Charcoal and Fuelwood Trends in Northern Haiti - USAID 2019.pdf

Charcoal and Fuelwood Trends in

Northern Haiti An Analysis of Charcoal Production and Supply in Cap Haïtien

Authors

Andrew Tarter, Ph.D. (lead author, Science & Technology Policy Fellow, American Association for the

Advancement of Science)

Rebecca Ciciretti (LAC Program Specialist, International Programs, U.S. Forest Service)

Kenson Théus (sociologist, Haiti Country Coordination Consultant, International Programs, U.S. Forest Service)

Melinda Miles (anthropologist, managing partner, Manifest Haiti (research logistics firm)

June 2019 http://www.andrewtarter.com/

Background

Funding for this report was supported by a Participating Agency Service Agreement (PASA)1 between the United States Department of Agriculture (USDA) Forest Service International

Programs (USFS IP) and the United States Agency for International Development (USAID) to provide data to inform the USAID Reforestation Project Activity, a 5-year cooperative agreement between the U.S. Government and Chemonics International, awarded in 2017, and focused on the

Cap Haïtien Development Corridor2 in northern Haiti. The Notice of Funding Opportunity (NFO) for the Activity3 called for two assessments to inform the Activity’s interventions and targets, and any necessary midcourse corrections:

‘USAID/Haiti is separately procuring two fundamental assessments to provide information critical to defining ambitious yet feasible results, and achieving impact. The USAID Reforestation Project will collaborate closely with the entity conducting these assessments to incorporate the findings into the project’s targets and interventions, as well as adapt the overall approach if necessary’. 4

The two fundamental assessments outlined in the NFO for the USAID Reforestation Project

Activity in Haiti were initially framed as:

1. ‘An assessment of the ability of existing stands of trees in the targeted areas to provide the functions and services of a forest’ and;

2. ‘A characterization of the wood-based fuel value chains and wood-based fuel “sheds” in the targeted areas’.5

The present report is based on the objectives of the latter assessment, but narrowed in scope as a result of the delayed initiation of both assessments and the commencement of the project with an existing and approved activity plan. Adjusting to this reality, members of the USAID/Haiti Mission and the U.S. Forest Service agreed that the methodology of the present report was best-suited to inform approved project activities that had already been initiated.

1 “306.3.2 FAA Section 632(b) Agreements Effective Date: 10/01/2002 Assuming that the requirements of OMB Circular A-76 either do not apply or are met (306.3.1.1), USAID may obtain the services of another agency under section 632(b) of the Foreign Assistance Act of 1961, as amended (FAA), which provides: Any officer of the United States Government carrying out functions under this Act, may utilize the services (including defense services) and facilities of, or procure commodities, defense articles, or military education and training from, any agency of the United States Government as the President shall direct, or with the consent of the head of such agency, and funds allocated pursuant to this subsection to any such agency may be established in separate appropriation accounts on the books of the Treasury. Under this authority, USAID may enter into agreements with other U.S. Government agencies to carry out functions under the FAA. This authority is the basis for entering into: A Participating Agency Service Agreement (PASA) (Form AID 306-2)…” – USAID ADS Chapter 306 Interagency Agreements, pages 15-16.

2 A year after the January 2010 earthquake in Haiti, the U.S. Government’s reconstruction strategy for Haiti was outlined in a report entitled, ‘Post-Earthquake USG Haiti Strategy: Toward Renewal and Economic Opportunity’. This strategy guided USAID interventions in Haiti for the next five years (FY 2011-2015), and focused activities in three geographically distinct ‘development corridors’.

3 United States Agency for International Development (USAID) Notice of Funding Opportunity Number: SOL-521-17-000011.

4,6 SOL-521-17-000011, pages 21-22.

5 See footnote 2, above.

https://www.usaid.gov/forms/aid-306-2 https://www.usaid.gov/sites/default/files/documents/1868/306.pdf https://www.grants.gov/web/grants/view-opportunity.html?oppId=292808

Acknowledgements

The authors wish to thank Abdel Abellard (Mission Environmental Officer, USAID/Haiti), Reginald Toussaint Juan (Food Security and Agroforestry Specialist, USAID/Haiti), J.W.

Camilien Saint-Cyr (Economic Growth & Agriculture Development, USAID/Haiti), Kénold

Moreau (Senior Natural Resources & Environment Adviser, USAID/Haiti), and Juan Carlos

Martínez-Sánchez (Senior Environmental Advisor, USAID/Haiti) for their thoughtful review and commentary on the research design and final report of this research study. We also wish to thank the many research enumerators, too numerous to list, for their contributions to the fieldwork data collection.

TABLE OF CONTENTS

BACKGROUND

ACKNOWLEDGEMENTS

EXECUTIVE SUMMARY

KEY FINDINGS OF THE REPORT

INTRODUCTION

NORTHERN HAITI

FUELWOOD USE IN HAITI

BUILDING ON RECENT FINDINGS

COMPLEMENTARY CONTRIBUTIONS OF THIS RESEARCH

RESEARCH QUESTIONS

METHODOLOGY

PRINCIPAL METHOD

ENUMERATOR TRAINING

Team-building Vehicle Typology Data Collection Data Quality Control

SAMPLING

Temporal Sampling Spatial Sampling

DATA TRANSCRIPTION AND STANDARDIZATION

DATA ANALYSIS

DIFFERENCES BETWEEN VEHICLE FLOWS BETWEEN THE TWO SAMPLING PERIODS

Differences Between Vehicle Flows in the Core and the Periphery

DIFFERENCES IN OVERALL FLOWS OF METRIC TONS BETWEEN THE TWO SAMPLING PERIODS

Differences in Metric Tons in the Core and Periphery

DAILY VARIATION IN THE CORE AND PERIPHERY

HOURLY VARIATIONS

DAILY AND HOURLY VARIATION

REGIONAL DIFFERENCES IN CHARCOAL SUPPLY TO CAP HAÏTIEN

General Regional Supply to the Core Broader Regional Charcoal Production in the Northern Corridor

