The return
A reef, a river mouth and a country, revisited through the modern record.
The thesis argued that environmental management fails when it separates ecology from people, and that the fix is participatory, integrated management. These three places have each now had thirty years to answer it. The answer is the same in all three, and it is not the one I was hoping for.
In Fiji, in Belize and in Latvia, most of the participatory architecture the thesis called for was actually built — in law, in institutions, in village-level management, in one case in a constitutional court. And in all three places the ecological indicators kept moving the wrong way.
It would be easy to write these pages as vindication, and easy to write them as disillusionment. Neither is honest. What the record actually shows is a split between process and outcome, and the split is consistent enough across three unrelated places that it looks like a property of the argument rather than an accident of any one case.
Fiji built it first and most thoroughly. The first locally-managed marine area was established at Ucunivanua in 1997, led by researchers at the university that had hosted my fieldwork; the FLMMA network formed in 2001; the Environment Management Act 2005 made a Resource Owners Committee and a public hearing after every environmental impact assessment mandatory. Then the best-designed national evaluation, using 146 matched villages, found that engagement improved participation, knowledge, management and funding — and had no effect on the perceived ecological health of a village’s fishing grounds (O’Garra et al. 2023). The Fiji page sets it out with its sources.
Belize turned a proposal into a reserve. The area I surveyed in 1994 as a baseline — the Port of Honduras — was declared a marine reserve in 2000, co-managed by the Fisheries Department together with a local organisation, TIDE. The mangrove inside it has since registered no change at all across seven satellite time-steps from 1980 to 2012, against national losses. Over the same period the seagrass at a least-impacted reference site inside the reserve fell from about 40% cover to about 10%, monitoring at a second station in the same district was suspended for poor visibility, and in 2015 the reserve’s water-quality monitoring ceased for lack of funding. No seagrass time series in Belize continues past 2012. The Belize page sets it out with its sources.
Latvia went furthest on paper and won in court. The Aarhus Convention was ratified in the middle of my fieldwork; the 2006 Environmental Protection Law gave organisations standing, information rights and a statutory comment period; and in April 2024 three environmental NGOs used the constitutional right to a benevolent environment to void a forest-felling regulation, partly on the ground that they had been given no real chance to comment on it. Over the same period Latvian forests went from a large net carbon sink to a net source, and land use became the country’s largest single emitting sector. The Latvia page sets it out with its sources.
The reading these pages take is narrow and, I think, defensible: participation distributes authority over the pressures a community controls. Sea temperature, a national sewerage budget, an EU-wide harvest economy and a demographic collapse are not among them. That is a limit on the argument, not a refutation of it — but it is a limit the thesis did not state, and the record since has stated it repeatedly.
The cross-case chapter that first tried to draw the three places together is Chapter Ten. These pages are its sequel, written with thirty years of evidence it did not have.
Each page rests on a research brief compiled in July 2026, one per place, in which every claim carries a citation and every unverified claim is marked as such. The pages reproduce those marks, and each carries its own Sources list; this hub asserts nothing that is not sourced on the page it links to. Where a figure was computed during the research pass rather than published by anyone, it says so; where a source was read only as a summary, the page either avoids it or flags it.
The most important caveat is the same in two of the three places and is stated at length on both pages: nobody has re-surveyed my transects or my plots. The Fiji transect coordinates were never written down and the site map has not been recovered from inside the Word file; the Belize plot coordinates were taken by differential GPS in 1994 and do not survive in any file that does. So there is no “then and now” chart of the same ground, and any that appeared here would be a fabrication. What can be compared is compared, and what cannot is named.
The original measurements are re-plotted from the surviving spreadsheets in the data section, and the field chapters themselves are Chapter Three and Chapter Four, with Latvia in Chapter Eight and Chapter Nine.