The law grew teeth.
The forest became the problem.

The design, the samples and the findings in this section are all from Chapter Nine itself, checked line by line against the surviving SPSS file, thesisdatwlabels.sav, during the 2026 research pass; every group count and mean matches the chapter’s prose exactly.

The instrument ran eighteen demographic items, 31 Likert-type items and eight open-ended ones, on a four-point scale from “very true” to “not at all true” with a separate “don’t know” option. Reliability across the scale was α = .80. Analysis was principal components with varimax rotation, yielding nine components explaining about 60% of variance, with group differences tested by ANOVA. I state plainly that this is exploratory and post hoc.

I also name my own design flaw: offering “don’t know” as a fifth option rather than a scale midpoint. It took roughly 15% of responses per item and approached half on the item about whether the government does enough to educate and include people. I ran three treatments — listwise deletion, imputation, and recoding to the midpoint — got near-identical factor structures and ANOVAs from all three, and adopted the midpoint recode.

The sample is the limit on everything downstream, and I am candid about it. Schools were recruited by permission of principals, the ministry by walking department to department with an inside contact, farmers through a colleague, and the Rīga and expatriate samples by approaching people in cafés and pubs. Four of the seven groups have fewer than twenty respondents. Seventy-three per cent of the whole sample were secondary-school students. There was no Russian-language arm at all, in a country where Russian-speakers were roughly a third to a half of the population, because my own Russian was not adequate to verify the translations. This is a convenience sample of seven purposive subgroups, not a probability sample of Latvia.

And an eighth group existed. I had collected surveys from NGO participants, and the notebook containing them was lost. In a study about participation, the NGO voice is the one that did not survive.

The chapters are Chapter Eight, on context and the daina, and Chapter Nine, the survey itself; the item means are re-plotted from the original SPSS file in the data section.

The hypothesis I set out to test broadly held: groups with a direct connection to the environment — farmers first, ministry staff second — scored highest on concern and on the cultural factors, with farmers often above the ministry, which I read as contact with real-world impacts beating theoretical engagement.

Three findings matter more than the hypothesis.

The ministry was systematically out of step with everyone else. Ministry staff scored significantly lower than other groups on the factor covering economic development as a threat to the environment and support for shutting down polluting factories, and lower on concern about global warming. At the same time they were the highest on believing the government does enough to educate and include the public, the highest on believing government is doing a good job protecting the environment, and the most satisfied with the state of the environment. My own summary calls this “particularly shocking.” The structural reading is the interesting one: the body responsible for participation was the body most convinced that participation was already adequate.

General disempowerment. On the item about whether there are actions the respondent could take that would effectively help the environment, every local group scored low. Only the expatriates — the people who did not have to live with the outcome — felt able to act. Read this one with the caveat below: it is the item whose Latvian wording diverges most from the English text printed in the thesis, and the divergence may make it a weaker claim than it looks — see a finding inside my own instrument.

The cultural factor was real and patterned. The factor covering culture and environmental protection was significantly higher for women, for married and cohabiting respondents than for single ones, for Lutherans and Catholics, and rose with education. That is the empirical hook for the daina argument in Chapter Eight: cultural identification and environmental concern were not independent. (Women p = .02; married and cohabiting against single p < .01; education p = .04 — Chapter Nine.)

The chapter also reports its non-findings, which is to its credit: two factors showed no group effect, number of languages spoken predicted nothing, and ethnicity could not be separated from group and so was not reported at all.

About 96% of respondents answered the Latvian version, so the Latvian wording is the operative instrument. Comparing it line by line with the English text printed in the thesis shows that several items are not translations of each other. The comparison was made for this edition from the two instruments as they survive in the fieldwork archive — Survey English final2.doc and SurveyLatviantranslation final2localchange.doc — and the English renderings below are literal translations made during that pass, not published ones.

Two consequences follow, and both are carried into the edition rather than tidied away.

The disempowerment finding rests on the most divergent item. Low Latvian scores may partly reflect modesty about having suggestions rather than a belief that action is futile. The finding is not destroyed — it is corroborated by the neighbouring item about knowing how to be more involved, which is a close translation and also scored low — but the strong reading should be softened.