ANNUAL CHARCOAL CONSUMPTION IN CAP HAÏTIEN

COMPARING CAP HAÏTIEN ANNUAL CONSUMPTION ESTIMATES TO NATIONAL ESTIMATES

SUMMARY AND CONCLUSIONS

POLICY IMPLICATIONS AND FUTURE RESEARCH

REFERENCES

Executive Summary

The research presented here builds on the findings of a recently-published nationwide assessment of charcoal production and consumption trends in Haiti (Tarter et al. 2018), financed by the World

Bank Group and the Haiti Takes Root initiative of J/P Haitian Relief Organization (J/P HRO).

The present study replicated aspects of the methodology of the previous study, using Cap Haïtien as the final urban destination for rurally-produced charcoal in the northern corridor of Haiti.

However, while the nationwide study was concerned with estimating national trends, the present report is focused more on regional trends and associated implications for the northern corridor of

Haiti.

The research design, methodology, and analysis of data are presented in the body of the report.

The key findings of the report are presented below.

Key Findings of the Report

The key findings of the research presented here provide data pertinent to contemporaneous and future U.S. Government programming related to wood-based fuels, agroforestry, reforestation, and tree-planting activities in the northern region of Haiti:

Flows of charcoal into Cap Haïtien increase by very small percentages between the final core stations and the preceding periphery stations. This result suggests that virtually all charcoal in the northern corridor is bound toward Cap Haïtien;

there is another, smaller charcoal entry-point into the city; there were slight enumeration discrepancies; or some combination of these and other influences.

Approximately 75% of charcoal entering Cap Haïtien comes from the east and south-east of the city and the remaining 25% enters largely from the west and northwest. Charcoal is emerging not from the forested6 areas identified in a landcover analysis produced by the GeoCenter of USAID’s Global Development

Lab7; rather, charcoal is emerging from dry, flat, coastal areas, historically used for agricultural production.

Charcoal production is likely not the principal driver of deforestation in the northern corridor of Haiti. If it were, we would expect to see much larger percentages of charcoal entering Cap Haïtien from areas to the west and northwest

6 As noted in the NFO for the USAID Haiti Reforestation Project, “There many definitions of forest, with one report listing hundreds. Forest structure, i.e., the number of trees or amount of land the tree canopy covers in a given area, is the most easily quantified definition of a forest, measured using estimates of tree cover from satellite images. USAID will follow the FAO FRA definition of a forest for this funding opportunity: “Land spanning more than 0.5 hectares with trees higher than 5 meters and a canopy cover of more than 10 percent. It does not include land that is predominantly under agricultural or urban land use.” -USAID-Haiti-521-17-1234-RFA.

7 USAID-Haiti-521-17-1234-RFA.

http://documents.worldbank.org/curated/en/697221548446232632/Charcoal-in-Haiti-A-National-Assessment-of-Charcoal-Production-and-Consumption-Trends http://documents.worldbank.org/curated/en/697221548446232632/Charcoal-in-Haiti-A-National-Assessment-of-Charcoal-Production-and-Consumption-Trends of the city, identified as ‘forested’ by the GeoCenter of USAID’s Global

Development Lab.

The amount of charcoal that may be entering northern Haiti overland from the Dominican Republic would be equivalent to 9-13% of the amount consumed in Cap Haïtien during the same period. This range reflects observations from the intersection of two major roads near the border between the two countries, although actual border crossings of charcoal were not observed.

We estimate that approximately 67,700 metric tons of charcoal are consumed annually in the Cap Haïtien metro area. This estimate is derived from extending our observations from high and low charcoal production periods to a 12-month period, assuming an equal division between these periods in a typical year (6 months of high production and 6 months of low production). This estimate suggests that per capita consumption of charcoal in Cap Haïtien is higher than in Port-au-

Prince.

There are clear preferences to transport charcoal in the early morning hours and the later afternoon/early evening hours. These trends likely reflect a combination of traffic, daylight, heat, and sleeping and waking patterns. Friday and

Saturday saw increases in the later circulation of charcoal, perhaps a function of the impending weekend or changes in traffic patterns.

Small pickup trucks deliver approximately 75% of the charcoal supplied to

Cap Haïtien. This result was notably different from a recent nationwide charcoal study using the same methodology8, where large trucks were the most prevalent and small trucks composed only approximately 39% of all unique observations.

There is a clear, weekly increase, from Sunday to Saturday, of small trucks entering Cap Haïtien. This weekly increase of small trucks entering the city suggests a predictable mobilization of charcoal trucks in the broader region of the

Northern Corridor on Mondays, with a general movement toward the Cap Haïtien throughout the week, peaking on Saturday before declining significantly by

Sunday.

8 (Tarter et al. 2018)

Introduction

Northern Haiti

Northern Haiti is well-known in contrast to the rest of the country for regional differences in flora and fauna, lexicon and syntax of Kreyòl (Haitian Creole), land tenure regimes, religious practices, and local governance structures—to name a few of the better-known differences.

