Impact of urbanization on natural ecosystem service values: a comparative study

Environ Monit Assess (2011) 179:575 588 DOI 10.1007/s10661-010-1764-1 Impact of urbanization on natural ecosystem service values: a comparative study Shuying Zang Changshan Wu Hang Liu Xiaodong Na Received: 22 February 2010 / Accepted: 21 October 2010 / Published online: 9 November 2010 Springer Science+Business Media B.V. 2010 Abstract With rapid population growth and ruralto-urban migration in many Chinese cities, a large amount of natural lands have been converted to urban and agricultural lands recently. During this process of land conversion, economic development and quality of life improvement are considered as major goals, and their influences on ecological systems have often been neglected. The degradation of natural ecological systems due to land use change, however, has become severe, and may require immediate attentions from urban planners and local governments. Taking HaDaQi industrial corridor, Heilongjiang Province, China, as a case study area, this paper examined the trend of land use changes during 1990 2005, and quantified their influences on natural ecosystem service values. In particular, this study applied two major valuation methods, and examined whether different valuation methods generate significantly S. Zang H. Liu X. Na (B) Key Laboratory for Remote Sensing Monitoring of Geographic Environment, Harbin Normal University, Harbin, Heilongjiang, 150025, People s Republic of China e-mail: naxiaodong_8341@163.com C. Wu Department of Geography, University of Wisconsin-Milwaukee, PO Box 413, Milwaukee, WI, 53201, USA different results. Analysis of results suggests that human dominated land uses (e.g., urban and agriculture) have expanded rapidly at the cost of natural lands (e.g., wetlands and forest). Due to these land use changes, the total ecosystem service value decreased 29% (2.26% annually) from 1990 to 2005 when the first method was applied, and this rate is estimated to be 15.7% (1.13% annually) with the second approach. Moreover, the annual rate of ecosystem service value decline during 2000 2005 is about four times higher than that in 1990 2000 with both methods, suggesting much more severe ecosystem degradation during 2000 2005. Keywords Ecosystem service values Urbanization HaDaQi industrial corridor Introduction Due to population growth and migration from rural to urban areas, Chinese cities have experienced rapid expansions at an unprecedented rate in the past decades (Zhang and Song 2003). Chinese population has increased significantly during the past 60 years. The total population of China was 552 million in 1950, and it reached 1.14 billion in 1990, and 1.31 billion in 2005 (National Bureau of Statistics of China 2010). As estimated by the United Nations Population Division, China s

576 Environ Monit Assess (2011) 179:575 588 population will continue to increase to 1.46 billion in 2030, and then decline slightly (United Nation Population Division 2008). Together with overall population growth, urban population also increased dramatically. Population resides in cities has grew from 302 million in 1990 to 560 million in 2005, and it is estimated that in 2020, there will be 900 million urban population in China (National Bureau of Statistics of China 2010). Moreover, the percent of urban population to the total population also increased rapidly. In particular, in 1990, only 26% of people lived in urban areas, and this number rose to 36% in 2000, and 43% in 2005, and it is projected that this trend will continue, and over 65% of Chinese population will reside in urban areas in 2050 (Song and Ding 2009). While urbanization brings economic benefits and improves the quality of life, ill-planned urban growth can also lead to adverse social and environmental effects (Newman and Kenworthy 1999). Many problems associated with rapid urbanization, such as congestion, excess commuting, urban heat island effect, and air and water pollution, have been well documented (Newman and Kenworthy 1999; Wangetal.2003). Among these problems, ecological system degradation is an essential one. Rapid urbanization leads to fast conversion from natural land uses (e.g., forest, rangeland, and wetlands) to human-dominated land uses (e.g., urban and agricultural lands), thereby affecting natural ecological systems profoundly. In particular, the changes of land use and land cover (LULC) modify the physical parameters of the earth surface, thus affecting material and energy interchanges between land and atmosphere. Moreover, such LULC change can also affect biochemical cycles, influencing nutrition transport between soil and vegetation. In addition, LULC change directly impacts biodiversity, modifying the composition and structure of ecosystems. In summary, the conversion from natural lands to urban lands degrades the functions, including goods and services, of natural ecosystems. Therefore, understanding the process of urban growth and exploring its effects on natural ecosystem is an essential research topic for scholars, planners, and policy makers. Traditionally, scholars and planners have examined the dynamics of urbanization based on a number of datasets in the geographic information system format. In particular, the boundaries between urban and rural areas have been identified through analyzing census data collected for every decade. In addition, the extent of urban areas and their spatial changes have also been quantified through examining multi-temporal land use/land cover data. Accurate and up-to-date census and land use datasets, however, are often unavailable to general public and researchers in developing