Following a request from the European Commission, an inventory of more than 500 current environmental organisations and almost 1 000 existing surveillance networks/programmes (ESNs) was established by an external contractor to provide an overview of existing networks and associated surveillance programmes at this point in time. The inventory is incomplete because all ESNs are not listed and because, for ESNs reported in the inventory, information was not available or not found during the survey (e.g. sampling protocols, statistical analysis methodologies and data validation). However, the inventory headings can serve as a checklist in the initial process of identifying potentially suitable ESNs prior to an in-depth analysis of their suitability for general surveillance (GS) of genetically modified plants (GMPs). It would therefore be desirable to complete, maintain and update this inventory as a resource for supporting GS.
The EFSA GMO Panel first defined the following assessment criteria to support the selection of ESNs suitable for post-market environmental monitoring (PMEM) of agricultural products (e.g. GMPs): the spatial resolution, the temporal resolution, a standard protocol for data collection, a survey carried out by professional surveyors and/or at least trained volunteers, data validation, the statistical analysis of collected data, and the availability and accessibility of collected data. When considering GS for GMPs, these criteria should be adapted on a case-by-case basis considering the geographical distribution of species/taxon relevant to the receiving environments for the GMP under consideration, the temporal resolution of an ESN depending on the biology (e.g. life cycle) and behaviour (e.g. migration) of the species/taxon relevant to the receiving environment covered by the ESNs and the type of data collection (‘continuous’ or ‘count’) in order to achieve good power to detect change. Moreover, the types of endpoints measured by the ESNs must be of relevance for GS; it is important that the selected biota occur in areas where GMPs may be cultivated.
The EFSA GMO Panel acknowledges that, in compliance with the aforementioned assessment criteria, several existing ESNs potentially suitable for GS of GMPs have been identified but considers that further analysis is needed to identify all the ESNs that could be used. In many cases, spatial resolution of ESNs does not cover agricultural landscapes where the GMPs may be cultivated. In addition, they only partly cover the protection goals identified by the EFSA GMO Panel in its 2011 Guidance Document on PMEM of GMPs in regions where GMPs might be cultivated. Further information on ESNs is still needed and direct contact with ESN organisers would be required to determine if the ESN fully meets requirements and to discuss options for access to data.
In this respect, raw data are only exceptionally available. Although problems currently exist in accessing data from ESNs, the move towards ‘open data’ policies may resolve these issues in the future. The EFSA GMO Panel is therefore of the opinion that GS of GMPs would benefit from open data policies applied by ESNs, as this would allow (re-)analysis and/or pooling of datasets collected by different ESNs, as well as the study of any interactions between datasets. Overall, the EFSA GMO Panel supports the centralisation and harmonisation of data recording according to European Union (EU) standards, such a those laid down in Directive 2007/2/EC establishing an Infrastructure for Spatial Information in the European Community (INSPIRE). However, the EFSA GMO Panel also recognises that technical support may be required by certain ESNs in order to transform their collected datasets to meet INSPIRE standards.
In addition, in the context of GS of GMPs, the monitoring sites or regions must be characterised for their level of exposure to GMPs to identify if there is a plausible link between the potential adverse effect and the cultivated GMP. This would require comparing the spatial and temporal resolution of the monitoring sites or regions with known locations of GMP cultivation. However, monitoring sites are not limited to single fields and usually cover a small agricultural area. Moreover, the uptake of GMPs may vary over time. Therefore, they cannot always be classified as either non-exposed (‘control’) or exposed (‘treated’) and would instead be characterised by the level of uptake of GMPs, which makes data analysis more complex. In such cases, an alternative approach, based on historical data to establish baselines and monitoring sites over time, may be required. The GMO registers could be the source for information for GMP cultivation. However, the availability of information on influencing factors (e.g. cropping systems) would provide added value to account for confounding factors and assess to what extent any adverse effect is associated with the GMP or with any other stressors. Ideally, the reporting of the locations should be the same for both monitoring sites and cultivation sites in terms of scale, format and projection system. Recording and reporting locations according to the INSPIRE standard for both monitoring sites and GMO registers would ensure interoperability.
The EFSA GMO Panel was also asked by the European Commission to further investigate data analyses by ESNs as well as on the sensitivity of these statistical analyses to detect change. The statistical method for data analysis is one of the assessment criteria listed herein for the selection of ESNs suitable for GS of GMPs. A decision tree is provided for selecting the optimal method for statistical analysis based on the study design, and the datasets available from ESNs in the case monitoring sites can be classified as either ‘exposed to GMPs’ or ‘non-exposed to GMPs’. A survey design with sufficient statistical power (> 70 %) is required to detect an effect for a particular indicator. A generic equation is also provided to estimate the power of a specific network to detect change considering such factors as number of sites, frequency of observations, missing data, data type and proportion of sites in areas of GM cultivation. This can be used during the case-by-case analysis to identify suitable ESNs.
For all data types, increasing the number or monitoring sites and/or the number of years of monitoring increases the power to detect an effect. Sample size is one of the main contributing factors in determining the power of any ESN to detect an effect of a product release into the environment. A different way to achieve a more powerful statistical analysis is to pool data collected by different ESNs covering the same protection goal(s). Although increasing the sample size of any ESN activity may have a positive effect on the power to detect any treatment effect, it also implies variable extra-costs depending on whether data collection is in the hands of volunteers or professionals. Moreover, combining results for different ESNs is not always appropriate, as there may be important covariates (e.g. receiving environments and/or stressors) leading to differentiated responses across geographical regions and different elements of variability from each constituent data supplier. Complex hierarchical models would be needed to fully investigate the advantages and disadvantages of combining data across ESNs. As this is an important issue to improve the efficiency of using ESNs for the purpose of GS, the EFSA GMO Panel recommends further investigation of the combination of datasets from different ESNs and conducting simulation exercises on selected case-studies.
Monitoring the environmental impacts of GMPs should be considered as a component of the environmental monitoring that is required to measure impacts of land use and management on biodiversity and the environment in the EU. In order to determine which human interventions are associated with environmental impacts, the EFSA GMO Panel recommends that all relevant environmental monitoring is fully integrated, so that data on all major agricultural and land use stressors (e.g. pesticides, cropping management practices) can be collated and analysed. Harmonisation and synchronisation of environmental monitoring would facilitate analysis and interpretation of monitoring reports and provide a strong scientific basis for supporting land use and environmental policy.