Eucalyptus productivity increase in China comes from newly afforested plantations

: Eucalyptus trees are a major fast-growing species in southern China. The ecological problems associated with constantly developing new Eucalyptus plantations have been the focus of extensive debate. In this study, we used spatial analysis and geostatistical methods along with four continuous national forest resource inventories and meteorological data to analyze dynamic changes in the distribution of Eucalyptus plantations in China. The productivity levels of Eucalyptus plantations were compared at different time periods by measuring annual mean productivity in permanent sample plots to provide baseline data related to the scientific management of Eucalyptus plantations. Results showed that the area of Eucalyptus plantations increased constantly in China from 1998 to 2013, expanding from 60.7 × 10 4 hm 2 in 1998 to more than 445.5 × 10 4 hm 2 in 2013. The productivity of Eucalyptus plantations was positively correlated with temperature and rainfall, but negatively correlated with elevation. However, these changes did not necessarily indicate an improvement in the management quality of Eucalyptus plantations, because they were mainly caused by an increased in the proportion of newly reclaimed areas for Eucalyptus afforestation and the constantly decreasing area of original Eucalyptus plantations, to which sufficient attention must be given.


Background
Eucalyptus is characterized by rapid growth, high yield, stress resistance, barren tolerance, good stem form, and extensive use. In many countries of the world, Eucalyptus is promoted for large-scale afforestation and is one of the four fast-growing genera used for afforestation in the world [1] . The total area of Eucalyptus and Pinus plantations accounts for approximately 30% of the global plantation area [2] . Eucalyptus species commonly used in afforestation include E. dunnii Maiden, E. globulus Labill, E. grandis Hill ex Maiden, E.
tereticornis Smith, and E. urophylla S. T. Blakely [3,4] . In 1990, Eucalyptus plantations covered a total of 1.34 × 10 7 hm 2 worldwide, which increased to 1.46 × 10 7 hm 2 in 1995, with more than 1.0 × 10 7 hm 2 of Eucalyptus plantations in 81 countries [5] . The total area of Eucalyptus plantations worldwide reached 2.0 × 10 7 hm 2 in 2010. In China, Eucalyptus plantations have been developed rapidly since the mid-1990s. According to the national forest resource inventory, the area of Eucalyptus plantations in China expanded from 6.07 × 10 5 hm 2 in 1998 to 4.455 × 10 6 hm 2 in 2013; the average annual rate of increase was 1.92 × 10 5 hm 2 , ranking first worldwide.
The large-scale establishment of Eucalyptus plantations, with their social and ecological benefits, has caused increasing concern and created substantial controversy [3,4,[6][7][8][9] . The development of Eucalyptus plantations have been debated from different perspectives, and 3 the controversial issues have mainly focused on: (1) an excessive consumption of nutrients (trees as 'fertility pumps'); (2) an excessive consumption of water (trees as 'water pumps'); (3) a reduction in biological diversity (creation of ecological 'green deserts'); and (4) poor ecological stability [9][10][11][12] . The following questions address these concerns: (1) How can ecological problems related to Eucalyptus plantations be addressed while improving their cultivation patterns and achieving technological innovation? (2) How can the adverse effects of Eucalyptus afforestation in the environment be minimized in a way that allows Eucalyptus plantations to play a role in substituting natural forest harvesting? How can the short supply of timber be augmented while maintaining forest resources, such as preserving important niches and rare species habitats? These practical problems urgently need to be addressed to achieve sustainable development of Eucalyptus plantations.
Chinese researchers have investigated the ecological effects of Eucalyptus plantations by setting up continuous observation points and comparative tests in local areas, which provided some quantitative results [6] . However, there have been few investigations of the dynamic changes in the distribution area of Eucalyptus plantations in China using data from multiple continuous inventories of permanent sample plots. Meanwhile, with the development of geographic information systems (GIS) and geostatistics, there has been increasing concern over the quantitative expression of the geospatial and geographical distribution patterns of Eucalyptus plantations, as well as remote sensing measurements of stand factors (height, density, and biomass). Remote sensing is a macroscopic, effective, and repeatable technology that provides an ideal tool for studies related to Eucalyptus resources. Good results have been obtained by remote sensing in studies on the dynamic changes in Eucalyptus resource inventory, harvesting, and reforestation [15][16] . Moreover, significant progress has been made using laser radar and hyperspectral remote sensing data in quantitative studies on the density, height, and aboveground biomass of Eucalyptus resources [17][18] . 4 In the present study, we analyzed the geospatial distribution of Eucalyptus productivity in southern China and its relationship with rainfall, temperature, and elevation, to reveal the spatial distribution and change patterns of Eucalyptus plantations in this country. The analysis was performed using geostatistical methods and Kriging interpolation by a combination of computer application technology and spatial analysis techniques, based on sample plot data from four national forest resource inventories (5 th -8 th ) and meteorological data. Meanwhile, we continuously measured the annual mean productivity of Eucalyptus plantations in permanent sample plots during multiple time periods, to reflect the dynamic changes in Eucalyptus productivity. This study will provide an important reference for scientific and rational management as well as for improving the productivity of Eucalyptus plantations.

