Neglect of the responses of individual species, in preference to collective properties such as species richness, is likely to mean failure to detect assembly rules. However, this is a single result from a vegetation type where almost the entire temperature gradient is dominated by a single genus Eucalyptus. The generality of the pattern needs to be investigated. Three features of these recent developments in vegetation theory need to be incorporated into the study of biodiversity. Firstly, the basic concept that species vary continuously in occurrence and abundance in a multidimensional environmental space. Failure to incorporate this into analyses is likely to obscure rather than clarify results. Secondly, environmental gradients are of different kinds and species responses to them will differ. Thirdly, the types of species niche responses postulated by Ellenberg are more appropriate for investigation than the Gaussian curve model. Methodology Lack of interest by many ecologists in plant community ecology and its theoretical concepts has been due to the preoccupation of some vegetation ecologists with methodology. Strong criticism of the exploratory multivariate approach has been made. Crawley (1997, p. 510) quotes May (1985) as saying “the wilderness of meticulous classification and ordination of plant communities, in which plant ecology has wandered for so long, began in pursuit of answers to questions but then became an activity simply for its own sake”. This criticism is not without substance though see van der Maarel(l989) and Kent and Coker (1992) for a spirited response. Ordination methodology continues to be an issue. Consider the papers presented at a symposium in Uppsala in 1985 on “Theory and models in vegetation science” by ter Braak (1987) (see also ter Braak 1986) on canonical correspondence analysis (CCA) (see also Jongman et al. 1987), and those by Faith et al. (1987) and Minchin (1987) (see also Minchin 1989) comparing correspondence analysis (CA) and multidimensional scaling (MDS). The two paradigms, one using canonical correspondence analysis based methods (ter Braak 1986) and the other, new forms of multidimensional scaling (Belbin 1991) continue to exist today. The basic ecological assumptions have still not been examined in depth, though as the above papers make clear the methods are very sensitive to the assumptions. The controversies continue (Wartenberg et al. 1987, Peet et al. 1988, James and McCulloch 1990), and continue (Belbin 1991, Tausch et al. 1995, Oksanen and Minchin 468 ECOGRAPHY 22.5 (1999) Fig. 1. The upper figure shows the hypothesised pattern of species response curves along an environmental gradient with possible explanatory processes (Austin 1990, reproduced with permission of Academic Press). The lower figure shows the results of fitting generalised linear models with a p-function for temperature to data from nine eucalypt species (Austin et al. 1994 reproduced with permission of Opulus Press). Note the pattern of skewness of the species response curves about the intermediate mean annual temperature of 11.5"C. Increasing skewness + Increasing role of competition t + Increasing role of physiological tolerance + 8 K ? 3 r .-. - 0.0 a 0.5 1997, Okland 1999). Failure to evaluate the issues in the vegetation literature has led some ecologists in other disciplines to continue to use linear models such as principal components analysis (PCA). A technique shown to be inadequate in the early 1970's (Austin 1985, 1999). Much of the controversy concerns the choice of ordination method. Buried under the methodology lies an ecological controversy which revolves around the nature of the non-monotonic species response shape and the robustness of the methods to differences in the response shape and species packing along gradients. The issue posed by May (1985) reECOGRAPHY 22.5 (1999) mains but the development of more detailed continuum propositions and more rigorous direct gradient techniques may lead to the development of a new paradigm for addressing the conceptual questions. The application of powerful new regression techniques, Generalised Linear Models (GLM) (Austin and Cunningham 1981, McCullagh and Nelder 1989, Nicholls 1989, Austin et al. 1990), and Generalised Additive Models (GAM) (Hastie and Tibshirani 1990, Yee and Mitchell 1991, Leathwick 1995, Austin and Meyers 1996) to vegetation analysis has expanded the opportunities to examine issues of vegetation organ469 isation (Austin et al. 1994). These methods relax many and the deeper deposits close to the eruption centre on of the constraints of traditional regression analysis. forest composition was important. The deep pumice Factors and variables can easily be combined as predic- deposits were interpreted as having favoured the curtors. Error functions other than normal are simply rent dominance of those sites by rapidly dispersing accommodated. Use of GAM means that the functional conifer species as opposed to the slower dispersing form of the predictive relationship does not have to be Nothofagus species. A statistical description (model) of predefined and a smoothing function is fitted (Yee and a complex regional pattern of species distributions was Mitchell 1991). Recognition that large sets of data obtained incorporating both environmental controls suitable for this type of statistical modelling could be and historical factors. The results support those of collated from vegetation surveys led to a number of Austin and colleagues (Austin et al. 1984, 1990) for studies of species response to environmental variables forests in south-east Australia that climate variables are (Austin et al. 1984, Margules et al. 1987, Leathwick and important predictors of regional species distributions Mitchell 1992). Presence/absence data from plots of and that the species response shapes are often complex specified size and location form a minimum data set for skewed surfaces. this type of analysis. Given the location, estimates of Leathwick (1995) extended his analysis to most of the the environmental predictors can be obtained from indigenous forest of New Zealand using ca 16000 plots geographic information systems (GIS). These methods of 0.4 ha and GAM. The results confirm the advantages have the potential to transform the regional analysis of of GAM (Yee and Mitchell 1991). The high residual vegetation, testing continuum concepts, the importance deviances associated with the models for the Nothofaof climate within regions, and the adequacy of environ- gus species provided support for the view that their mental models which ignore history and geographic distribution may be limited by volcanic destruction and barriers. As regression methods, GLM and GAM suffer slow rates of reinvasion. from all the well-known statistical limitations e.g. Following the approaches of Legendre (1993), Smith multi-collinearity, inadequacies of stepwise regression (1994), Borcard and Legendre (1994), Leathwick et al. procedures and that it is correlation not causation (1996) extended the GAM modelling approach incorpo(James and McCulloch 1990). These methods provide a rating a water balance model, spatial autocorrelation, statement of the observable pattern of relationships and Nothofagus competition (cf. Austin and Cunningbetween a species and the environment. They also ham 1981) and used it to evaluate the response of the provide a relative measure of the importance of differ- forests to global warming. Leathwick (1998) examined ent components of the environment e.g. climate versus the extent of spatial autocorrelation in the residuals history, and a statistical test of the interpretation. after fitting climatic and other environmental predictors Hopefully they will also contribute to resolving the to consider whether existing forests were in equilibrium ordination and classification issues in vegetation with the current environment. These regression models confirmed the relative importance of spatial autocorreecology. The statistical modelling work of Leathwick on the lation for Nothofagus species as compared with conifers distribution of forest trees in New Zealand provides a and broad-leaved angiosperm trees. The differences beclear example of the power of these methods (Leath- tween predictions of the environmental model with and wick and Mitchell 1992, Leathwick 1995, 1998, Leath- without spatial effects highlighted the areas of the well wick et al. 1996, 1998). New Zealand has a varied tree known “beech gaps” in New Zealand. These gaps have flora, large climatic gradients, and a turbulent history similar climates to adjacent areas with Nothofagus and of glaciation, volcanism and human invasion. The dif- the lack of invasion is ascribed to poor dispersal and/or ferential regional processes of disturbance, climate the requirement for ectomycorrhiza. For a given envichange, local elimination due to volcanic eruptions and ronment, the full range of densities of Nothofagus can subsequent migrations are all present together with be found including natural absences. Leathwick and local and regional control of forest composition due to Austin (unpubl.) have therefore been able to examine present climate, topography and lithology (Leathwick the influence of the major dominant Nothofagus on the and Mitchell 1992). Leathwick and Mitchell (1992) realised environmental niche of other tree species. investigated the distribution of forest trees in an area of There are significant improvements in the statistical ca 20000 km2 in the central North Island of New models of species when the competitive effects of Zealand using presence/absence species data from 2172 Nothofagus species are included. plots of 0.4 ha and GLM. They concluded that the The studies of Leathwick and colleagues demonstrate primary predictors of the eleven species were mean an observational methodology which can detect the annual temperature and solar radiation. Of secondary relative importance of environmental variables, historic importance were mean annual rainfall, topography and processes and biotic effects such as competition. There drainage. The depth of the Taupo pumice deposits was is now a developing literature on statistical modelling of also a significant predictor for several species. The species responses (Franklin 1995) with a variety of influence of the large Taupo Pumice eruption (130 AD) applications (Margules and Stein 1989, Lenihan 1993, 470 ECOGRAPHY 2 2 5 (1999) Beerling et al. 1995, Brzeziecki et al. 1995, Austin and Meyers 1996, Bio et al. 1998, Franklin 1998, Guisan et al. 1998, Lehmann 1998). It is not without controversy (Austin and Nicholls 1997, Oksanen 1997) and the need for performance evaluation (Austin et al. 1995). Statistical models of this kind can indicate the relative importance of variables and test assumptions about whether climate is a secondary factor or not (cf. Lawton 1982, Harrison et al. 1992, Srivastava 1999). The adoption of these types of methods under the stimulation of the continuum concept provides a paradigm of immediate relevance to the study of other forms of biodiversity and particularly current concern for a collective property such as species richness. Species richness Patterns of species richness (number of species per unit area) are a major focus of biodiversity studies of many diverse taxa (Huston 1994, Rosenzweig 1995). The current issues arise at three distinct scales. l) What determines the observed latitudinal gradients in species richness at global or continental scale? There is a voluminous literature of which Rosenzweig (1995) provides one recent summary. 