Global Environmental Change 50 (2018) 133–141

Contents lists available at ScienceDirect

Global Environmental Change

journal homepage: www.elsevier.com/locate/gloenvcha

Global phosphorus flows through agricultural trade T

a,b,c,⁎ e,1 c,d
Thomas Nesme , Geneviève S. Metson , Elena M. Bennett
a
Bordeaux Sciences Agro, Univ. Bordeaux, UMR 1391 ISPA, 33175, Gradignan Cedex, France
b
INRA, UMR 1391 ISPA, 33882, Villenave d'Ornon Cedex, France
c
McGill School of Environment, McGill University, Montreal, Quebec, Canada
d
Department of Natural Resource Sciences, McGill University, Sainte Anne de Bellevue, Montreal, Quebec, Canada
e
Division of Theoretical Biology, Department of Physics, Chemistry & Biology, and Center for Climate Science and Policy Research (CSPR), Linköping University, 581 83
Linköping, Sweden

A R T I C LE I N FO A B S T R A C T

Keywords: The global phosphorus cycle has been transformed in recent decades through increased use of mineral phos-
Phosphorus cycle phorus fertilizer in agriculture and losses to water bodies, leading to risks of fossil phosphorus resource depletion
International trade and freshwater eutrophication. By moving phosphorus resources across world regions, international trade of
Global food security agricultural products (food, feed, fiber and fuel) may contribute to these changes in the global phosphorus cycle,
Anthropocene
including critical nutrient imbalances. However, we lack a comprehensive, quantitative understanding of the
role of agricultural trade in the global phosphorus cycle. By combining detailed data on international trade and
the phosphorus content of agricultural products, we demonstrate that phosphorus flows through trade increased
nearly eight-fold from 0.4 Tg P/yr in 1961 to 3.0 Tg P/yr in 2011, leading to an increase in the fraction of
phosphorus taken up by crops that is subsequently exported from 9% in 1961 to 20% in 2011. The P flows in
traded agricultural products was equivalent to 27% of the P traded in mineral fertilizers in 2011. Agricultural P
flows were mostly driven by trade of cereals, soybeans and feed-cakes, with 28% of global phosphorus traded in
human food, 44% in animal feed and 28% in crops for other uses in 2011. We found a strong spatial pattern in
traded phosphorus in agricultural products, with most flows originating from the Americas and ending in
Western Europe and Asia, with large amounts of phosphorus moving through trade within Western Europe, in
strong contrast with the pattern of the mineral P fertilizer trade. We demonstrate that international trade of
agricultural products has affected the domestic phosphorus cycle within many countries, making phosphorus
exporters susceptible to the volatility of the mineral phosphorus fertilizer market. Overall, these results highlight
the importance of trade as key component of the global phosphorus cycle.

1. Introduction mineral P fertilizer use (Ringeval et al., 2014; Tilman et al., 2002;
Vitousek et al., 2009); this large increase in mineral fertilizer use has
Phosphorus (P) is a critical element for all living organisms; its been driven by increasing population and shifts towards more meat-
availability drives the productivity of many aquatic and terrestrial based diets, which are known to have low P use efficiency (Metson
ecosystems worldwide (Mueller et al., 2012; Peñuelas et al., 2013). In et al., 2012), in affluent, industrialised, and emerging countries
agricultural systems, additional P can be supplied to soils as mineral (Lassaletta et al., 2014b; Tilman and Clark, 2014). In other words, the
fertilizer or manure to support crop growth and sustain high yields. modern farming systems we depend on for food globally are dependent
However, mineral P fertilizer production is dependent on the physical on, and affect, the extraction and movement of phosphate rock re-
and economic availability of mined rock phosphate resources, a non- sources around the planet. In addition, P losses to water bodies through
renewable, diminishing, and geopolitically concentrated resource runoff and erosion from fertilized agricultural soils and from the in-
(Cordell et al., 2009; Van Vuuren et al., 2010). The P cycle has been adequate management of animal manure or human excreta has led to
greatly transformed since the pre-industrial era (Bennett et al., 2001; aquatic eutrophication (Carpenter et al., 1998; Schindler et al., 2008).
Liu et al., 2016; Maavara et al., 2015) through increased agricultural These issues of scarcity and pollution reflect the need to consider the