Data sources
The data used in this study mainly included data from the 5 th -8 th national forest resource inventories, a national 1:250,000 digital elevation model (DEM), and meteorological data related to temperature and rainfall in southern China over the period of 1981-2010. The forest resource inventory data included statistical report, sample plot, and sample tree databases. The sample plot database included 48 factors, e.g., plot number, vertical and horizontal coordinates of the plot, mean age, mean diameter at breast height (DBH), mean tree height, canopy density, and stand volume; the sample tree database included 11 factors, e.g., tree number, type of standing trees, DBH, and volume. Meteorological data mainly included digital grid data of annual mean temperature and rainfall in Chongqing, Fujian,  The stand volume and mean age were recorded in each plot and the mean productivity of Eucalyptus plantations was calculated as follows: where P is the annual mean productivity of the stand (m 3 hm −2 a −1 ), V is the stand volume per unit area (m 3 hm −2 ) and T is the mean age of the stand (a, where a is the stand age).

1) Area estimation
The plantation area was calculated using the following systematic sampling formula: where n is the total number of sample plots, mi is the plot number for type i (including land type, vegetation type, and forest type, and other types of land classification attributes); pi is 6 the estimated area percentage for type i; SPi is the standard deviation of estimated area percentage for type i. The PAi can be calculated as: where PAi is the sampling accuracy of area estimates for type i and ta is the reliability index; pi and SPi are the same as denoted above.
2) Volume estimation ① Sample mean: where V i is the mean volume of sample plots; Vij is the volume of sample plot j for type i. ② Sample variance: where 2 i V S is the sample variance; i V S is the sample standard deviation; ij V is the volume of sample plot j for type i; and n is the total number of sample plots for type i.
③ Sampling accuracy: where i V P is the sampling accuracy of volume estimates for type i; t a is the reliability index; V i and i V S are the same as denoted above.

Extraction of meteorological and elevation information from sample plots
First, the spatial data including point vector data of sample plots, DEM data, and 7 meteorological data of annual mean temperature and rainfall were integrated and projected into less deformed Albers latitude and longitude projection and the Krasovsky ellipsoid coordinate system; we established a central meridian of 105°E and an origin latitude of 0°.
Because the point vector database of sample plots lacked climatic factors such as annual mean temperature and rainfall, we used a GIS to assign each vector plot. For example, the annual mean rainfall was assigned to each plot by GIS spatial overlay analysis of the sample plot data and digital raster maps of rainfall. The information for annual mean temperature and elevation was extracted following the same procedure.

