Reconciling policy instruments with drivers of deforestation and forest degradation: cross-scale analysis of stakeholder perceptions in tropical countries

Cross-scale studies combining information on policy instruments and on drivers of deforestation and forest degradation are key to design and implement effective forest protection measures. We investigated the scale and country dependency of stakeholder perceptions about future threats to tropical forests (e.g. agriculture, logging, woodfuel) and preferred policy instruments (e.g. reforestation, protected areas, combat illegal logging), by interviewing 224 representatives of forest-related institutions. We conducted analysis of variance and principal component analysis for eighteen variables across three countries (Zambia, Ecuador and the Philippines) and four spatial levels (from international to local). We found that the overall alertness about commercial drivers and the confidence in policy instruments are significantly lower at subnational levels and also in Zambia. Stakeholder expectations about the most important drivers and the most effective policies in the coming decade follow regional narratives, suggesting that there are no one-size-fits-all solutions in international forest policy. However, we found an unexpected consensus across scales, indicating potential for collaboration between institutions operating at different geographical levels. Overall, agriculture remains the driver with the highest expected influence (43%), while a strong favoritism for reforestation and forest restoration (38%) suggests a paradigm shift from protected areas to a stronger focus on integrative approaches.

Supplementary Figure S7. Number of responses ranked within the top-5 per policy instrument category, regarding the influence on forest protection in the next 10 years, for the total sample (a) and the country subsamples (b, c, d).
Note: the range of the x-axis varies between samples. Blank answers are not shown or considered for the total N in the graph. 12 Supplementary Figure S8. Number of responses ranked within the top-5 per policy instrument category, regarding the influence on forest protection in the next 10 years, for the four spatial level subsamples. Note: the range of the xaxis varies between samples. Blank answers are not shown or considered for the total N in the graph. 13

Summary statistics for al variables across countries and across spatial scales 14
Supplementary Table S4. Summary statistics (count, average and standard deviation) of the studied variables for the total sample and the country subsamples. 14 Supplementary Information Ferrer Velasco et al. Reconciling policy instruments with drivers of deforestation and forest degradation: Cross-scale analysis of stakeholder perceptions in tropical countries Supplementary Table S5. Summary statistics (count, average and standard deviation) of the studied variables for the spatial level subsamples. 15

Variables and transformations: analyzing univariate and multivariate normality 16
Supplementary Table S6. Skewness, histograms, boxplots, Shapiro-Wilk and Mardia normality results for the variables related to drivers of deforestation, before and after transformation (we selected the method which brought skewness the closest to zero, between square-root, log or inverse). 16 Supplementary Table S7. Skewness, histograms, boxplots, Shapiro-Wilk and Mardia normality results for the variables related to policy instruments, before and after transformation (we selected the method which brought skewness the closest to zero, between square-root, log or inverse). 18

Principal component analysis (PCA) 20
Supplementary Figure  Supplementary Figure S11. Results of the PCA with all the variables: biplots of the individuals grouped by country (a) and spatial level (b) and of the variables (driver categories) (c) for the two first components. 23

Non-parametric analysis of variance: one-way Kruskal-Wallis and Dunn tests 24
Supplementary

Descriptive statistics of stakeholders and institutions
Supplementary Table S1. Number of interviews conducted per country (ZMB: Zambia, ECU: Ecuador, PHL: Philippines) and spatial level of institution (INT: International, NAT: National, REG: Regional, LOC: Local), grouped by characteristics of the respondents and by the characteristics of the stakeholders' institutions. Note: More than three fourths of the respondents (77%) were male representatives. We only obtained information about age and education in the Philippines, where the majority of the interviewed stakeholders (75%) were older than 45 years of age and possessed a university degree (95%), generally an undergraduate one (57%). Our sample was largely dominated by institutions related to central governments (50% of the conducted questionnaires). Most of these central institutions belonged to the national or the regional spatial levels (60 and 41 out of 110, respectively). The other six stakeholder groups studied (e.g., academia, private enterprises, indigenous associations) represented similar smaller shares of the total sample, ranging from 6% (i.e., international organizations) to 10% (i.e., local governments). Approximately half of the interviewed representatives belonged to large organizations with more than 50 workers. Moreover, eight out of ten respondents declared gaps or insufficient resources for their forest institutions to perform their tasks. This perception of not having adequate resources was especially recurrent in the Philippines, with 96% of the participants acknowledging gaps.  Supplementary Table S2. Pre-selection of answers (and extra answers provided by the respondents) to the questions: "Please score for each of the following drivers their influence on deforestation and forest degradation in the next 10 years" and "Which of the presented deforestation and forest degradation drivers will be the TOP 3 to 5 in the next 10 years?", in each country and grouped by cross-country driver categories.

Lists of driver and policy instrument answers/categories
Country [1][2][3][4] Cross-country driver categories [5][6][7][8][9] Zambia Ecuador Philippines  Table S3. Pre-selection of answers (and extra answers provided by the respondents) to the questions: "Please score for each of the following Reforestation and Conservation measures their influence on reforestation or forest conservation in the next 10 years?" and "Which of the Reforestation and Conservation measures will be the TOP 3 to 5 measures that can stop deforestation and increase forest areas in the next 10 years?", in each country and grouped by cross-country policy instrument categories (second section of the questionnaire).
Country 2,4,10 Cross-country policy instrument categories 11,12 Zambia Ecuador Philippines  Supplementary Table S7. Skewness, histograms, boxplots, Shapiro-Wilk and Mardia normality results for the variables related to policy instruments, before and after transformation (we selected the method which brought skewness the closest to zero, between square-root, log or inverse).