Agrobiodiversity underpins sustainable, nutrition-sensitive food systems, yet its taxonomic and functional dimensions remain poorly integrated in spatial planning. We introduce two composite indicators—the Agrobiodiversity Index Score (ABDIS) and the Agrobiodiversity Potential Score (ABDPS)—to assess realized and latent crop diversity across 906 districts in South Asia. These indices integrate species diversity with dietary and nutritional functional diversity and are analyzed using spatial statistics. Results reveal a systematic taxonomic-functional mismatch: moderate to high species diversity (median SD = 0.56) often fails to translate into dietary (DFD = 0.44) and nutritional (NFD = 0.49) diversity, with only 4% of districts serving as simultaneous SD–DFD hotspots, while 24% of SD districts align with high ABDPS, indicating substantial unrealized functional potential. Functional outcomes are more strongly associated with crop composition than total diversity, with cereal-dominated systems (74% of production) showing constrained nutritional potential. Notably, 44% of districts with very high stunting overlap with ABDPS hotspots, though this overlap is descriptive. Our framework provides a scalable spatial diagnostic to inform the geographic prioritization of nutrition-sensitive agricultural strategies, while recognizing its production-side focus and lack of causal inference.