2) What is the relationship between local and regional species richness? Interest in this topic has developed more recently but there is an expanding literature (Ricklefs 1987, Cornell and Lawton 1992, Ricklefs and Schluter 1993, Cornell and Karlson 1997, Zobel 1997). 3) At the local scale, the current question is the role of species richness in determining ecosystem function, particularly its influence on productivity following the early work of Al-Mufti et al. (1977) and Grime (1979). There is now a varied observational and experimental literature (Naeem et al. 1994, 1995, 1996, Tilman et al. 1996, 1997, Huston 1997, Hodgson et al. 1998, Lawton et a]. 1998, Thompson and Hodgson 1999). At each scale, recent results in vegetation ecology both in terms of concept and methodology can contribute to a more comprehensive analysis of the questions. Global species richness Numerous hypotheses have been advanced to account for the observed gradient of species richness from the equator to the poles, including area (Rosenzweig 1995) and energy (Currie 1991). These two hypotheses provide examples of the type of approach which can be adopted and of the problems which arise. The work of Currie and colleagues (Currie and Paquin 1987, Currie 1991, Francis and Currie 1998) has established relationships between tree species richness and actual evapotranspiration (AET) and between vertebrate richness and potential evapotranspiration (PET) for North ECOGRAPHY 2 2 5 (1999) America. The data are based on species ranges recorded as presence in geographical grid cells of 2.5" lat. x 2.5" long. south of 50"N, and 5" lat. x 2.5" long. north of 50"N. Currie and Paquin (1987) found a nonlinear regression relationship between tree species richness and AET with an r2 of 0.76. An additional predictor which explained a further 8% of the variance was elevation variation within the gridcell. Currie (1991) extended the analysis to various vertebrate taxa including amphibians, reptiles and mammals. Potential evapotranspiration explained at least 79% of the variance in species richness of these taxa. Other variables which were found to be important in a multiple regression analysis after taking account of PET included elevation, AET, and spatial variation in elevation and in PET. There was a very strong curvilinear relationship between PET and latitude (Currie 1991: Fig. 8a). An energy measure, a potentially direct gradient accounted for as much if not more than the indirect gradient, latitude. Environmental heterogeneity within the gridcells accounted for additional variability in species richness for both trees and vertebrates. Adams and Woodward (1989) following the approach of Currie and Paquin (1987) examined the patterns of tree species richness in Europe, North America and East Asia. They tested a modelled estimate of net primary production rather than AET and found a very strong non-linear relationship between tree species richness and primary production in each continent. They concluded that glacial history could not explain richness patterns and that differences could be explained by current climate and its impact on productivity. These results and those of Currie and Paquin (1987) and Currie (1991) were criticised by Latham and Ricklefs (1993). They indicate that the data of Adams and Woodward (1989) were not representative of east Asia. They assembled a more extensive data set, finding that though there was a relationship between species richness and AET, mean species richness was 1.9 times higher in Asia than eastern North America. Using a second data set of species richness for approximately one hectare plots, they found after taking account of differences between five regions that AET did not explain any significant additional vari- . ance. The small plot size data set was used as it was argued that large ecologically heterogeneous areas were not suitable for testing the energy-diversity hypothesis. Francis and Currie (1998) responded that for the large area data set the most appropriate model was one with area, AET and region. AET and region were strongly collinear in the small plot data set. As a consequence, either interpretation of species richness as a function of AET or region was equally plausible and could not be distinguished statistically. Kerr and Packer (1997) reanalysed the mammal data of Currie (1991), breaking the data into two sections. Below 1000 mm PET, there is a linear relationship 47 1 between mammal species richness and PET; above 1000 mm species richness and PET are independent. This is consistent with the curvilinear relationship found by Currie. The data above 1000 mm PET were then examined in relation to topographic and PET heterogeneity, and other predictors selected to test other hypotheses concerning species richness. Kerr and Packer (1997) rejected climatic favourability, climatic stability and glacial history ,as explanatory hypotheses on the basis of their regression analyses. Above 1000 mm PET, a significant regression model was found explaining 76.7% of the variance with the terms topographic heterogeneity, PET variability and coastal location. The authors state the result “contradicts previous studies of large scale richness patterns that dismissed the importance of habitat heterogeneity”. The analysis uses PET, a direct gradient sensu Austin (1980), plus a measure of its within gridcell variability and topographic heterogeneity, a measure of variability of a complex gradient, altitude. Potential evapotranspiration is strongly correlated with altitude. Variation in altitude will be associated with variation in PET. If the relationship between species richness and PET holds, cells with high variation in PET will have higher species richness than those with low variance. Two possibilities arise from this consideration of the environmental data. Firstly, only one environmental variable, PET may be associated with richness but through two variables, directly through PET and indirectly through topographic heterogeneity. Secondly, the confounding of PET and its variability above 1000 mm means we cannot be sure of the shape of the relationship above 1000 mm PET. Is there another variable associated with topographic heterogeneity apart from PET that is correlated with species richness? This brief review of the energy-diversity hypothesis raises many issues regarding the analysis of observational data sets particularly those based on large area enumerations of species distributions. Data quality is often unknown and variable. Data also need to be representative of the full range of conditions, for example the criticism of Adams and Woodward (1989) by Latham and Ricklefs (1993) regarding the absence of Asian moist-temperate forests. The use of large area gridcells or polygons must result in massive environmental heterogeneity within some gridcells. It is then necessary to incorporate some measure of heterogeneity for efficient analysis (Currie 1991, Kerr and Packer 1997). Ecological interpretation of such measures can prove difficult as indicated for Kerr and Packer’s (1997) result. Is it environmental heterogeneity in PET or some other variable or both? For patterns of species richness to have a biotic explanation species must interact (Latham and Ricklefs 1993, Huston 1999). For this reason the advantages of large area polygons over small plots distributed over large areas needs to be considered 472 carefully. The analysis of Francis and Currie (1998) emphasises the need to include multiple predictors in any regression model. The problem of multi-collinearity is then clearly apparent as with region and AET in Francis and Currie’s analysis. This is not a statistical problem but an ecological one: are the variables representing a complex indirect gradient or a functional potentially causal relationship? In contrast, Rosenzweig (1995) does not distinguish between latitude, elevation or mean July temperature in terms of the nature of their likely influence. He also asserts that “latitudinal gradients must arise because the tropics cover more area than any other zone. Their greater area stimulates speciation and inhibits extinction.” (Rosenzweig 1995 p. 284). No attempt is made by Rosenzweig (1 995) to examine more than one variable at a time. Yet both his area hypothesis and the energy hypothesis of Currie (1991) could be complementary. Neither PET or area is a pure regulator or resource gradient. There is a complex correlative path between these variables and any causal process determining species richness. A description of species richness in relation to these two variables with estimates of the correlation between PET and area for different continents might yield fresh insight into global richness patterns. For example, using the Currie data set the following model might be fitted: SR = f(PET, H,,, E, He, Area, C) (1) where SR is species richness; PET potential evapotranspiration; HFt a measure of PET heterogeneity within the grid cell; E another environmental variable; Area the area of the latitudinal zone for the continent, and C a correction factor for different sized grid cells. Statistical modelling at this scale needs to incorporate both alternative hypotheses and the influence of data heterogeneity. All ecological paradigms using observational analysis face similar problems. A concensus on the appropriate formulation of hypotheses and suitable methodology has yet to be reached. The approach used by Leathwick (1998) for analysis of forest patterns in New Zealand can be applied to species richness (Leathwick et al. 1998) and indicates a potential direction for development. Regional species richness The current interest in the relationship between local and regional species richness (Ricklefs 1987, Cornell and Lawton 1992, Ricklefs and Schluter 1993, Cornell and Karlson 1997, Zobel 1997) raises similar issues. Species richness is assumed to be a result of local biotic interactions such as competition and