The anchor drift is the exact bias I worried about. I wrote that I sometimes felt people were answering with the response they thought was “correct.” The Latvian anchors literally said correct and incorrect. That is a demand characteristic built into the instrument, and I could not see it, because I was reading my own English text. The same is true of the nature-versus-environment distinction in item 16: I had flagged that framing hazard in the chapter itself, citing Lang (1999) for the finding that Estonians associated nature with life and beauty and environment with waste and degradation — and it is present, uncontrolled, inside my own questionnaire.

My survey is not a baseline that Eurobarometer continues, and the temptation to set the two side by side has to be refused. The sampling is different: seven purposive subgroups against roughly a thousand nationally representative respondents. The instrument is different, and no item in my questionnaire has a Eurobarometer twin. The language is different: Eurobarometer fields in Latvian and Russian in Latvia, and my study has no Russian arm. The mode is different: pen and paper, often in a supervised classroom, administered by a visible foreign researcher, against surveys that in Latvia in 2023 were conducted entirely online (Eurobarometer environment and climate series, Latvia from 2004). And the population is different, because it has changed underneath the question.

So there is no “X per cent of Latvians in 2002 against Y per cent in 2024” anywhere on this page, for any item. What can be compared, and is, are institutions and law; physical outcomes measured on continuous official series that span both periods; and qualitative claims — the belief that the public was adequately included, set against the later documented disputes about whether it was. That last is a comparison of claims, not of numbers, and it is offered as one.

The starkest change is not environmental. I wrote of nearly two and a half million people, on the ministry’s own 2001 figures (MEPRD 2001, cited in Chapter Eight). At the beginning of 2026 Latvia’s population was 1.845 million, down 0.83% — about 15,500 people — in a single year, from negative natural change and negative net migration (Centrālā statistikas pārvalde, 2 June 2026). A study of how a public relates to its environment was conducted on a population that has since lost roughly a quarter of itself. The teenagers in the Valmiera and Skrīveri samples are in their late thirties and early forties now, and a substantial share of that cohort left.

The institution whose staff produced my most striking finding has been dismantled and redistributed. The timeline below is from the Cabinet of Ministers’ own announcement of the transfer of functions and from the ministry’s announcement of its renaming (Ministru kabinets; VARAM, 21 May 2024); the 1993 entry is from Chapter Eight.

In 2003 I described a country with environmental information rights in its constitution, a statutory fifteen-day response time for information requests that was better than the Aarhus recommendation — and an actual rate of involvement I called extremely low. Officials refused information until told what it was for; data was fragmented across institutions with nobody able to publish it; the public typically learned of decisions after they were made and could only picket (Zaļā Biblioteka 2002, as cited in Chapter Eight).

Latvia signed the Aarhus Convention on 25 June 1998 and ratified it on 14 June 2002 — in the middle of my fieldwork, while I was administering the survey (UN Treaty Collection, chapter XXVII-13). The constitutional right I noted, Article 115 of the Satversme, had been added by the amendment of 15 October 1998 and in force from 6 November 1998, and made, as I observed, no practical difference. The date on which the Convention entered into force for Latvia is not asserted here: the research pass behind this page found no source stating it.

The substantive change came in 2006. The Environmental Protection Law, adopted on 2 November 2006 and in force from 29 November, gives the three Aarhus pillars statutory form. Section 6 defines “the public” to expressly include associations, organisations and groups of persons — so NGOs have standing. Section 7 gives information rights with no need to justify the request. Section 8 requires participation before a decision is adopted, with at least thirty days for comment. Section 9 provides for contest and appeal. (All four sections read from the official English translation on likumi.lv.) Note the direction of travel on one point: the fifteen-day deadline the chapter praised became one month, aligned with the Convention.

Section 14 of the same law established the Environmental Advisory Council (Vides konsultatīvā padome), governed by Cabinet Regulation No. 209 of 27 March 2007 and still in force: six to twenty members, nominated by associations and foundations each year and elected by secret ballot, one vote per organisation, with the stated purpose of promoting public participation in environmental policy. That is structurally close to the representative-public NGO I proposed in Chapter Ten — an organisation whose function is to carry public views into the decision process.