countries like China (Harris and Longley 2000). On the contrary, remote sensing technologies can provide a synoptic view over a large geographical area with multiple temporal intervals. Therefore, remote sensing imagery can serve as a better alternative for quantifying urban extent and monitoring urban land use changes (Lu et al. 2004). Due to these benefits, many studies have examined the process and patterns of urbanization, and achieved reasonable accuracies (Kreuter et al. 2001;Luetal.2004;Zhaoetal.2004). With the knowledge of urban land use changes, the next step is to evaluate their impacts on natural ecosystems. Since 1970s, scientists have attempted to understand the functions (including goods and services) that ecosystems have provided to human societies (Holdren and Ehrlich 1974; Odum 1977; Westman 1977), and evaluate the impact of land use changes on these functions (Kreuter et al. 2001). In particular, Costanza et al. (1997) identified 17 ecosystem services (e.g., gas regulation, climate regulation, disturbance regulation, etc.) provided by 16 biomes globally, and estimated the monetary value of each service generated by every biome quantitatively. Through adopting the valuation coefficients calculated by Costanza et al. (1997), a few applications have been developed to evaluate the impacts of urbanization on the ecosystem service values (ESVs) (Hu et al. 2008; Kreuter et al. 2001; Li et al. 2007, 2010; Tong et al. 2007; Zhaoetal.2004). While simply applying the valuation coefficients provided by Costanza et al. (1997) is straightforward and efficient, many economists criticized the valuation method (e.g., Pimm 1997; Toman 1998), and argued that extreme caution should be taken when transferring these coefficients to a different study area, especially under a different cultural system (Turner et al. 1998, 2003; Zhang

Environ Monit Assess (2011) 179:575 588 577 et al. 2010). When applied in China, Xie et al. (2001) argued that the valuation method proposed by Costanza et al. (1997) underestimates the service value of agricultural lands, and overestimates other biomes, such as wetlands and forest. In particular, based on surveying results of Chinese ecologists, agricultural lands should be able to provide essential services for climate regulation and soil formation, while these service functions are considered negligible in Costanza et al. s (1997) method. For wetlands and forest, Chinese ecologists agree that they can provide important services, but the willingness to pay (price) should be much lower (Xie et al. 2001). Consequently, Xie et al. (2003) calculated valuation coefficients for ecosystems in China through a contingent valuation analysis based on a survey of 213 ecologists. This valuation method has been applied in many studies that evaluate the impact of urbanization on ecosystem service values in Chinese cities recently (Gu et al. 2009;Zhaoetal.2009). Studies on examining the impacts of urbanization on ecosystem functions in China have adopted either the valuation coefficients developed by Costanza et al. (1997) or those modified by Xie et al. (2003), and there have not been any comparative studies. Different valuation methods, however, may produce significantly different results (Bao et al. 2007). As an example, the total terrestrial ecosystem service value (ESV) in China was estimated to be 148 trillion RMB Yuan according to a study conducted by Ouyang et al. (1999); and this value became 5.6 trillion RMB Yuan when the Costanza et al. s (1997) parameters were adopted (Zhang et al. 2005). Therefore, there are remarkable differences when different valuation methods were applied. Because of this, Turner et al. (1998, 2003) pointed out that the valuation method is less meaningful to estimate the total value of an ecosystem, and may be more appropriate for marginal change analysis. Although many researchers agree with this argument (Kreuter et al. 2001; Bao et al. 2007), there has not been any empirical evidence to support this. The purpose of this paper, therefore, is to evaluate the impact of urbanization on natural ecosystem service values, and examine whether different valuation methods generate significantly different results. Materials Study area HaDaQi industrial corridor is located in the western part of Heilongjiang Province, and the longitude of this area is between East 122 48 and 127 15 and its latitude is between North 45 31 and 47 51. This region includes three major cities: Harbin, Daqing, and Qiqihar, and two smaller cities: Zhaodong and Anda (see Fig. 1). Located in the eastern part of HaDaQi industrial corridor, Harbin is the capital city of Heilongjiang Province, with a geographical area of 53,100. Daqing is in the central part of HaDaQi industrial corridor, with an area of 22,161 km 2. Qiqihar is located in the far northern part of the industrial corridor, with an area of 4,310 km 2. These three major cities are connected by the Bingzhou Railway. HaDaQi industrial corridor has a population of 6.7 million, or 17.6% of the total population in Heilongjiang Province. The gross domestic product (GDP) was about US $53.6 billion in 2005, or 49% of the total GDP of Heilongjiang Province. This area is an important industrial center of China, with industrial sectors including petroleum mining; electric, electronic, and chemical engineering; construction, medicine, etc. This area has experienced rapid growth in the past decades, and it is important to examine the influences of human dominated developments on natural ecological systems. For examining the influences of HaDaQi corridor on surrounding natural ecosystems, a slightly larger study area, including the HaDaQi corridor and adjacent rural areas, was employed (see Fig. 1). Data Major dataset used in this research includes remote sensing imagery and socio-economic information. Landsat Thematic Mapper (TM) data acquired in 1990, 2000, and 2005 were utilized to derive land use maps for these 3 years. These Landsat TM images were provided by the China Remote Sensing Satellite Ground Station and Heilongjiang Academy of Agricultural Science. Each image consists of six spectral bands (bands 1 5 and 7) with a spatial resolution of 30 m. These images