predation, and ECOGRAPHY 22:5 (1999) regional or historic processes such as dispersal and speciation; no intra-regional environmental control is envisaged. Local communities may be saturated with species if competition and predation are the dominant processes, or unsaturated if intra-regional dispersal and speciation are determining local richness. These alternatives are distinguished by examining the plot of local richness against regional richness; saturated regions should show a hyperbolic relationship and unsaturated a linear relationship (Cornell and Lawton 1992, Fig. 2) This comes as a surprise to a vegetation ecologist familiar with the complexity of species niches in relation to the numerous environmental gradients within a region. Do any species reach an environmental limit within a region? The results of Harrison et al. (1992) are instructive on this topic. They tested whether high beta diversity (turnover in species composition) along transects of 50 x 50 km squares was associated with poor powers of dispersal. They did this for a wide range of invertebrates, birds, mosses and vascular plants. Both p and a diversity were tested in relation to distance along transects but climate was strongly related to both the north-south and east-west transects. The authors conclude “Our results demonstrate that turnover with distance may be a relatively minor component of regional diversity, especially in the presence of strong environmental gradients,. . . Gradient-driven patterns such as we see in Britain may be characteristic of many temperate biotas at comparable scales.” This result is consistent with what is expected from continuum concept propositions and the statistical models of Austin (1987), Margules et al. (1987), Pausas (1994), Pausas and Carreras (1999, Austin et al. (1996), and Leathwick et al. (1998). These statistical models demonstrate that species richness within a region shows complex patterns of relationships with both climatic and local environmental gradients (see below). Cornell and Lawton (1992) provide a theoretical perspective on the topic of whether “local ecological communities” are ever saturated with species. They refer to local and regional processes in the following terms “ . . . predation, parasitism, competition, and abiotic fluctuation or disturbance are played out within local arenas, whereas long-distance dispersal, speciation, wide-spread extinction, and fluctuation in species’ distributions take place across broad geographic regions.” There is no mention of climatic gradients within or between regions possibly controlling species distribution. A few studies have incorporated an environmental component e.g. depth and habitat type (Cornell and Karlson 1996). The properties of the local and regional floras responding to climatic gradients across regions and to substrate heterogeneities (e.g. peat vs chalk) within regions need to be considered in this type of analysis. Fauna also respond to climate and lithology. The methodology used in this paradigm, particularly the statistical aspects, has received little attention ECOGRAPHY 2 2 5 (1999) (Cresswell et al. 1995, Srivastava 1999). The conceptual framework is represented in Fig. 2a (Cornell and Lawton 1992). An example analysis (Hawkins and Compton 1992) is presented in Fig. 2b where the maximal possible line is shown together with the quadratic curve which was fitted to the data. The effect of fitting a dependent regression (i.e. y = f(x + y)) is mentioned by Cresswell et al. (1995) but not considered in any of the examples of local versus regional relationships presented in Cornell and Karlson (1997). Of necessity, the regression line is constrained below the 45” line (Fig. 2). Several studies (Cornell and Karlson 1997) fit a quadratic regression to demonstrate possible saturation of the local species richness. A hyperbolic function as shown in Fig. 2a cannot be represented by a quadratic function which must decline. A linear function will give a better fit to the saturation curve drawn in Fig. 2a than a quadratic. Null models are used by very few authors (Partel et al. 1996, Caley and Schluter 1997), yet they may have significant implications. Note that in Fig. 2b, the data fill almost all possible combinations of local richness. One null model states that all possible values of local richness are equally probable (Partel et al. 1996). The expected value is then half the maximal local richness or half the regional richness. An alternative null model based on a binomial expansion of the possible combinations of all species, singly, in pairs etc. gives the same result (Nicholls pers. comm.). The expected relationship is then a straight line with a slope of 22.5” (Fig. 2c). Partel et al. (1996) used such a null model to test the strength of the correlation between the actual species pool and the regional species pool and between species richness. They concluded that the relationships tested differed significantly from that expected under the null model. The null model of Caley and Schluter (1997) assumes the local individuals are randomly sampled from a regional pool of species whose abundances have a canonical log-normal distribution, a further complicating assumption. Srivastava (1999) criticises Partel et al. (1996) for making between habitat comparisons (e.g. grassland with forest) rather than within a habitat where interpretation is less ambiguous. In spite of this, the work (Partel et al. 1996) represents a major step forward in the use of data. They defined the regional and actual species pools for the community on the basis of phytosociological information. The regional flora capable of growing in the community was determined from the ecological amplitude of each species recorded as indicator values in Ellenberg et al. (1992). The actual species pool was recorded from community descriptions and actual species lists from surveys, then compared with local small-scale richness in 1 m2-quadrats. Comprehensive data of this kind are only available where intensive vegetation studies have been done. Few animal taxa or communities could be analysed in this way. It would be interesting to have a similar analysis done within phytosociological associations. 473 Maximum possible Type I Type II Regional richness Maximum possible I - ii I I I Regional richness l5 Maximum possible Mean when all possible local richness values are equi-probable Regional richness Fig. 2. (a) shows the proposed relationships between local and regional species richness from Cornell and Lawton (1992 reproduced with permission of Blackwell Science). Type I is unsaturated and Type I1 is for a fully saturated local fauna. (b) shows results for fig wasps from Cornell and Karlson (1997) based on data from Hawkins and Compton (1992 reproduced with permission). The fitted curve is a quadratic. (c) shows the line expected under a particular null model where all values of local richness are equally likely. ECOGRAPHY 2 2 5 (1999) Environmental predictors of regional patterns of plant species richness have been reviewed by Pausas and Austin (unpubl.). Margules et al. (1987) showed that eucalypt species richness for a region of ca 40000 km2 in south-eastern Australia could be modelled using climatic variables, mean annual temperature, mean annual precipitation and their interaction, plus a relative measure of solar radiation. Austin et al. (1996) in a more extensive study using eight predictors, both regional climatic variables and local topographic factors, showed that various measures of species richness (total tree species, eucalypt species, rainforest species and richness of Eucalyptus subgenera) could all be described by curvilinear GLM models based on 7208 plots. The models for species richness of the eucalypt subgenera Monocalyptus and Symphomyrtus raise significant issues of interpretation (Austin et al. 1996, Austin 1998). The models suggest that there is an optimum environment where Monocalyptus species richness is maximal. Species richness for Symphomyrtus is highest in other environments and the maxima are complimentary with reference to climatic variables. The processes generating optimal environments for the species richness of a subgenus with associated displacement of another are obscure (Austin 1998). The predicted patterns are however consistent with independent circumstantial evidence (Noble 1989). Pausas (1994) and Pausas and Carreras (1995) showed that different growth forms in pine forests in the Pyrenees had predictable patterns of species richness in relation to environmental variables. A model has also been developed for arboreal marsupial species richness (Pausas et al. 1995). The model contains climatic predictors as well as specifically faunal predictors such as food quality and potential presence of hollows for nesting. These papers and others cited in Pausas and Austin (unpubl.) support the contention that species richness patterns within regions vary strongly with environment. This is in marked contrast with the regional richness control hypothesis adopted by animal ecologists (Cornell and Karlson 1997, Gaston and Blackburn 1999, Lawton 1999). Srivastava (1999) comments in discussing regional effects on herbivore insect communities on bracken Pteridium aquilinum in South Africa " . . . two species appear to have range limits within the region under study". Perhaps bracken herbivore communities have a continuum type of organisation and show individualistic responses to climate. The two paradigms being contrasted here differ in several aspects. Some vegetation ecologists use plot data assumed to be homogeneous, GLM or GAM statistical models and fit descriptive correlative models with environmental variables. Some other ecologists working at the same organisational level use various units (geographic gridcells, ponds, communities or a single fig fruit, Cornell and Karlson 1997), classical linear often uni-variate regression but with a specific ECOGRAPHY 2 2 5 (1999) hypothesis (Austin 1999). Srivastava (1999) in discussing the regional richness approach accepts ponds as a suitable local unit. Ponds are highly environmentally diverse internally. An acid pond will have a different flora from an alkaline one and may also differ in size, depth and hydrology. Use of heterogeneous units will obscure more than it reveals. A model similar to eq. (1) might be used to distinguish between the hypotheses but agreement on units would be necessary first. The real question is not which paradigm is right, it is what is the relative importance of different processes in determining the composition and collective properties of vegetation and faunal assemblages? Local species richness While regional and global patterns of species richness are actively researched, it remains true