Beyond the environment specifically, Latvia has had a standing Cabinet regulation on public participation in development planning since Regulation No. 970 of 2009, replaced by Regulation No. 639 of 15 October 2024, and draft legislation with its consultation record now sits on a public portal (tapportals.mk.gov.lv) with a dedicated participation module. The information fragmentation I described has an infrastructural answer it did not have then.

This is the single most important development for the thesis’s argument, and it turns explicitly on public participation.

In June 2022 a Cabinet regulation — No. 374 of 21 June 2022, amending Annex 7 of the felling regulation No. 935 of 2012 — lowered the minimum diameters at which trees may be taken in main felling: pine from a range of 39 to 31 cm to a uniform 30, spruce from 31 to 29 down to 26. Three environmental NGOs — Latvijas Dabas fonds, Pasaules Dabas Fonds and Latvijas Ornitoloģijas biedrība — challenged it. On 8 April 2024 the Constitutional Court, in case No. 2023-01-03, declared the reduced diameters incompatible with Article 115 of the Satversme — the same right I had noted as making no practical difference — and void from the moment they took effect. Felling permits issued on the invalidated diameters were cancelled (Valsts meža dienests; LV portāls).

The grounds were procedural as much as substantive. The amendment reached the Cabinet agenda one day beforehand and was discussed for under an hour, leaving organisations no time to submit written comment; it was adopted without a strategic environmental assessment; and it breached the precautionary principle by not analysing the impact on old-growth habitat and on birds dependent on large-diameter trees (Satversmes tiesa, case No. 2023-01-03).

Read against Chapter Six, this is the structured-participation argument winning in a court: a decision taken without genuine opportunity to comment, and without assessing alternatives and uncertainties, is not merely impolite but unconstitutional.

What followed is an argument about compliance, and it is not settled. An agriculture ministry consultation running from 9 to 22 October 2024 drew eight submissions, listed on the public consultation portal; the package proposed, among other things, a 20% tolerance on diameter measurement, which five organisations argued in an open letter to the Prime Minister would in effect restore what the court had struck down (LV portāls, 2024). What actually took effect on 1 July 2025 was narrower: diameter-based clear-cutting was barred where the stand had been thinned in the preceding three years (Valsts meža dienests). Whether the same package’s proposal to raise the maximum clear-cut area from two hectares to five was adopted is unverified. A further round in September 2025 proposing lower main-felling ages drew objections from more than twenty organisations, and the State Chancellery noted that it had received more than thirty-five private submissions requesting a public consultation which the ministry did not adequately justify refusing. That last round is recorded from a single secondary news source (NRA, 2025) and its outcome is unverified.

The Latvian main-line environmental impact assessment concluded in 2016, and the route was fixed by Cabinet order No. 467 of 24 August 2016. A parliamentary inquiry commission’s final report in 2024 established the structural problem, and the account below is drawn from it. Alternatives for the approach to Rīga were cut from eight to three at a steering group meeting on 28 November 2014, and then to one on 23 December 2014 — before the consultants’ own interim report of 6 January 2015, and against the consultants’ advice favouring a different option. The commission’s sixth conclusion is that only one Rīga-entry variant ever went into the assessment. The public consultation of March 2015 therefore had no genuine choice of Rīga-entry alternatives to comment on.

The same commission found that the transport ministry had concealed the true cost increase from the Cabinet by re-expressing it in 2016 prices, and that scope changed more than a hundred times without a fixed budget or a responsible official. The state audit office put the Rīga Central Station node at €430 million rising to €888 million, and the whole project at €5.8 billion in 2017 rising to €23.8 billion in 2023 (Valsts kontrole, June 2024). The first phase was cut to a single track (Satiksmes ministrija, 29 October 2024). Whether a strategic environmental assessment was ever carried out for Rail Baltica in Latvia is unverified — only the project-level assessment is confirmed — so no claim is made about one here.

