Module 2 of BioMAT provides guidelines for the analysis of data and assessment of trends of biodiversity from existing monitoring schemes. Users are guided to recommendations by navigation through a decision tree. The decision tree covers the main strategic choices that users need to consider in the analysis of monitoring data. The recommendations consist of key messages regarding suitable approaches and will identify key reference and support material or software tools that can facilitate the analysis of monitoring data. Both, the analysis of data from single schemes and the integration of information from different schemes are covered.
Module 2 of BioMAT provides background information for the statistical analysis of monitoring data. Recommended methods are explained and, if available, links to free software for data analyses are provided. It covers both species and habitat monitoring but currently does not cover monitoring of genetic diversity or ecosystem properties. However, it does cover, albeit to a limited extent, changes in species community composition. For a complete list of topics currently available see Table of Content of Module 2.
Knowing whether there is a trend of a certain size is important first information derived from monitoring schemes. When making inferences from statistical tests about trends, we can make two errors: a) we may falsely conclude that there is a trend, when in fact there is none. This error is called type I error. Most of the biological literature focuses on this type of error. For applied biodiversity monitoring, however, it may be more critical to falsely conclude that a population is stable, when in fact it shows a trend, especially a decline. This error is called type II error and is related to the power of an analysis. Module 2 also covers recommendations regarding power analysis.
Detectability is an important issue in biodiversity monitoring that too often is ignored, as can be easily seen when conducting a search for monitoring schemes that account for detectability using module 1 of BioMAT. Without accounting for detectability, we also may make incorrect inferences from our data regarding the presence of a trend. Read more at What's the point about detectability - what is it and why is it important?.
In long-term monitoring schemes across geographically large areas it is often difficult to sustain monitoring continuously. Thus, missing data are not uncommon. BioMAT provides recommendations what to do if data points are missing in a time series.
Detecting a trend is a first important step for applied biodiversity conservation. However, to react in an effective way, we need to diagnose the cause(s) of a change in trend. A prerequisite for causal inferences is to monitor also factors that may contribute to a change in trend and, whenever feasible, one should use for this purpose an experimental design. While this is primarily an issue addressed by module 3 of BioMAT, module 2 provides some basic recommendations regarding the potential and limits of causal inferences from monitoring data.
Few monitoring schemes cover a large spatial area and a broad range of biodiversity (taxonomic groups/habitat types). Combining data or information derived from several schemes may be a powerful way to expand the generality and applicability of the conclusions possible from monitoring schemes. Scientists, managers, and policy makers may greatly benefit from such broader generality. Therefore, the recommendations of module 2 focus especially on approaches of combining several schemes in the analysis and assessment of monitoring data. Read more about the advantages of integrating different monitoring schemes and principle approaches available for such an integration (combination of raw data, estimates and trend information).
The integration of different monitoring schemes to evaluate changes in conservation status as required by the Habitats Directive is currently not directly covered as the rules for assessing favorable conservation status across schemes are partially political and insufficient experience exists on the effects of different alternative rules. However, the basic principles for integration of monitoring data also can be used as a guideline for future improvements in assessing the conservation status of species (and habitats) across schemes.
Types of data covered
For species the following data types are covered:
- demographics (usually using capture-mark-recapture - CMR - methods)
- species assemblages, species richness, diversity
- causes of change
For habitats both data from remote sensing and from field surveys are considered. The following data types are covered:
- abundance (number of patches; area)
- habitat type distribution
- patch size distribution
- patch spatial structure
- habitat quality
- causes of change
Usual monitoring data are made of three elements: the measure, the site, the year. The measure characterises the state of the entity considered (e.g., actual abundance of a given species) at a particular site in a particular year. When the monitoring goals include inferences about the causes of change, the measure must also comprise potential drivers of change. These measures are systematically repeated in time. Most analytical methods rely on the assumption that this is indeed the case. Indeed, ideal monitoring data should be made of four components: the measure, the site, the year, but also uncertainty. That is, the sampling design should allow the estimation of measurement error. The validity of these assumptions depend on the field methods, which are not considered here. Afterwards, data analysis can hardly compensate for poor field methods.
The most commonly available data in applied biodiversity conservation, grid-based or dot-based distribution maps identify areas or locations where a species or a particular habitat type has been recorded. They usually lack systematic repeated surveys, the area covered may change through time, and it often is unclear whether a species/habitat has been searched for at all within a particular grid. Analyses of status and trends from such mapping data are complicated and limited and were not covered by EuMon. Therefore, they are only briefly touched in the current version of BioMAT but we plan to address these types of data in-depth within a new project (www.scales-project.net).
- Elzinga CL, Salzer DW, Willoughby JW, Gibbs JP (2001) Monitoring plant and animal populations: a handbook for field biologists. Blackwell Science, Malden, MA
- Lengyel, S., A. Kobler, L. Kutnar, E. Framstad, P.-Y. Henry, V. Babij, B. Gruber, D. Schmeller & K. Henle (2008): A review and a framework for the integration of biodiversity monitoring at the habitat level. Biodivers. Conserv. 17: 3341-3356.
- Henry, P.-Y., S. Lengyel, P. Nowicki, R. Julliard, J. Clobert, T. Celik, B. Gruber, D.S. Schmeller, V. Babij & K. Henle (2008): Integrating ongoing biodiversity monitoring: potential benefits and methods. Biodivers. Conserv. 17: 3357-3382.
- Pollock, K.H., Nichols, J.D., Simons, T.R., Farnsworth, G.L., Bailey, L.L. & Sauer, J.R. 2002. Large scale wildlife monitoring studies: statistical methods for design and analysis. Environmetrics 13: 105-119.
- Thomas, L. 1996. Monitoring long-term population change: why are there so many analysis methods ? Ecology 77: 49-58.
EuMon core team; April 2009