that little is known about the processes determining local richness patterns in vegetation. Al-Mufti et al. (1977) and Grime (1979) discovered the humped-back curve of species richness in response to a gradient in productivity/fertility. This has assumed importance in recent years (Huston 1993, 1994, Tilman and Pacala 1993) and generated controversy concerning its application to biodiversity conservation and sustainable agriculture (Huston 1993, Margules and Gaston 1994). The postulated mechanism to account for the pattern is that under high productivity, superior competitors suppress other species while under low productivity few species can tolerate the extreme conditions. It is only under intermediate conditions that higher numbers of species can maintain themselves in a community (Grime 1979). This has led to a number of developments (Grime et al. 1988, Tilman and Pacala 1993, Huston 1994). Tilman and Pacala (1993) outline a number of explanations for the unimodal pattern of species richness including changes in nutrient competition, soil spatial heterogeneity, light competition and increasing plant size along the productivity gradient. They also present a number of examples of unimodal graphs where productivity is equated with biomass, drainage, biomass plus litter, soil PO,, a normalised phosphorus plus potassium index and a climatic moisture index. It is assumed that a productivity gradient has an identical effect on species richness regardless of the type of gradient which may show an increase in biomass as the variable increases. This has been challenged by Austin and Gaywood (1994). They point out that a universal pattern of that kind cannot be distinguished from a more restricted response to one environmental gradient conditional on a second, see Fig. 3. The ring response of species richness to biomass equivalent to hump-back response occurs when maximal biomass is assumed to occur under intermediate conditions for two environmental gradients. This pattern (Fig. 3b) cannot be 475 distinguished from Fig. 3d unless the observations or the experiments are conducted over the full two-dimensional environmental space. Such studies do not seem to have been done. Goldberg and Novoplansky (1997) review field competition experiments along gradients with respect to competitive intensity and the pulsed nature of resources. They distinguish between resources which may be accumulated ‘by plants (nutrients) and those less likely to accumulate (water). Water may act as a resource under low rates of supply and as a toxic direct gradient sensu Austin (1980) when creating waterlogged conditions. Goldberg and Novoplansky (1997) expect a “greater difference in competition intensity between xeric and mesic environments than between infertile and fertile environments. However, the few field experiments along water gradients show highly variable results that are not consistent with this prediction.” Their hypothesis implies that these gradients should be different. The inconsistent results indicate that the gradients or the experimental designs for the different gradients differ sufficiently to warrant further study. If competitive intensity differs, then species richness along gradients might be expected to differ too. Goldberg and Miller (1990) investigated the impact of increased watering and nutrients (nitrogen, phosphorus and potassium) on a weed community. The addition of water led to a greater increase in biomass than did addition of nitrogen. but species diversity declined only with nitrogen addition. Light levels were similar in both treatments, so the difference was not explained by differential canopy closure and competition for light. Mortality was consistently higher in the nitrogen treatments. Observational analyses of natural communities and field experiments, both using a two-dimensional gradient design, are needed to resolve this issue (see next section). Because studies of the humped-back curve have focussed on a collective property of vegetation, we are left with another problem. What we do not know is whether the humped-curve reflects the co-occurrence of species maxima (realised niche optima) or the zone of overlap of species limits. Experimenta1 community ecology The dominant research paradigm in plant ecology in the 1960’s and 1970’s was that of the experimental study of plant competition using pair-wise species comparisons (Harper 1977). The accepted methodology was the de Wit replacement series (de Wit 1960). Annual weeds were the preferred experimental material. Diffuse competition between perennial species in multispecies communities was not commonly examined. If environmental conditions were studied, analysis of variance was the standard statistical procedure. Tests were made of the significance of two or three levels of a factor. The experimental testing of hypotheses derived from multivariate vegetation analysis was rare, though see Goldsmith (1973a, b). The use of resource gradients and estimation of species response curves was not done. Since the late 1970’s, a more truly community-level experimental ecology has developed where perennials, multispecies mixtures and environmental gradients are common place (Grime 1979, Tilman 1982, 1988, Grime et al. 1988, Keddy 1989). A wide variety of experimental designs and methods are now used. Various schools of experimental ecology have developed associated with the names of Grime, Tilman and Keddy. Most recently debate between these schools and others has centred upon the relationship between species richness and ecosystem function (Tilman and Downing 1994, Naeem et al. 1994, 1995, 1996, Johnston et al. 1996, Tilman 1996, Tilman et al. 1996, Garnier et al. 1997, Huston 1997, Hodgson et al. 1998, Lawton et al. 1998, Thompson and Hodgson 1999). A large and growing literature has developed on the topic (Lawton et al. 1998) which will not be reviewed in detail here. Some observations pertinent to the thesis being developed here of the relevance of vegetation ecology to biodiversity studies will be made. Huston (1997) drew attention to a difficulty with the Tilman experiment (Tilman and Downing 1994, Tilman 1996). Tilman and Downing claimed that increasing species richness made plant ECOGRAPHY 2 2 5 (1999) @‘* Envimnmenm!gradient. €1 Fig. 3. Possible biomass and species richness patterns in a two-dimensional environmental space (dark shading indicates maximum values). (a) Biomass with a central maximum: (b) consequent species richness response with a maximum at intermediate biomass values: (c) projection of environmental space onto a biomass ( = productivity) gradient, where all radial gradients (AX to AY and AX to AZ) are projected onto a single biomass gradient AB: (d) an alternative pattern of species richness: if the pattern were projected onto the biomass gradient AB in (c), it would be indistinguishable from the pattern in (b). Reproduced with permission of Opulus Press. communities more resistant to the effects of drought, see also Tilman (1996). Huston (1997) points out that the level of species richness was a result of a nitrogen treatment and the response to drought could be interpreted as an interaction between rainfall and the additional biomass produced on the high nitrogen/low richness plots. In the experiments by Naeem et al. (1996) and Tilman et al. (1996), a selection probability effect ensures that species mixtures with higher numbers of species will contain the larger more productive species. Comparison of the mean biomass for 1, 2, 4, 8, etc. species mixtures must result in increasing biomass with increasing species richness (Huston 1997). The concentration on biological variables to the exclusion of the environment led to problems of interpreting the results. The choice of analysis in the Naeem et al. (1996) experiment led to results where their Fig. 2 shows that maximum biomass in certain monocultures was greater than in any of the 8 or 16 species mixtures. If biomass production increases with species richness, how does this relate to the hump-backed curve of species richness with biomass? Subsequent studies (Tilman et al. 1997, Lawton et al. 998) have focused on the influence of functionally ifferent types of plants e.g. grasses and legumes. The behaviour of individual species in relation to these richness gradients has not usually been examined, though see Naeem et al. (1996) and Tilman et al. 7). It must be of some concern that the relations between ecosystem functions and species richness do not have an r2 > 0.22, or for functional richness and composition > 0.36 (Tilman and Downing 1994, Tilman et al. 1996, 1997). Some unknown factors are exercising a greater influence on ecosystem production and function than is diversity. If the species response patterns of Ellenberg (Mueller-Dombois and Ellenberg 1974, Austin 1999) as part of a vegetation continuum (Austin and Smith 1989) are correct, then all biotic properties are functions of environment. The ultimate role of environmental gradients in determining the biological composition and functioning of ecosystems has yet to be investigated. An independent approach to the study of plant species competition was developed by Ellenberg (1953, 1954) in Germany concentrating on competition along gradients. An experiment with six grass species along a water-table gradient was carried out. When grown together in mixture the species showed marked shifts in their ecological (realised niche) compared to their physiological (fundamental niche) maxima in monoculture (Ellenberg 1953). Each species came to dominate a different segment of the water-table gradient. Subsequent work extended the approach to a nutrient gradient, with different levels of species richness (five and ten grass species), and developed a statistical analysis procedure for the experimental design (Austin and Austin 1980, Austin 1982). Species response under multispecies ECOGRAPHY 22:5 (1999) competition was shown to be a function of the particular species present in the mixture and the position along the nutrient gradient. Maximal biomass varied with the number of species depending on the position along the gradient (Austin and Austin 1980). Other studies have been carried out with thistles (Austin et al. 1985) and grass/legume mixtures under ambient and elevated C 0 2 (Navas et al. in press). Microcosms are becoming part of plant community ecology (Fraser and Keddy 1997). The experimental design usually consists of different treatments with a limited number of levels which are tested for whether they have a significant effect on species or vegetation properties. Compare Weiher and Keddy (1995) with seven treatments each at two levels with Austin and Austin (1980) with one treatment and 16 levels. The types of questions and the possible