This is the felling-diameter case in transport form. The dispute is not about what the public wanted but about whether a real choice was ever put to it — which is exactly the failure mode the five-step process in Chapter Six begins with objectives and constructed alternatives in order to prevent.

I described Latvia in 2003 as 44% forested, comparatively clean, with the risk lying in future development. What actually happened is stranger. Latvia’s land use, land-use change and forestry sector has gone from a large net carbon sink to a net source. The series below is from Latvia’s own National Inventory Document 1990–2024, Tables ES.2a and ES.2b, submitted to the UNFCCC on 15 April 2026.

Latvia’s own national inventory attributes the change mostly to an increased harvest rate, with the ageing of forests raising natural mortality and reducing increment. The largest single source within the sector is organic soils, at 6,322 kt CO₂ eq in 2024, including emissions from peat extracted for horticulture. Land use is now Latvia’s largest single emitting sector, at 23% of emissions, ahead of transport and agriculture. Emissions excluding land use are down about 63% on 1990; including it, net emissions rose 146% between 2005 and 2024 (European Commission DG CLIMA, Latvia country profile, January 2026). The European Environment Agency reaches the same verdict independently, recording land-use emissions up 137% between 1990 and 2023 and moving in the opposite direction from the national target (EEA 2025).

Latvia’s headline decarbonisation success and its headline failure are the same country.

Two cautions belong with those numbers. Figures move between inventory vintages: the European Commission’s January 2026 factsheet, working from the 2025 submission, gives 2023 net emissions as 4.6 Mt, while the 2026 submission revises the same year to 1.65 Mt. Always state the submission year. And forest cover itself is stable, not expanding: the international assessment series records a rise from 51.0% in 1990 to 54.9% in 2020 (FAO Global Forest Resources Assessment 2020), but the National Forest Inventory team’s own analysis of the 2004–2025 series reports cover stable from 2005 to 2025 and total standing stock apparently reaching equilibrium with mortality and felling in 2025 (Šņepsts et al. 2026). The forest sector’s own promotional framing of a constantly expanding forest is not supported by the inventory team’s numbers.

And the science is contested, and the contest is domestic. State forest research institute authors, comparing 84 old-growth stands against 266 managed ones, find that although old-growth carries about 20% more biomass carbon, the total mitigation effect of managed stands is 41 to 66 per cent higher once harvested wood products and material substitution are counted (Jansons et al. 2025). Conservation NGOs read the same inventory and reach the opposite policy conclusion. This is exactly the expert-disagreement structure Chapter Six predicted — a disagreement that is value-laden, not merely technical — and the edition flags it as such rather than picking a side.

Water

Latvia is covered by four river basin districts — Daugava, Gauja, Lielupe and Venta — each with an approved management plan on the Water Framework Directive’s six-year cycle (LVĢMC). The European Environment Agency’s country profile, written by national Eionet experts, states the outcome bluntly: only one third of surface water bodies are in good or high ecological status. The main pressures are diffuse nutrient pollution and hydromorphological alteration — straightening, barriers, water-level change — a direct legacy of the Soviet-era drainage network (EEA 2025).

My questionnaire carried an item on the importance of protecting the Baltic Sea. For the Gulf of Rīga, Latvia’s own basin, HELCOM’s nutrient-input assessment finds that fulfilment of the maximum allowable input cannot be judged for nitrogen — the remaining reduction of 2.0% sits inside statistical uncertainty — while for phosphorus it is not met, with 11% still to cut for the Gulf and 91% for the Baltic Proper, given the uncertainty on inputs (HELCOM, inputs-of-nutrients indicator, assessment period 1995–2022). Twenty years on, the honest status of the Gulf of Rīga is indeterminate.

Protected areas

The system I described — three national parks, four strict nature reserves, 211 nature reserves, 22 nature parks and six protected landscape areas, covering 8.53% of the territory (MEPRD 2002, cited in Chapter Eight) — has roughly held its shape and grown: four, four, 239, 37 and nine respectively within the Natura 2000 network in 2024, plus seven marine protected areas it did not have at all, for 333 sites and 11.56% of land territory (Dabas aizsardzības pārvalde), on a source page dated December 2020. Three caveats before that becomes a trend: that date; the 2024 counts are of sites in the Natura 2000 network, which is not by definition the full national register; and the 8.53% is of the whole country while the 11.56% is of land territory. The category-by-category counts are the safer comparison, and they tell the story anyway.