Historically, most wood-fuel and charcoal studies in Haiti have focused on the capital city of Port-au-Prince and other regions; there is a paucity of concurrent data for the northern corridor of Haiti.

Even the recent nationwide study published by the World Bank (Tarter et al. 2018) had limited coverage in the northern regions of Haiti. The lack of reliable data about woodfuel consumption in the north of Haiti is notable, considering the area is marked by the most-extensive expanses of mangroves, multiple watersheds, some of the most tree- and forest-covered areas of the country, and is located just over the border from a large national park in the Dominican Republic.

Cap Haïtien is the former capital of colonial St. Domingue, and historically was the second largest city in the country. That place has since been subsumed by the city of Gonaïves, rending Cap

Haïtien to the third most-populated city in Haiti.

Knowledge about which areas of the north of the island are supplying the charcoal consumed in

Cap Haïtien, and by what percentages, is crucial for effective programming that attempts to mitigate tree cutting and other activities related to the promotion of more sustainable wood management practices. Such activities are of significant interest to the U.S. Government, as evidenced by ongoing United States Congressional earmarks for reforestation and tree-planting activities, as illustrated in the Haiti Reforestation Act of 2016 and the expanded Haiti and Armenia

Reforestation Act of 2018.

Fuelwood Use in Haiti

Charcoal in Haiti is largely produced in rural areas and transported to urban areas for consumption.

Fuelwood9 (firewood) in Haiti is also largely produced in rural areas but consumed domestically by rural households—which utilize tree branches, unproductive trees, cleared arboreal fallow, trees-on-farms, or the same coppice that is typically transformed into charcoal—to prepare household meals. The ratio of firewood to charcoal use is not well established, but is believed to follow population dynamics, and is thus believed to follow the rural-to-urban population divide in

Haiti.

Firewood is also used by bakeries, drycleaners, and distilleries, although there is a paucity of reliable figures on the present-day number and extent of these businesses. A World Bank report from over a decade ago referenced a study from the 1990s indicating some 1,300 bakeries operate in Haiti, and another study in 2001 suggested the country has some 170 laundry (dry cleaning) facilities (ESMAP 2007: 10). Finally, some industrial production enterprises, such as vetiver oil

9 We retain the distinction employed by FAO that ‘wood-fuel’ is a term that encompasses fuelwood (firewood and other woody biomass) and charcoal.

and sugarcane distilleries use wood to power their operations, although these appear to be on an historical decline.

Remote sensing methodologies in combination with land-based consumption models would be most appropriate for measuring firewood use across rural Haiti.

The research and methodology here and in the aforementioned nationwide study focus on urban-bound charcoal, which contributes comparable data by replicating a method and that has been utilized in charcoal estimations for Haiti over several decades. We also focus on charcoal production and consumption because recent shifts in rural-urban population dynamics and a large rural out-migration into the global diaspora has resulted in a clear urban majority population in present-day Haiti. While the dynamics of wood-fuel and charcoal use in the northern corridor of

Haiti may vary slightly from national trends, they are understood to be generally consistent with national trends.

Building on recent findings

The research presented here builds on the findings of a recently-published nationwide assessment of charcoal production and consumption trends in Haiti (Tarter et al. 2018), financed by the World

Bank Group and the Haiti Takes Root initiative of J/P HRO. That report addressed several important and unresolved questions concerning charcoal production and consumption in Haiti, including: (1) How much charcoal is consumed annually in the capital city of Port-au-Prince; (2)

Which geographical regions (including the neighboring Dominican Republic) produce the charcoal consumed in the capital; (3) How do these production areas variably supply charcoal; and

(4) In what ways have the charcoal production areas supplying Port-au-Prince changed over time?

The aforementioned study considered the capital city of Port-au-Prince—containing approximately one-fourth of the entire population of Haiti—as a ‘magnet’ city, used to estimate the incoming flows of charcoal produced in other locations throughout the country. In the methodology of that study, teams of enumerators were stationed at twenty-three different enumeration stations that were strategically located at key roads, intersections, and maritime wharfs across the country, where they both counted and categorized charcoal trucks in-bound toward the capital. This approach permitted researchers to express the total amount of charcoal consumed in the capital as respective percentages with production origins in different regions of the country, including the neighboring Dominican Republic. Since the enumerators sampled during three different high or low periods of charcoal production, the researchers were also able to produce annual estimates based on known seasonal fluctuations of charcoal production across the year. Annual estimates were established based exclusively on ‘core’ flows of charcoal that entered the capital, and then extrapolated out to the entire country of Haiti using a ratio of the total metric tons of charcoal consumed by the population of Port-au-Prince. This tons-to-population ratio was then applied to population estimates for the other major cities and urban areas throughout Haiti to arrive at national charcoal consumption estimates.

The present study replicated aspects of the national study methodology, principally the enumeration of charcoal produced regionally, using Cap-Haïtien on the northern coast of Haiti as the principal urban destination of charcoal. However, while the nationwide study was concerned http://documents.worldbank.org/curated/en/697221548446232632/Charcoal-in-Haiti-A-National-Assessment-of-Charcoal-Production-and-Consumption-Trends http://documents.worldbank.org/curated/en/697221548446232632/Charcoal-in-Haiti-A-National-Assessment-of-Charcoal-Production-and-Consumption-Trends with estimating national trends, the present report is focused more on regional trends and associated implications for the northern corridor of Haiti.