578 Environ Monit Assess (2011) 179:575 588 Fig. 1 Location map of the HaDaQi industrial corridor were georeferenced to Gaussian Projection, with the 1980 Xi an Datum. In addition to the remote sensing imagery, socio-economic data, such as the gross domestic products and market prices of agricultural produce, were also obtained from Heilongjiang Statistics Yearbooks (Xuan 1990;He 2000;Li2005). Methods Land use land cover change analysis With the Landsat Thematic Mapper imagery acquired in 1990, 2000, and 2005, a hybrid unsupervised/supervised classification technique and a post-classification were applied to classify these three images into six land use land cover types, including urban, agriculture, forest, grassland, wetlands and water, and barren lands. In particular, urban land uses include commercial, residential, industrial, and transportation infrastructures; wetlands and water refer to lake basin bog, swamp, river, reservoir, pond, and floodplain; and barren lands include sands and alkaline and salinize lands. In this study, we use a single land use/cover class to represent wetlands and water, due to the difficulty to separate them using medium resolution remote sensing data. With the results from the hybrid classification, a post-classification was carried out through manual interpretation and digitization. To ensure satisfactory classification accuracy, local land use paper maps, historical aerial photographs, and other ground information were employed as references for manual digitization. In addition, the classification accuracies for these three land use maps were evaluated through examining 100 randomly generated field samples, and the overall accuracies for each map are higher than 85%. Ecosystem service valuation Firstly, the valuation method developed by Costanza et al. (1997) was applied, and each land cover type extracted from the Landsat TM imagery was compared with the biomes proposed by Costanza et al. s valuation method. Because it is difficult to have a perfect match, the representative biome was employed as a proxy for every land use/cover type, including cropland for agriculture, temperate/boreal forest for forest, grassland/rangelands for grassland, wetlands and lakes/rivers for wetlands and water, urban for ur-

Environ Monit Assess (2011) 179:575 588 579 ban, and desert for barren lands (see Table 1). In particular, agricultural lands, with major crop types of rice, wheat, corns, soybeans, vegetables etc., are interpreted as a single biome: cropland, as identified by Costanza et al. (1997). Forest, including evergreen, deciduous, and mixed types, is considered to represent the temperate/boreal forest biome. Moreover, due to the difficulty of classifying wetlands and water, we interpreted them as a single land use class, representing a combination of two biomes: wetlands and lakes/rivers, and the mean value of the coefficients for wetlands and lakes/rivers was assigned. In addition, urban land uses, including commercial, residential, industrial, etc., are considered as the urban biome, and barren lands, which include alkaline and salinize land, sandy areas, surface extraction, etc., are interpreted as the desert biome. With the ecosystem service coefficients for each land use/land cover type, the total ecosystem service values in HaDaQi industrial corridor in 1990, 2000, and 2005 were calculated as follows (Eq. 1). ESV = n (A k VC k ) (1) k=1 Where ESV is the total ecosystem service value in the study area for a particular year; A k is the geographic area (ha) for land use type k, VC k is the value coefficient (US $ ha 1 year 1 ) for land use type k, andn is the total number of land use types (n = 6 in this study). Xie et al. (2003) valuation method While Costanza et al. s (1997) method has been widely applied, Xie et al. (2003) argued that, when applied in China, the valuation method proposed by Costanza et al. underestimates the service value of agricultural lands, while overestimates other biomes, such as wetlands and forest. Consequently, Xie et al. (2003) developed a contingent valuation method based on the survey of more than 200 Chinese ecologists. In particular, this method includes four steps for calculating service values for each ecosystem, including (1) calculating the food production service value for agricultural lands, which is one-seventh of the market price of agricultural produce, (2) with the food production service value of agricultural lands as the unit, estimating the ratio between service values of all other ecosystem functions to the food production service value of agricultural lands using surveying techniques, (3) deriving the service values for each function of each ecosystem through multiplying the agricultural land service value with the corresponding ratio, and (4) generating a final ecosystem service value for each land use type through summing the values for each function, and multiplying by the its respective geographical area. Applying this method, the first step is to calculate the food production service value of agricultural produce per hectare. According to Xie et al. (2003), this value is