Inventory-based estimation of aboveground net primary production in Japan ’ s forests from 1980 to 2005

Introduction Conclusions References


Introduction
Net primary production (NPP) is defined as the rate of accumulation of organic matter by vegetation and equals the difference between carbon assimilated by plants through photosynthesis and the carbon consumed by plant autotrophic respiration (Chapin et al., 2002).NPP therefore represents the efficiency of CO 2 fixation by plants, determines the amount of materials and energy available for heterotrophic organisms, and represents the activity of vegetation in the global carbon cycle (Jenkins et al., 2001).Recent results from satellite remote sensing and carbon process models have suggested that NPP has increased in the middle and high northern latitudes since the 1980s (Myneni et al., 1997(Myneni et al., , 2001;;Hicke et al., 2002;Fang et al., 2003;Nemani et al., 2003;Piao et al., 2005) and that this increase was crucial for explaining the increased terrestrial carbon sink.However, regional NPP estimates based on significant field data are rare, especially over long time scales (Turner et al., 1999;Fang et al., 1996;Brown and Schroeder, 1999;Kurz and Apps, 1999;Jenkins et al., 2001;Brown, 2002).
Biomass-based NPP estimation has been considered an effective method to assess NPP variations on a national scale (Whittaker and Marks, 1975;Fang et al., 1996;Jenkins et al., 2001;Brown, 2002).Whittaker and Marks (1975) detected a linear relationship between biomass and NPP for the first time, and Fang et al. (1996) developed this method by establishing several allometric biomass-NPP relationships for major forest types and using the allometries to estimate NPP for China's forests.Jenkins et al. (2001), using plot-level forest inventory data, also reported a linear relationship between biomass and NPP for the mid-Atlantic region of the United States.A positive biomass-NPP relationship is useful for estimating NPP because biomass can be easily obtained from forest inventory data, which have become increasingly complete and available for many countries and regions in recent decades.
Japan has a typical marine climate with abundant rainfall, and two-thirds of its land area is covered with forest (Kira, 1991).Since the International Biological Program (IBP, 1965(IBP, to 1974)), biomass and NPP have been measured extensively for the main forest types in Japan (Fang et al., 2005).In addition, a systematic forest inventory has been conducted in Japan at about 5 yr intervals since 1947, and the surveys have intensified since the 1980s (Fang et al., 2005).These direct field measurements and systematic forest inventories make it possible to investigate the historical changes in NPP for Japan's forests.
In order to estimate changes in biomass in Japan's forests over time, we reviewed the research literature on Japan's forests and compiled a database for Japan's forest biomass (Fang et al., 2005).Based on this database and forest inventory data, we investigated changes in the biomass of Japan's forests from 1947 to 1995 in the previous study (Fang et al., 2005).In this study, we further complemented the database with NPP datasets, developed allometric relationships between aboveground biomass (AB) and aboveground NPP (ANPP) for Japan's major forest types, and investigated the changes in ANPP in Japan's forests from 1980 to 2005.

Data and methods
We estimated the ANPP of Japan's forests based on biomass-NPP relationships and data from forest inventories.Therefore, we used two datasets in this study: direct field measurements, including both biomass and NPP, and data from forest inventories.

Field measurement data
Since the IBP, studies of biomass and NPP have been conducted for Japan's major forest types.Cannell (1982) compiled a global database of forest biomass and NPP, including some data from Japan.Fang et al. (2005) developed a database of Japan's forest biomass that included 945 sets of data.We started with this database, then collected additional NPP data for Japan's forests from all the available literature to establish a new database.This new database contains the forest type, stand age, stand density, total basal area, stand volume, stem biomass, AB, belowground biomass, biomass of forest-floor vegetation, and 572 sets of NPP data.The NPP data generally included stem NPP (NPP S ), branch NPP (NPP B ), and leaf NPP (NPP L ), and some of them also include root NPP (belowground NPP, BNPP) or total NPP (TNPP = ANPP + BNPP).Unfortunately, there Examples of the relationships between the components of net primary production (NPP) and aboveground net primary production (ANPP).(a) The relationship between total net primary production (TNPP) and ANPP for C. japonica forests.(b) The relationship between stem net primary production (NPP S ) and ANPP for evergreen broadleaf forests.(c) The relationship between leaf net primary production (NPP L ) and ANPP for needle-leaf and broadleaf mixed forests.The relationships between these NPP components and ANPP for other forest types are shown in Table 1.   was insufficient longer-term BNPP data for us to reconstruct BNPP changes during the study period.We defined ANPP as the sum of NPP S , NPP B , and NPP L .All the NPP data were estimated by means of destructive sampling and were defined as the difference in biomass and litterfall between two survey years divided by the number of years between measurements, thereby providing the net biomass and litterfall change per year.In this study, we defined "biomass" as the standing crop (Mg ha −1 ).In this study, biomass and ANPP are all dry-matter based.

