Available (semi)field methods

AppendixK gives an overview of (semi)field studies evaluated by Brown et al. (2009) and Sch€affer et al. (2010) with some additions from literature, but has no claim of being exhaustive.

In current regulatoryfield-test protocols, tests are performed in-field on replicated plots and study effects on ‘natural’ in-soil populations present at the time of the experiment. In other field-study methods, specimens were added to the system. In some cases, test systems are chosen on an uncontaminated soil (e.g. grassland), and the test substance is applied to the test soil or system. In the current procedure, laboratory tests are performed with arthropods, non-arthropods and microorganisms. Some field-study methods (e.g. the earthworm field study) are aimed at refining the particular first-tier test. In Section4, it is proposed to study the natural assamblages of the soil community. It is clear from Appendix K that most field studies are aimed at effects on natural assemblages of the soil community.

Semifield studies are defined as controlled, reproducible systems that attempt to simulate the processes of and interactions between components in a portion of the terrestrial environment, either in the laboratory (small scale) or in the field, or somewhere in between. Single-species field studies with earthworms are submitted for the peer-review process in a standardised way.

The mobility of the test species in semifield studies is artificially limited, using e.g. enclosures or TMEs. TMEs can be conducted in the field, but it is also possible to move the TMEs to the laboratory and to study effects under controlled conditions.

Sch€affer et al. (2010) classify the different types of semifield studies according to a number of ecological and performance criteria (Table 38). The evaluation of the criteria was based on expert knowledge. They distinguish three main types of semifield studies: the assembled system, the terrestrial models ecosystem, andfield enclosures.

Table 37: Main features of Terrestrial model ecosystems (TMEs) (Sch€affer et al., 2010) Guidelines ASTM (1993), UBA (1994), USEPA (1996), Knacker et al. (2004)

Principles Interaction of soil properties and the natural community of microorganisms, animals, plants

Species Natural soil-organism community

Substrate Undisturbed soils fromfield sites

Duration Usually about 16 weeks

Parameter Wide variety of fate and effect endpoints

Experience Fungicides, contaminatedfield soil, or pharmaceuticals in dung

Table 38: Rough classification of three semifield methods according to eight ecological and performance criteria. The grey shading indicates whether the criteria are fulfilled or not (from Sch€affer et al., 2010)

Assembled System TME (TerrestrialModel

Ecosystem) Field enclosures

Ecological criteria Relevance Artificial food chain, but no real competitors/prey-predators

Natural community Addition of organisms possible

Endpoints No community measures All‘known’ parameters can

be measured

All‘known’ parameters can be measured Flexibility No crop simulation possible Most soils, except very

sandy/very dense soils

All soils Sensitivity

Practicability Performance

criteria

Reproducibility/repeatibility Exact results not reproducible

because of natural plasticity

Several studies, but few publications

Experience If similar approaches

are combined

EU ring test

Standardisation None ASTM guideline and UBA

draft available

IOBC guideline available

High/good/many Medium/fair/numerous Low/poor/few

As a result of the classification which was agreed at the workshop in 2007, Sch€affer et al. (2008) conclude that TMEs are at promising method for assessing effects of pesticides.

Below the different types of studies are discussed in more detail.

Assembled systems

In assembled systems defaunated (sieved) soil is used, to which lab cultured organisms or parts of the natural community organisms are added, so that e.g. food chain effects can be studied. Their role is limited because inter alia it is hard to predict what happens when an additional species is added to the system. Studies with assembled systems can be very useful to answer specific scientific questions.

Their role for risk assessment, however, could be in the intermediate tier B.

In the intermediate tier B of the proposed risk assessment flowchart (see Section7.8 and Figure13), an option is given to the applicant to provide (or to the risk assessor to require) additional information on effects to particular soil-organism groups of interest, gaining relevance by assessing mainly indirect effects of the substance of concern towards these groups.

The proposed test for this tier is a microcosm set up using natural, defaunated soil to which fragments of the natural community of the group of in-soil organisms of interest is added (e.g.

microarthropods, nematodes, mesofauna including microarthropods and enchytraeids). These systems resemble the existing gnotobiotic tests (Sch€affer et al., 2010), but have a major difference; instead of assembling the tested community with a limited number of fully known species, usually those existing in laboratory cultures, these microcosms contain a true fraction of the natural community collected in the field, thus increasing the number of species in the system and therefore the possible interactions between them (e.g. either via competition or predation).

In brief, samples are collected in the field, the natural community is extracted (by adopting standard methods for each organism group – see R€ombke et al. (2006) for the sampling and extraction methods for different organism groups) and then it is added to the spiked natural soil on the microcosms. Depending on the group of interest, different methods can be adopted to decrease the variation on the number and composition of the community in each replicate microcosm.

Although the experiment is set up to focus on a specific group of in-soil organisms (e.g.

microarthropods), species from other groups could be added to the system to increase the realism (e.g. nematodes or enchytraeids). Microorganisms are always included by re-inoculating the soil with a microbial suspension extracted from the same natural soil.

This means that the microcosm setup can be tailored according to data needs and to the type of substance. For instance, if focusing on an insecticide, the group of interest could be the microarthopods. This means that, although other species from other groups (e.g. enchytraeids) could be added to the microcosm to increase realism, only the microarthropods could be assessed at the end. Organisms can be identified at species or at life-form group level, or using any other trait-based typology of interest.

Besides having the advantage of working with a relevant fraction (in terms of number of species) of the natural community of interest, assessing not only direct effects to several species but also indirect effects via interactions among them, these systems have the advantage to ally the reproducibility of laboratory single species tests (mainly regarding the possible number of replicates) with a gain in ecological realism.

There is limited experience of assessing effects of chemicals, especially PPPs, using this approach.