one-seventh of the market price of agricultural produce. For this study, the average market price of agricultural produce per hectare was calculated using the 2005 data of major crops, including rice, wheat, corns, soybeans, and vegetables. The estimated monetary value of agricultural produce per hectare is US $915 (RMB Yuan 6,247), and therefore, the value of the food production service provided by agricultural lands per hectare is US $131 (RMB Yuan Table 1 Equivalent biomes and their ecosystem service coefficients (revised from Costanza et al. 1997) Land cover categories Equivalent biome identified Ecosystem service coefficient (classified from remote sensing imagery) in Costanza et al. (1997) ($ ha 1 year 1 ) a Agricultural Cropland 126 Forest Temperate/boreal forest 413 Grassland Grassland/rangelands 318 Wetlands and water Wetlands and lakes/rivers 15,937 Urban Urban 0 Barren Desert 0 a The coefficients are adjusted to 2005 US $ according to the Consumer Price Index (CPI) of China

580 Environ Monit Assess (2011) 179:575 588 Table 2 Ratio between ecosystem service value and that of food production service value provided by agricultural lands (revised from Xie et al. 2003) Service types Forest Grassland Agriculture Wetlands and water Barren Gas regulation 3.50 0.80 0.50 0.90 0.00 Climate regulation 2.70 0.90 0.89 8.78 0.00 Water regulation 3.20 0.80 0.60 17.94 0.03 Soil formation and retention 3.90 1.95 1.46 0.86 0.02 Waste treatment 1.31 1.31 1.64 18.18 0.01 Biodiversity 3.26 1.09 0.71 2.50 0.34 Food production 0.10 0.30 1.00 a 0.20 0.01 Raw material 2.60 0.05 0.10 0.04 0.00 Recreation and cultural 1.28 0.04 0.01 4.95 0.01 Total 21.85 7.24 6.91 54.34 0.42 a The value of food production service provided by agricultural lands is utilized as a unit 892), one-seventh of the agricultural produce market price. With the ecosystem service value of agricultural lands, the next step is to estimate the ratio of service values between other lands and agricultural lands. For this estimation, we employed the results obtained from Xie et al. (2003), but with some minor adjustments. In particular, for this research, wetlands and water are considered as one single class, and therefore, the average value of wetlands and water was assigned to this class. In addition, we assigned the ecosystem service values of desert to barren lands. This assignment is reasonable as barren lands include sand and alkaline and salinize lands, which provide comparable services. The ratio of each ecosystem service value to the food production service provided by agricultural lands is shown in Table 2. With the ecosystem service value provided by the food production service of agricultural lands ($131) and the ratio shown in Table 2, the third step is to multiply them to obtain the ecosystem service value for each function of each land use type. The results of this calculation (see Table 3) indicate that wetlands and water have the highest ecosystem service value per hectare (US $7,103), followed by forest ($2,856), grassland ($946), agricultural ($903), barren land ($55), and finally urban ($0). After obtaining the ecosystem service value for each land use type, the final step is to calculate the overall ecosystem service value for the HaDaQi industrial corridor. With the geographic areas of each land use in 1990, 2000, and 2005, the overall ecosystem service value for each land use type was obtained. Sensitivity analysis Following the method applied by Kreuter et al. (2001) and Wang et al. (2006), a sensitivity analy- Table 3 Ecosystem service value coefficients for each land use category ($ ha 1 year 1a ; revised from Xie et al. 2003) Service types Forest Grassland Agriculture Wetlands and water Barren Gas regulation 458 105 65 118 0 Climate regulation 353 118 116 1,148 0 Water regulation 418 105 78 2,345 4 Soil formation and retention 510 255 191 112 3 Waste treatment 171 171 214 2,376 1 Biodiversity 426 142 93 326 44 Food production 13 39 131 26 1 Raw material 340 7 13 5 0 Recreation and cultural 167 5 1 646 1 Total 2,856 946 903 7,103 55 a The coefficients are adjusted to 2005 US $ according to the exchange rate between RMB Yuan and US $

Environ Monit Assess (2011) 179:575 588 581 sis was applied. In particular, the coefficient of sensitivity (CS) was obtained to represent the elasticity of ESV, in response to the change of VC for a particular land use type k (see Eq. 2). CS k = ESV/ESV (2) VC k /VC k Where CS k is the coefficient of sensitivity for a land use type k; ESV indicates the change of the ESV due to the change of VC; and VC k is the change of VC (e.g., ±50%) for land use type k. The coefficient of sensitivity represents whether the overall ESV is sensitive to the variation of the ecosystem value coefficient provided by a particular land use type. The higher the CS value, the more important an accurate ecosystem VC for a land use category should be used. Results and discussion Land use change analysis Land use land cover maps in 1990, 2000, and 2005 derived from Landsat TM imagery were illustrated in Fig. 2, and the results are summarized in Tables 4 and 5. Analysis of these results suggests that, during these 15 years, humandominated land uses (i.e., urban and agriculture) have expanded rapidly at the cost of natural lands (wetlands, forest, and grassland). In particular, as can be discerned from Table 4, the area of urban land uses increased from 417,000 hectares in 1990 to 629,000 hectares in 2005, with a total increment of 51% (2.78% annually). Similarly, the geographic areas of agricultural and barren lands have increased 214,848 acres (or 3.2%) and 62,020 acres (13.5%), respectively. Conversely, natural