The following observations were made on the implementation of the approach outlined above.
Observations from a community perspective:
Improved livelihoods and alternative livelihood strategies are the key drivers for community members to engage in structured forest monitoring.
Community members can acquire the necessary skills and knowledge to collect data, with minimal technical supervision, but with high investments needed in the
44 design, initial training, and
technical support / backstopping.
Communities cannot process the data and discuss the implications for different management options without substantial technical support.
Motivation to work and contribute to the project comes from the incentives provided, but also from exploring the potential for forest production.
Observations from the advisors’
perspective:
A high amount of technical support is needed to analyse data and interpret information.
Community-based forest monitoring is promising for different stages of the REDD+
process, including for the baseline and MRV (measurement, reporting and verification) processes, but it should be approached cautiously and within a context of different forest management options.
Observations from community level facilitators’ perspective:
The community level facilitators find themselves in a difficult position when facing government-approved inventory processes and methods that may be less appropriate and the pragmatic approaches being proposed as an outcome of the testing of protocols/methods under the project (discussed below).
Both experience with the methodology and community level facilitation skills are critical for their role in the project.
Boundary demarcation 3.5
During the community launch meeting half a day was spent identifying two potential trial blocks (one deciduous, one evergreen/semi-evergreen) to be used as training areas for all forest management techniques including inventory, harvesting and silviculture. A set of criteria were suggested (e.g. target sizes, presence of harvestable timber species, accessibility, overlap between village territories where possible to promote collaboration) and participants added others (e.g. avoidance of overlap with proposed shifting cultivation areas). After initial discussions, large scale laminated maps based on the 2005 aerial photos and 2010 Landsat imagery were laid out and provisional areas were marked with pens. It was agreed for a subset of the group to inspect the areas and select suitable natural boundaries.
The boundaries of the proposed areas were inspected by motorbike and on foot by the WCS team. A demarcation team (three people from each village) then walked the entire boundaries, recording waypoints on handheld GPS devices and
The deciduous block (east) covers 68.9 ha, of which almost all is deciduous. It contains good stocks of locally important timber species. It is bisected by a small seasonal stream that has been buffered according to FA guidelines and removed from the inventory area by GIS.
The evergreen/semi evergreen block (west) covers 264.6 ha. It contains patches
45 of good commercial timber, most notably
stands of the Grade I species Lagerstroemia calyculata (sralao in Khmer) but has evidently been quite heavily logged in the past, so it is expected to have lower stocks than some remoter parts of the CBPF area. It was selected because it is accessible, suitable for training purposes and was proposed by the communities.
Table 3.2 shows the composition of each block according to a JICA national forest cover assessment. The JICA assessment indicates some parts of the evergreen/semi evergreen training block are 'Bamboo and secondary forest'.
Observations during this survey show that this category corresponds quite well on the ground with heavily degraded patches smothered in vines and with a very low stock of trees (most presumably non-commercial). It is recommended that these patches be excluded from the proposed trial inventory as they are of low potential (unless there is great investment in silviculture) and very great logistical difficulty for inventory. The small area of deciduous forest should also be excluded, leaving 189 ha of area to be inventoried.
Table 3.2: Forest type according to the JICA national land-use cover map (2002)
Land Use Type (JICA 2002)
CBPF Deciduous Forest Training Area
CBPF Evergreen Forest Training Area Deciduous forest 68.9 14.8 Evergreen broad
leafed forest 0.00 189.5 Bamboo and
Secondary forests
0.00 60.3
Grand Total 68.9 264.6
Figure 3.1: Location of the two training areas in relation to the JICA 2002 land-use interpretation
Sampling design 3.6
3.6.1 Basic approach
The basic approach for determining carbon stocks in the project area is to derive carbon estimates from regular forest inventory data as prescribed under the FA Community Forestry (CF) Guidelines (which are an annex to MAFF Prakas 219).
Two kinds of modifications were considered by the project - protocols for measuring additional carbon pools/sub-pools and efficiency improvements.
46
Protocols for additional pools
The CF Guidelines do not require the measurement of dead wood, and do not specify exact protocols for bamboo, which was felt to be the only NTFP sub-pool with a potentially significant carbon stock.
These additions are discussed in the Carbon Pools section below.
Efficiency improvements
The CF Guidelines, while statistically correct, are unlikely to provide optimal statistical efficiency (that is, optimal level of precision for a given price). An inventory following the guidelines will be prohibitively expensive across the whole 10,000-plus ha of the CBPF forest area, whether for carbon alone or for all forest values. This issue has been raised a number of times in recent years (e.g.
Brofeldt (2009), Blomley et al. (2010)). It is also recognised as an issue by FA (E Payuan and Sok Srun, pers. comm.). It was therefore considered timely to test alternative methods.
Brofeldt (2009) gives a systematic account of the main issues and many smaller points of technique. In the current trials we chose to focus on the two major issues: plot size/number and estimation of tree heights. Plot design is discussed here and tree heights in a later section.
The FA regulations foresee 50 x 100 m operations over periods of five to 10 years, such large plots are not the ideal choice for covering larger areas as in the case of the Seima project area.
It was therefore decided to explore alternatives for the management planning inventory, a forest resources assessment that is intended to cover all operable areas within the 13,000 ha CBPF project area - especially as digital measuring equipment was used, meaning no demarcation was required. As this approach has already been tested, it was decided to test a further potential enhancement, the 6-tree method. Also called k-tree sampling, this approach was suggested due to its two major advantages of (1) allowing an automatic adjustment to tree spacing and (2) being far less time consuming since the outer perimeter of the plots do not have to be marked. This was considered a further major advantage in situations with dense understory vegetation.
K-tree sampling designs have an inherent bias by overestimating (up to 15%) plot volume since the kth tree is always located on the plot perimeter. The degree of bias depends on the pattern of spacing of the trees, which varies between sites. Various methods for partially correcting this bias have been suggested in the literature, but none is universally effective. Given the expected sampling efficiency of the method, and the fact that it is inventory designer; 6 is often found to be an optimal number and was chosen here.
The design for 6 trees is shown in Fig. 3.2.
47 In the EF, teams also recorded distance
and diameter at breast height (DBH) for the 7th tree, to allow us to test one of the methods proposed for correcting bias.
Figure 3.2. Plot lay-out of k-tree sampling