Complementary contributions of this research

This study also provides complementary outcomes beyond addressing the specific research questions outlined subsequently:

● The research may provide valuable information for future and contemporaneous efforts beyond the USAID Haiti Reforestation Project—including efforts of other governments, donors, bi- and multi-lateral development institutions, NGOs, and

Haitian civil society—toward understanding the complexity of wood-fuel production and consumption in northern Haiti;

● The replication of the methodology used for the earlier national study can be used to predict charcoal consumption in Cap Haïtien, and then be compared to the final counts from this study, to determine how effective the national study data are in predicting the consumption trends of individual urban centers;

● The data from the national study also permits an assessment of the tons-to-population ratio used to estimate charcoal consumption at the national level in

Haiti, by extrapolating results from a new urban center; and

● The data from the northern study, combined with the data from the earlier national study, establishes a complete, nationwide dataset for Haiti, spanning 15 months.

Research Questions

While these results are complementary to the research, the primary objectives of the research are addressed through the following research questions:

(1) How much charcoal is consumed annually in Cap-Haïtien;

(2) Which geographical regions produce the charcoal consumed in Cap-Haïtien;

(3) How do these production areas variably supply charcoal;

(4) How much of the charcoal consumed in Cap-Haïtien originates in mangroves; and

(5) What percentage of charcoal originates from the neighboring Dominican Republic?

Methodology

Principal Method

The principal method of data collection for this study is parsimonious and replicates the method of the national study (Tarter et al. 2018): the enumeration and classification of charcoal transport vehicles (trucks) entering the city of Cap Haïtien and/or passing through important crossroads and roads en route to the city. In preliminary scoping trips to the regions, the team determined that charcoal transport by maritime route directly into the city was not occurring on a large enough scale to merit a wharf enumeration station.

Enumerator Training

Prior to the commencement of data collection, the research team leader (third author) organized and facilitated a one-day training that provided an overview of the study rationale, methodology, location of the enumeration stations, a survey schedule, data collection standards, methods of data quality control, and communication and security plans.

Several modules in the training, discussed below, instituted a system of accountability to ensure enumerators were at their stations at the required times, and reliably classifying truck sizes, locations, and time of vehicle passage.

Team-building

At the onset of the training, enumeration teams were formed, and each team was assigned three enumerators. The training provided team-building and morale-boosting activities that emphasized collaborative teamwork; in order for research to be successful, each person had to participate, adhere to cooperatively developed (and sometimes adaptive) schedules, and follow protocols to meet the data collection requirements.

Enumerators worked to develop overlapping schedules, which served several complementary roles, including: as a security function to ensure there were always two enumerators present; to permit 24-hour enumeration at each station while still allowing for meals, restroom breaks, and sleep schedules; and to act as data quality mechanism for verifying the categorization and accurate recording of relevant details concerning passing vehicles.

Vehicle Typology

Given variances in the size and carrying capacity of different trucks, a typology of vehicles was created. The typology of vehicles was based on a simple, tripartite classification scheme, (small, medium, or large trucks) developed and field-tested during the national charcoal study in Haiti, funded by the World Bank (Tarter et al. 2018). A series of ‘carrying capacity ranges’ were also established for each of these three truck categories during the national study, and were based on trucks at full capacity with large bags of charcoal (ibid.).

This classification method permits all subsequent data to be analyzed as: (1) counts of the different categories in the tripartite vehicle typology; and (2) total weight based on the average (midrange) of the metric ton ranges for each type of truck, as displayed in Figure 1.

FIGURE 1. ENUMERATION STATIONS FOR CAP HAÏTIEN CHARCOAL STUDY

VEHICLE

SIZE

VEHICLE CARRYING CAPACITY

(at full)

TRUCK WEIGHT RANGE (metric tons)

Low-end range Mid-range High-end range

Small Vehicle 25-50 (16)10 0.75 1.125 (.48)11 1.5

Medium

Vehicle

100-200 3 4.5 6

Large Vehicle 300-400 9 10.5 12

Data Collection

Enumerators were trained to uniformly record the type of vehicle (based on a tripartite typology), the direction the vehicle was traveling from, and the time of the observation. Enumerators recorded entries on enumeration sheets, and also by photographs shared in real-time through group chat platforms (discussed subsequently).

At roadside stations that were not intersections, the enumerators were only required to record the category of the truck and time of passage. At intersections, enumerators were required to record the category of truck, time of passage, and location of origin.

Data Quality Control

Robust data collection was a major theme of the enumerator training. Beyond the rigorous training on how to properly record data, several mechanisms were set in place to ensure ongoing quality data collection, and are discussed below.

WhatsApp Groups

During the training, every single enumerator was assigned to and trained on how to utilize two different mandatory WhatsApp12 groups, each with a unique function.

The first WhatsApp group was used by enumerators check in and out of their predetermined shifts, which allowed the research team leader to monitor the progress of data collection by ensuring that all enumerators were adhering to their schedules and were at their stations at the appropriate times.

The first WhatsApp group also acted as a security monitoring mechanism—any lapse in reporting shift changes might signal a potential security situation, although no situations occurred during data collection. The first WhatsApp group was also used to disseminate any pertinent information

10 See section entitled ‘Adjusting the Carrying Capacity for Small Trucks’ under methodology, for an explanation of the adjusted carrying capacities (in parentheses) used for small trucks in the results of the study.

11 See section entitled ‘Adjusting the Carrying Capacity for Small Trucks’ under methodology, for an explanation of the adjusted carrying capacities (in parentheses) used for small trucks in the results of the study.

