Wula Nafaa (WN) is a natural resource management program funded by USAID and implemented by International Resources Group. One of WN’s main goals is to put large forest areas under community-based management in partnership with the Senegalese Forest Service. Another is to market forest products in a sustainable way.
As these tasks have certain technical requirements, WN has signed protocols with various technical collaborators, of which USDA Forest Service is one.
In December 2006, the US Forest Service assisted WN to complete a step-by-step procedural guide on developing community-based forest management plans (PAFs). The guide is based on WN experiences in Tambacounda and Kolda Regions. It acknowledges Senegal’s legal requirement that the PAF “establishes the maximum standing wood that can be cut each year, as a function of the regeneration of the forest stands” (Code Forestier Titre I, Ch 2, R.17). This requirement is met by producing a table of volumes resulting from inventory fieldwork. Thus “carrying out an inventory” is one of the 13 steps in the guide. Below is the pertinent excerpt from the newly developed “Procedural Handbook for Natural Resource Management Plans for Rural Communities and Community Forest Management”.
The main objective of this mission was to flesh out Step 7 in the Procedural Handbook to reduce ambiguity and clarify how foresters, consultants, and decentralized administrative bodies can fulfill legal requirements in a timely way as they begin a more localized control and harvest of forests. The objective was accomplished through several activities: verification of how the SIEF was developed and how it is currently used; describing how the inventory system is extrapolated for new areas with few or no plots; describing how its results are incorporated into PAFs; conversion factors between calculated and true volumes; how one may use the inventory system for non-charcoal forest product quantification; and the appropriateness of the 8-year rotation prescribed for Tambacounda forest cuts.
FOREST INVENTORY PROCEDURES AND MAPS USED IN THE FOREST MANAGEMENT PLANS (PAFs)
DESCRIPTION OF THE SYSTEM IN USE
Overview of the “Système d’Information Ecologique, Forestière, et Pastorale” (SIEF)
Since Wula Nafaa (WN) started assisting with writing forest management plans (or PAFs) in 2005, it has depended on the Système d’Information Ecologique, Forestière, et Pastorale (SIEF) inventory software that was developed by the Programme de Gestion Durable et Participative des Energies Traditionnelles et de Substitution (PROGEDE). The software has a pastoral component and a forestry component. The forestry component is the one used to produce allowable cut tables for PAFs written with WN’s assistance. Therefore our investigations began with gaining a deeper understanding of how the system works, whether it is being used appropriately, and how it can be applied to forests and forest products targeted for new PAFs.
SIEF is based on 1999 orthophotography-based Yangambi classifications whose main forest strata are: woodland, shrub, tree, and wooded savannas, fallows, and riverine forest. The inventory is carried out using an established field method (described below) and data are entered into the SIEF. The software analysis of inventory data incorporates regression equations developed around 2003 for above-ground biomass volumes. Because the software outputs tables of volumes in specified areas, it can -- when put together with a cutting protocol -- fulfill the legal requirements in the Forestry Code cited above. The software also summarizes field data into tables used to make thematic maps of the forest in ArcView mapping/GIS software.
There are two consultants in Tambacounda who understand the use of the SIEF sufficiently to perform all the steps (detailed in Appendix B) that lead up to drawing management maps. The maps divide a forest’s area, based on per-hectare estimated wood volume, into a predetermined number of blocks with 8 parcels each (based on a commonly-accepted rotation age of 8 years). The availability of the consultants and their ability to use the GIS together with the SIEF software has sped up the management plan writing process for WN and given more credibility to the PAFs themselves.
Using the SIEF software requires a level of understanding and expertise that goes beyond the field technician level and is taught in two-week mini courses to university-educated foresters in the Senegalese Forest Service. Combining the SIEF output with the GIS use takes even longer, especially if the agents are uninitiated in GIS. Fortunately, each region has been allocated the needed computer and software for the GIS aspect of interfacing with the SIEF, which is just part of its greater mapping and inventory needs.
An undeniable characteristic of the SIEF software is that it is geared for charcoal-based management plans, as opposed to all-species, timber, or non-timber products. There has been an attempt to make it into a more inclusive, national inventory software by matriculating the original program (termed SIEF-1) into another complementary version (SIEF-2) that uses a broader Landsat-based vegetation-classification process and has more plots throughout the country.