degree of resolution of the response are quite different, see also Keddy et al. (1994). If the responses of species and vegetation to gradients results are complex response curves, these will not be detected with less than five treatment levels. Campbell and Grime (1992) in a novel experimental design examined multispecies responses in relation to two environmental gradients. They constructed continuous gradients of nutrient concentration and disturbance in the same container. This allowed two gradients each with five levels to be studied together with the interactions in one container i.e. 25 combinations. Monocultures for seven species and the seven species mixture were replicated three times. The spatial autocorrelation between residuals for the neighbours was not significant. This experimental design (see also Burke and Grime 1996) offers the opportunity to explore species and plant community responses to an environmental space in a logistically efficient manner. The equivalent large-scale field experiments at the mercy of the weather and with the necessity of weeding are costly and logistically demanding. Gradient experiments with numerous levels and response surface analysis rather than ANOVAs offer advantages. The complex responses of individual species and of collective vegetation properties can then be examined in detail. A synthesis between observational vegetation ecology and experimental plant ecology will then be possible. Conservation studies Vegetation ecology has seen the establishment of large data bases of species by sites. Three distinct types can be recognised: a) phytosociological association tables where the site is a selected relevee of varying size (Westhoff and van der Maarel 1978); b) vegetation survey data where the site is a quadrat or plot of a specific size and the survey design may have been stratified or purposive; c) geographic distribution data 477 where species occurrence is recorded for large contigu- may contain species that are different from those in ous gridcells, and may be based on survey data of an species rich areas and have high conservation value. unspecified type or herbarium specimens. In general, Hotspots may occur for two main reasons. An unusual fauna databases on geographical distribution are re- environment may support endemic species or relict stricted to type (c). The limitations of such data for species from a previous climatic regime. An area may conservation and biodiversity studies are discussed by have high species richness due to steep local environmental gradients. Species richness may result from the Margules and Austin (1994) (see also Austin 1998). Very large phytosociological databases can be devel- artificial combination of species e.g. counting Koalas oped e.g. 35000 samples for the account of British and Grey kangaroos together when one is an arboreal Plant Communities (Rodwell 1991). The conceptual folivore and the other a ground dwelling herbivore. framework of subjectively identifiable associations and There is a strong case for combining the vegetation hierarchical syntaxonomic classification has practical paradigm ecology with conservation evaluation apvalue but remains controversial (Westhoff and van der proaches, while reducing the emphasis on species Maarel 1978, Rodwell 1991). The primary data have richness. immense value for conservation independent of the conceptual framework providing the subjective sampling bias is recognised. Quadrat surveys can also gen- Discussion erate large data bases. Grime et al. (1988) use a data base of > 10000 quadrats for a region in central Eng- The view of plant community ecology presented here, land of 3000 km2. Information on species preferences of a vegetation continuum embedded in environmental can also be built up from such survey data, for example space using statistical models to describe patterns in the indicator values of species for various environmen- species and collective properties such as species richness tal gradients developed by Ellenberg et al. (1992, see or dominance, is a very partial and personal view. A also Grime et al. 1988). The results of analysing these synthesis of vegetation ecology paradigms is needed. At data bases so far have reflected the particular paradigm the present time an increased dialogue between animal adopted by the authors. Australian experience has and plant community ecologists is probably of greater shown that a minimum data set of plots with presence/ priority. Lawton (1999) has provided a rather pessimistic reabsence data collated from various surveys, plot locaview of community ecology. “[It] is a mess, with so tion and a GIS can make a significant contribution to conservation evaluation (Austin 1998). Statistical mod- much contingency that useful generalisations are hard to find.” He is concerned that any laws, rules and elling of such data allows various different approaches mechanisms will be contingent i.e. “only true under to conservation evaluation to be adopted, see for examparticular or stated circumstances.” A more extreme ple Margules and Nicholls (1987), Margules and Stein quote is “the rules are contingent in so many (1989), and Margules and Austin (1991). State agencies ways. . . as to make the search for patterns unworkin New South Wales have adopted these methods in a able.” He is similarly concerned about population dymajor study of forest conservation requirements in the namics and advocates the search for pattern at the State (e.g. Anon. 1994a, b). Several hundreds of thou- macro-scale, that is the study of macroecology (Brown sands of hectares of forest were reserved as a result of 1984, 1995, Brown and Maurer 1989). However, as the the study. Much of the Australian work in this field has reviews above of species richness at various scales have been put together as a manual and package of com- indicated, even global patterns of species richness are puter programs called BIORAP (Margules et al. 1995). contingent on glacial history, continental patterns of It represents the collation of experience in vegetation mountain barriers, and current climate. Most results of ecology, climate estimation, environmental analysis, ecological studies are contingent on environment and and conservation evaluation using complementarity. history. What can be done? Examining and developing the links between paradigms There is a more intimate interdependency between in conservation ecology and mainstream ecology would concept, analytical methods and data than is commonly repay a detailed review. recognised in ecology (Austin 1998, 1999). Lawton There is an extensive literature on conservation eval- (1999) quotes as an example of useful generalisations in uation which is based on a very different paradigm but ecology, a diagram from Whittaker (1975) which shows also using databases and GIs. Vegetation communities the relationship between plant biomes and two climatic are mapped and wildlife habitat models are developed variables, mean annual temperature and mean annual based on the assumption of homogeneous plant com- precipitation. Whittaker’s earlier work on regional vegmunities over large areas (Scott et al. 1993). Yet an- etation patterns using direct gradient analysis is not other conservation focus is on species richness with the mentioned. This work on the vegetation of the Great detection and conservation of species rich “diversity Smokey Mountains (Whittaker 1952, 1956), the hotspots” (Prender@st ef al. 1993). However, this ap- Siskiyou Mountains (Whittaker 1960) and the Santa proach neglects the possibility that species poor areas Catalina Mountains. (Whittaker and Niering 1965) was 478 ECOGRAPHY 2 2 5 (1999) of the pioneering works which established the ,.ontinuurn concept. The results also show the existence of strong within-region patterns. There are general patterns of community organisation at regional scale with continuous change in composition, structure and collectiva properties contingent on environmental gradients. Whittaker (1952) demonstrated that similar patterns oc~uf for individual insect species with no species having identical patterns in a two-dimensional environmental Wace of elevation and topographic moisture. Brown (1984) mentions the work of Whittaker but then proceeds to emphasise geographic ranges rather than environmental gradients. One possible explanation is the difficulty of collecting suitable data for such analysis for animals. Another contributing factor may be the traditional concern with the temporal patterns of faunal abundance at a location rather than spatial distribution patterns. Statistical models capable of establishing quantitatively the existence of these patterns now provide a firm basis for their existence (Austin et al. 1984, 1990, 1994, Margules et al. 1987, Leathwick and Mitchell 1992, Franklin 1995, 1998, Leathwick 1995, 1998). Given the work of Whittaker (1952) and others (e.g. Bond 1957, Beals 1960, Rotenberry and Wiens 1980 and Sabo 1980), it is perhaps time that vegetation ecology paradigms were re-examined and re-applied to faunal data. Lawton (1999) discusses Schoener’s (1986) attempt to develop a contingent theory for community ecology. Schoener defines six primitive organismic axes (body size, generation time, etc.) and six environmental axes (severity of physical factors, spatial fragmentation etc.). These axes create a twelve-dimensional space within which community assembly patterns might be found. Lawton (1999) is sceptical: “whether it is worth the effort to shoe-horn every study into a point in n-dimensional space defined by combinations of at least 12 primitive axes, I have my doubts.” Yet, vegetation properties and species distributions are now being regularly examined in a seven-dimensional environmental space (Austin et al. 1994, Austin and Meyers 1996, Leathwick et al. 1998, see Pausas et al. 1995 for faunal example). The organismic axes can be regarded as equivalent to the functional attributes of plant ecologists (Grime et al. 1988, Ellenberg et al. 1992, Smith et al. 1997, Westoby 1998). These are now a significant concern for those plant ecologists investigating climate change (Smith et al. 1997). The pioneering work of the Sheffield school (Grime and Lloyd 1973, Grime 1979, Grime et al. 1988) in this respect represents a standard against which to judge the importance of functional attribute patterns. Perhaps the vegetation studies in organismic and environmental space need to be reviewed before the topic is dismissed. The expression of any biotic component is contingent on the environment. A first step would therefore be to examine patterns of biotic properties and processes in a relevant ECOGRAPHY 225 (l999) multidimensional environmental space. By relevant I mean not indirect gradients