Natura 2000 also arrived in Latvia as a participation problem before it arrived as a conservation fact. The pre-designation assessment I cite in Chapter Eight — Lang (2000), 113 interviews across the three Baltic states in January 2000, and a copy of it sits in this project’s own fieldwork archive — found every target group fearing the financial consequences of EU-driven sustainability, with Latvians specifically naming fear of more bureaucracy, and recommended that the regional level and the sectors, rather than the national administration, needed far more information in the local language.

The daina

Chapter Eight’s argument that Latvian oral poetry encodes a non-separated relation to nature was, in 2005, an idiosyncratic move. It has since acquired company. The Cabinet of Folksongs was inscribed on the UNESCO Memory of the World Register in 2001; the Baltic song and dance celebrations were proclaimed a masterpiece of oral and intangible heritage in 2003 and inscribed on the Representative List in 2008 (UNESCO). The university institute that holds the Archives of Latvian Folklore now lists environmental humanities among its research areas and runs a department of culture, society and environment (LU LFMI). And there is now scholarship reading life and nature in the dainas as grounding an affirmative environmental ethics (Sauka 2024) — the closest thing to a direct successor to my chapter, though it is paywalled and was verified only at abstract level for this project.

The correctives matter as much as the company. A disciplinary history of Latvian mythology documents how the field was constructed (Ķencis 2012), so any claim of unbroken continuity from daina to modern belief has to pass through it; and the modern reconstructed paganism is framed in the literature around indigeneity and nationalism rather than environmentalism (Stašulāne 2019), so a green Dievturība should not be asserted. Nor is the right of access to the forest what the folk story suggests: the Law on Forests, sections 5 and 16, applies free movement to state and municipal forest, allows an owner to restrict access, and permits gathering berries and mushrooms only where the owner has set no restriction. This is not Nordic allemansrätt. And no Eurostat or Eurobarometer dataset was found in the 2026 research pass that would allow a Latvia-against-EU comparison of time spent in nature, so none is asserted.

It won — but by the route I did not propose

In Latvia the participatory architecture did more than get built. The 2006 law gave organisations standing, information without justification and a statutory comment period before decisions. The Environmental Advisory Council gave them a permanent seat. And in April 2024 three of those organisations — Latvijas Dabas fonds, Pasaules Dabas Fonds and Latvijas Ornitoloģijas biedrība — used the constitutional right I had described as inert to void a government felling regulation, partly on the ground that they had been given a single day and under an hour of Cabinet discussion in which to comment (Satversmes tiesa, case No. 2023-01-03).

But notice how it arrived: by litigation. Chapters Six and Ten are almost entirely about facilitated deliberation. Access to justice — the third pillar of the Aarhus Convention — barely figures in the argument at all. In practice it is the pillar that has done the work here. And it was the domestic route rather than the international one: across 233 communications to the Aarhus Compliance Committee since 2004, Latvia has never once appeared, while its own Constitutional Court struck down a forestry regulation on participation grounds. That count is not a published statistic: the UNECE case lists were enumerated end to end for this project on 31 July 2026, and a search of the communications list for “Latvi” returned nothing. Latvia is likewise absent from the four submissions by Parties and the five requests for advice; no referral has ever been made by the secretariat. That silence carries two readings and the page does not choose between them. It may mean the domestic route worked well enough that nobody needed Geneva; it may equally mean the international route is not reached for. Neighbouring Lithuania has appeared before the committee once, in ACCC/C/2006/16, which is a comparator but not an answer.

The caution that belongs with that: the European Commission has an infringement open since 2019 over an incomplete Natura 2000 network in Latvia — INFR(2019)2304, letter of formal notice 27 November 2019, still open six and a half years later on a query of the Commission’s infringement database run for this project — a finding that the country has generally and persistently failed to set site-specific conservation objectives and measures, and a 2020 finding that it had not correctly transposed the Seveso III directive in relation to public consultation and participation (European Commission, May 2020 infringement package). Participation improved. Compliance is another matter.