In the following paragraphs we will describe how the field and office components of the SIEF works, and verify its effectiveness as used in the writing of PAFs by Wula Nafaa in its community-based forest management program.
In order to verify the effectiveness of SIEF, it is necessary to
review the sample and plot designs,
evaluate the volume equations, and even
review PROGEDE’s ability to perform supporting services to projects such as Wula Nafaa to ensure consistent and correct use of the SIEF system.
Below is a synopsis of PROGEDE’s work from a number of sources including a visit to their office and a discussion with Dr. Cheikh Dieng, the PROGEDE coordinator, followed by a review of Wula Nafaa project’s use of the SIEF program.
Special attention is given to the procedure used for stratification of forest types before fieldwork began, because the method of stratification affects the accuracy and precision of the reported average volumes per hectare. Since SIEF averages are being applied over thousands of hectares at a time, it is important to understand how these inventories were executed in order to assess the implications of extrapolating this data for the Wula Nafaa project.
Aerial photography was taken in 1999 at a scale of 1:30,000 over 1,165,000 ha in the Tambacounda – Kolda regions of eastern and southern Senegal. In 2001, another 130,000 ha over the forests of Diambour, Guimara, Nétéboulou and Thiéwal was procured. Both sets of photos were ortho-rectified. From these aerial photos, a vegetation polygon map based on the Yangambi classification system (woodland, shrub, tree, and wood savannas, fallows, and gallery forests) was developed across all lands.
Figure 1 shows Tambacounda and Kolda Regions, where the Wula Nafaa project operates, and the extent of the aerial photography used by PROGEDE.
Inventory work was concentrated in forests of interest to the Senegalese Forest Service and to PROGEDE (an energy project assisting in seeking sustainable fuel sources for urban centers). Some of these forests are already legally protected as classified forests; others are designated “community forests” with rough boundaries, ostensibly under the control of local decentralized government structures. Four of these forested areas were chosen for a pre-inventory in 2001 comprised of 247 plots used to estimate the variance of the basal area of forest populations distributed across the forested Yangambi classes of shrub savanna, tree savanna, wooded savanna, woodland, and fallow. The three savanna types had approximately equal numbers of plots; the jachère received the most and woodlands received the fewest plots.
After the variance was estimated in each Yangambi class, the number of additional plots required per class was calculated (1,050) based on the goal of estimating basal area within ± 10 percent of the average square meters per hectare at a 90% confidence level (Cheikh Dieng, 2006 and personal communication).
Within the WN work regions of Tambacounda (Tamba) and Kolda, a 1,000 meter by 1,000 meter grid of potential sample centerpoints was constructed, creating approximately 5 to 6 times the number needed for the required confidence interval. A subset of the intersections was selected without replacement by Yangambi classes across the photo-covered part of the Tamba-Kolda regions, exceeding the estimated total number of plots needed, over 1,200. Some of these additional plots were more concentrated than others, within Yangambi classes, in several forests of interest. These differences in plot concentration within the same Yangambi class must be taken into account to produce unbiased estimates across forests. Thus at this point in 2003, we believe that over 1,600 plots were measured in the field. There are differing accounts of how many pre-inventory plots were carried out; it varies between 246 and 273. See Figure 2. (Note: there were some 800 plots measured for pasture characteristics only; although these plots are contained in parts of the SIEF, the analysis here is restricted to plots used in forest wood volume calculations.)
A map of the Tamba region’s areas where the plot intensity was considerably higher than surrounding areas is shown in Figure 3: the greatest density is within Nétéboulou and Missirah forests.
In 2004, PROGEDE measured an additional (approximately) 1,300 forest and pasture plots in all regions of Senegal as part of a national survey. With the exception of the Bakor classified forest, these additional plots were outside of the extent covered by the aerial photography indicated in Figure 2.
Every plot in the SIEF is associated with two classification systems: one based on the Yangambi system, and one based on satellite imagery which is more generalized. The satellite classes are weak, somewhat rich, rich, and very rich.
It is not clear if the 2004 national inventory plots were pre-stratified with differing probabilities of selection, or post-stratified into satellite classes, as the Tamba-Kolda inventory surely was since it started out in Yangambi classes. These additional national inventory plots are only available in the SIEF-2 database.