where physiologically influential variables are confounded in a location specific manner. The formulation of Schoener (1986) resembles a simpler formulation of Jenny (1941), the functional factorial approach to the study of soil properties (s): s = f(c1, p, r, 0,t...) where the five principal factors are regional climate (cl), parent material (p), topography (r), biota (0)and time (t). An array of soil properties can then be studied as a function of the environmental factors. Major (1951) suggested it could equally well be applied to vegetation. (In fact it could apply to any other dependent variable in an ecosystem.) Billings (1952) and Crocker (1952) commented on the mathematical limitations, the need to account for interactions of plants and the sheer complexity of the functional relationships which were known to occur (cf. Lawton 1999). Jenny (1941) intended the equation to be a guide. One would study a soil property, say soil nitrogen, as a quantitative function of rainfall while controlling as far as possible for the other factors i.e. Many of the studies quoted in this review do not bother to examine whether environmental variables may influence or confound their analysis. Is climate uniform within the region? Is parent material confounded with climate in the study? Does topographic heterogeneity within geographic pixcels contribute to the residual deviance? Can these experimental results obtained on sandy glacial till be extrapolated to heavy clay soils? This check list could materially improve analysis of community ecology issues. Vegetation ecologists have not adopted this approach with the exception of one neglected study. Perring (1958, 1959, 1960) explicitly adopted Major’s (1951) suggestion to investigate climatic and topographic gradients in chalk grassland. He examined species patterns in relation to aspect and slope gradients stratifed by four climatic areas while controlling for parent material, history and recent disturbance (Perring 1958). It was a graphical analysis without the benefit of statistics or computers. It does clearly show that patterns of species distribution in relation to topographic gradients are contingent on climate (Perring 1959). This contingency is however a regular ordered response to a climatic gradient probably associated with evaporative stress (Perring 1960). The towns nearest the different climatic study areas were York, Cambridge, Dorchester and Rouen. The inferences which can be drawn from this old study are that any study of biotic patterns covering England and Northern France cannot consider that the region is 479 environmentally homogeneous or that species response patterns across the region are insensitive to the local climatic or environmental gradients. This would also apply to a region containing any two of the adjacent areas. The framework incorporates currently important factors such as the regional flora as a potential determinant of the dependent variable. The questions then and today are what is the relative importance of the different factors and does their importance vary in a predictable way with environment or taxa? In other words, are the values of local contingent predictors of biotic patterns predictable from intra-regional gradients? Chapin et al. (1996) have recently focussed on this conceptual framework in the context of the sustainability of ecosystems. They use the Jenny equation as the framework within which interactions within the ecosystem between the dependent variables, vegetation and soils and their feedback to processes such as soil erosion and nutrient cycling can be considered. The weakness of the vegetation ecology paradigm presented here and of some other vegetation ecology studies is the absence of any allowance for spatial pattern. The book edited by Tilman and Kareiva (1997) illustrates the potential importance of the spatial arrangement of organisms and their environment. The key questions are again, what is the relative importance of spatial patterns and how does that importance change with environment? The proximal causes of ecological patterns and temporal behaviour are highly contingent as maintained by Lawton (1999). There is another question to be asked, are there any patterns in the importance of different proximal causes as a function of the distal environmental causes? A synthesis of the questions posed by animal ecologists with the environmental framework and statistical methods of vegetation ecologists may yield a less pessimistic outcome than that implied by Lawton (1999). Gaston and Blackburn (1999) present a defensive review of the role and importance of macroecology in ecology. They contrast observational analysis with reductionist experimental analysis. This has been an ongoing debate throughout the history of ecology (McIntosh 1985). The increasing importance of plant community experiments (Campbell and Grime 1992, Tilman and Downing 1994, Naeem et al. 1996, Tilman 1996) and their informed criticism (Huston 1997, Grime 1998, Hodgson et al. 1998) is leading to better experiments (Tilman et al. 1997), and a greater interface between observational plant ecology and experimentation. A more valid criticism of macroecology is the failure to adopt a sufficiently critical conceptual framework and analytical methodology when addressing large scale ecological patterns. Three strategies of analysis may be adopted. A collective biodiversity property may be correlated with a single predictor. A significant result can be regarded as a robust result having been obtained against a background of numerous other sources of variation. If the predictor is a complex indirect variable e.g. latitude, then a biological explanation of the relationship may still be elusive. A second strategy is multiple regression with well-defined direct or resource gradients as potential predictors, testing a particular hypothesis. The analysis of Currie (1991) testing the energy hypothesis for species richness is a clear example of this approach. The difficulty of distinguishing between regional and energy hypotheses due to collinearity problems remains (Francis and Currie 1998). The potential complementarity of hypotheses such as the correlation of terrestrial area with latitude (Rosenzweig 1995) and the energy relationship of Currie (1991) could also be addressed. A third as yet unrealised strategy would be to use a functional factorial approach similar to that of Jenny (1941) to both design the collection of data and to stratify the data for statistical analysis in order to test multiple hypotheses. This would require that the species data are such that the species actually co-occur and interact at a local scale (Huston 1999). It is only in this way that regional or global scale patterns can have ecological meaning. The environmental gradients selected would also need to be consistent with the biology of the taxa studied. One reason for the potential contribution of vegetation ecology to addressing these issues is that plants are easy to sample and measure; they do not run away. Large stratified surveys and the resulting data bases can be established relatively cheaply. Plot data at a scale relevant to species or individual interactions can be collected and within plot environmental heterogeneity relatively easily allowed for. To collect equivalent unambiguous data for fauna will be much more difficult and costly. Although the continuum concept seems to be generally accepted by vegetation ecologists its potential for organising questions about the community level of ecology has yet to be realised. If the continuum concept (Austin and Smith 1989) has any validity, then characteristics of species’ responses and patterns of collective properties such as species richness, dominance, standing biomass and assembly rules should be detectable in appropriate environmental spaces. Ecological processes are contingent on environment, space and history. This is true whether we study local, regional or global patterns. The research issues are how to determine the relative importance of the different distal causes and to determine how contingent the processes are on environment etc. This is no easy task given the high degree of correlation which exists among potential predictors. The concepts and methods of vegetation ecology have a role to play if better communication between the different paradigms can be established. ECOGRAPHY 2 2 5 (1999) Austin, M. P. et al. 1994. Determining species response functions to an environmental gradient by mealis of a P-func. tion. J . Veg. Sci, 5: 21 j-228. Austin. M. P. et al. 1995. Modelling of landscape patterns and processes using biological data. Subproject 5: simulated data case study. Rep. ERIN. CSIRO Wildlife and Ecol.. Canberra. Beak. E. W. 1960. Forest bird communities in the Apostle References Islands of Wisconsin. - Wilson Bull. 72: 156-181. Beerling. D. J.. Huntley, B. and Bailey. J. P. 1995. Climate and Adams, J. M. and Woodward, F. I. 1989. Patterns in tree the distribution of Fullopiu , j q w ? i a i : use of an introduced species richness as a test of the glacial extinction hypothespecies to test the predictive capacity of response surfaces. sis. - Nature 339: 699-701. - J. Veg. Sci. 6: 269-282. Anon. 1994a. Fauna of north-east NSW forests. North-east Belbin, L. 1991. Semi-strong hybrid scaling. a new ordination forests biodiversity study report No. 3. Int. Rep., NSW algorithm. - J. Veg. Sci. 2: 491-496. National Parks and Wildlife Serv.. Sydney. Billings, W. D. 1952. The environmental complex in relation Anon. 1994b. Flora of north-east NSW forest. North-east to plant growth and distribution. Quart. Rev. Biol. 27: forests biodiversity study report No. 4. Int. Rep., NSW 251 -265. National Parks and Wildlife Serv.. Sydney. Bio, A. M. F., Alkemade, R. and Barendregt, A. 1998. DeterAl-Mufti, M. M. et al. 1977. A quantitative analysis of shoot mining alternative models for vegetation response analysis: phenology and dominance in herbaceous vegetation. J . a non-parametric approach. J. Veg. Sci. 9: 5-16. EcoI. 65: 759-791. Austin, M. P. 1980. Searching for a model for use in vegeta- Bond, R. R. 1957. Ecological distribution of breeding birds in the upland forests of southern Wisconsin. - Ecol. Monogr. tion analysis. - Vegetatio 42: l 1-21. 