Aquatic freshwater mesocosms are higher tier studies designed to increase the realism of ecotoxicology testing by determining the effects of a chemical on a community, in a system replicating the natural environment in an edge-of-field waterbody. These complex studies can be difficult to interpret, and without reliable data it is almost impossible to determine the true effects of the chemical on the aquatic community.
Historically, mesocosm studies have been criticised for having low statistical power and high variability as their labour intensive nature may limit the size of the study. However, there are other ways of increasing the reliability of data to derive robust endpoints without having to increase replication.
Brock et al (2015)  provide further recommendations to the EFSA guidance (2013)  for the improvement of the statistical power of mesocosm studies by improving sampling efficiency and intensity. Sampling techniques that increase the number of individuals and reduce variability can reduce the Minimum Detectable Difference (MDD) values. MDD’s are useful in the interpretation of mesocosm data, providing a measure of the differences between the means of the treatment and control data that can be detected as statistically significant. Using historical mesocosm data, we have demonstrated how improvements to mesocosm study designs, based on the recommendations by the EFSA guidance and Brock et al., affect the MDD values, and thus the reliability of the resulting endpoints.
In the remainder of this article, we present a series of case studies that highlight the improvements we’ve made to our mesocosm study design. We recently presented this work at the SETAC conference in Helsinki (May 2019), with the presentation available for download here.
Case study 1: Multiple plant health and growth measures
A major design feature at CEA is the use of sloped mesocosms to assess the effects of a herbicide on the macrophyte community, allowing species to be planted together along the gradient. This increases the realism of the test system, and uniquely allows for inter-species competition and other secondary effects to be accurately monitored to determine the most sensitive endpoints.
We have also developed methods to assess macrophyte health and growth by recording chlorosis, necrosis, plant height and the number of stems at regular sampling occasions for the duration of the study. By increasing the number of parameters measured for each plant, more relevant and robust data can be obtained to accurately monitor the effect of the chemical.
Case study 2: Increased counting of phytoplankton samples
Increasing sampling intensity reduces variability and increases the statistical power, and it is argued in Brock et al (2015) that this has a greater benefit to improving MDD values than increasing the replication.
To investigate this, the control data from two mesocosm studies were compared where 30 or 50 whipple grid views had been enumerated for phytoplankton cells under the microscope.
In the 50 whipple grid view samples, a larger proportion of the sample was assessed, a higher number of taxa were identified and there were more reliable endpoints with good %MDD values. Although this is slightly more labour intensive, it is still less time-consuming than increasing replication and the data collected is more representative and less variable.
Case study 3: emergent insect traps
It is recommended in Brock at al. (2015) to use two emergent insect traps per mesocosm rather than just one, in order to increase numbers of sampled organisms for enumeration and to reduce variability.
To investigate this, the control data from two mesocosm studies were compared where one study had used two traps per mesocosm and the other just one. Although a higher number of different taxa were identified when using two traps compared to just one, in our facility the number of reliable endpoints with acceptable %MDD values was the same for both study designs. It is thought that this is because samples are dominated by a number of key taxa, namely Chironomid sp. and Chaoborus sp..
In these studies, the increased sampling effort did not yield more reliable results. Another option is optimise the trap design to increase the area for capturing emergent insects, which has been demonstrated to improve %MDD values by Brock at al. (2015). We are planning to investigate this within our own facility.
Case Study 4: Increasing Efficiency of macroinvertebrate mesocosms
Our mesocosm design has evolved over the years in accordance with the EFSA guidance. Optimisation techniques developed for macroinvertebrates include: an extended pre-application phase with monitoring in place to seed and mix mesocosm water in order to establish similar communities between replicates, using different types of traps and techniques to capture macroinvertebrates and increasing the time taken for identification and enumeration.
To investigate the impact of these optimisation techniques on the reliability of the results, we compared macroinvertebrate control abundance data from an older study design (pre-optimisation) with a more recent study design (post-optimisation). There was a higher number of identified taxa in the more recent study and a higher number of reliable endpoints with good %MDD values, indicating that the multiple improvements implemented had a positive impact on organism abundance and the reliability of data.
These case studies demonstrate that mesocosm study designs can have a significant impact on the statistical power and reliability of the resulting endpoints. In our opinion it is also important that mesocosm design is flexible in order to meet the needs of specific regulatory issues by focussing sampling effort in a targeted and informed manner depending on the type of effects expected and the critical taxa. Mesocosm design should continue to evolve and develop in order to derive more reliable and robust population and community level endpoints. The findings from these case studies can be used to inform the experimental design of future mesocosm studies such that the results meet the EFSA requirements for 8 sensitive/vulnerable taxa with acceptable %MDD values, in order to maximise regulatory acceptability.
For further information about our mesocosm design, please contact Marie Brown (Marie.Brown@cea-res.co.uk).
You can find the other platform presentation and 6 posters that CEA presented at the SETAC conference in Helsinki (May 2019) on our website here.
 Brock, T. C. M., Hammers-Wirtz, M., Hommen, U., Preuss, T. G., Ratte, H. T., Roessink, I., and Van den Brink, P. J. (2015). The minimum detectable difference (MDD) and the interpretation of treatment-related effects of pesticides in experimental ecosystems. Environmental Science and Pollution Research, 22(2), 1160-1174.
 EFSA (2013). Guidance on tiered risk assessment for plant protection products for aquatic organisms in edge of field water bodies. EFSA Panel on Plant Protection Products and their Residues (PPR).