Effects of conservation policy on China’s forest recovery

Comment on “Effects of conservation policy on China’s forest recovery”
Yong Pang1, Guangyu Wang2, Shari L. Mang2, Zhongqi Xu3, Hongbo Zhai4, Zhijian Yang5, and John L. Innes1*

1Institute of Forest Resource Information Techniques, Chinese Academy of Forestry, Beijing China
2Faculty of Forestry, University of British Columbia, Vancouver, BC, Canada
3College of Forestry Agriculture University of Hebei, Baoding, China
4Chinese Academy of Forestry Inventory and Planning, Beijing, China
5College of Forestry, Fujian Agriculture and Forestry University

Vina et al. (Reports, 18 March 2016) use MODIS VCF to assess change in forest cover in China in an attempt to verify government statistics regarding forest cover and to evaluate the effectiveness of the government’s Natural Forest Conservation Program (NFCP). However, they fail to verify their data against the government’s own data in a quantitative way, and their evaluation of NFCP ignores the five other forest conservation programs run simultaneously by the government. Furthermore, they use a limited number of parameters to evaluate success and rely on MODIS, which lacks the spatial resolution to address the authors’ objectives.

The authors aim to verify Chinese government statistics on forest recovery, an element they correctly highlight as being crucial in ensuring reliable and transparent information for conservation and climate change action; however, this verification is not performed in depth. Both parties conclude that forests have been recovering over the period of 2000-2010, but the authors do not quantitatively compare their findings against the government’s own statistics. Vina et al.’s analysis indicates that 1.6% of China’s territory experienced a significant increase in percent tree cover from 2000-2010, which is substantially lower than the value reported by the Chinese government. From 2003-2008 alone, the government reports an increase in total forest cover from 18.21% (1) to 20.36% (2) - a change of 2.15%. The 10-year change presented by Vina et al. is much lower than that reported by the Chinese government, but the authors do not acknowledge this disparity.

The authors relate their findings to the Natural Forest Conservation Program (NFCP). However, from 2000-2010, China had six key programs aimed at addressing forest loss and ecosystem degradation. Four of these programs overlap with NFCP’s implementation area – there is 82.4% overlap with Conversion of Cropland to Forest Program (CCFP) (referred to as Grain-to-Green in the paper), 29.4% with Three North Shelterbelt Development Program, 50.8% with Shelterbelt Development Program along the Yangtze River Basin (3Ns&YRB) (3), and 11.8% with Afforestation Projects in the plain areas (4). Although the authors acknowledge the other programs, they do not account for the substantial overlap in the analysis. Based on their methods, it is not feasible to evaluate the success of the NFCP in isolation from other government environmental programs as there are no steps taken to separate the source of forest cover change. This is of concern as the authors aim to “evaluate the effectiveness of… NFCP” and state that “… the implementation of NFCP exhibited a significant relationship with forest gain…”. Although NFCP contributed to increased forest cover, it is unreasonable to claim that the changes in forest cover are significantly related to NFCP as the contributions of each program are not evaluated. All programs carried out forest preservation and restoration activities between 2000-2010, such as logging bans and afforestation. However, the authors assume all changes are a result of NFCP activities when the program was responsible for only 18.7% of the area reforested by government programs from 1998-2010 (2). Furthermore, statements such as “NFCP seems to significantly contribute to carbon sequestration…” and claims that the dramatic increase in forest cover is in response to NFCP implementation undermines the contribution of other programs. If the authors wished to evaluate the NFCP, they should have focused on the locations where NFCP activities were undertaken and excluded areas where the other programs were active. As such, the primary objective to ‘evaluate the effectiveness of the NFCP at the national scale… and its contribution to changes in net primary productivity…’ is not achieved by this research.

The parameters used for the county-based spatial autoregressive model are insufficient to evaluate “the relation between forest loss and gain and the implementation of the NFCP”, which is the authors’ stated purpose. Although several industry- and demographic-related factors were used, several factors critical for an analysis at the county level are not included. To fully understand county level factors that relate to the NFCP and change in forest cover, the authors should have considered additional parameters related to local livelihoods and industry. The NFCP program led to the layoff and replacement of over 910,000 forest workers during 2001-2010 (5); as such, factors such as household income, government compensation, change in labor in various industries, and change in non-timber forest products (e.g. tea, fruit) need to be considered (6, 7). This analysis misses these critical elements, and thus is not robust enough to draw meaningful conclusions.

The authors’ assessment using Vegetation Continuous Fields (VCF) tree cover products derived from MODIS is a significant advance for forestry. However, MODIS lacks the spatial resolution to accurately evaluate changes in forest cover to the degree required to draw the desired conclusions. Forestry activities in China are carried out in management compartments that range from 1-3 ha in size. The spatial resolution of the MODIS data used is 6.25 ha; thus the analysis is unable to capture the diversity of activity occurring within a single pixel, reducing the accuracy of the perceived changes in forest cover across the landscape. Although the authors point out that evaluation using LANDSAT can often be hindered due to cloud cover, there is cloud-free LANDSAT data at 30X30m resolution available for the entire area of mainland China (8, 9). Analyses with these data would yield pixels that are 0.09 ha (smaller than forestry compartments), which would generate more accurate forest cover data compared to the authors’ analysis with MODIS.

Reference
1. State Forestry Administration, “Forest Resource Statistics of China, 6th Report” (China Forestry Publishing House, Beijing, China, 2005).
2. State Forestry Administration, “Forest Resource Statistics of China, 7th Report” (China Forestry Publishing House, Beijing, China, 2010).
3. G. Wang, J.L. Innes, J. Lei, S. Dai, S.W. Wu, China’s Forestry Reforms. Science, 318, 1556-1557 (2007).
4. China Sustainable Forest Management Strategy Research Task Force, “China Sustainable Forest Management Strategy Research” (China Forestry Publishing House, Beijing, China, 2003).
5. State Forestry Administration, “Forest Statistical Yearbook” (China Forestry Publishing House, Beijing, China, 2010).
6. E. Uchida, J. Xu, Z. Xu, S. Rozelle, Are the poor benefiting from China’s land conservation program? Environ. Dev. Econ., 12, 593-620 (2007).
7. E. Uchida, S. Rozelle, J. Xu, Conservation payments, liquidity constraints, and off-farm labor: Impacts of the Grain-to-Green Program on rural households in China. Am. J. Agric. Econ., 91, 70-86 (2009).
8. Global Land Cover Facility, “Landsat Imagery” (2014) (available at <http://glcf.umd.edu/data/landsat/).
9. GlobeLand30, “GlobeLand30” (2016) (available at http://www.globallandcover.com/home/Enbackground.aspx).