Nepal Livestock Risk Map Finds 37-District Cluster Beyond Hazard Hotspots
Nearly half of Nepal’s districts sit above a new compound-risk threshold for livestock, but the most important result is not the count. It is the mismatch between where climate hazards are most intense and where animals and livestock-dependent households face the greatest combined risk.
A peer-reviewed district assessment published on October 22 in npj Natural Hazards combines five physical hazards with four measures of social and livestock-sector vulnerability. Thirty-seven of Nepal’s 77 districts, or 48%, exceeded the study’s compound-risk threshold. The five highest-priority districts were Achham, Doti, Bajura, Bajhang and Pyuthan.
The resulting pattern matters for investment and preparedness. Central-western mountain districts recorded the strongest physical hazard exposure, while the highest combined livestock risk shifted west, where moderate exposure overlapped high poverty, deep reliance on livestock, limited veterinary access and social marginalization. A hazard-only map would therefore direct attention away from several places where losses may be hardest for households to absorb.
A 77-district view of hazards and vulnerability
The authors — Suhel K.C., Aabhash Poudel and Sikesh Manandhar — work across Nepal’s agriculture and livestock institutions. They describe the study as the country’s first nationwide, livestock-specific district risk assessment. Its hazard baseline spans 2000 to 2023.
The Multi-Hazard Exposure Index covers flood, drought, landslide, high-temperature frequency and low-temperature frequency. The Livestock Vulnerability Index adds four different questions: how dependent a district is on livestock, the incidence of poverty, access to veterinary services and the degree of social marginalization.
Those ingredients are important because they represent different parts of the loss chain. A drought or flood is an external shock. Herd density and livelihood dependence determine how much economic value is exposed. Veterinary coverage, poverty and marginalization influence whether a household can prevent disease, buy feed, move animals, rebuild shelter or recover after losses.
The underlying district-level files and annotated R workflow have also been released. They include normalized component scores, the composite indices, local spatial-cluster classifications and sensitivity checks for all 77 districts. The documented inputs draw on ERA5, CHIRPS, NASA GPM-LHASA, JRC surface-water data, Nepal’s agriculture and population censuses, livestock-service records and poverty indicators.
Why hazard intensity alone missed the worst hotspots
The physical exposure index was highest in parts of the central-western mountains. Yet the compound-risk map placed the strongest concentration in Sudurpashchim and the mid-western hills. In practical terms, the study finds that an area does not need Nepal’s most extreme hazard score to become a top livestock priority. Moderate, repeated exposure can become more consequential when animals are central to income and nutrition and the services that protect them are thin.
This changes how a limited preparedness budget might be allocated. If authorities rank districts only by rainfall anomalies, flood area, landslide frequency or temperature extremes, they can miss the places where the same shock is more likely to become a livelihood emergency. The paper’s bivariate approach makes that trade-off visible instead of collapsing exposure and vulnerability into an unexplained national average.
It also separates two policy problems that require different tools. High exposure with lower measured vulnerability calls for hazard monitoring, infrastructure protection and contingency planning. High vulnerability under moderate exposure calls for veterinary coverage, feed reserves, shelter, finance and inclusive extension. Where both are high, those measures have to work together. These are operational implications of the index, not interventions tested by the study.
The cluster is statistically strong; the loss test is modest
The compound index was not randomly scattered across Nepal. Global Moran’s I was 0.41, compared with an expected value of -0.013; the reported z-score was 4.87 and p was below 0.001. That indicates a pronounced spatial cluster, meaning neighbouring high-risk districts are more common than chance alone would suggest.
For planners, clustering can justify shared systems across district boundaries: mobile veterinary teams, vaccine and medicine cold chains, emergency fodder procurement, livestock shelters and routes for moving animals. A connected western cluster is unlikely to be managed efficiently as five isolated district problems.
The authors also compared the composite scores with historical livestock losses recorded in Nepal’s BIPAD disaster portal. The Spearman correlation was 0.239, with p=0.037. That direction is supportive, but the relationship is modest. It does not show that the index can precisely forecast losses, nor does it establish that any individual vulnerability factor caused a recorded death.
That distinction matters. Statistical significance says the observed positive association is unlikely to be a random result under the test’s assumptions. It does not say the model explains most of the variation in losses. Unrecorded incidents, differences in herd size, local disease outbreaks, reporting quality and household coping actions may all affect the loss record.
What decision-makers can use now
The immediate value is triage. Achham, Doti, Bajura, Bajhang and Pyuthan emerge as the five places where multiple indicators warrant closer examination. The 37-district threshold offers a broader screening group, while the component scores show why each district entered it.
A district with weak veterinary access needs a different response from one driven mainly by flood exposure. Publishing the components alongside the composite score makes the framework more useful than a simple league table. Ministries, provincial governments and development partners can identify whether the binding constraint is surveillance, animal-health staffing, household purchasing power, fodder security or physical exposure before committing funds.
The same design could sharpen early warning. A weather alert has more decision value when it reaches livestock keepers with advice matched to local herd systems and access constraints. In the western cluster, that could mean combining drought and temperature monitoring with pre-positioned fodder, vaccination continuity and accessible animal-health services rather than treating each as a separate programme.
For agribusinesses, the map is a signal about service gaps rather than a sales forecast. Feed suppliers, insurers, dairy buyers, veterinary distributors and digital advisory platforms would still need local demand, road access, herd-composition and affordability data before making commercial decisions.
What the assessment cannot yet answer
The index is a district-scale spatial model, not an animal-level trial and not a forecast of a specific future disaster. Its 2000–2023 exposure baseline describes historical frequency. It does not project how hazards will change under a particular emissions scenario or predict losses in the next season.
Composite results also depend on indicator selection, normalization and weighting. District averages can hide large differences between municipalities, production systems and households. A single veterinary-access measure cannot capture service quality, medicine availability or whether marginalised keepers can afford to use a clinic.
The modest correlation with recorded losses is useful as an external check, but stronger validation would require more complete, comparable livestock-loss and production data over time. The article is being shared as a peer-reviewed accepted manuscript and may receive editorial changes before the final version of record.
The next useful step is therefore not simply another national map. It is to test the ranking at municipal and household level, update the indicators as services and hazards change, and measure whether risk-weighted preparedness actually reduces mortality, productivity loss and recovery time.
Sources and method
Primary research: “Multi-hazard climate hazard exposure and livestock vulnerability in Nepal: a district-level spatial risk assessment,” npj Natural Hazards, published October 22, 2026.
Data and reproducible workflow: Government of Nepal authors’ Zenodo record, including district scores, spatial files and the annotated R analysis pipeline.
Agrilinkage reviewed the paper’s methods, reported statistics and public data documentation. The operational implications above are editorial analysis; they should not be read as results of an intervention trial.
Comments