The method I actually prescribed has not been tried

No verified evidence exists that structured decision aiding in the sense Chapter Six means it — trained facilitators, objectives elicited before alternatives, explicit characterisation of tradeoffs, asking what participants can support and why — has ever been used in Latvian environmental policy. What Latvia built instead is procedural participation: notice periods, comment windows, written submissions, hearings. That is Convention-compliant and it is not the same thing.

The gap shows in an unexpected place. Latvijas Dabas fonds’s 2025 position on wind energy asks for precisely what Chapter Six asks for — a national development plan with strategic environmental assessment and sensitivity mapping, rather than project-by-project fights — which is an organisation in 2025 independently arriving at my 2005 prescription because the procedural machinery is not delivering. The specific method has not been tried and found wanting. It has not been tried.

The structure recurs, and it has moved between ministries

The pattern I found in seventeen ministry employees — the body responsible for participation being the one most satisfied with it — recurs across the documented record as a dispute about whether consultation was genuine rather than about what the public wanted. Rail Baltica’s single Rīga-entry variant (Saeima inquiry commission 2024). A felling amendment discussed for under an hour (Satversmes tiesa 2024). An agriculture ministry declining a public consultation in September 2025 despite more than thirty-five private submissions requesting one (NRA 2025, secondary and unverified). A collective submission signed by 10,652 citizens, “Par Latviju bez vēja parkiem”, rejected by the Saeima without stated reasoning.

But it has also sharpened in a way I could not have predicted. In 2002 the discord I measured ran between the environment ministry and the public, over how much environmental protection to want. In the 2020s it runs between ministries: agriculture pushing felling deregulation against objections from climate and energy, culture, transport, finance and the smart administration ministry — which is the September 2025 round again, and rests on the same single secondary account. The environment portfolio I surveyed has meanwhile been split and renamed out of existence.

The one I did not see coming

The ecological trend went the other way regardless. The forest that was 44% of the country and read as an asset is now the largest single emitting sector, having flipped from sink to source. Only a third of water bodies reach good or high status. The Gulf of Rīga misses its phosphorus target and cannot be adjudicated on nitrogen at all. The population that would have to do the participating has fallen by roughly a quarter.

And the collision is precise. My questionnaire asked whether protecting Latvia’s forests was important to the respondent, and it drew one of the highest agreement scores in the whole instrument. Nothing in the 2005 document anticipates that this consensus would run into a carbon accounting rule and become the country’s most contested environmental politics.

My corruption worry, by contrast, has aged well — and as the epistemic claim I made, not as an accusation. I argued that a widespread belief that everything is corrupt undermines the legitimacy of environmental data itself, which is a first-order problem for participation. The state audit office now reports a forest sector operating without approved medium-term policy guidelines since 2021 and a state forestry company holding €405 million in free funds (Valsts kontrole, April 2026); the parliamentary inquiry reports Rail Baltica’s cost increase concealed from the Cabinet by re-basing prices. The complaint was well founded, for reasons better documented than the single anecdote I had.

And the deep background Chapter Eight does not tell

One thing changes the shape of my central finding. On 17 October 1986 two writers — Dainis Īvāns and Artūrs Snips — published an article in a Rīga literary weekly opposing a planned hydroelectric station on the Daugava. The campaign succeeded: the government announced cessation on 6 July 1987, and the USSR Council of Ministers decision was published on 5 November 1987. The Environmental Protection Club was founded on 25 February 1987, four months after the article, and Īvāns, the first of its two authors, became the first chairman of the Popular Front on 9 October 1988 (Daugavpils University Latgales dati; Latvijas Radio 1). Latvia’s environmental movement and its independence movement are the same event, and the first won before the second existed.

Which makes my 2002 finding stranger, not simpler. I found a public that did not report much confidence in its own capacity to act — and, on the most careful reading the instrument will bear, may have been reporting modesty about having specific suggestions rather than a belief that action is futile — in a country where, within living memory, acting had worked spectacularly, once, against an empire.