Geostatistics-based Kriging interpolation
Geostatistics is a branch of applied statistics that deals with the spatial distribution of variables. The basic goal of using geostatistics is to achieve local optimal estimation of geological variables by linear weighting. The Kriging method, also known as spatial local estimation or spatial local interpolation, is one of the two main topics in the field of geostatistics [19][20][21][22][23] . The Kriging method is essentially an approach for linear optimal estimation of regionalized variables at unknown sampling points using the original data of regionalized variables and structural characteristics of a variogram. The Kriging method maximizes the use of information provided by spatial sampling. When estimating an unknown sampling point, this method takes the data at the unknown point into consideration as well as data for the adjacent points. Moreover, it considers the spatial position of the target point and its adjacent known points, as well as the positional relationship between adjacent samples [19] . In this study, geostatistics-based Kriging interpolation was adopted to analyze the spatial distribution pattern of productivity of Eucalyptus plantations in southern China, with data from four forest resource inventories.

Results and discussion
3.1 Productivity dynamics 8 The mean productivity of Eucalyptus plantations was analyzed based on the data from four forest resource inventories. Table 1 shows that over the period 1998-2013, the mean productivity exhibited a continuously growing trend in the distribution areas of Eucalyptus plantations in southern provinces of China: the productivity increased from 5.34 m 3 hm −2 a −1 in 1998 to 9.88 m 3 hm −2 a −1 in 2013. Among these provinces, Guangxi and Fujian showed the most significant growth trend, which is closely related to the increased development of Eucalyptus plantations in these regions.   Note: 1) "No." refers to the sample plot number; "M." refers to mean productivity; "Max." refers to maximum productivity; 2) "--" indicate areas with no sample plots.
Based on the analysis of permanent sample plots in four continuous inventories, we

Geospatial distribution pattern of productivity
We obtained the geospatial distribution of productivity of Eucalyptus plantations in the four inventory periods by Kriging interpolation. From 1998 to 2013, the spatial range of productivity distribution gradually developed from the Leizhou Peninsula in Guangdong and Hainan towards the north (Guangxi, Hunan, and Guizhou), east (Fujian and Jiangxi), and west (Yunnan and Sichuan) (Fig. 3). This is generally consistent with the development of  13 Based on vector data for the spatial distribution of sample plots, we built a spatial database for the sample plots of Eucalyptus plantations using GIS software, with geographical data related to latitude and longitude, elevation, annual mean temperature, annual mean rainfall, and other information extracted from the DEM and meteorological data. The spatial database was then used to analyze the relationship between annual mean productivity as well as the area distribution of Eucalyptus plantations with annual mean temperature, annual mean rainfall, and elevation. Using a statistical analysis of the spatial    [2,3] , which reported that an annual mean temperature above 20 °C is the most suitable  Table 2 shows the area and volume of Eucalyptus plantations in southern provinces

Conclusions
This study generated geospatial distribution maps of the mean productivity of Eucalyptus (2) The distribution of productivity in Eucalyptus plantations was closely related to temperature and rainfall. The regions with a temperature above 19 °C and rainfall of 1400-1600 mm constituted the main and highest Eucalyptus plantation production areas in China.
The production operation of Eucalyptus plantations was closely related to elevation, and Eucalyptus plantations were mainly distributed at the elevations below 300 m. (4) Eucalyptus trees are excellent fast-growing species. Because Eucalyptus species feature vigorous growth, strong competitiveness, short maturity time, and high growth requirements of nutrients and water, they are difficult to cultivate with other tree species and thus are suitable for planting in regions with relatively high annual mean temperature and rainfall. While the present evaluation of Eucalyptus plantations as it related to the poor ecological stability of the genus is objective, it is also one-sided. We must respect the biological characteristics of Eucalyptus trees, take advantage of their rapid-growth, and select suitable tree species for each site, to better enhance China's ability to supply wood.
Additionally, the use of remote sensing technology has the advantages of low cost and high efficiency when compared with traditional ground surveys. Therefore, the monitoring and analysis of dynamic change in the spatial extent and volume of Eucalyptus plantations using remote sensing will be the direction of our next research study.