Sustainable Indicators Conference - An Approach Towards the Quantitative Identification of Environmental Indicators

References

An Approach Towards the Quantitative Identification of Environmental Indicators

Ockie Bosch, Landcare Research NZ Ltd, Alexandra

The use of indicators in vegetation condition assessment is a well-known practice in the grazing industry (Hurt et al, 1993; Bosch and Janse van Rensburg, 1987). A knowledge of the indicators of vegetation and soil health is becoming increasingly important with the growing interest of land managers in community-based monitoring programmes. Monitoring of species that are not highly responsive to management, decrease the sensitivity of the condition and trend assessment techniques that are available. The exclusive use of indicator species can also simplify monitoring, in that land managers do not have to have a knowledge of all the species that can occur in a particular area. In this brief talk, the approach published by Bosch and Gauch (1991) is used to demonstrate the identification of vegetation and soil health gradients, and their use in defining plant indicator species for different conditional states. The approach can also be used to quantitatively determine various other environmental indicators that can be used in the monitoring of environmental quality.

The Approach

A methodology was required to quantitatively define vegetation and soil health indicators, using the presence or abundance of plant species occurring in an area. For this, modified ordination techniques (Bosch and Gauch, 1991) are used in which a data matrix is analysed and continually sub-divided on the basis of the main factors responsible for the variation in the data, in that way creating different gradients present in the data.

Data Set Requirements

The technique requires a data set that consists of sites X species abundances X environmental factors (fixed, e.g., topography) and soil condition factors (non-fixed, e.g., soil compaction), and known management histories (e.g., OSTD, stocking rates, time of grazing, burning, etc.) of at least 10 percent of the sites in the matrix.

Undisplayed Graphic

Identifying Major Habitat Environmental Groups

The sites are initially ordinated with any technique that deals well with data representing long gradients (e.g., TWINSPAN, DECORANA - Gauch, 1982), in order to subdivide the data set into major groups such as climatic zones, topo-graphical units, aspects, soil types, etc. (Figure 1).

Establishing Gradients

Each of the habitat/environmental groups are treated as a new data set and is further analysed with modified versions of Reciprocal Averaging, and centred and standardised Principal Components Analysis (Bosch and Gauch, 1991). These techniques are appropriate for data sets with shorter gradients (as could be expected after the initial analyses to subdivide the data set into habitat groups). The modification includes focusing on the first axis (main factor), while the second and higher axes are mathematically combined in a single value (residual). The size of the residual is used to refine the ordination model in order to eliminate sites in the data matrix that are still more influenced by other factors than the main factor of interest (on the first axis). Sites with a residual value larger than 50 percent of the Euclidean distance of the x-axis are removed from the matrix, in order to refine the gradient.

The process of developing gradients are continued with a continual grouping and re-analysis of sub-groups (Figure 2).

Undisplayed Graphic

Each step provides a gradient on the first axis, that could represent any gradient such as soil fertility, grazing impact, soil moisture conditions, soil compaction and erosion, soil acidity, etc.

Identifying Indicators

At any level of analysis the species abundances are plotted against the gradient on the first axis and tested statistically for the significance of the regression. The goodness of fit is used to assess the value of each of the species as an indicator. In this way, plant indicators for different vegetation and soil conditions are identified for each of the major habitat' environmental groups in the landscapes being studied.

The example in figure 3 illustrates that species 1 is an indicator of high grazing impact, while species 3 is associated in the low grazing impact. Species 2 increases with an increase of grazing impact, but does not occur in vegetation subjected to a high impact. Species 4 does not form any relationship with the grazing impact gradient, and is therefore not identified as an indicator.

Undisplayed Graphic

Concluding Remark

Depending on the nature of available or obtainable data sets, the approach can potentially be used for the identification of other types of indicators than plant species. For example, soil and fauna indicators can be identified through the analysis of data sets that consist respectively of Site/Sample X soil characteristics, or Site/Sample X population compositions and abundances of fauna (e.g., soil fauna, moths and other insects). In addition, a knowledge of management or treatment histories will be required to interpret and confirm the gradients obtained through the analyses.

References

Bosch, O J H, and Gauch, H R, (1991). The use of a degradation gradient for the assessment and ecological interpretation of range condition. Journal of the Grassland Society of Southern Africa 8:138-146.

Bosch, O J H, and Janse van Rensburg, F P, (1987). Ecological status of species on grazing gradients on shallow soils of the western grassland biome in South Africa. Journal of the Grassland Society of Southern Africa 4:143-146.

Gauch, H R, (1982). Multivariate Analysis in Community Ecology. Cambridge, Cambridge University Press, p289.

Hurt, CR, Hardy, M B, Tainton, N M, (1993). Identification of key grass species under grazing in the Highland Sourveld of Natal. African Journal of Range and Forage Science 10:96-102.