27: 351-384. Austin. M. P. 1982. Use of relative physiological performance value in the prediction of performance in multispecies Borcard, D. and Legendre, P. 1994. Environmental control and spatial structure in ecological communities: an exammixtures from monoculture performance. - J. Ecol. 70: ple using oribatid mites (Acari, Oribatei). - Environ. Ecol. 559-570. Stat. 1: 37-61. Austin, M . P. 1985. Continuum concept, ordination methods, Brown, J. H. 1984. On the relationship between abundance and niche theory. Annu. Rev. Ecol. Syst. 16: 39-61. Austin, M. P. 1987. Models for the analysis of species response and distribution of species. - Am. Nat. 124: 255-279. to environmental gradients. Vegetatio 69: 35-45. Brown, J. H. 1995. Macroecology. Chicago Univ. Press. Austin, M. P. 1990. Community theory and competition in Brown, J. H. and Maurer, B. A. 1989. Macroecology: the vegetation. - In: Grace, J. B. and T'ilman, D. (eds), division of food and space among species on continents. Perspectives on plant competition. Academic Press, pp. Science 243: 1145-1150. 215-238. Brzeziecki, B., Kienast, F. and Wildi, 0. 1995. Modelling Austin, M. P. 1998. An ecological perspective on bidiversity potential impacts of climate change on the spatial distribuinvestigations: examples from Australian eucalypt forests. tion of zonal forest communities in Switzerland. - J. Veg. - Ann. Missouri Bot. Gard. 85: 2-17. Sci. 6: 257-268. Austin, M . P. 1999. A silent clash of paradigms: some incon- Burke, M. J. W. and Grime, J. P. 1996. An experimental study sistencies in community ecology. - Oikos 86: 170-178. of plant community invasibility. - Ecology 77: 776-790. Austin, M . P. and Austin, B. 0. 1980. Behaviour of experi- Caley, M. J. and Schluter, D. 1997. The relationship between mental olant communities alone a nutrient gradient. - J. local and regional diversity. Ecology 78: 70-80. Ecol. 68': 891-918. Campbell, B. D. and Grime, J. P. 1992. An experimental test Austin, M. P. and Cunnineham. R. B. 1981. Observational of plant strategy theory. - Ecology 73: 15-29. analysis of environmenh gradients. Proc. Ecol. SOC. Chapin, F. S. 111, Torn, M. S. and Tateno, M . 1996. Principles Aust. 11: 109-119. of ecosystem sustainability. - Am. Nat. 148: 1016-1037. Austin, M. P. and Gaywood, M. J. 1994. Current problems of Cornell, H. V. and Lawton, J. H. 1992. Species interactions, environmental gradients and species response curves in local and regional processes, and limits to the richness of relation to continuum theory. - J. Veg. Sci. 5: 473-482. ecological communities: a theoretical perspective. J. Austin, M . P. and Meyers, J. A. 1996. Current approaches to Anim. Ecol. 61: 1-12. modelling the environmental niche of eucalypts: implications for management of forest biodiversity. - For. Ecol. Cornell, H. V. and Karlson, R. H. 1996. Species richness of reef-building corals determined by regional and local proManage. 85: 95-106. cesses. J. Anim. Ecol. 65: 233-241. Austin, M. P. and Nicholls, A. 0. 1997. To fix or not to fix the species limits, that is the ecological question: response to Cornell, H. V. and Karlson, R. H. 1997. Local and regional processes as controls of species richness. - In: Tilman, D. Jari Oksanen. - J. Veg. Sci. 8: 743-748. and Kareiva, P. (eds), Spatial ecology: the role of space in Austin, M. P. and Smith, T. M. 1989. A new model for the population dynamics and interspecific interactions. Princecontinuum concept. Vegetatio 83: 35-47. ton Univ. Press, pp. 250-268. Austin, M. P., Cunningham, R. B. and Fleming, P. M. 1984. New approaches to direct gradient analysis using environ- Crawley, M. J. 1997. The structure of plant communities. In: Crawley, M. J. (ed.) Plant ecology, 2nd ed. Blackwell, mental scalars and statistical curve-fittine mocedures. pp. 475-531. Vegetatio 55: 11-27. Austin. M. P.. Nicholls. A. 0. and Mareules. C. R . 1990. Cresswell, J. E., Vidal-Martinez, V. M. and Crichton, N . J. 1995. The investigation of saturation in species richness of Measurement of the realised qualitative iiche: environmencommunities: some comments on methodology. - Oikos tal niches of five Eucalyptus species. - Ecol. Monogr. 60: 72: 301-304. I61 177. Austin, M. P., Pausas, J. G. and Nicholls, A. 0. 1996. Patterns Crocker, R. L. 1952. Soil genesis and the pedogenic factors. Quart. Rev. Biol. 27: 139-168. of tree species richness in relation to environment in southeastern New South Wales, Australia. Aust. J. Ecol. 21: Currie, D. J. 1991. Energy and large-scale patterns of animaland plant-species richness. Am. Nat. 137: 27-49. 154- 164. Austin, M . P. et al. 1985. Relative growth of six thistles along Currie, D. J. and Paquin, V. 1987. Large-scale biogeographical patterns of species richness of trees. - Nature 329: 326a nutrient gradient with multispecies competition. - J. Ecol. 73: 667-684. 327. ~~ ~ ~ k ~ o w l e d ~ c n l eI nthank A . B. Wellington. J . R. W. Reid. t.~ D, M. Sprdtt and A . J. Kearns for useful discussions and on the manuscript. D. Grice for assistance with the figures and preparation of the manuscript and the Divisional library staff for assistance with references. ECOGRAPHY 22 5 (1999) Curtis, J. T. 1959. The vegetation of Wisconsin: an ordination of plant communities. - Univ. Wisconsin Press. de Wit, C. T. 1960. On competition. - Versl. Landbouwkonderz. 66: 1-82. Ellenberg, H. 1953. Physiologisches und okologisches Verhalten derselben Pflanzenarten. - Ber. Deutsch Bot. Ges. 65: 351-362. Ellenberg, H. 1954. Uber einige Fortschritte der kausalen Vegetationskunde. - Vegetatio 5/6: 199-21 1. Ellenbere. H. 1988. Vegetation ecologv of central EuroDe. Camcridge Univ. Press. Ellenberg. H. et al. 1992. Zeigerwerte von Pflanzen in Mitteleuropa: - Scr. Geobot. 18: Faith, D. P., Minchin, P. R. and Belbin, L. 1987. Compositional dissimilarity as a robust measure of ecological distance. - Vegetatio 69: 57-68. Francis, A. P. and Currie, D. J. 1998. Global patterns of tree species richness in moist forests: another look. - Oikos 81: 598-602. Franklin, J. 1995. Predictive vegetation mapping: geographic modelling of biospatial patterns in relation to environmental gradients. - Prog. Physical Geogr. 19: 474-499. Franklin, J. 1998. Predicting the distribution of shrub species in southern California from climate and terrain-derived variables. - J. Veg. Sci. 9: 733-748. Fraser, L. H. and Keddy, P. 1997. The role of experimental microcosms in ecological research. - Trends Ecol. Evol. 12: 46-49. Garnier, E. et al. 1997. A problem for biodiversity-productivity studies: how to compare the productivity of multispecific plant mixtures to that of monocultures? - Acta Oecol. 18: 657-670. Gaston, K. J. and Blackburn, T. M. 1999. A critique for macroecology. - Oikos 84: 353-368. Giller, P. S. 1984. Community structure and the niche. Chapman and Hall. Goldberg, D. E. and Miller, T. E. 1990. Effects of different resource additions on species diversity in an annual plant community. - Ecology 71: 213-225. Goldberg, D. E. and Novoplansky, A. 1997. On the relative importance of competition in unproductive environments. - J. Ecol. 85: 409-418. Goldsmith, F. B. 1973a. The vegetation of exposed sea cliffs at South Stack. Anglesev. I. The multivariate aDDrOaCh. - J. ._ Ecol. 61: 787-8fi. Goldsmith. F. B. 1973b. The vegetation of exDosed sea cliffs at South Stack, Anglesey. 11. gxperimental studies. - J. Ecol. 61: 819-830. Grime, J. P. 1979. Plant strategies and vegetation processes. Wiley. Grime, J. P. 1998. Benefits of plant biodiversity to ecosystems: immediate, filter and founder effects. - J. Ecol. 86: 902910. Grime, J. P. and Lloyd, P. S. 1973. An ecological atlas of grassland plants. - Edward Arnold. Grime, J. P., Hodgson, J. G. and Hunt, R. 1988. Comparative plant ecology: a functional approach to common British species. - Unwin Hyman, London. Guisan, A., Theurillat, J.-P. and Kienast, F. 1998. Predicting the potential distribution of plant species in an alpine environment. - J. Veg. Sci. 9: 65-74. Harper, J. L. 1977. Population biology of plants. - Academic Press. Harrison, S., Ross., S. J. and Lawton, J. H. 1992. Beta diversity on geographic gradients in Britain. - J. Anim. Eco~.61: 151-158. Hastie, T. and Tibshirani, R. 1990. Generalised additive models. - Chapman and Hall. Hawkins, B. A. and Compton, S. G. 1992. African fig wasp communities: undersaturation and latitudinal gradients in species richness. - J. Anim. Ecol. 61: 361-372. Hodgson, J. G. et al. 1998. Does biodiversity determine ecosystem function? The Ecotron experiment reconsidered. - Funct. Ecol. 12: 843-848. I _ Huston, M. A. 1993. Biological diversity, soils and economics. - Science 262: 1676-1680. Huston, M. A. 1994. Biological diversity: the coexistence of species on changing landscapes. - Cambridge Univ. Press. Huston, M. A. 1997. Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. - Oecologia 110: 449-460. Huston, M. A. 1999. Local processes and regional patterns: appropriate scales for understanding variation in the diversity of plants and animals. - Oikos 86: 393-401. James, F. C. and McCulloch, C. E. 1990. Multivariate analysis in ecology and systematics: panacea or pandora’s box. Annu. Rev. Ecol. Syst. 21: 129-166. Jenny, H. 1941. Factors of soil formation. - McGraw-Hill. Johnston, K. H. et al. 1996. Biodiversity and the productivity and stability of ecosystems. - Trends Ecol. Evol. 11: 372-377. Jongman, R. G. H., ter Braak, C. J. F. and van Tongeren, 0. F. 1987. Data analysis in community and landscape ecology. - Pudoc, Wageningen. Kareiva, P. 1994. Diversity begets productivity. - Nature 368: 686-687. Keddy, P. A. 1987. Beyond reductionism and scholasticism in plant community ecology. - Vegetatio 69: 209-21 1. Keddy, P. A. 1989. Competition. - Chapman and Hall. Keddy, P. A. 1992. Assembly and response rules: two goals for predictive community ecology. - J. Veg. Sci. 3: 157-164. Keddy, P. A. 1994. Applications of the Hertzsprung-Russell star chart to ecology: reflections on the 21st birthday of geographical ecology. - Trends Ecol. Evol. 9: 231-234. Keddy, P. A,, Twolan-Strutt, L. and Wisheu, I. C. 1994. Competitive effect and response rankings in 20 wetland plants: are they consistent across three environments? - J. Ecol. 82: 635-643. Kent, M. and Coker, P. 1992. Vegetation description and analysis: a practical approach. - Belhaven Press, London. Kerr, J. T. and Packer, L. 1997. Habitat heterogeneity as a determinant of mammal species richness in high-energy regions. - Nature 385: 252-254. Krebs C. J. 1994. Ecology: the experimental analysis of distribution and abundance, 4th ed. - Harper and Row. Kuhn, T. S. 1970. The structure of scientific revolutions, 2nd ed. - Univ. Chicago Press. Latham, R. E. and Ricklefs, R. E. 1993. Global patterns of tree species richness in moist forests: energy-diversity theory does not account for variation in species richness. Oikos 67: 325-333. Lawton, J. H. 1982. Vacant niches and unsaturated communities: a comparison of bracken herbivores at sites on two continents. - J. Anim. Ecol. 51: 573-595. Lawton, J. H. 1994. What do species do in ecosystems. Oikos 71: 367-374. Lawton, J. H. 1999. Are there general laws in ecology? Oikos 84: 177-192. Lawton, J. H.’et al. 1998. Biodiversity and ecosystem function: getting the Ecotron experiment in its correct context. Funct. Ecol. 12: 848-852. Leathwick, J. R. 1995. Climatic relationships of some New Zealand forest tree species. - J. Veg. Sci. 6: 237-248. Leathwick, J. R. 1998. Are New Zealand’s Nothofagus species in equilibrium with their environment? - J. Veg. Sci. 9: 719-732. Leathwick, J. R. and Mitchell, N. D. 1992. Forest pattern, climate and vulcanism in central North Island, New Zealand. - J. Veg. Sci. 3: 603-616. Leathwick, J. R., Whitehead, D. and McLeod, M. 1996. Predicting changes in the composition of New Zealand’s indigenous forests in response to global warming: a modelling approach. - Environ. Softw. 