Calculation of ANPP from other NPP components
Since some of the original sets did not provide ANPP data and instead presented only data on NPP S , NPP L , or TNPP, we estimated the ANPP for each dataset using allometric relationships between the various NPP components and ANPP.We developed three allometric relationships to perform this analysis: (1) the TNPP-ANPP relationship, (2) the NPP S -ANPP relationship, and (3) the NPP L -ANPP relationship.Figure 1 shows typical examples of these relationships, and Table 1 lists the regression results for the sets of data that required this allometric approach.As shown in Table 1 and Fig. 1, strong and significant linear correlations were found between the NPP components (such as NPP S , NPP L ) or TNPP and ANPP, suggesting that our approach is an acceptable way to obtain ANPP data from studies that presented only NPP components.

Relationship between biomass and ANPP
Biomass is generally considered to represent the accumulation of plant organic matter during a plant's life cycle.During a forest's growth stages, ANPP tends to increase with increasing AB (Whittaker and Marks, 1975;Fang et al., 1996;Jenkins et al., 2001;Brown, 2002).Therefore, previous studies established the biomass-NPP relationship based on field measurements and used biomass to estimate the changes in NPP (Whittaker and Marks, 1975;Fang et al., 1996;Jenkins et al., 2001).In the present study, we used our new database to establish a series of relationships between AB and ANPP for Japan's main forest types.To document the contribution of various forest types to Japan's total ANPP, we classified the country's forests into 10 major types: Cryptomeria japonica forests, Chamaecyparis obtusa forests, Pinus forests, Larix leptolepis forests, Abies and Picea forests, other needle-leaf forests, Quercus forests, other deciduous broadleaf forests, evergreen broadleaf forests, and needleleaf and broadleaf mixed forests (For the details of the dominant species in each of the forest types, see Table S1).   1. Regression parameters for the relationships between aboveground net primary production (ANPP) and total NPP (TNPP), stem NPP (NPP S ), and leaf NPP (NPP L ) for Japan's major forest types, where a and b are the regression constants for each forest type.The units of all NPP components are Mg ha −1 yr −1 .All regressions were statistically significant (p < 0.001).broadleaf forests, other needle-leaf forests, and needle-leaf and broadleaf mixed forests.For these three forest types, we used the mean value from the field data for our further analysis of ANPP.

Forest inventory dataset
Although Japan's Forest Resources Statistics are available from 1947 to 2005, only those since 1980 are complete and report both the forest area and total timber volume for each age class and each major forest type by prefecture (for the 47 prefectures in Japan).The data were compiled from 10 000 statistically representative, adequately replicated, permanent sample plots across the country.The area of each plot is 1000 m 2 .In the present study, forest was defined as land with 20 % or more crown cover of government-owned forests and more than 30 % crown cover of community and privately owned forests.The data recorded included the forest group (planted and natural forests), owner, dominant tree species, age class, diameter at breast height, tree height, and stem volume.Biomass was estimated for each kind of forest from the timber volume.For a detailed description of forest inventories in Japan and the methods of biomass estimation, see Fang et al. (2005).

Results
Because of limitations in the earlier data from Japan's Forest Resources Statistics, we have focused on the changes in ANPP of Japan's forests only from 1980 to 2005.Table 3 summarizes the mean and total ANPP of Japan's forests for this period.The mean ANPP for all forest types combined averaged 10.5 Mg ha −1 yr −1 during the 25 yr, and ranged from 9.6 to 11.5 Mg ha −1 yr −1 (Table 3, Fig. 3).From 1980 to 2005, the mean ANPP increased markedly, by 1.9 Mg ha −1 yr −1 , with a mean annual increment of 0.076 Mg ha −1 yr −1 (0.79 %).
The mean ANPP of needle-leaf forests increased steadily throughout the study period (Fig. 3), with a mean annual increment of 0.10 Mg ha −1 yr −1 (1.12 %), giving a total increment of 2.6 Mg ha −1 yr −1 (Table 3).In addition, the mean ANPP of most types of needle-leaf forest increased.The largest increase was for C. japonica forests, with a total increase of 7.3 Mg ha −1 yr −1 over the 25 yr, followed by Pinus forests, L. leptolepis forests, Abies and Picea forests, and C. obtusa forests, with increases of 3.3, 2.5, 1.5, and 0.8 Mg ha −1 yr −1 , respectively (Table 3).Because the AB-ANPP relationship for other needle-leaf forests was not significant (Table 2), we used the mean value of the field data in Table 3.Therefore, it is difficult to determine the changes in mean ANPP of this forest type during the study period.
In contrast, the mean ANPP of broadleaf forests showed a more complex pattern of changes.ANPP declined from 10.9 Mg ha −1 yr −1 in 1980 to 9.7 Mg ha −1 yr −1 in 1995, then   broadleaf forests in Table 3 represent the mean values for other deciduous broadleaf forests and evergreen broadleaf forests (Table 2), the changes in this forest type were not visible during the study period.Therefore, the majority of the change of mean ANPP of broadleaf forests resulted from changes in the mean ANPP of Quercus forests.Moreover, we noticed that the forests in two categories, "other needle-leaf forests" and "other broadleaf forests", did not show a clear trend in temporal variation both in area and other variables, compared with other forests.A possible error may exist in the inventory data due to land-surface classification differing in different periods.Further attention to the two forest types should be paid in the future because of their significant area, biomass, and productivity.The total ANPP of Japan's forests (all types combined) averaged 249.1 Tg yr −1 during the study period and ranged from 230.0 to 271.4 Tg yr −1 (Table 3).Simultaneous with the increase of mean ANPP, the total ANPP of Japan's forests increased by 41.4 Tg yr −1 over the 25 yr, representing a mean annual increase of 1.66 Tg yr −1 (0.72 %).However, in contrast with the trends for mean ANPP, the total ANPP of needle-leaf forests decreased by 2.2 Tg yr −1 from 1980 to 2005 (a mean annual rate of 0.06 %), whereas the total ANPP of broadleaf forests increased by 33.7 Tg yr −1 (a mean annual rate of 1.62 %) during the same period (Table 3).For specific forest types, the total ANPP increased most for the C. japonica forests, followed by other broadleaf forests, Quercus forests, C. obtusa forests, and L. leptolepis forests, which increased by 34.2, 21.0, 14.8, 7.2, and 1.5 Tg yr −1 over the 25 yr, respectively (Table 3).The reductions in total ANPP were greatest for Pinus forests, followed by Abies and Picea forests and by other needle-leaf forests, with decreases of 9.9, 8.2, and 3.4 Tg yr −1 over the 25 yr, respectively (Table 3).