lands, including forest, grassland, and wetlands, have diminished substantially. The geographical area of wetlands and water, in particular, has decreased from 1.13 million hectares in 1990 to 780,000 ha in 2005, with an overall area change of 351,000 ha, or 31.0% (2.45% annually). Similarly, forest also diminished significantly (about 15.6%), while the area of grassland only decreased slightly (about 2.11%). These results suggest that human-dominated land uses have expanded significantly, with the costs of natural lands. In addition, the rapid expansion of barren lands indicates the severe degradation of natural ecosystems. After examining the general trend of land use change from 1990 to 2005, we also performed a detailed analysis for two different periods: 1990 2000 and 2000 2005. Results (see Table 5) indicate that although the trend of most land use changes during these two periods is consistent, more significant changes have happened from 2000 to 2005. From 1990 to 2000, urban land uses only grew about 0.40% annually, while the growth rate of urban land uses became 7.72% annually during the period of 2000 and 2005, indicating a much more rapid urbanization process. In addition, a similar trend can be found for the growth of barren land. Together with the significant expansions of urban and barren lands from 2000 to 2005, wetlands and water bodies diminished rapidly during this period (5.16% annually), much higher than the rate (1.06% annually) of diminishment from 1990 to 2000. For agriculture lands, the expansion rate (0.13% annually) during 2000 2005 is slower than that (0.25% annually) between 1990 and 2000. Simultaneously, we can also discern an expansion of grassland (1.49% annually) from 2000 to 2005, while grassland had a trend of diminishment ( 0.95 annually) from 1990 to 2000. The lower expansion rate of agriculture land and the increment of grassland area during 2000 2005 can be attributed to the ecological environmental protection project Converting agricultural land to forest and grassland implemented in late 1990s. These findings are consistent with the results of Wang et al. (2006) and Li et al. (2010). In particular, Wang et al. (2006) found that, in Sanjiang Plain, Northeast China, the annual growth rate of agriculture lands dropped from 1.6% before 1996 to 0.2% afterwards. Moreover, the decrement of grassland has also slowed down in late 1990s ( 0.4% annually before 1996 and 0.1 afterwards) (Wang et al. 2006). Comparative analysis of ecosystem service valuation methods With the knowledge of land use changes, it is necessary to quantify the influences of such land use conversion on natural ecosystems. Through applying the Costanza et al. s (1997) and Xie et al. s

582 Environ Monit Assess (2011) 179:575 588 (a) (b) (c) Agriculture Wetlands and water Grassland Forest Barren Urban Fig. 2 Land use maps derived from Landsat TM images of a 1990, b 2000, and c 2005 Table 4 Land use land cover classification results in 1990, 2000, and 2005, and their area changes from 1990 to 2005 Land use type 1990 2000 2005 1990 2005 (ha) (ha) (ha) (ha) % % per year Agricultural 6,743,859 6,914,748 6,958,707 214,848 3.19 0.21 Forest 815,726 772,114 688,724 127,002 15.57 1.12 Grassland 529,647 481,408 518,455 11, 192 2.11 0.14 Wetlands and water 1,131,233 1,016,609 780,092 351,141 31.04 2.45 Urban 416,794 433,793 629,261 212,467 50.98 2.78 Barren 460,838 479,425 522,858 62,020 13.46 0.85

Environ Monit Assess (2011) 179:575 588 583 Table 5 Land use land cover changes during 1990 2000 and 2000 2005 Land use type 1990 2000 2000 2005 Change (ha) % change % per year Change (ha) % change % per year Agricultural 170,889 2.53 0.25 43,959 0.64 0.13 Forest 43,612 5.35 0.55 83,390 10.80 2.26 Grassland 48,239 9.11 0.95 37,047 7.70 1.49 Wetlands and water 114,624 10.13 1.06 236,517 23.27 5.16 Urban 16,999 4.08 0.40 195,468 45.06 7.72 Barren 18,587 4.03 0.40 43,433 9.06 1.75 (2003) valuation methods, the total ecosystem service values in HaDaQi industrial corridor were obtained for 1990, 2000, and 2005 (see Tables 6 and 7). In particular, with the Costanza et al. s (1997) method, the total ESV in the study area was approximate $19.4 billion in 1990, and this value decreased to $17.5 billion in 2000, and further decreased to $13.8 billion in 2005. From 1990 to 2005, the total ESV decreased about 29%, or 2.26% annually. This is majorly due to the reduced ESVs of wetlands and water, as the ESV of wetlands and water decreased approximate $5.6 billion, or 31% (2.45% annually). Moreover, the ESVs provided by forest and grassland during 1990 2005 decreased about $52.2 million and $3.5 million, respectively, thereby contributing to the decline of total ESV. On the other hand, agricultural lands provided a higher ESV in 2005 ($876.8 million) than that in 1990 ($849.7 million), with an overall increase of $27.1 million. This increment, however, cannot offset the negative influences of the decline of wetlands and water, forest, and grassland. Comparatively, with the Xie et al. s (2003) method, the total ESV was approximately $17.0 billion in 1990, reduced to $16.2 billion in 2000, and finally reduced to $14.3 billion in 2005 (see Table 7). The decline rate of the total ESV from 1990 to 2005 (1.13% per year) is about the half of that estimated using the Costanza et al. s (1997) method (2.26% per year). This is due to the lower coefficient value assigned to wetlands and water, as well as the higher coefficient value assigned to agriculture. For individual land use types, the decline of total ESV for wetlands and water during these 15 years ($2.5 billion) is much lower than that ($5.6 billion) obtained from the Costanza et