12 WhatsApp is the world’s largest, free, chat phone application, which permits the sharing of messages, images, and video conferencing.

that should be shared with the larger team (e.g., impending weather conditions, security concerns, etc.). Finally, the first WhatsApp group promoted coordination and communication among enumerations teams, fostering a spirit of collaboration and boosting overall morale.

The use of the second WhatsApp group was mandated by all on-shift enumerators to share the images of passing trucks that they were instructed to photograph, alongside their classification of the vehicle within the tripartite categorization scheme. The second WhatsApp group served several complementary functions, including: a backup system for data collection—complete with an image of the truck, displaying the direction it emerged from (for intersections) and an automatic timestamp—in the event that enumeration sheets were lost; verification that enumerators were at their posts and actively enumerating; and a system of ongoing inter-coder reliability assessments, whereby other enumerators at other stations could contest a mis-categorization of the vehicle size.

Sharing photographs also alerted enumerators at subsequent stations on the same road that a truck would soon be passing, and facilitated the tracking trucks from one station to the next. In this sense, the second WhatsApp group also fostered a spirit of collaboration and helped to break up long stretches of time by keeping enumerators actively engaged in the data collection process.

Surprise Site Visits

The research team leader (third author) also made a series of unannounced, surprise site visits during both sampling periods. During each surprise visit, he took a team photograph with the enumerators present at the station and this image was shared in the first WhatsApp group. The visits evolved into a fun, team-building activity, as enumerators speculated when and where the research team leader might show up next, and made largely unsuccessful attempts to track or anticipate his sporadic movements. These surprise visits and speculation on the group chat platform provided motivation for teams to stay engaged and at their station in case they should be the recipient of such a visit.

Every single enumeration station received at least one surprise site visit during the course of data collection across two sampling periods. Because of the relatively large geographic scope of the study, the managing partner of the research logistics firm (fourth author) also participated in this activity and conducted more-frequent field visits to the core stations that controlled for flows of charcoal entering the city of Cap Haïtien.

Sampling

The design of a proper sampling strategy was crucial to the collection of robust data that would permit reliable, subsequent extensions of these data to regional and national estimates of charcoal production and consumption. The sampling strategy considered both temporal and spatial considerations, discussed below.

Temporal Sampling

The costs associated with primary education are widely-recognized and predictable periods of bi-annual financial expenditure for most rural Haitian families. While charcoal is increasingly produced year-round in Haiti, production increases prior to the commencement of the first semester of the school year around September, and a second time in preparation of the start of the second semester in January, in anticipation of school enrollment fees and material costs.

School commencement dates are not concretely fixed and start dates may vary between public and private school systems. Haitian public-school commencement periods may also vary annually based on the receipt or delay of government funding.

Rural Haitians also harvest trees for charcoal production during droughts. In such instances, charcoal production intensifies when it becomes apparent that seasonal rains are insufficient to produce a marketable crop, although droughts are regionally variable due to unpredictable rain distribution patterns (Tarter 2015).

Despite these annual variances, at least three intervals of time are known periods of high charcoal production across rural Haiti: (1) August/September; (2) December/January; and (3) May/June.

The first two periods correspond to the payment to educational expenses, and the latter corresponds to agricultural downtime associated with the dry season. It is also generally accepted that the first range is the highest production period, but regional differences should not be discounted. The remaining intervals of time throughout the year are considered as lower charcoal production periods.

The strategy for sampling in the northern area of Haiti around Cap Haïtien was designed to capture these temporal fluctuations in charcoal production periods. Enumerators sampled: (1) one week

(September 2-9, 2018) during the high season of charcoal production; and (2) one week (October

21-28, 2018) during the low-production season.

Both of these sampling ranges overlap two of the three low and high seasons sampled in 2017 in the nationwide study (Tarter et al. 2018), which permit for the control and comparison of annual differences, and will allow a more seamless collation and combination of datasets to produce a complete nationwide record.

Data collected during these two temporal sampling periods are subsequently compared to the nationwide data set to gauge similarities and differences between the two periods, and then to appropriately extrapolate data to make annual estimates13.

Spatial Sampling

Several principal roads connect the northern coastal city of Cap Haïtien with other regions to the west, east, and south. These main roads later split or meet other roads leading to regions more remote from the city. In order to adequately answer research questions 2-5, the strategic placement of enumeration stations at key roads and intersections was a crucial decision.

13 All figures are estimates, based on conservative mid-ranges on a variety of variables, including average charcoal bag weights, average truck carrying capacities, and the assumption of equal high and low charcoal production seasons across a year (12 months).

Prior to data collection, two initial scoping periods of approximately one week each were undertaken by the 3rd author, to determine the best locations to place enumeration stations. During this scoping period, it was determined that while mangroves contribute to charcoal production in the region, such charcoal does not enter the city in large amounts by maritime vessel, but instead by trucks on roads that trace long extents of the northern coast.

Ultimately, eight (8) enumeration stations were selected to control flows of charcoal into Cap

Haïtien from multiple directions, including the neighboring Dominican Republic (see Figure 2).

FIGURE 2. ENUMERATION STATIONS FOR CAP HAÏTIEN CHARCOAL STUDY

STATIONS INTERSECTIONS

1. Ennery NR#1 and DR 306

2. Limbe (new/revised core station)14 NR#1 and DR 14

3. Barrière Bouteille (original core station) NR#1 (Route principale)

4. Quartier Morin (core station) NR#3 and NR#6

5. Limonade NR#6 and Trou du Nord (Old NR#6)

6. Ouanaminthe NR#6 and DR 602

7. Grande Riviere du Nord NR#3 and Route Grande Rivière du Nord

8. St Raphael NR#3 and Route St Michel de L’Attalaye

Seven of the eight enumeration stations displayed in Figure 1 were located at key intersections that controlled for a total of nine different roads (National Roads #1, #3, and #6; Departmental

Roads #14, #306, and #602; and several smaller roads including Route Trou-du-Nord [Old

National Road #6], Route Grand Rivière du Nord, and Route St. Michel de L’Attalaye).

