Evolution of the global virtual water trade network

Abstract

Global freshwater resources are under increasing pressure from economic development, population growth, and climate change. The international trade of water-intensive products (e.g., agricultural commodities) or virtual water trade has been suggested as a way to save water globally. We focus on the virtual water trade network associated with international food trade built with annual trade data and annual modeled virtual water content. The evolution of this network from 1986 to 2007 is analyzed and linked to trade policies, socioeconomic circumstances, and agricultural efficiency. We find that the number of trade connections and the volume of water associated with global food trade more than doubled in 22 years. Despite this growth, constant organizational features were observed in the network. However, both regional and national virtual water trade patterns significantly changed. Indeed, Asia increased its virtual water imports by more than 170%, switching from North America to South America as its main partner, whereas North America oriented to a growing intraregional trade. A dramatic rise in China's virtual water imports is associated with its increased soy imports after a domestic policy shift in 2000. Significantly, this shift has led the global soy market to save water on a global scale, but it also relies on expanding soy production in Brazil, which contributes to deforestation in the Amazon. We find that the international food trade has led to enhanced savings in global water resources over time, indicating its growing efficiency in terms of global water use.

Fig. 1.

Evolution of global variables relative to 1986 values (percent difference). GDP and population information was from ref. . Virtual water flow and number of links information was from this study. Total crop yield of five crops information was from ref. .

Fig. 2.

Networks statistics and best functional fit in 1986 (A, C, and E) and 2007 (B, D, and F). (A and B) Volume of virtual water traded (i.e., node strength) vs. number of trade partners (i.e., node degree) and power law fit s(k) ∝ k^α (fitting with least squares method). (C and D) Node degree exceedance probability distribution with its mean value and corresponding exponential fit. (E and F) Node strength exceedance probability distribution with its mean value and corresponding stretched exponential fit.

Fig. 3.

Virtual water flows between the six world regions: Africa (Af), North America (NA), South America (SA), Asia (As), Europe (Eu), and Oceania (Oc). (A) Regional VWT network in 1986. (B) Regional VWT network in 2007. Numbers indicate the volume of VWT in cubic kilometers, and the links’ colors correspond to the exporting regions. The regional map at the bottom left provides a key to the color scheme and acronyms of the regional VWT networks. The circles are scaled according to the total volume of VWT. Note the large difference between total VWT in 1986 (A; 259 km³) and 2007 (B; 567 km³).

Fig. 4.

Evolution of important features of VWT. (A) China's VWI associated with soy over time broken down into the corresponding exporting countries. (B) Global water savings over time. The shaded area shows the total global water savings from crops and livestock (beef, poultry, and pork) trade. Individual lines show the global water savings associated with trade of that particular crop. Note that the dramatic increase in global water savings from soy trade corresponds to the increase in China's imports from more efficient countries of production (the United States, Argentina, and Brazil).