Journal of Experimental Agriculture International

Background: Natural-farming assessment is complicated by heterogeneous practices and by the frequent conflation of implementation, ecological proxies and livelihood outcomes.

Objective: To identify empirically used indicators and develop a traceable candidate framework without statistical pooling.

Methods: An unregistered mixed-studies review used ranked live web searches, including AGRIS- and PubMed-restricted web queries, and openly retrievable primary reports. Eligible first publications were dated 1 January 2000-15 July 2026; searching occurred on 15 July 2026. One reviewer screened, extracted and appraised studies using design-specific MMAT domains and additional economic-accounting checks. Findings were synthesised by domain; proposed indicators were separated from observed evidence.

Results: Of 102 records, 26 were duplicates and 47 of 76 unique reports were excluded at initial screening. Twenty-nine full reports were sought, eight were not retrieved and two of 21 assessed reports were excluded. Nineteen primary reports represented 18 studies: 17 in India and one in Japan. Yield responses were context dependent; soil biological improvements did not consistently establish crop nutrient sufficiency. Profitability depended on labour, own inputs, subsidies and market assumptions. Social evidence mainly identified constraints rather than causal wellbeing effects. Twenty-four candidate indicators were organised into six domains, without weights or validated thresholds.

Conclusions: A multidomain dashboard is more defensible than a single natural-farming score. The proposed framework requires participatory and field validation. Bounded web retrieval, inaccessible reports, single review and primary-study limitations constrain confidence; climate, clinical-health and long-term resilience benefits were not established.