Enumerator teams located at such intersections recorded flows from two directions of origin. Map

1 displays the locations of these eight enumeration stations across the northern corridor of Haiti.

Data Transcription and Standardization

All data were transcribed into data matrices and standardized when necessary. Most standardization reflected slight differences in how data were initially transcribed. For instance, some enumerators noted the highway of a passing vehicle as ‘Route Nationale 3’, ‘RN#3’, ‘RN3’, or other slight variations, which were standardized to facilitate data sorting and analysis. Likewise, there was an instance in which the time of truck passing was not recorded, so the midpoint of the previous and the next truck recorded were taken.

14 See section entitled “Delineating Core and Periphery of the Metropolitan Area” for an explanation of the new core station.

Map 1. Location of Eight Enumeration Stations

For a period of approximately 30 hours spanning two dates (Monday, 9/3 & Tuesday, 9/4) in

Sampling Period 1, traffic at the Limbe enumeration station slowed from the direction of Route

Nationale #1 (RN#1) due to political protests. This same time period of hours and days was examined in Sampling Period 2, and it was noted that even fewer trucks passed, despite the lack of protests. When Periods 1 and 2 were both compared, excluding the 30 hours affecting RN#1 during Period 1 (high season), the overall flow of trucks through Limbe from RN#1 in Period 1 was 75% higher than the overall flow of trucks through Limbe in Period 2 (low season). An application of this ratio—reflecting different seasonal flows—to the observed values on Monday and Tuesday in Period 2 and Period 1 would have added only 1-3 new data points (1-3 trucks) to the dataset, respectively. Given that the weight of observed trucks between Monday and Tuesday of both periods approached the 75% difference noted for the other days of the week, when considering weight, no interpolation of missing data were deemed necessary.

Delineating Core and Periphery of the Metropolitan Area

The dataset was constructed to permit analyses of both ‘core’ and ‘periphery’ enumerations in the northern corridor. Initially, two enumeration stations were considered as the core stations to control charcoal entry into Cap Haïtien from the west (Barrière Bouteille) and from the east (Quartier

Morin) (see Image 1).

Image 1. Original and Adapted ‘Core’ Enumeration Station to the West of Cap Haïtien

In order to capture urban areas west and south of the Barrièrre Bouteille enumeration station (see

Image 1) and newer urban areas in Vaudreuil (see Image 2) a post data-collection decision was made to use the Limbè enumeration station—located approximately 14.5 kilometers (9 miles) by road to the east of our delineation of the Vaudreuil area—as the new western ‘core’ control point of charcoal entering the Cap Haïtien metropolitan urban area (see Image 2, below; and Figure 2, above).

Image 2. The Vaudreuil urban area added to the larger metropolitan area of the study.

Delineating Core and Periphery of the Metropolitan Area

With the core stations controlling charcoal flows into the Cap Haïtien metropolitan (urban) area established, we then estimated the entire population of the metropolitan area. Since charcoal is consumed principally in urban areas, we started with the entire urban population for the

Commune15 of Cap Haïtien, which was estimated by the Government of Haiti at 269,036 persons

(IHSI 2015: 35). The urban estimate of the commune includes the City of Cap Haïtien, composed of the following two Section Communale:

● 2e Section Haut du Cap, which adjoins Cap Haïtien to the south and southwest, with an urban population of 149,893 (IHSI 2015: 35); and

15 The Commune is an administrative unit that is a carry-over from the colonial period; a series of Section Communale, including the Grande Ville or La Ville (historical city and typically the present-day downtown area) make up a Commune.

● 1re Section Bande du Nord, which adjoins Cap Haïtien the west and northwest, with an urban population of 21,101 (ibid.).

The total population of the City of Cap Haïtien (pop. 170,994) also includes the adjoining Quartier de Petite Anse (3e Section Petite Anse), located due east of the Rivière Mapou, with an urban population of 98,042 (IHSI 2015: 35), bringing the total urban population within the Commune of

Cap Haïtien to 269,036 (ibid.).

In order to include the urban population of the neighboring Vaudreuil area to the southwest of the

Commune of Cap Haïtien, we consulted population densities from a recent World Bank report, which stated:

Outside of Cap Haïtien, population is relatively higher along Route Nationale 1 from Vaudreuil to

Moustique to the southwest, in Quartier Morin, Limonade, and Trou-du-Nord to the southeast, and in

Milot to the south. Arguably, most of those regions are not part of the metropolitan area of Cap Haïtien.

Population density in those areas fluctuates between 500 and 1,000 people/km2 and is particularly higher in Vaudreuil and Trou-du-Nord. Outside of those satellite regions, density drops dramatically, reaching below 500 people/km2 (Lozano-Gracia and Lozano 2017: 126).

Since the authors quoted above note that population density is ‘particularly higher in Vaudreuil’, we took a conservative estimate (the higher end of their range) of 1,000 people/km2 and used satellite imagery to estimate the urban area of Vaudreuil (6.3 km2) at 6,300 people. Thus, the final population estimate for the broader Cap Haïtien metropolitan area used in this study is 275,336 people.

