Patrol records and community perceptions around Koh Rong

Research account. This page covers the source sections identified below.

Roig-Boixeda, P., Chea P., Brozovic, R., You R., Neung S., San T., Teoh, M. & West, K. (2018). Using patrol records and local perceptions to inform management and enforcement in a marine protected area in Cambodia. Cambodian Journal of Natural History 2018(1):9–23.

Original article · Saved paper · Selected structured findings

What this tells us

This study combines patrol records with household perceptions and key-informant interviews for the Koh Rong Archipelago Marine Fisheries Management Area. It demonstrates why maps of recorded offences also reflect where patrols operated. Areas with no recorded offence are not automatically areas with no illegal fishing.

Source coverage

Bibliographic material on PDF p.1, English abstract/context on p.2, methods/results on pp.4–6 and results/discussion portions on pp.8–9 were examined as text. Table 1 on PDF p.5 was visually checked. Printed page numbers are PDF page + 8. The Khmer abstract, PDF p.3, p.7, pp.10–15, remaining figures, Appendix 1, full reference list, corrections, supplements and funding/conflict statements are not yet fully examined. The source account is deliberately partial.

Study period and sampling

The analysis uses 161 patrols between 16 June 2016 and 16 March 2017 and a March 2017 household survey. The authors report randomly surveying 163 of approximately 546 households across five villages. There were 33 fisher respondents among the household sample. Twenty-one key informants supplied additional interviews; seven fishers and four tourism operators were also household-survey respondents. Do not add 163 and 21 as if these were 184 independent participants. Methods: printed pp.12–13 / PDF pp.4–5.

Table 1: printed p.13 / PDF p.5. The household estimates came from village chiefs, rather than an independently verified census.

Selected results and negative findings

• Patrols recorded 38 fishing offences during the study interval. These are recorded infractions, not an estimate of every offence in the protected area. Printed p.14 / PDF p.6.
• The spatial analysis used 511 one-square-kilometre cells. Infractions per patrol could be calculated for 151 cells; 360 cells were unpatrolled and could not receive that rate. Missing effort is not a zero-offence rate. Fifteen patrols were excluded from the per-hour analysis because recorded durations were under one hour or over five hours. Printed p.13 / PDF p.5.
• No statistically significant overall spatial association was reported between perceived illegal-fishing distribution and patrol-based distribution: Spearman rho = 0.15, p > 0.05. Hotspot agreement was reported separately: phi = 0.39, p < 0.001. These are different comparisons. Printed p.16 / PDF p.8.
• Perceived and recorded patrol effort were more closely associated: Spearman rho = 0.54, p < 0.001; hotspot phi = 0.69, p < 0.001. This does not demonstrate a causal effect of patrols on offending. Printed p.16 / PDF p.8.

Interpretation and consistency checks

The authors warn that patrols were not randomly distributed and that patrol records alone can mislead management (printed p.17 / PDF p.9). Household answers describe perceptions of others' behaviour, not respondents' admissions. Key-informant recruitment included snowball selection and repeat interviews of selected household respondents. These design features limit representativeness and independence.

The methods report an effective mapping sample of 95 after four exclusions from 99 respondents (printed p.12); the results refer to 99 respondents (printed p.16). Table 2 gives 143 responses about local fishers, while accompanying text reports excluding 17 from a sample of 163, which would leave 146. These discrepancies require reconciliation; no corrected denominator is invented here. Funding, author relationships, later findings and all remaining tables/appendices still need review.

Atlas use and verification

Retain distinct evidence types for patrol observations, household perceptions and interview interpretations. Treat them as components of one study rather than three independent studies. Do not turn the printed maps into precise contemporary offence locations. Underlying SMART records, interview responses and geographic grids have not been recovered in this batch.

This report contributes evidence on Cambodian marine management and livelihoods. The source-specific uncertainties are recorded below.