Preventing Zoonotic Spillover at Agricultural Frontiers: A Systematic Review of Geospatial Surveillance, Predictive Risk Mapping, and Biosecurity Implementation

Background Agricultural expansion into natural habitat creates contact through which animal pathogens reach people. We asked how zoonotic spillover at agricultural frontiers can be detected, predicted and prevented before it reaches people, and whether published studies connect these steps. Methods Following PRISMA 2020, we searched Scopus and PubMed/MEDLINE (2015–2026; searched 24 September 2026) with citation searching (25 September 2026). Two independent reviewers (Reviewer 1 and Reviewer 2) screened 8,698 records and extracted data under pre-established eligibility criteria; discrepancies were resolved through independent adjudication by a third senior reviewer (Adjudicator). Synthesis was narrative, structured as a detection–prediction–prevention chain, with adapted GRADE-CERQual grading. Results Fifty-nine studies were included; 31 addressed detection, 28 prediction and 14 prevention, and 45 only one of the three. Detection relied on land-cover products (29 of 31); none reported routine surveillance use. Ten predictive models were independently validated, only 3 in new places or years, and 8 reported no validation. Twenty-four of 28 predictive studies stated a prevention use (14 naming a specific action), yet none evaluated the outcome of an intervention allocated by a risk map (0 of 59). Prevention studies, 8 of 14 in high-income countries, measured outcomes no further along the spillover pathway than livestock infection. Confidence was moderate for the gap in map uptake and low or very low otherwise. Conclusions Across peer-reviewed studies of agricultural frontiers, detection, prediction and prevention are rarely connected; the gap is widest between a risk map and an evaluated intervention. Transfer-tested validation and prospective evaluation of map-targeted interventions are needed before risk maps guide prevention.