The original Tamba-Kolda sample design (SIEF-1) kept Tambacounda and Kolda regions separate and stratified each by five Yangambi classes. The analysis for each region was completed on a stratum by stratum basis. This approach reduced the analysis to a series of random samples for each stratum within each of the two regions.
With the SIEF-2 program, the user can display data from only the Tamba-Kolda inventory (over 1,200 plots), from all the “bassins d’approvisionnement” (1700+ plots), or from all the inventories that are contained in the database including plots outside WN project areas (over 2,900 plots). Sub-population analysis is conducted by choosing Departments, Arrondissements, classified forests, or community forests. SIEF-2 can display information on a given forest and stratum, using all the plots in the chosen boundary/stratum combination to calculate mean values as in a simple random sample.
There are provisions in SIEF-2 for producing results for non-wood products, but these options currently do not work since the basic information is not available in the data base.
In Figure 4, one can see the different inventories that make up the total of over 2,900 plots in the SIEF-2 database.
Plot execution in the field
The initial plot used in the Tamba-Kolda inventory was a nested design with 4 subplot sizes for live trees: a 10-meter radius subplot for trees greater than or equal to 3 cm and less than 10 cm in diameter; a 15-meter subplot for trees greater than or equal to 10 and less than 20 cm in diameter; and a 20-meter subplot for all trees greater than or equal to 20 cm in diameter. The plot design was eventually changed to the simpler 3-subplot size design used in the national inventory (see box above). Still in use are the four 1-meter radius subplots for trees less than 3 cm in diameter located in the four cardinal directions straddling the inner 10-meter subplot (see Figure 5). All trees 3 cm or less in diameter at 1.3 meters above the ground are recorded by group tally by species.
Standing dead trees at least 3 cm in diameter are recorded on the entire 20 meter subplot. Down woody stems are also recorded if the mid-point of the length that is at least 3 cm in diameter is within the subplot. There are also stump and length criteria for dead wood to be entered on the inventory sheet.
Development of the volume regression equations
A subsample of 102 of the pre-inventory plots was chosen for destructive sampling for the development of volume equations. The first tree in each diameter class per targeted species was selected to cut. The volume of stems to a 10 cm top diameter limit was determined by cutting the tree into sections and using Smalian’s formula and summing the sections. For branches or stems less than 10 cm, they were cut into 1-meter lengths and tied into fagots. One fagot was chosen at random to immerse in water to determine the displaced volume, from which the specific gravity was calculated and applied to the rest of the weighed fagots. The Smalian’s and the weight-based volumes were added together to produce the total volume of above ground biomass for each tree.
One- and two- parameter models (dbh and total height) were fitted to the data to produce several predictive functions of volume, and that with the best fit was retained for the SIEF. A total of 439 stems covering 14 species were measured using these procedures (Diop, 2002). Since only one or 2 stems of Combretum nigricans were cut, it was combined with Combretum glutinosum. In the end, 13 valid volume equations were produced. They are incorporated into the SIEF so that when inventory field data are entered OR when specific plots are chosen, the volume estimates per hectare are automatically calculated along with the basal area and stems per hectare.
Establishment of permanent growth plots and research parcels
PROGEDE established 57 permanent plots throughout Senegal in different ecogeographical zones. Some of these plots (around 20) are in areas of intervention of interest to PROGEDE and Wula Nafa. The permanent plots are arranged in clusters of 4 subplots separated by 1km (see Figure 6). The standing living and dead wood and stumps were first measured in 2004. The more dense the vegetation, the more concentrated the clusters were.
The subplot centers were marked with cement markers and metal rods so they could be relocated with a metal detector and GPS.
he permanent plot protocol was set up in 2002 with the cooperation of the Centre National de Recherches Forestières (CNRF) and the Centre de Suivi Ecologique (CSE). Plots that were measured were incorporated into the SIEF.
There are also four fenced research-oriented parcels 20m x 40m set up within the charcoal producing areas in the vicinity of Tambacounda. A France-based researcher was to oversee the cutting of all stems within the parcels and all stems in a paired plot with no fence, and compare the rate of regeneration between the two cut areas. Although Wula Nafaa is not directly involved in these plots yet, it could take advantage of the setup (see recommendations).
