DYNAMIC PATTERN OF CARBON BALANCE IN THE FORESTS OF FEDERAL DISTRICTS OF THE RUSSIAN FEDERATION

According to the data of the State Forest Registry and the archive materials of the State Forest Fund Account carbon budget of the forests of federal districts (FDs) of Russia was calculated for the years 1988–2015. The total carbon pool of the forested lands of Russia amounted to 123.77±18.93 Gt C as of 01.01.2015. The biggest contribution to the national forest carbon pool was made by Siberian (36.4%) and the Far Eastern (35.2%) FDs followed by the Northwestern (11.3%), Ural (9.3%), Volga (4.4%), the Central (2.8%), North Caucasian (0.3%) and the Southern (0.2%) FDs. Contribution to the national forest carbon sink (206.10±66.86 Mt C yr ) by districts: the Siberian district 39.3%, the Far Eastern 15.1%, the Northwestern 12.4%, Volga 12.1%, Ural 11.0%, and the Central 8.8%. Two groups of districts stand out in terms of the average value of forest carbon sink: 1) with 0.64–0.85 t С ha year (the Central, the Southern, North Caucasian, Volga FDs), 2) with 0.11–0.33 t C ha year (the Northwestern, Ural, Siberian, the Far Eastern FDs). The magnitude of the carbon sink in the forests of the FDs was at its lowest in 1988–1993. The reduction of losses due to felling in 1993–2000 resulted in increased carbon sink in the majority of federal districts. According to the level of this increase the forests of FDs can be divided into 2 groups: 1) with a significant increase in carbon sink (by 86% and more) – the Northwestern, Ural, Siberian, the Far Eastern FDs; 2) with a moderate increase in carbon sink (by 4–46%) in the Central, the Southern, North Caucasian and Volga FDs.

2017; Shvidenko et al., 2014;Dolman et al., 2013, etc.). After the adoption of the Paris Agreement this situation led to an increased discussion on the role of forests in the formation of the national greenhouse gas budget in the context of feasibility of reducing greenhouse gas emissions in industrial economic sectors. Some of the debating parties argue that there's no need for the Russian economy to reduce emissions since the Russian forests absorb more carbon than is emitted by the industry. It is only necessary to recognize this fact at the conference venues of the UN Framework Convention on Climate Change (UNFCCC) with possible introduction of the mechanisms of international compensations payment for carbon sink in Russia. According to the other position, anthropogenic emissions in Russia are higher than carbon absorption by its forests, so it is necessary to improve both the energy efficiency to reduce emissions and the management of national forests to increase carbon sink. Both points of view are corroborated by existing scientific publications.
The first point of view does not take into account the need to comply with a number of requirements when shaping national reporting on the greenhouse gas budget. These requirements are set out in the guidelines of the International Panel on Climate Change (IPCC) (Guidelines..., 2003). In particular, it notes that estimated carbon balance in the managed forests is derived from official forest inventory data using a very specific set of equations, the coefficients of which can be determined on the basis of the available national data. A significant part of regional and national assessments for Russian forests (Bobkova et al., 2015;Vaganov et al., 2005;Vedrova, 2011;Dolman et al., 2013) with respect to data generalization (eddy covariance etc.) is an extrapolation of a limited number of field studies to larger areas, which is inconsistent with the IPCC approach.
The RECBF system (Regional Evaluation of Carbon Budget of Forests) developed by the authors of this article (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013 is fully in line with the recommendations of the IPCC. This calculation system integrates the results of 2 decades of dwelling on forest-carbon issue in the Center for ecology and forest productivity of RAS, launched on the initiative of the member of the Academy of Sciences A. S. Isaev (Isaev et al., 1993(Isaev et al., , 1995Zamolodchikov et al., 2003Zamolodchikov et al., , 2005Zamolodchikov, 2009;Chestnykh et al., 1999Chestnykh et al., , 2004Chestnykh et al., , 2007Utkin et al., 2001, etc.).
