Percent of studies that measured interventions as “Effective,” “Ineffective,” or “Counter-productive” in reducing livestock loss to large carnivores, as measured by four independent reviews in 2016–2018.

The sample sizes inside disks represent the number of studies or tests, as some studies reported more than one test of the same or different interventions. Darker colors represent reviews that included experimental or quasiexperimental controls; lighter colors represent reviews that also included comparative or correlative studies (see S1 Table for details). “Deterrents” include nonlethal interventions such as audio or visual deterrents, fladry, and livestock protection collars. “Enclosure/barrier” includes electrified and nonelectrified fencing and corralling. “Guarding” includes human shepherding and livestock guardian animals. “Lethal removal” includes hunting, poison baiting, and other lethal methods. “Non-lethal removal” refers to translocation of carnivores. “Other” includes carnivore sterilization and diversionary feeding. Eklund and colleagues measured effectiveness using RR and classified Effective as RR 1.10. Treves and colleagues measured effectiveness as significant change in livestock loss. Note that Treves and colleagues initially contained 12 studies with 14 separate tests using gold or silver standards, but one test was subsequently removed after review of the methods found it impossible to draw strong inference [17]. van Eeden and colleagues measured effectiveness as Hedges’ d and classified Effective as d d d > 0.05. Miller and colleagues measured effectiveness as percentage change in livestock loss (or carnivore behavior change) and classified Effective as d > 0% change, Ineffective = 0%, and Counterproductive < 0%. RR, relative risk.