Table 3 Variables assessed and criteria used for critical appraisal of included studies

Study design

Long (>3 months) study period, multiple seasons

Short (<3 months) study period, winter measurement

Long-term intervention maintained for multiple years

Intervention in place for <2 years

Replication at level of intervention, large sample size (>3), some degree of randomisation in sample selection

Pseudoreplicated, low sample size (<3), no randomisation

Control and treatment/exposure samples well-matched (i.e. close in proximity but low chance of spillover effects)

No evidence of matching, potentially influential differences between treatment and control

General study design very clear and repeatable

Some missing information

Specific methodology

Full description of methodology, accounting for wind direction/speed

Some missing methodological detail, no accounting for wind direction/speed

Full description of methodology, measurement disturbance mitigation measures

Some missing methodological detail, no mitigation for measurement disturbance

Full description of methodology, representative sampling

Some missing methodological detail, sampling unlikely to be representative of variability in environment

Full description of methodology, representative sampling

Some missing methodological detail, sampling unlikely to be representative of variability in environment

Bias

Measurement bias unlikely or evidently not present

Bias likely as a result of methodology

No obvious confounders stated or evident, or stated but adequately accounted for

Confounders stated and unaccounted for or likely to be present

Unclear classification given to any study where substantial details within the methods are either unclear of missing.