Since 2010 the RECBF has been used in the National Greenhouse Gas Emissions Inventory for reporting about the forestry sector (National report..., 2017 and earlier). As part of the National Inventory the RECBF system is regularly examined by the experts of the UNFCCC. These tests are successfully passed which confirms that the system is robust and complies with the IPCC approaches.
The previously published findings obtained with RECBF considered either the territory of entire Russia (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013a(Zamolodchikov et al., , 2013b(Zamolodchikov et al., , 2017 or its territorial entities and smaller territorial units (Zamolodchikov, Ivanov, 2016;Zamolodchikov et al., 2018). The objective of this paper is to describe the spatial variability of forest carbon budget at the level of federal districts (FDs). This level, on the one hand, is sufficient to identify regional features of the forest carbon cycle determined by both natural conditions and forest management aspects. On the other hand, the results obtained for the FDs can be sufficiently detailed for scientific generalizations, as opposed to a more detailed level for the territorial entities of the Russian Federation.

MATERIALS AND METHODS
The RECBF system implies using the State Forest Registry (SFR) entries as input data. The SFR was introduced by the Forest Code of the Russian Federation (Forest Code, 2007), replacing the previous system of the State Forest Fund Account (SFFA). Fortunately, methods underpinning the SFR and the SFFA do not differ much, which makes it possible to recover agreed data series needed to assess the dynamic pattern of the forest carbon budget. Our paper (Zamolodchikov et al., 2011) gives a detailed description of the procedures for constructing data series, of the particular characteristics of presenting accounting information associated with the change of authorities of forest management as well as the dynamic pattern of accounting categories of the forest fund. In this paper, we used the SFFA databases as of January 1, 1988January 1, , 1993January 1, , 1998January 1, -2006  The full set of RECBF equations and parameters can be found in the articles (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013b, so here only a brief summary of the approach is given. The initial part of the calculations is to estimate the carbon pool for different age groups of forest stands by predominant species. The calculation of the carbon storage in the pools of phytomass and dead wood is carried out based on the data on volume stocks of stem wood from the SFR or SFFA using sets of conversion factors from the works of (Zamolodchikov et al., 2003;Zamolodchikov, 2009).
The calculation of the carbon storage in the pools of litter and soil is carried out based on the information about the areas of forest-forming species stands from the SFR or SFFA using standard average values from the works of (Chestnykh et al., 2004(Chestnykh et al., , 2007. The obtained estimation of the carbon pool in stand groups of different age makes it possible to calculate the gains for all carbon pools by using information about age ranges in the age groups. By applying the data on annual areas of destructive disturbances (felling, forest fires, other reasons causing the death of forest stands) to the found carbon pools in different categories of the forest stands we can estimate the annual carbon losses. The annual scale of destruction can be estimated in two ways (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013b The federal districts differ significantly both in total area (the Far Eastern FD is the largest one with 616.9 million ha, the North Caucasian FD is the smallest with 17.0 million ha) and in forest cover (maximum in the Siberian FD -53.4%, minimum in the Southern FD -3.5%). It is obvious that the contribution of the FDs to carbon pools and carbon balance will largely depend on the total area of forested land. In 2015, the Far Eastern FD had the largest forest area (292.67 million ha), the smallest area was in the Southern FD (1.58 million ha).
All FDs showed an increase in forested areas from 1988 to 2015 ranging from 1.1 (Siberian FD) to 5.7% (Northwestern FD). This growth is connected with the decrease of the forest fund areas that were temporarily not covered with forests, e.g. clear сut areas in the first place. In its turn, it is caused by more than 200% decrease of logging during the socio-economic reforms of the early 1990s (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013b. Even now the level of logging is still about 2 times lower than in the late 1980s. A modest increase in the areas of forested lands in the Siberian FD reveals the dire situation around forest fires in this district. As of 01.01.2015, the total carbon pool of the forested lands of Russia amounted to 123.77±18.93 Gt C (billion tons of C). The Siberian (36.4%) and Far Eastern (35.2%) FDs made the largest contribution to the national carbon pool of the forests, containing 71.6% of the national carbon pool of the forests (Table 1). This makes sense if we think about huge forested areas in these districts. The Southern (0.2%) and North Caucasian (0.3%) FDs account for the smallest percentage of carbon pools in their forests.