al. s approach. For agricultural lands, however, the increment of the ESV ($194 million) is significantly higher than that ($27 million) estimated via the Costanza et al. s method. Similarly, the changes of ESVs for forest and grassland are also substantially higher than those derived from the Costanza et al. s method. In addition to this general analysis, we also performed a comparative analysis for two periods: 1990 2000 and 2000 2005 (see Tables 8 and 9). Examination of Table 8 reveals that, with the Costanza et al. s (1997) approach, the decline rate of the total ESV during 2000 2005 (4.75% annu- Table 6 ESVs in 1990, 2000, and 2005, and their changes during 1990 2005 (with Costanza et al. 1997 s valuation method) Land use type 1990 2000 2005 1990 2005 (10 6 $) (10 6 $) (10 6 $) Change (10 6 $) % change % per year Agricultural 849.7 871.3 876.8 27.1 3.19 0.21 Forest 336.9 318.9 284.4 52.5 15.57 1.12 Grassland 168.4 153.1 164.9 3.5 2.11 0.14 Wetlands and water 18,028.5 16,201.7 12,432.3 5,596.1 31.04 2.45 Urban 0 0 0 0 0 0 Barren 0 0 0 0 0 0 Total 19,383.5 17,544.9 13,758.4 5,625.1 29.02 2.26

584 Environ Monit Assess (2011) 179:575 588 Table 7 ESVs in 1990, 2000, and 2005, and their changes from 1990 to 2005 (with Xie et al. 2003 s valuation method) Land use type 1990 2000 2005 1990 2005 (10 6 $) (10 6 $) (10 6 $) Change (10 6 $) % change % per year Agricultural 6,089.7 6,244.0 6,283.7 194.0 3.19 0.21 Forest 2,329.7 2,205.2 1,967.0 362.7 15.57 1.12 Grassland 501.1 455.4 490.4 10.6 2.11 0.14 Wetlands and water 8,035.1 7,221.0 5,541.0 2,494.2 31.04 2.45 Urban 0 0 0 0 0.00 0 Barren 25.3 26.4 28.8 3.4 13.5 0.85 Total 16,981.0 16,151.9 14,310.9 2,670.1 15.72 1.13 ally) is about four times higher than that during 1990 2000 (0.99% annually). The major reasons for this trend are the rapid decline of the ESV provided by wetlands and water, as well as forest. In fact, the ESV of wetlands and water decreased 5.16% per year during 2000 2005, significantly higher than the annual decrement rate (1.06%) during 1990 2000. Similarly, the annual decline rate of forest is 2.26% during 2000 2005, about three times higher than that (0.55%) during 1990 2000. Comparatively, with the Xie et al. s (2003) approach, consistent results have been found (see Table 9). For example, from 2000 to 2005, the total ESV decreased approximately 11.4% (2.4% annually), while it only decreased around 4.9% (0.5% annually) from 1990 to 2000. Similarly, for individual land use types, there has been a much higher rate of decline during 2000 2005 than that during 1990 2000. These results indicate that for marginal change analysis of ecosystem service values, these two methods can generate consistent results. After calculating the ESV according to the Costanza et al. (1997) andxieetal.(2003) valuation methods, we also performed a sensitivity analysis, and obtained the coefficient sensitivity (CS) value through applying the Eq. 2 discussed in Sensitivity analysis. Results (see Table 10) indicate that, with the Costanza et al. s (1997) approach, the CS values of wetlands and water for 1990, 2000, and 2005 are the highest (0.92 0.94), while the values of agricultural lands are about 0.04 0.06, and all others are near zero. With the Xie et al. s (2003) method, however, the CS values for wetlands and water are around 0.4, while the values for agricultural lands are also around 0.4, and the CS values for forest are about 0.14 (see Table 11). These results indicate that, with the Costanza et al. s (1997) method, an accurate estimate of the coefficient value of wetlands and water is essential, as wetlands and water is a major land cover type in the study area, and its coefficient value is significantly higher than others. While with the Xie et al. s (2003) method, Table 8 ESV changes during 1990 2000 and 2000 2005 (with Costanza et al. 1997 s valuation method) Land use type 1990 2000 2000 2005 Change (10 6 $) % change % per year Change (10 6 $) % change % per year Agricultural 21.5 2.53 0.25 5.5 0.64 0.13 Forest 18.0 5.35 0.55 34.4 10.80 2.26 Grassland 15.3 9.11 0.95 11.8 7.70 1.49 Wetlands and water 1,826.8 10.13 1.06 3,769.4 23.27 5.16 Urban 0 0 0 0 0 0 Barren 0 0 0 0 0 0 Total 1,838.6 9.49 0.99 3,786.5 21.6 4.75

Environ Monit Assess (2011) 179:575 588 585 Table 9 ESV changes during 1990 2000 and 2000 2005 (with Xie et al. 2003 s valuation methods) Land use type 1990 2000 2000 2005 Change (10 6 $) % change % per year Change (10 6 $) % change % per year Agricultural 154.3 2.53 0.25 39.7 0.64 0.13 Forest 124.6 5.35 0.55 238.2 10.8 2.26 Grassland 45.6 9.11 0.95 35.0 7.70 1.49 Wetlands and water 814.2 10.13 1.06 1,680.0 23.27 5.16 Urban 0 0 0 0 0 0 Barren 1.0 4.03 0.40 2.4 9.06 1.75 Total 829.0 4.88 0.50 1,841.0 11.40 2.39 accurate information about wetlands and water is not that essential, and the accuracies of the valuation coefficients of wetlands and water, agricultural lands, and forest collectively decide the quality of the total ESV estimates for the study area. Ecosystem service value and economic development While we have examined the influence of urbanization on the changes natural ecosystem service values, it is necessary to further examine the relation between economic growth and the changes of natural ecosystem services. To achieve this goal, we obtained the gross domestic products for the study area in 1990, 2000, and 2005 from the Heilongjiang Statistics Yearbooks, adjusted these values according to the consumer price index, and converted them from RMB Yuan to US dollars. Results (see Table 12) show that the GDP for the HaDaQi industrial corridor was $10.7 billion in 1990, then rose to $35.8 billion in 2000, and finally rose to $53.6 billion in 2005. Comparatively, it indicates that during 1990 2005, the GDP growth was about 11.3% annually, while the annual decline rate of total ESV was 2.26% according to the Costanza et al. s method and 1.13% according to the Xie et al. s method. When comparing the two periods: 1990 2000 and 2000 2005, we found that the GDP growth in 1990 2000 (12.8%) was much higher than that (8.4%) during 2000 2005. The annual decline rate of ecosystem service values during 1990 2000, however, was only about one-fourth to one-fifth of that in 2000 2005. This result might allude that there may not be a necessary connection between economic development and ecosystem degradation, and during 2000 2005, significant ecosystem degradation was discerned while at the same time the GDP growth is much slower. Further analysis, however, is required to examine the relationship between economic development and ecosystem degradation for this study area. The results also suggest that the current economic growth is associated with natural ecosystem degradation, and a better planning practice that taking ecosystem service into consideration may be necessary. Table 10 Coefficients of sensitivity (CS) obtained with the Costanza et al. s (1997) valuation method 1990 2000 2005 Agriculture 0.04 0.05 0.06 Forest 0.02 0.02 0.02 Grassland 0.01 0.01 0.01 Wetlands and water 0.94 0.92 0.92 Urban 0.00 0.00 0.00 Barren 0.00 0.00 0.00 Table 11 Coefficients of sensitivity (CS) obtained with the Xie et al. s (2003) valuation method 1990 2000 2005 Agriculture 0.36 0.39 0.44 Forest 0.14 0.14 0.14 Grassland 0.03 0.03 0.03 Wetlands and water 0.47 0.45 0.39 Urban 0.00 0.00 0.00 Barren 0.00 0.00 0.00

586 Environ Monit Assess (2011) 179:575 588 Table 12 Comparisons between the ESV estimated by the Costanza et al. s method, the Xie et al. s method, and the gross domestic products (GDPs) in 1990, 2000, and 2005 1990 2000 2005 1990 2005 1990 2000 2000 2005 (10 9 $) (10 9 $) (10 9 $) (10 9 $) % % year 1 (10 9 $) % % year 1 (10 9 $) % % year 1 ESV c 19.38 17.54 13.76 5.62 0.29 2.26 1.84 9.49 0.99 3.78 21.55 4.74 ESV x 16.98 16.15 14.31 2.67 0.16 1.13 0.83 4.89 0.50 1.84 11.39 2.39 GDP 10.69 35.77 53.58 42.89 4.01 11.34 25.08 234.62 12.84 17.81 49.79 8.42 ESV c indicates the ESV estimated with the Costanza et al. s (1997) method, ESV x indicates the ESV estimated with the Xie et al. (2003) method Conclusions Taking HaDaQi industrial corridor, Heilongjiang Province, China, as a case study area, this paper examined the impact of urbanization on natural ecosystem service values, and evaluated whether two valuation methods widely applied in China, including (1) the Costanza et al. (1997) method and (2) the Xie et al. (2003) approach, may generate significantly different results. With the estimated ecosystem service values in 1990, 2000, and 2005, this paper also analyzed the relationship between ecosystem function degradation and economic developments. Analyses of results suggest several conclusions. Firstly, through analyzing the land use land cover maps in 1990, 2000, and 2005 derived from Landsat TM imagery, we found that during these 15 years, human-dominated land uses (e.g., urban and agriculture) have expanded at the cost of natural lands (wetlands, forest, and grassland). The geographical area of urban lands increased approximate 212,000 ha, with a total change of 51%, and wetlands and water diminished about 351,000 ha, representing an area change of 31.0%. In addition, when comparing the periods of 1990 2000 and 2000 2005, we found that much more significant changes have happened during 2000 2005. The annual urban growth rate was 7.72% during 2000 2005, much higher than that (0.40%) during 1990 2000. Moreover, from 2000 to 2005, wetlands and water diminished at an annual rate of 5.16%, significantly higher than that (1.06%) from 1990 2000. Secondly, through comparing the results from the ecosystem service valuation methods developed by Costanza et al. (1997) and Xie et al. (2003), we found that the total ecosystem service changes estimated by the Costanza et al. s method are significantly higher than those estimated by Xie et al. s approach. In particular, with the Costanza et al. s method, the total ESV in the study area decreased about 29% (2.26% per year) from 1990 to 2005, while the total decrease is 15.7% (1.13% annually) according to the Xie et al. s (2003) method. These differences are majorly due to the different valuation coefficients assigned to wetlands and agricultural lands. When compared the changes for the two different periods: 1990 2000 and 2000 2005, however, these two models produced consistent results. Both models estimated that the annual rate of ESV decrease during the period of 2000 2005 is approximately four times higher than that in 1990 2000. This result indicates that ecosystem service valuation models are more appropriate for marginal change analysis and comparison among different periods. Finally, through exploring the relationship between economic development and change of ecosystem service values, we could not find the expected association between economic development and ecosystem degradation in the study area. In fact, the annual GDP growth during 1990 2000 (12.8%) was much higher than that during 2000 2005 (8.4%). The annual decline rate of ecosystem service values during 1990 2000, however, was only about one-fourth to one-fifth of that in 2000 2005. Although rigorous statistical tests should be carried out to examine this relationship, the results do suggest that the current economic growth with the cost of natural ecosystem is unsustainable, and a better planning practice should be taken into effect.