Adjusting the Carrying Capacity for Small Trucks

The initial analysis of vehicles indicated that small trucks dominated counts in both sampling periods. This result was notably different from the nationwide charcoal study, where large trucks were the most prevalent and small trucks composed only 38.77% of all unique observations (Tarter et al. 2018: 15). Conversely, in Cap Haïtien, small trucks dominated by approximately 71% of all vehicles observed in the core and periphery in Period 1, and by 77% in Period 2. Both the Cap

Haïtien and the nationwide study showed slightly more small trucks during the low season.

The predominance of small trucks in approximately 75% of the observations across the combined sampling periods prompted the authors to revisit the data, particularly the photographs of small vehicles taken by enumerators and shared in the WhatsApp groups.

The photos revealed widely-ranging differences across small trucks, with some used exclusively for carrying charcoal, while others were used for a varied mix of transportation for people, other types of produce, and charcoal bags (see Image 3). This phenomenon was markedly different from the national charcoal study (Tarter et al. 2018), where small trucks entering the core of the capital were filled with charcoal.

Image 3. Examples of differences in charcoal loads of small trucks.

Cognizant that widely-ranging charcoal loads of small trucks might affect the end results and subsequent extrapolations, the research team took action to appropriately adjust the dataset. The team randomly selected a subset of photographs of small trucks from both sampling periods in Cap

Haïtien, representing approximately 7.5% of the total number of small trucks that were observed entering the core metropolitan area of Cap Haïtien. A few photographs were removed due to poor image quality or because the image failed to capture the entire vehicle.

The remaining photographs were assigned to three parties (including the first author; an intern of the data collection firm who was closely involved with the entire study; and a champion enumerator who participated in both sampling periods) who individually estimated and recorded the number of charcoal bags of the small truck in each photograph. All three individuals produced averages that were within approximately 1.5 charcoal bags of each-other (15.6, 16, and 17.1), with a shared average of 16.22 and a standard deviation of 11.3.

When all the data were bifurcated into categories above and below the shared average, the inter-rater reliability (IRR) was 89.3%. The shared average was applied using 30kg per bag of charcoal to arrive at .48 metric tons as the midpoint carrying capacity for small trucks used in this study16.

All future calculations of metric tons use midpoint estimates for large, medium, and small truck carrying capacities.

Data Analysis

Differences Between Vehicle Flows Between the Two Sampling Periods

A total of 2,455 unique observations of vehicles were made across the core and the periphery stations during both sampling periods, with 1,326 observations in Sampling Period One (high season) and 1,129 observations in Sampling Period 2 (low season), making unique observations of vehicles in Sampling Period 1 17.45% higher than those of Sampling Period 2.

When the Barrière Bouteille enumeration station was removed from the dataset to adjust for the new core station17, the total of unique observations of vehicles in the dataset was slightly reduced to 2,367, with 1,270 observations occurring during Sampling Period 1 and 1,097 observations occurring in Sampling Period 2 (see Figure 3), reflecting known high and low charcoal production seasons, respectively.

In terms of the distribution of these vehicles across the three categories of the truck typology, small trucks dominate in both sampling periods in the core and periphery (see Figure 4).

16 This adjustment is registered in parentheses in Figure, which references this section of the report.

17 See section above entitled “Delineating Core and Periphery of the Metropolitan Area”.

However, since trucks were inevitably counted more than once along some of the same routes, these figures are misleading without considering flows into the metropolitan (urban) core of Cap

Haïtien separately from flows in the core and the periphery.

Differences Between Vehicle Flows in the Core and the Periphery

In Figure 5, where vehicle counts are displayed disaggregated by sampling period, by core and periphery, and by size, three trends become apparent: (1) periphery and core counts in Sampling

Period 1 are higher across the board, for all sizes, than counterparts in Sampling Period 2; (2) core

1 2

C o u n t

Sampling Period

Figure 3. Total Unique Observations of Vehicles

Total

119 96

1 2

O b se rv a ti o n s

Sample Period

Figure 4. Truck Sizes by Sample Period

Large

Medium

Small counts are higher than periphery counts in both sampling periods; and (3) small trucks dominate both the core and the periphery, in both sampling periods (see Figure 5).

Differences in Overall Flows of Metric Tons between the Two Sampling Periods

The 2,367 unique observations of vehicles noted in the preceding sections registered 4,986.9 metric tons of charcoal when applying the midrange tonnage estimates for small, medium, and large charcoal trucks. To correct for double-counting of the same vehicles, we considered flows into the metropolitan (urban) core of Cap Haïtien and flows in the periphery separately.

Differences in Metric Tons in the Core and Periphery

In Figure 6, Sampling Period 1 remains the higher period, but exaggerated differences between the number of vehicles is significantly reduced when considering metric tons, as a function of numerous counts of small trucks (Figure 5) averaging out with the larger carrying capacities of medium and large trucks.

L ar ge

M ed iu m

Sm al l

L ar ge

M ed iu m

Sm al l

L ar ge

M ed iu m

Sm al l

L ar ge

M ed iu m

Sm al l

Periphery Core Periphery Core

1 2

C o u n t

Sampling Period

Figure 5. Vehicle Flows Disaggregated by Core/Periphery, Sampling Period, and Size

Total

When considering charcoal entering into the core during Sampling Period 1 (1,485.48 metric tons) and Sampling Period 2 (1,379.75 metric tons), a combined 2,606.58 metric tons entered the metropolitan area of Cap Haïtien during the data collection periods. In both Sampling Periods, total flows of metric tons into the core were higher than those observed circulating in the periphery.

