← All insights

Evidence standards in a sector without a research tradition

The Alliance for Responsible Aquaculture asked us to lead India operations across Andhra Pradesh. The work covered water quality and stocking density practices with fish farmers, and engagement with the Animal Welfare Board of India and the Bureau of Indian Standards. We knew when we accepted that aquaculture welfare was a new policy area. We did not fully appreciate how new.

Andhra Pradesh produces roughly 70% of India's farmed fish and shrimp and is one of the world's largest aquaculture regions. The sector employs several million people, most of them in the Krishna and Godavari delta districts. The measurement systems that welfare policy depends on barely exist in it.

In human welfare research, anyone setting a standard for a school, a health facility or a social protection programme can draw on prior studies, established indicator sets, regulatory precedent and peer-reviewed literature. Methodological arguments are possible because the people making them share a body of evidence. Indian aquaculture has little of this. The Animal Welfare Board of India has jurisdiction, but its regulations were written for terrestrial animals and have been extended to aquatic species in ways the scientific literature does not fully support. BIS has published some standards. Their empirical basis varies a good deal, and few farmers know about them.

In practice, evidentiary questions that are routine in human welfare work have no established answer here. The right stocking density for rohu in a grow-out pond under Andhra conditions depends on dissolved oxygen management, feeding regime, pond age and season, and no single standard covers all of them. Crowding stress in fish can be measured through cortisol proxies in the laboratory literature, but there are no farm-level measurement protocols. The line between a welfare-improving intervention and normal good husbandry matters for policy claims, and no definition of it is agreed in an Indian regulatory context.

Before any study, our first contribution was a measurement vocabulary. It set out operational definitions for the welfare indicators we intended to track, written clearly enough that farmers, enumerators and policy audiences could argue about them. The argument that followed, over whether our definition of "chronic stress" in a stocking density protocol was correct, was more useful than any amount of data collected under an undefined standard would have been.

Some patterns from the field visits were consistent enough to record. Farmers stocking at higher densities were often aware of the risks and stocked densely because the economics of their lending arrangements required it. A farmer carrying input loans at 24-36% annual interest over a four-month grow-out cycle weighs risk very differently from a welfare policy document written for a general audience. Information transfer is a small part of improving welfare in this sector. Most of the work lies in changing the incentives under which farmers decide how densely to stock.

Engagement with AWBI and BIS was more productive than we expected. Both bodies were interested in aquaculture welfare standards and aware that their existing rules were inadequate, and neither was hostile. Progress was limited by technical capacity and by the lack of an evidence base to draw on, and those problems are easier to solve than hostility would have been.

For research practice, this means a sector without a research tradition needs a different entry point from one that has it. The first contribution is definitional and the second descriptive. Causal work comes third and depends on both. Causal findings produced in a measurement-poor environment cannot be acted on, because there is no agreed standard to evaluate them against.

FAO aquaculture production statistics and MPEDA's India aquaculture data provide sector-level context. Welfare science grounding draws on Braithwaite (2010), Do Fish Feel Pain? and subsequent peer-reviewed literature on fish sentience and stress indicators.