Environ Monit Assess (2011) 179:575 588 587 One future research direction could be the spatial explicit change analysis of the ecosystem service values. In this research, a single ESV value was obtained for a particular biome in the study area, and the detailed geographic locations with significant ESV changes are not specified. For planning purposes, however, the locations with remarkable ESV changes are important, as necessary remediation methods could be applied. Another future research may be incorporating the ESV change analysis into urban growth modeling approaches, thereby predicting the changes of ecosystem service values under different urban planning policies. Acknowledgements This research was supported by the National Natural Science Foundation of China (no. 40771195), the National Natural Science Foundation of China (no. 40871082), and the Excellent Youth Foundation of Heilongjiang Province (no. JC200714). We would also like to acknowledge the anonymous reviewers for their constructive and valuable suggestions. References Bao, Y., Wu, W., Wang, M., & Liu, W. (2007). Disadvantages and future research directions in valuation of ecosystem services in China. International Journal of Sustainable Development & World Ecology, 14, 372 381. Costanza, R., d Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., et al. (1997). The value of the world s ecosystem services and natural capital. Nature, 387, 253 260. Gu, X., Zhou, S., Zhang, H., Li, S., & Geng, Z. (2009). Temporal variation and regional differences of ecosystem service value of Nanjing City. Chinese Journal of Ecology, 28(3), 497 502 (In Chinese with English abstract). Harris, R. J., & Longley, P. A. (2000). New data and approaches for urban analysis, modeling residential densities. Transactions in GIS, 4(3), 217 234. He, Q. (2000). Statistics yearbook (2000) of Heilongjiang. Beijing: China Statistics Press. Holdren, J. P., & Ehrlich, P. R. (1974). Human population and the global environment. American Scientist, 62, 282 292. Hu, H., Liu, W., & Cao, M. (2008). Impact of land use and land cover changes on ecosystem services in Menglun, Xishuangbanna, Southwest China. Environmental Monitoring and Assessment, 146, 147 156. Kreuter, U. P., Harris, H. G., Matlock, M. D., & Lacey, R. E. (2001). Change in ecosystem service values in the San Antonio area, Texas. Ecological Economics, 39, 333 346. Li, Z. (2005). Heilongjiang statistics yearbook (2005). Beijing: China Statistics Press. Li, R., Dong, M., Cui, J., Zhang, L., Cui, Q., & He, W. (2007). Quantification of the impact of landuse changes on ecosystem services: A case study in Pingbian County, China. Environmental Monitoring and Assessment, 128, 503 510. Li, T., Li, W., & Qian, Z. (2010). Variations in ecosystem service value in response to land use changes in Shenzhen. Ecological Economics, 69, 1427 1435. Lu, D., Mausel, P., Brondízio, E., & Moran, E. (2004). Change detection techniques. International Journal of Remote Sensing, 25(12), 2365 2407. National Bureau of Statistics of China (2010). http://www. stats.gov.cn/english/. Accessed 02 September 2010. Newman, P., & Kenworthy, J. R. (1999). Sustainability and cities: Overcoming automobile dependence. Washington DC: Island Press. Odum, E. P. (1977). The emergence of ecology as a new integrative discipline. Science, 195(4284), 1289 1293 Ouyang, Z., Wang, R., & Zhao, J. (1999). Ecosystem services and their economic valuation. Chinese Journal of Applied Ecology, 10(5), 635 640 (In Chinese with English abstract). Pimm, S. L. (1997). The value of everything. Nature, 387, 231 232. Song, Y., & Ding, C. (2009). Smart urban growth for China. Cambridge: Lincoln Institute of Land Policy. Toman, M. (1998). Why not calculate the value of the world s ecosystem services and natural capital. Ecological Economics, 25, 57 60. Tong, C., Feagin, R. A., Lu, J., Zhang, X., Zhu, X., Wang, W., et al. (2007). Ecosystem service values and restoration in the urban Sanyang wetland of Wenzhou, China. Ecological Engineering, 29, 249 258. Turner, R. K., Adger, W. N., & Brouwer, R. (1998). Ecosystem services value, research needs, and policy relevance: A commentary. Ecological Economics, 25(1), 61 65. Turner, R. K., Paavola, J., Cooper, P., Farber, S., Jessamy, V., & Georgiou, S. (2003). Valuing nature: Lessons learned and future research directions. Ecological Economics, 46, 493 510. United Nation Population Division (2008). World urbanization prospects, the 2007 revision. http://www.un. org/esa/population/publications/wup2007/2007wup_ Highlights_web.pdf. Accessed 02 September 2010. Wang, S., Chi, G., Jing, C., Dong, X., Wu, C., & Li, L. (2003). Trends in road traffic crashes and associated injury and fatality in the People s Republic of China, 1951 1999. Injury Control and Safety Promotion, 10(1 2), 83 87. Wang, Z., Zhang, B., Zhang, S., Li, X., Liu, D., Song, K., et al. (2006). Change of land use and of ecosystem service values in Sanjiang Plain, Northeast China. Environmental Monitoring and Assessment, 112, 69 91. Westman, W. (1977). How much are nature s services worth. Science, 197, 960 964. Xie, G., Lu, C., & Cheng, S. (2001). Progress in evaluating the global ecosystem services. Resource Science, 23(6), 5 9.

588 Environ Monit Assess (2011) 179:575 588 Xie, G., Lu, C., Leng, Y., Zheng, D., & Li, S. (2003). Ecological assets valuation of the Tibetan Plateau. Journal of Natural Resources, 18(2), 189 196 (in Chinese with English abstract). Xuan, D. (1990). Yearbook (1990) of economic statistics of Heilongjiang. Beijing: China Statistics Press. Zhang, Y., Li, Y., & Pan, Y. (2005). Critique on nonmarket valuation of natural and environmental resources. Journal of Natural Resources, 20, 453 460 (in Chinese with English abstract). Zhang, B., Lo, W., & Xie, G. (2010). Ecosystem services research in China: Progress and perspective. Ecological Economics, 69, 1389 1395. Zhang, K. H., & Song, S. (2003). Rural-urban migration and urbanization in China: Evidence from time-series and cross-section analyses. China Economic Review, 14(4), 386 400. Zhao, Y., Jiang, Y., & Deng, W. (2009). Environmental impact analysis of land-use change in Jinkouhe District, Leshan City. Resource Development & Market, 25(5), 406 409 (in Chinese with English abstract). Zhao, B., Kreuter, U., Li, B., Ma, Z., Chen, J., & Nakagoshi, N. (2004). An ecosystem service value assessment of land-use change on Chongming Island, China. Land Use Policy, 21, 139 148.