Daily Variation in the Core and Periphery

An examination of the daily flow of vehicles in the periphery and the core during both sampling periods (see Figure 7) reveals some general trends: (1) Friday and Saturday is the highest charcoal flow day in the core of both sampling periods; (2) Monday is the highest charcoal flow day in the periphery of both sampling periods; and (3) Sunday and Thursday are the lowest charcoal flow days in both the periphery and the core of both sampling periods.

1,349.64

1,030.68

1,485.48

1,121.10

0.00

200.00

400.00

600.00

800.00

1,000.00

1,200.00

1,400.00

1,600.00

1 2

M e tr ic

T o n s

Sampling Period

Figure 6. Metric Tons in Periphery and Core

Periphery

Core

78.18

115.38

33.18

71.58

310.62

253.98 277.98

142.62

225.18

193.44

116.58

194.52

213.54 220.14

152.64

95.04 114 115.38

131.04

277.14 257.94

162.12

222.12

149.94

330.6

133.44

206.58 periphery core periphery core

1 2

Sampling Period

Figure 7. Daily Differences Flows of Metric Tons

Sunday

Monday

Tuesday

Wednesday

Thursday

Friday

Saturday

In Figure 8 (below), the trends noted in Figure 7 (above) are contextualized when considering the disaggregated flow of vehicles in the combined periphery and core during both sampling periods.

There is a clear increase of small truck activity across the week toward Saturday (see Figure 8).

The same generalized trends in Figure 8 remain—whether controlling for the core, the periphery or both.

Hourly Variations

To examine a wide range of timestamps across the dataset, all observations from both sample periods in the core and periphery were grouped into six intervals of four hours each, and approaches bi-modal distribution (see Figure 9). The period of with the highest traffic occurs from 4 pm to 8 pm, followed by 4 am to 8 am. This general trend does not change when controlled for core and periphery or for the two sampling periods.

30 39

41 32

73 68

Sunday Monday Tuesday Wednesday Thursday Friday Saturday

Figure 8. Daily Differences of Vehicles Entering the Core

Large

Medium

Small

Linear (Large)

Linear (Medium)

Linear (Small)

250.92

1237.8

553.26

835.92

1422.12

686.88

M e tr ic

T o n s

Time Intervals

Figure 9. Hourly Variation in Flows of Charcoal Tons

Midnight to 4 am

4 am to 8 am

8 am to 12 noon

12 noon to 4 pm

4 pm to 8 pm

8 pm to midnight

Daily and Hourly Variation

The highest intervals of charcoal flows in tons noted in Figure 9 (4 pm to 8 pm, followed by 4 am to 8 am) shift when considering the daily movement of all vehicles (small, medium, and large) in the core and periphery (see Figure 10). In Figure 10, noon to 4pm is the highest aggregated charcoal vehicle traffic period, for Wednesday, Friday, and Saturday, followed by 4 pm to 8 pm.

Across all days, traffic slows considerably from 8 pm to midnight, and even more so from midnight to 4 am

Regional Differences in Charcoal Supply to Cap Haïtien

General Regional Supply to the Core

Of the 2,606.58 metric tons that entered the urban, metropolitan core of Cap Haïtien during both sampling periods combined, 76.3% (1,987.98 metric tons) entered the city from the south/south-east, through the Quartier-Morin enumeration station; the remaining 23.7% entered from the south/south-west, through the Limbè enumeration station (see Figure 11).

12 15 7 10 6

75 71

94 93

54 56

113 113

19 16 19

Sunday Monday Tuesday Wednesday Thursday Friday Saturday

Figure 10. Daily Time Interval Differences In S, M, and L Charcoal Truck Traffic

Midnight to 4 am

4 am to 8 am

8 am to 12 noon

12 noon to 4 pm

4 pm to 8 pm

8 pm to midnight

68.28 16.74

305.4 171.9

850.26

660.42

261.54 272.04

Limbé Quartier-Morin Limbé Quartier-Morin

Sample Period 1 Sample Period 2

M e tr ic

T o n s

Sample Period

Figure 11. Metric Tons of Charcoal Entering the Core

RD14

RN6

RN3

RN1

West of Cap Haïtien

Further disaggregating the 23.7% (618.6 metric tons) of charcoal that entered at Cap Haïtien through the Limbé enumeration station, 86.3% had origins along Route Departmental #14, which heads north-west toward Port-Margot. Stated differently, 20% of the total charcoal entering Cap

Haïtien during both sampling periods passed on Route Departmental #14.

East of Cap Haïtien

Further disaggregating the 76.3% (1,987.98 metric tons) of charcoal that entered at Cap Haïtien through Quartier-Morin, over three-fourths (1,510.68) had origins along Route Nationale #6, which heads east from Cap Haïtien along Haiti’s northern coast toward the neighboring Dominican

Republic. Stated differently, 57.95% of the total charcoal entering Cap Haïtien during both sampling periods passed on Route Nationale #6.

South of Cap Haïtien

Approximately 18% (477.3 metric tons) of the charcoal entering Cap Haïtien during both sampling periods comes from Route National #3, which heads south away from the city.

Broader Regional Charcoal Production in the Northern Corridor

While the ‘core’ stations—and the immediate roads that join them at their intersections—control for charcoal entering the metropolitan urban area of Cap Haïtien, the ‘periphery’ stations can be used to estimate the regional origins of production zones of charcoal across the northern corridor.

Figure 12 displays, as percentages, the 2,380.32 metric tons of charcoal circulating in the…