11: 81-90. Leathwick, J. R., Burns, B. R. and Clarkson, B. D. 1998. Environmental correlates of tree alpha-diversity in New Zealand primary forests. - Ecography 21: 235-246. ECOGRAPHY 225 (1999) Lehmann, A. 1998. GIS modeling of submerged macrophyte distribution using generalized additive models. - Plant Eco]. 139: 113-124. Legendre, P. 1993. Spatial autocorrelation: trouble or new paradigm? - Ecology 74: 1659-1673. Lenihan, J. M. 1993. Ecological response surfaces for North American boreal tree species and their use in forest classification. - J. Veg. Sci. 4: 667-680. Major, J. 1951. A functional, factorial approach to plant ecology. - Ecology 32: 392-412. Margules C. R. and Austin, M. P. (eds). 1991. Nature conservation: cost effective biological surveys and data analysis. - CSIRO, Melbourne, Australia. Margules, C. R. and Austin, M. P. 1994. Biological models for monitoring species decline: the construction and use of data bases. - Philos. Trans. R. SOC.Lond. B Biol. Sci. 344: 69-75. Margules, C. R. and Gaston, K. J. 1994. Biodiversity and agriculture. - Science 265: 457. Margules, C. R. and Nicholls, A. 0. 1987. Assessing the conservation value of remnant habitat ‘islands’: mallee patches on the western Erye Peninsula, South Australia. In: Saunders, D. A. et al. (eds), Nature conservation: the role of remnant vegetation. Surrey Beatty and Sons, pp. 89-102. Margules, C. R. and Stein, J. 1989. Patterns in the distributions of species and the selection of nature reserves: an example from Eucalyptus forests in south-eastern New South Wales. - Biol. Conserv. SO: 219-238. Margules, C. R., Nicholls, A. 0. and Austin, M. P. 1987. Diversity of Eucalyptus species predicted by a multi variables environmental gradient. - Oecologia (Berl.) 71: 229232. Margules, C. R. et al. 1995. BioRap: guidelines for using the BioRap methodology and tools. - CSIRO, Canberra. May, R. M. 1985. Evolutionary ecology and John Maynard Smith. In: Greenwood, P. J., Harvey, P. H. and Slatkin, M. (eds), Essays in honour of John Maynard Smith. Cambridge Univ. Press, pp. 107- 116. McCullagh, P. and Nelder, J. A. 1989. Generalized linear models, 2nd ed. - Chapman and Hall. McIntosh, R. P. 1985. The background of ecology. - Cambridge Univ. Press. Minchin, P. R. 1987. Simulation of multidimensional community patterns: towards a comprehensive model. - Vegetatio 71: 145-156. Minchin, P. 1989. Montane vegetation of the Mt. Field Massif, Tasmania: a test of some hypotheses about properties of community patterns. - Vegetatio 83: 97-1 10. Mueller-Dombois, D. and Ellenberg, H. 1974. Aims and methods of vegetation ecology. - Wiley. Naeem. S. et al. 1994. Declining biodiversity can alter the performance of ecosystems. - Nature 368: 734-737. Naeem. S. et al. 1995. Empirical evidence that declining species diversity may alter the performance of terrestrial ecosystems. - Proc. R. SOC.Lond. B Biol. Sci. 347: 249262. Naeem. S. et al. 1996. Biodiversity and plant productivity in a model assemblage of plant species. - Oikos 76: 259-264. Navas, M.-L. et al. Effect of competition on the responses of grasses and legumes to elevated atmospheric CO, concentration along a nitrogen gradient: differences between isolated plants, monocultures and multi-species mixtures New Phytologist, in press. Nicholls, A. 0. 1989. How to make biological surveys go further with generalized linear models. - Biol. Conserv. 50: 51-76. Noble, I. R. 1989. Ecological traits of Eucalyprus L‘Herit. subgenera Monocalyptus and Symphyomyrtus. - Aust. J. Bot. 37: 207-224. Okland, R. H. 1999. On the variation explained by ordination and constrained ordination axes. - J. Veg. Sci. 10: 131136. Oksanen, J. 1997. Why the beta-function cannot be used to estimate skewness of species responses. J . Veg. Sci. 8: 147- 152. Oksanen, J. and Minchin, P. R. 1997. Instability of ordination results under changes in input data order: explanations and remedies. - J . Veg. Sci. 8: 447-454. Partel, M. et al. 1996. The species pool and its relation to species richness: evidence from Estonian plant communities. - Oikos 75: 11 1- 117. Pausas, J. G. 1994. Species richness patterns in the understory of Pyrenean Pinus syluestris forests. - J. Veg. Sci. 5: 5 17-524. Pausas, J. G. and Carreras, J. 1995. The effect of bedrock type, temperature and moisture on species richness of Pyrenean Scots pine (Pinus syluestris L.) forests. - Vegetatio 116: 85-92. Pausas, J. G., Braithwaite, L. W. and Austin, M. P. 1995. Modelling habitat quality for arboreal marsupials in the south coastal forests of New South Wales, Australia. For. Ecol. Manage. 78: 39-49. Peet, R. K. et al. 1988. Putting things in order: the advantages of detrended correspondence analysis. Am. Nat. 131: 924-934. Perring, F. 1958. A theoretical approach to a study of chalk grassland. - J. Ecol. 46: 665-679. Perring, F. 1959. Topographical gradients of chalk grassland. J. Eco~. 447-481. 47: Perring, F. 1960. Climatic gradients of chalk grassland. - J. Ecol. 48: 415-442. Prendergast, J. R. et al. 1993. Rare species, the coincidence of diversity hotspots and conservation strategies. - Nature 365: 335-337. Ricklefs, R. E. 1987. Community diversity: relative roles of local and regional processes. Science 235: 167-171. Ricklefs, R. E. and Schluter, D. (eds). 1993. Species diversity in ecological communities: historical and geographical perspectives. - Univ. Chicago Press. Rodwell, J. S. 1991. British plant communities. Vol. 1. Woodlands and scrub. Cambridge Univ. Press. Rosenzweig, M. L. 1995. Species diversity in space and time. Cambridge Univ. Press. Rotenberry, J. T. and Wiens, J. A. 1980. Habitat structure, patchiness, and avian communites in North American steppe vegetation: a multivariate analysis. - Ecology 61: 1228- 1250. Sabo, S . R. 1980. Niche and habitat relations in subalpine bird communities of the white mountains of New Hampshire. Ecol. Monogr. 50: 241-259. Schoener, T. W. 1986. Overview: kinds of ecological communities - ecology becomes pluralistic. - In: Diamond, J. and Case. T. J. (edsl. Communitv ecolow. Haruer and Row. pp. 333-343: Scott, J. M. et al. 1993. Gap analysis: a geographic approach to protection of biological diversity. - Wildlife Monogr. 123 1-41. Shipley, B. and Keddy, P. A. 1987. The individualistic and community-unit concepts as falsifiable hypotheses. - Vegetatio 69: 47-55. Shugart, H. H. 1998. Terrestrial ecosystems in changing environments. - Cambridge Univ. Press. Smith, P. A. 1994. Autocorrelation in logistic regression modelling of species’ distributions. - Global Ecol. Biogeogr. Lett. 4: 47-61. Smith, T. M. and Huston, M. A. 1989. A theory of the spatial and temporal dynamics of plant communities. - Vegetatio 83: 49-69. Smith, T. M., Shugart H. H. and Woodward, F. I. 1997. Plant functional types. - Cambridge Univ. Press. Srivastava, D. S. 1999. Using local-regional plots to test for species saturation: pitfalls and potentials. - J. Anim. Ecol. 68: 1-16. Tausch, R. J. et al. 1995. Patterns of ordination and classification instability resulting from changes in input data order. - J. Veg. Sci. 6: 897-902. I - ” ECOGRAPHY 22:5 (1999) ter Braak, C. J. F. 1986. Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. - Ecology 67: 1167-1179. ter Braak, C. J. F. 1987. The analysis of vegetation-environment relationships by canonical correspondence analysis. - Vegetatio 69: 69-77. Thompson, K. and Hodgson, J. G. 1999. Response to Lawton et al. - Funct. Ecol. 12: 852-853. Tilman, D. 1982. Resource competition and comunity structure. - Princeton Univ. Press. Tilman, D. 1988. Plant strategies and the structure and dynamics of plant communities. - Princeton Univ. Press. Tilman, D. 1996. Biodiversity: population versus ecosystem stability. - Ecology 77: 350-363. Tilman, D. and Downing, J. A. 1994. Biodiversity and stability in grasslands. - Nature 367: 363-365. Tilman, D. and Kareiva, P. 1997. Spatial ecology: the role of space in population dynamics and interspecific interactions. - Princeton Univ. Press. Tilman, D. and Pacala, S. 1993. The maintenance of species richness in plant communities. - In: Ricklefs, R. E. and Schluter, D. (eds), Species diversity in ecological communities: historical and geographical perspectives. Univ. Chicago Press, pp. 13-25. Tilman, D., Wedin, D. and Knops, J. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. - Nature 379: 718-720. Tilman, D. et al. 1997. The influence of functional diversity and composition on ecosystem processes. - Science 277: 1300- 1302. van der Maarel, E. 1989. Theoretical vegetation science on the way. - Vegetatio 83: 1-6. Wartenberg, D., Ferson, S. and Rohlf, F. J. 1987. Putting things in order: a critque of detrended correspondence analysis. - Am. Nat. 129: 434-448. Weiher, E. and Keddy, P. A. 1995. The assembly of experimental wetland plant communities. - Oikos 73: 323-335. Westhoff, V. and van der Maarel, E. 1978. The Braun-Blanquet approach. - In: Whittaker, R. H. (ed.), Classification of plant communities. Junk, The Hague, pp. 287-399. Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. - Plant Soil 199: 213-227. Whittaker, R. H. 1952. A study of summer foliage insect communities in the Great Smokey Mountains. - Ecol. Monogr. 22: 1-44. Whittaker, R. H. 1956. Vegetation of the Great Smoky Mountains. - Ecol. Monogr. 26: 1-80. Whittaker, R. H. 1960. Vegetation of the Siskiyou Mountains, Oregon and California. - Ecol. Monogr. 30: 279338. Whittaker, R. H. 1975. Communities and ecosystems, 2nd ed. - Macmillian. Whittaker, R. H. 1978. Direct gradient analysis. - In: Whittaker. R. H. (ed.). Ordination of ~ l a n tcommunities. Junk,’ The Hague, pp. 7-51. Whittaker. R. H. and Niering. W. A. 1965. Vegetation of the Santa Catalina Mount&, Arizona: a grdient analysis of the south slope. - Ecology 46: 429-452. Yee, T. W. and Mitchell, N. D. 1991. Generalised additive models in plant ecology. - J. Veg. Sci. 2: 587-602. Zobel, M. 1997. The relative role of species pools in determining plant species richness: an alternative explanation of species coexistence? - Trends Ecol. Evol. 12: 266-269. ECOGRAPHY 2 2 5 (1999)

Journal

EcographyWiley

Published: Dec 1, 1999

There are no references for this article.