The average carbon pool in the phytomass per unit area is the largest in the North Caucasian FD (85.05±9.66 t C ha -1 ) and the smallest in the Far Eastern FD (31.26±2.93 t C ha -1 ) ( Fig. 1).
Therefore the average carbon pool of phytomass shows almost a 3-fold variation across the districts, which is primarily due to differences in natural conditions. About 70% of the forest area in the North Caucasian FD is covered by mountain stands of hardwoods with large timber resources. Also, hardwoods have high wood density and therefore are more carbon-intensive given the same stands volume as compared to other groups of species. Before presenting the results related to the assessment of the carbon balance, let us clarify the terminology used. By carbon absorption we mean its annual growth in all the pools of the forest ecosystem (phytomass, dead wood, litter, and soil). These gains are the difference between natural flows of carbon replenishment and outflow from each pool. For example, for the phytomass pool, it is the difference between net primary production (NPP) and the tree mortality and litter fall, bringing carbon to the pools of dead wood and litter, respectively. For dead wood, this is the difference between the tree mortality (replenishment) and emission flows from the decomposition of dead wood and its transition to the soil pool (withdrawal). We emphasize that the current version of RECBF software does not give an opportunity to assess the entire set of carbon flows and is limited to the calculation of the increments in carbon pools only. Non-destructive disturbance of forest cover (low intensity ground fires, thinning, selection felling) cause a certain reduction in the volume of growing stocks, at the same time the disturbed areas remaining part of the forested land.
Therefore, non-destructive disturbances affect the size of average stocks of stands in a particular forest region, so they are taken into account in the assessment of carbon absorption by RECBF. positive (absorption exceeds losses) or negative (absorption is smaller than losses). In the first situation, the term "sink" is used, in the second we are talking about a "source" of carbon.
The distribution of the total carbon pool of forests by districts largely corresponds to the areas covered with forest lands, the distribution of carbon sink values, however, demonstrates visible changes ( Table 2). The total contribution of the Siberian and Far Eastern FDs to the national forest carbon sink (206.10±66.86 Mt C yr -1 ) is 54.5%, which is significantly less than 71.6% of their contribution to the carbon stock. At the same time, contributions of western FDs to the forest carbon sink are increasing and amount to 12.1% in the Volga district and 8.8% in the Central district, which is noticeably more than their contributions to of carbon pool (4.4% and 2.8%). to the national features of forest fire protection (Lupyan et al., 2017). It is fires that are the main  4A). This decline is caused by logging drop during the socio-economic reforms of the early 1990s. The relationship between this trend and the increase in carbon sink in the Russian forests was repeatedly discussed in our previous papers (Zamolodchikov et al., 2011(Zamolodchikov et al., , 2013a(Zamolodchikov et al., , 2013b(Zamolodchikov et al., , 2017. The development of remote forests requires significant costs for the construction of logging roads and transportation of wood. In the non-market economy of the Soviet Union, these costs were financed from the state budget. Transition to a market economy made it necessary to consider costs and profits, therefore the development of logging on a large part of the area of operational forests became unprofitable. can be used to qualitatively check the values of felling-determined carbon losses calculated in RECBF. Fig. 5 demonstrates the dynamic pattern of the volume of harvested wood in the forests currently located on the territory of the relevant FDs, which was compiled using the forms of statistical survey of forestry from the information archives of the Center for Forest Ecology and Productivity of the RAS. In 1988 to 1998 the volume of logging decreased by 2-4 times in all the districts, in Russia in general it decreased by about 3 times. We can see that the dynamic pattern of carbon losses due to felling (Fig. 4A) is falling behind by about 3 years as compared to the volume of logging (Fig. 5). This delay is perfectly understandable as it takes some time to make changes to the regional forms of SFR and further reflect it at the national level. Losses from fires are high in two districts -Siberian and the Far Easternand they are 2.0-2.5 times higher than the loss due to felling in these FDs (Fig. 4B). In other districts, carbon loss due to fires is small compared to the loss due to felling. As was noted above, it is associated with the national features of forest fire protection. In Siberian FD fire losses do not show a pronounced temporal variability, while in the Far Eastern FD there is a decrease in these losses by about 2 times from 1988 to 2005. Unfortunately, these trends are difficult to compare with actual data on forest burning. Official forest fire statistics have been available since the mid-1950s, but they do not cover much of the forest area. In Soviet times, these areas were called "unguarded part of the forest Fund", now they are "forest fires control zones" (Order..., 2015). The development of remote monitoring methods for forest fires revealed multiple discrepancies between statistical data and independent estimates of the forest area covered by fire (Lupyan et al., 2017;Shvidenko, Schepashchenko, 2013;Soja et al., 2004, etc.). Satellite forest assessments cover mainly the period from the early 2000s, which, taking into account high temporal and spatial variability of forest fires, is not sufficient for the analysis of long-term regional trends determined by changes in the organization of forest protection and climatic variations.