Effects of carbofuran have been assessed on nematode and microarthropod communities from temperate and sub-tropical soils by Chelinho et al. (2011, 2014). Data on nematodes showed not only the decrease in abundance and richness, but a shift in community composition of these organisms (although no shifts were detected on feeding groups). Data on microarthropods clearly showed that the direct effect on Collembola (an overall reduction in abundance and species richness together with shifts in community composition favouring epigeic over euedaphic species) clearly affected the abundance of competitor-mite species (an increase in oribatids was observed) and originated a decrease in the abundance of predatory mites.

While not aiming to replace a TME or a field study, where a higher level of ecological realism can be obtained, these test setups could help to obtain more realistic information on both direct and indirect effects on specific communities of interest with much less workload and a higher level of statistical power. Nevertheless, due to the limited experience available, further research is needed to refine some methodological aspects, namely on the size of the test vessels according to the community of interest, number of replicates, and the duration of the experiment to accommodate the variation in the life cycles between species.

Terrestrial Model Ecosystem (TME)

The experience with TMEs is increasing. Protocols are available from ASTM (1993), UBA (1994) and an updated protocol was presented at the PERAS workshop (see Section 2.2), see Table39. Knacker et al. (2004) describe a number of TMEs exposed to carbendazim in different regions, comparing the results from the TMEs with those in the field. In his thesis, Scholz-Starke et al. (2013), describes the results of TME studies with lindane. The pros and cons of four types of TME (defined by Knacker et al.

(2004)) are discussed: closed, homogenous TMEs; closed, intact TMEs; open, homogenous TMEs; and open, intact TMEs. TMEs are suited to studying effects on a number of invertebrate species, including some earthworm species. It is also possible to include effects on microorganisms, including functional endpoints, by e.g. including litterbags or bait lamina tests in the TME. Different aspects of TMEs correlated with their use as higher tier studies related to the risk assessment flowchart presented in Section 7.8, see also Figure 13, are discussed below in Section 9.7.2.

Field enclosures

Sch€affer et al., 2010 define field enclosures as systems with undisturbed soil, where migration of species is prevented by barriers. Enclosure studies can focus on natural occurring communities, but also studies with added organisms where found. However, according to Sch€affer et al. (2010), the added organism in practice are non-target arthropods living on the soil surface.

Field studies

Field studies without enclosures have been carried out with specific groups of organisms.

Earthworm field studies are being carried out and used in the risk assessment since many years according to guidelines by BBA (1994), ISO (1999, 2014) and Kula et al. (2006). Field studies with in-soil microarthropods are being carried out since many years with a design based on the paper of R€ombke et al. (2009).

Efforts are being made to address potential drawbacks of thosefield test (analysis of the statistical power of the field test, external recovery, dose–response design, etc.). For the ISO guideline for earthworm field studies, a revision is currently under way (SETAC GSAG/OECD expert group). Work on the standardisation of the in-soil microarthropod field study is also under way. Standardisation needs and options are also under discussion (Pieper et al., 2016).

Litter bag study

In the SANCO Guidance (European Commission, 2002), the litter bag method is always recommended in case of substances with DT₉₀ higher than 365 days. The test is conditional for substances with a DT₉₀ between 100 and 365 days and/or high risk is identified at lower tiers on soil fauna (earthworms, collembolans, mites) and microorganisms.

The method was considered an appropriate higher tier study at the time of the SANCO Guidance development since a wide range of in-soil organisms is involved in organic matter degradation.

According to the test design, litterbags containing dried organic material are buried in the soil of an arable field site which is treated with the test substance according to the representative uses reported in the good agricultural practice (GAP) table. The litter bags are sampled by removing from the soil after certain time periods using at least three sampling dates for a total duration of the test of minimum 6 months. With regard to the exposure, the annual rate (with crop interception) is applied on top of the plateau concentration, before the litter bags are buried. The mass loss of the organic material in the control and treatment groups of litter bags are determined for each sampling date as relevant endpoints. In addition, the breakdown (mass loss) rate between each individual sampling date Table 39: Main features of Terrestrial model ecosystems (TMEs) (Sch€affer et al., 2010)

Guidelines ASTM (1993), UBA (1994), USEPA (1996), Knacker et al. (2004)

Principles Interaction of soil properties and the natural community of microorganisms, animals, plants

Species Natural soil-organism community

Substrate Undisturbed soils fromfield sites

Duration Usually about 16 weeks

Parameter Wide variety of fate and effect endpoints

Experience Fungicides, contaminatedfield soil or pharmaceuticals in dung

and between the start of the study and the last sampling date should be reported for the control and the treatment. The test is considered valid if at least 60% mass loss has occurred in the control plots at the end of the study (European Commission, 2002; OECD, 2006). In the risk assessment scheme, risk for soil organisms is considered acceptable if no significant effects on organic matter decomposition are detected at study end.

However, although recommended in the SANCO Guidance, the usefulness of this study design can be questioned. In the risk assessment proposed by the SANCO Guidance, a litter bag test is triggered either by a high risk identified in one of the single species test (earthworms, collembolans or mites) or by an effect on nitrate formation of more than 25% as proxy for the activity of soil microorganisms compared to control. In the case of both soil fauna and soil microorganisms, the litterbag test method is not considered appropriate as an higher tier approach for several reasons. Firstly, the link between the outcome of this test and the SPGs as defined in 6 is confounded by the fact that organic matter decomposition is performed jointly by microorganisms and soil fauna with different activity shares during the breakdown processes. Secondly, being an integrated measurement of activity litter biodegradation can be observed even if some species or functional group has been lost or their abundance has been highly reduced. Therefore, this test is not considered appropriate to refine risk to populations of soil organisms at the higher tier.

9.7.2. Addressing specific protection goal in (semi)field studies