Discussion
Our analysis revealed different patterns of change in the mean ANPP between needle-leaf and broadleaf forests (Fig. 3).This can be attributed to differences in the ratio of planted and natural forests for needle-leaf and broadleaf forests.In Japan, most of the planted forests are needleleaf forests.Moreover, due to the economic benefits and the pinewood nematode, the large area of deforestation of natural needle-leaf forests happened in recent decades, making the ratio of planted to natural forests increase from 1.5:1 to 4.1:1 (Table S2).Compared to natural forests, the wellmanaged plantations exhibited rapid early growth, resulting in a significant ANPP increment in needle-leaf forests from 1980 to 2005.Conversely, more than 97 % of the broadleaf forests were natural forests (Table S2).The area of old natural forests with high biomass density decreased rapidly, accompanied by a rapid increase in the area of young secondary forests with low biomass density, thereby decreasing the mean ANPP of broadleaf forests.
Interestingly, the directions of mean ANPP change for needle-leaf and broadleaf forests were the opposite of the directions for total ANPP.Because total ANPP was estimated by combining the mean ANPP with the total forest area, the forest area did play an important role in this difference.During the study period, the area of needle-leaf forests declined by 3.716 × 106 ha (22.9 %), whereas that of broadleaf forests increased by 3.516 × 106 ha (46.0 %) (Table 3).
To compare mean forest ANPP in Japan with values from elsewhere, we summarized the mean forest ANPP for several key northern countries or regions in the Northern Hemisphere (Table 4).The overall mean ANPP of Japan's forests averaged 10.5 Mg ha −1 yr −1 from 1980 to 2005, and those of broadleaf forest and needle-leaf forest were 10.4 and 10.5 Mg ha −1 yr −1 , respectively.The mean ANPP of Japan's forests was more than two times the mean TNPP of China's forests (4.4 Mg ha −1 yr −1 ; Fang et al., 1996).Because TNPP equals the sum of ANPP and BNPP, the actual ANPP Biogeosciences, 8, 2099Biogeosciences, 8, -2106Biogeosciences, 8, , 2011 Figure1

Fig. 2 .
Fig. 2. Examples of the relationships between aboveground biomass (AB, Mg ha −1 ) and aboveground net primary production (ANPP, Mg ha −1 yr −1 ) for three of the main forest types in Japan.(a) C. japonica forests, (b) other deciduous broadleaf forests, and (c) Abies and Picea forests.Table 2 presents details of the regression analysis for all forest types.
Figure 2 presents typical examples of the relationships between AB and ANPP, and

Fig. 3 .
Fig. 3. Changes in mean aboveground net primary production (ANPP) of Japan's forests from 1980 to 2005: (a) needle-leaf forests; (b) broadleaf forests; (c) overall forests.The vertical lines extending from the top of boxes indicate standard errors of the means for different forests.
Table 2 presents the regression results for all of these forest types.Table 2 and Fig. 2 both show strong and significant (p < 0.05) relationships between AB and ANPP for most forest types, with the exception of evergreen

Table 2 .
Relationships between aboveground biomass (AB, x) and aboveground net primary production (ANPP, y) for the major forest types in Japan.SD = standard deviation.

Table 4 .
Estimates of mean aboveground net primary production (ANPP) for Japan's forests and those of other northern countries or regions.