Forests in most of the districts were atmospheric carbon sinks throughout the time period under review (Fig. 6). The forests of the Far Eastern district turned out to be a carbon source for the atmosphere with 7-8 Mt C in 1988 and 1993. This occurs when gains do not compensate for losses.
In this case, there is a possibility of calculated overestimation of carbon losses due to felling, as was mentioned above. The reduction in felling losses in the 1990s led to an increased carbon sink in the majority of the federal districts. By the nature of changes in carbon sink into forests, the FDs can be divided into 2 groups. The first group consists of the districts with a significant increase in carbon sink (86% and above). It is the Northwestern, Ural, Siberian, and the Far Eastern FDs with the reversed balance sign. All these districts are characterized by high forest cover and relatively low population density. The reduction of logging and the concomitant rise of the carbon sink were therefore very significant. The second group includes the districts with a moderate increase in carbon sink (4-46%). This includes the Central, the Southern, North Caucasian and Volga FDs.
These districts show fairly high population density and denser road network which makes it possible to develop any forests without significant costs. Therefore, the loss of carbon during felling decreased less than in the first group of FDs.
Since early 1990s, a large number of Russian works have been published considering the carbon budget items of forest areas. The most popular spatial levels considered are the forests of the territorial entities of the Russian Federation (Zamolodchikov et al., 2018;Sopiga, Bogoslovskaya, 2011;Safonov et al., 2016;Tulokhonov et al., 2006;Usoltsev et al., 2008, etc.) and Russia in general (Zamolodchikov et al., 2013b(Zamolodchikov et al., , 2017Fedorov et al., 2011;Filipchuk et al., 2016Filipchuk et al., , 2017 Shvidenko, Shchepashchenko, 2014, etc.). The intermediate spatial level of the federal districts was largely overlooked, probably due to the relatively low impact on forestry of administrative structures at this level compared to federal and regional authorities. The exception is a series of papers on forest carbon of Ural FD published by the team under the supervision of V. A. Usoltsev (Usoltsev et al., 2005(Usoltsev et al., , 2006(Usoltsev et al., , 2008. According to these papers, the carbon pool in the phytomass of forests of the Ural FD is 2.68 Gt C for an area of 62.29 mln ha (43.0 t C ha -1 ). Our estimation is 2.92±0.18 Gt C for an area of 68.51 million ha (42.6±2.7 t C ha -1 ). Average values of phytomass carbon unit area are quantitatively similar and confirm the trend (which has been observed since 1990s) toward the convergence of estimates of forest carbon pools by different authoring teams. in an underestimation of carbon absorption by the Russian forests, therefore the Russian forests absorb less carbon than the foreign ones (Tikhomirova, 2017). The materials of this article show, however, that the Russian forests are diverse in terms of their involvement in the carbon cycle, which is clearly evident at the level of FDs. Let's compare (Fig. 7)  protection is less effective, therefore more carbon loss from fires) and biological properties of wood species. Consideration of these factors is necessary to achieve the optimal balance between using forests as suppliers of valuable raw material, i. e. wood, and using them as a sink of greenhouse gases.