Guiding the management of an agricultural pest: Indexing abundance of California meadow voles in artichoke fields

• Voles damage California artichokes where ∼100% U.S. production occurs.
• Tested standard artichoke bract vs nontoxic bait block chewing mediums for indexing.
• Tested binary presence-absence chew measures vs continuous % chewed for index calcs.
• Bait block index correlated well with number known to be alive, bract index marginal.
• All indices with binary measures did not correlate with abundance.

Nearly 100% of U.S. artichoke production comes from California and is concentrated in Monterey County. California meadow voles are damaging rodent pests that can threaten the profitability of growing artichokes. A practical population monitoring method can be invaluable to integrated pest management programs for guiding when and where control is needed and assessing control efficacy. The standard method for indexing vole populations in artichoke fields has been based on observing chewing on artichoke bracts placed throughout the field. Because toxicants are delivered on artichoke bracts, bias for population indexing is potentially introduced. We therefore compared artichoke bracts to nontoxic grain-based wax bait blocks as an alternative chewing medium for eliciting chewing observations for indexing abundance. We also compared the use of binary (presence-absence) observations of chewing to continuous measures (percent chewed). We considered the effect of three sizes of observation grids (4 × 4, 5 × 5, 6 × 6) for indexing. We conducted intensive trapping to determine number of voles known to be alive (KTBA) at each site as a basis for assessing which of the 12 indexing approaches (2 chewing mediums, 2 measurement types, 3 grid sizes) best tracked population abundance. The percent chewed on artichoke bracts for all grid sizes only marginally correlated with KTBA (∼0.5), whereas percent chewed on bait blocks correlated very well with KTBA for all grid sizes (∼0.9). Reducing continuous data to binary observations produced indices only weakly or negatively correlated with KTBA. Available resources would probably determine whether smaller grid sizes would be used for obtaining chewing observations.