The method of incorporating SIEF data into PAFs
The SIEF program is well accepted within Wula Nafaa and the Senegalese Forest Service. It has become essential to completing forest management plans (PAFs) since, by law, the management plans must include information on the volume of wood that can be cut each year on a sustainable basis. The volume to be cut is determined by a combination of measuring field plots, analyzing the data in the SIEF, and interfacing SIEF output with ArcView GIS.
The steps used to get SIEF output into the PAF can be summarized as follows (from the manual used by Canadian International Development Agency’s PAEFK project in Kolda):
Yangambi strata within a selected forest area are synthesized into a “Série de Production” map; using overlaid or borrowed plot data, total forest values are calculated from averages per hectare.
On the forest map, a fine grid composed of one hectare squares is created and intersected with the vegetation map containing the five Yangambi classes. (According to the manual, this step is the most complex and the heaviest.)
Each hectare square is assigned a volume that is the sum of (the average volume for each Yangambi stratum as calculated from all plots selected) X (the area of the square in each stratum) (This step requires multiple interactions with the GIS).
A management block of volume (total forest WOOD VOLUME) divided by (number of blocks foreseen) is defined. The block is created on the computer screen by adding or subtracting hectare squares with their associated individual volumes, until all the blocks have an equal number of cubic meters.
Each block is further divided into 8 parcels; one parcel a year is assigned to villages in charge of the block for harvest of wood for charcoal production. The parcel boundaries are formed in the same way as the blocks: clicking on hectare squares until they add up to one-eighth the volume c
STEPS FOR THE INCORPORATION OF SIEF DATA INTO THE PAF
ontained in the block.
CREATE THE NECESSARY DIRECTORIES AND SUBDIRECTORIES
CREATE A NEW PROJECT IN ARCVIEW
GET DATA FROM EXISTING THEMES: Forest boundaries, Roads, Hydrology, Villages, Inventory plot locations, Occup Sol, Mares; Clip all to one forest boundary and make a map
MAKE THE 3 SERIES OUT OF THE BASE MAPS: Agriculture (from OccupSol); Protection (Mares + Hydrol with buffers); Production (ForetClaire + Savanes Arbustive/Arborée/Boisée + jachères)
ANALYZE, SUMMARIZE AREA AND LENGTH DATA FROM SERIES MAPS
DISPLAY VARIOUS INVENTORY PLOT CHARACTERISTICS THAT DESCRIBE THE TERRAIN
MAKE A 1-HECTARE GRID FOR THE BLOCK AND PARCEL CALCULATIONS
DRAW OUT DATA ON YOUR FOREST FROM THE SIEF AND SAVE OR WRITE DOWN THE DATA TO ADD TO ACCESS: Stems/Basal area/volume per hectare living and dead; no. of plots; Total for forest and for utilization type especially Bois d’énergie
SYNTHESIZE AND INCORPORATE DATA BY STRATUM INTO FOREST POLYGON TABLES (for later estimation of stems, basal area, and volume in the forest, the blocks, and the parcels)
EXTRACT AND GROUP DATA FROM FOREST POLYGON TABLES FOR THE FOREST, FOR EACH STRATUM
DO CROSS-ANALYSES IN ACCESS TO GET STEMS/BASAL AREA/VOLUME BY DIAMETER CLASS AND BY STRATUM -- PER HECTARE, PER UTILIZATION
CREATE BLOCKS WITHIN FOREST BOUNDARY BY CLICKING 1-HA SQUARES AND APPLYING THE AVERAGE “SERIE PRODUCTION” VOLUME TO EACH (based on one criterion: total of forest volume of bois d’énergie divided by the number of blocks)
CONSTRUCT PARCELS SIMILARLY BASED ON BOIS D’ENERGIE VOLUME IN EACH BLOCK DIVIDED BY NUMBER OF PARCELS (=8 in Tamba; =12 in Kolda); MAP THE SEQUENCE OF EXPLOITATION WITH COLORFUL POLYGONS (Criterion: no parcel is connected to another being worked in the same year)
The following box further expands the subtasks involved.
An important recommended change to this process is to draw block and parcel boundaries together with community members rather than computerizing the entire process (see below). Then the average volumes per hectare can be simply applied to the number of hectares in each stratum as with the more complex grid-creation method.
Appendix B contains a more detailed description from the PAEFK manual of how the tables from SIEF output are prepared and placed in the management plan. Alternative block-and parcel-drawing procedure is given in more detail there.