Optimising survey effort to monitor environmental variables: A case study using New Zealand kiwifruit orchards

• Published in: Journal of environmental management Vol. 183; pp. 13 - 21
• Main Authors: MacLeod, Catriona J., Green, Peter, Tompkins, Daniel M., Benge, Jayson, Moller, Henrik
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01-12-2016; Academic Press Ltd
• Subjects: Actinidia; Agricultural ecosystems; Agriculture; Animals; Berries; Birds; Change detection; Composition; Cost analysis; Decision making; Design analysis; Design optimization; Ecosystem; Ecosystem studies; Ecosystems; Environmental changes; Environmental management; Environmental monitoring; Environmental Monitoring - methods; Environmental Monitoring - statistics & numerical data; Farm management; Fruit; Indicators; Indigenous animals; Indigenous species; Kiwifruit; Nectar; New Zealand; Pests; Plant ecology; Polls & surveys; Power analyses; Risk analysis; Risk reduction; Sample size; Sampling frequency; Software packages; Soil sciences; Surveys; Sustainability assessment; Sustainable agriculture; Trends; New Zealand; PSE, New Zealand; Power analyses; Sample size; Indicators; Sustainability assessment; Sampling frequency
• ISSN: 0301-4797; 1095-8630
• DOI: 10.1016/j.jenvman.2016.08.033

Environmental monitoring is increasingly used to assess spatial and temporal trends in agricultural sustainability, and test the effectiveness of farm management policies. However, detecting changes in environmental variables is often technically and logistically challenging. To demonstrate how survey effort for environmental monitoring can be optimised, we applied the new statistical power analysis R package simr to pilot survey data. Specifically, we identified the amount of survey effort required to have an 80% chance of detecting specified trends (−1 to −4% pa) in 13 environmental variables on New Zealand kiwifruit orchards within an 11-year period. The variables assessed were related to soil status, agricultural pests (birds), or ecosystem composition (birds). Analyses were conducted on average values (for each orchard and year combination) to provide a consistent scale for comparison among variables. Survey frequency varied from annual (11 surveys) to every 5 years (3 surveys). Survey size was set at either 30, 60, 150 or 300 orchards. In broad terms, we show the power to detect a specified range of trends over an 11-year period in this sector is much higher for ‘soil status’ than for ‘agricultural pest’ or ‘ecosystem composition’. Changes in one subset of native bird species (nectar-feeders) requiring a particularly high level of relative survey effort to detect with confidence. Monitoring soil status can thus be smaller and less frequent than those which also want to detect changes in agricultural pests or ecosystem composition (with the latter requiring the most effort) but will depend on the magnitude of changes that is meaningful to detect. This assessment thus allows kiwifruit industry in New Zealand to optimise survey design to the desired information, and provides a template for other industries to do likewise. Power analyses are now more accessible through the provision of the simr package, so deploying and integrating them into design and decision-making should be routine to reduce the risk of inefficiencies and opportunity costs.
•New statistical power analysis R package simr informs environmental survey design.•Identifies design with an 80% chance to detect specified trends within 11 years.•Soils require less frequent surveys and fewer kiwifruit orchards than birds.•Optimal design depends on what magnitude of changes and timeframes are meaningful.•Integrate power analyses into decision-making to reduce risk and opportunity costs.