1. Biological control (biocontrol) of invasive alien plants is a widely utilised weed management tool. Prospective biocontrol agents are typically assessed through host-specificity testing and pre-release efficacy studies performed in quarantine. However, rearing of the potential biocontrol agents and/or test plants is often difficult or impossible under quarantine conditions. Moreover, practitioners may attain laboratory-artefacts in quarantine, which may result in the potential agent being needlessly rejected. Field-based studies in the weed’s indigenous distribution could overcome these issues.

2. Sporobolus pyramidalis and Sporobolus natalensis (giant rat’s tail grass; Poaceae) are indigenous in Africa but have become problematic invasive alien plants in Australia. A previous biocontrol programme was terminated because the candidate agent could not be reared and tested in quarantine. We performed field-based host-specificity and efficacy studies for prospective biocontrol agents in South Africa (indigenous distribution). Forty-seven non-target grass species were sampled during host-specificity assessments. Candidate agent efficacy was estimated based on damage to the target weeds, for each host-specific candidate individually and in combination with other host-specific candidates.

3. Three species of endophagous wasps were deemed host-specific. Efficacy assessments identified an undescribed stem-boring wasp (Tetramesa sp.) species as the most damaging candidate. A second Tetramesa species was much less damaging alone but had a cumulative impact on the plant in combination with the more damaging Tetramesa species. Both Tetramesa species are recommended for importation into quarantine in Australia for confirmatory host-specificity testing with a significantly reduced test plant list.

4. Synthesis and applications: Similar field-based assessments in the indigenous distribution of weeds targeted for biocontrol could be included in future programmes. Where rearing of potential agents and/or test plants is difficult or impossible under quarantine conditions, our field-based method provides an alternative. Where quarantine-based testing is feasible, this method ensures that only candidates that have passed an ecologically realistic host-specificity and potential efficacy screening are imported into quarantine. This may reduce the number of agents that are imported and the length of time each agent is kept in quarantine. This is advantageous because quarantine space is highly valuable and is usually a limiting factor in pre-release assessments of biocontrol agents.