On the measurement of diversity-productivity relationships in a northern mixed grass prairie (Grasslands National Park, Saskatchewan, Canada)

For the investigation of diversity-productivity relationships under natural conditions, we present an operationally feasible measurement scheme explicitly considering the spatial organization of vegetation. We hypothesised that the spatial arrangement of the coexistence of species influences patch-level productivity. To characterise diversity, co-occurrences of species were recorded along oval transects allowing scaling by aggregation between 5 cm and 25 m. Productivity was characterised by field radiometric measurements, calibrated for leaf area and biomass, arranged in a sampling scheme scalable between 20 cm and 50 m. All data were collected along a slight resource gradient in the Stipa-Bouteloua (upland) community of the northern mixed-grass prairie in Grasslands National Park, Saskatchewan. We found a wide range of correlations (Kendall’s τ between -0.2 and 0.9) between various measures of diversity (species richness, local species combinations) and productivity (average and variability of leaf-area index) as a function of sampling unit size. For field assessment of patch-level composition and functioning, we recommend to use samples at the spatial resolution corresponding to the maximum number of local species combinations as an appropriate scale for comparison. We demonstrate how our sampling methodology can be considered for possible process-oriented inference about diversity and productivity. To characterise diversity-productivity relationships for long-term monitoring and prediction of plant community structure and functioning, scalable, repeatable, non-destructive observations should be applied. Nomenclature: Budd et al. 1987. Abbreviations: NS number of species; SC species combination; NRSC number of realised (observed) species combinations; NRSCmax maximum of NRSC vs. resolution (i.e., sampling unit size); LNM resolution at which NRSCmax occurs; FPARFraction of absorbed photosynthetically active radiation; LAI leaf-area index; LAIave average of LAI tioning have developed due to the need for generalization from a limited number of (potentially incompatible) observations (Ricklefs and Schluter 1993). Powerful and widely accepted concepts, such as the “humped-back” (Grime 1973), the “dynamic equilibrium” (Huston 1979), or the “resource ratio” (Tilman 1982) models, although their assumptions are somewhat different, all predict a unimodal relationship between diversity and productivity (Grace 2001). In general, it is difficult to translate generalised concepts of diversity and productivity into operationally feasible measurement schemes. There are many potential discrepancies with regard to what to measure (e.g., species richness, number of functional groups, connectivity of the food-web to quantify diversity; total or aboveground biomass, leaf-area index, etc. to quantify productivity), where to measure (e.g., number, size and arrangement of sampling units) and when to measure (e.g., should the measurements be repeatable within one or over several growing seasons). Spatial structure is one of the inherent characteristics of vegetation and, as such, it influences diversity-productivity relationships (Rosenzweig 1995). The vast literature of vegetation pattern analysis (Greig-Smith 1983, Dale 1999) has provided much less stimulus for spatial scaling of diversity-productivity models than might have been expected (Wiens 1989). In spite of seminal results emphasizing the “importance of being discrete and spatial” (Durrett and Levin 1994, Bartha et al. 1997, Czárán 1998) and operational measurement specifications for taking it into account (Juhász-Nagy and Podani 1983), there is no consensus about how to handle spatial structure explicitly and, as a result, it is frequently excluded from the model parameters (Loreau 1998, Grace 2001). Our proposition here is to match the “complexity” of a conceptual model and its field implementation. We suggest that productivity be characterised by the areal and temporal integral of production, and that diversity explicitly comprises the pattern of species coexistence as a spatial component. That is, productivity is a function of resources and neighbourhood effects: where i runs for all resources (e.g., light, nutrients, etc.) and j runs for all neighbourhoods (e.g., species spatial configurations). Even without assuming any specific form of the functions (f and g), the relationship described above is scale dependent, since it explicitly contains second-order (“neighbour-dependent”) effects (Bailey and Gatrell 1995, p. 77, Csillag et al. 2001). The objective of this paper is to investigate the scaledependence (or more strictly, the resolution-dependence) of patch-level diversity-productivity relationships in a natural grassland ecosystem where, when considering productivity, neither resources nor neighbours can be neglected as control factors. Instead of attempting to prove a theory we would like to provide the methodological tools to characterise simultaneously two aspects of an ecosystem: composition and functioning. To do this, we focus our attention to the relationships between parameters of distributions describing first-order and second-order effects (Figure 1). We emphasise that this approach is suitable for in situ studies. The rest of the paper is organised as follows. In Section 2, a brief description of the sites is provided. In Section 3, we summarise the field sampling methodology and the associated statistical tools. Section 4 summarises the results of the diversity-productivity data analysis. Finally, in Section 5 we conclude with a discussion of the relevance of these results, with some potential extensions and implications. area time i j ij f g z z ∑ production resource neighbor ~ [ ( , )] Figure 1. Relationships between measures of diversity (on the left) and productivity (on the right) and the focus of this paper. NS stands for “number of species” (or species richness), NRSC stands for “number of realised species combinations”, LAI stands for leaf-area index (our surrogate for productivity) and γ(LAI) stands for the semi-variogram of LAI. The pair on the top represents first-order characetristics, the pair on the bottom are second-order ones. Grey arrows indicate well-documented relationships, black arrows indicate relationships in the focus of this study. 146 Csillag et al.