⁎
Corresponding author at: Bordeaux Sciences Agro, Univ. Bordeaux, UMR 1391 ISPA, 33175 Gradignan Cedex, France.
E-mail address: [email protected] (T. Nesme).
1
Present address: National Research Council, National Academies of Science, Washington, DC 20001and School of the Environment, Washington State Univ., Vancouver, WA 98686,
United States.

https://doi.org/10.1016/j.gloenvcha.2018.04.004
Received 21 August 2017; Received in revised form 26 January 2018; Accepted 3 April 2018
0959-3780/ © 2018 Published by Elsevier Ltd.
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

Fig. 1. Total P flows through international trade of agricultural products (in Tg P/yr) from 1961 to 2011.

sustainability of the global P cycle (Cordell et al., 2012) and call for the about food security and agriculture, globally (MacDonald et al., 2015).
parsimonious use of mined P resources, limited P accumulation in Here we quantify the magnitude of P flows through agricultural product
agricultural soils, increased P use efficiency, and recycling wherever trade at the global scale from 1961 to 2011, as well as the spatial
possible (Withers et al., 2015). patterns of these trade flows in 2011 using FAO data to start to fill the
The modern era is characterized by intensive trade of agricultural existing knowledge gap related to the importance of trade on the global
products as food, feed, fibre and fuel, which has accelerated dramati- P cycle. More specifically, we focus on a descriptive and quantitative
cally during the last decades (Krausmann et al., 2008) due to falling analysis as a first step towards understanding the impact of modern
transportation costs, trade liberalisation, and the shift of most agri- agricultural trade and we:
cultural production systems towards economies of scale in in-
dustrialised and emerging countries (Frankel and Rose, 2005). For in- i) estimate changes in global P flows through trade
stance, the global wheat and rice trade increased by 42% and 90% in ii) assess the relative contribution of food, feed and other agricultural
just the 17 years from 1992 to 2009, respectively (Puma et al., 2015). product trade to global P flows
The effect of such trade on rock phosphate resource conservation is iii) compare agricultural product trade P flows to the magnitude and
likely to be large because it can lead to major displacement of P re- patterns of mineral P fertilizer trade flows
sources embedded in agricultural products, limiting opportunities to iv) preliminary assess the consequences of agricultural product trade
recycle P resources locally, e.g. through recycling of P in animal manure for the global P cycle.
back to croplands that produce feed (Schipanski and Bennett, 2012).
However, although there are studies of the intensity of agricultural
2. Methods
product trade (Kastner et al., 2014; Krausmann et al., 2008; MacDonald
et al., 2015), and of phosphate rock trade (Jasinski and USGS, 2011),
Not all agricultural products exhibit the same P concentration making
the effects of agricultural trade on the global P cycle have never been
biomass flow a poor indicator of the effect of trade on the global P cycle.
quantified. Two decades ago, Bouwman and Booij (1998) examined P
For instance, P concentration in palm oil is very low but palm oil is one of
feed in trade, and more recently Wang et al. (2017) examined P flows
the most heavily traded products. In other words, converting all biomass
associated with agricultural trade in a few selected countries, but we
flows into a single unit is necessary to estimate the consequences that
still lack a comprehensive and quantitative understanding of how P
global trade has for the global P cycle. We calculated P flows through
flows associated with agricultural trade have affected the global P cycle
agricultural product trade among countries by multiplying the weight of
or the impact this has on overall P sustainability. In particular, an ac-
agricultural products traded (imported and exported) among countries by
curate estimation of how much P is displaced between world regions
their respective P contents (Food Standards Agency, 2002). We obtained
and countries as traded products is needed. Comparing the magnitude
the quantities traded among countries for 397 agricultural products from
and pattern of P flows in traded harvested crop products with the use
the FAOSTAT trade module (http://faostat3.fao.org/download/T/*/E,
and trade of mineral P fertilizers can help us to better understand the
accessed on July 3rd, 2017). The FAOSTAT trade module provides annual
relative contribution of agricultural product trade on the modern global
data on total import and export of crop and livestock products for each
P cycle. This, in turn, is important for highlighting the spatial patterns
FAO country (n = 203) from 1961 to 2011 and on detailed import/export
of such flows and their drivers, and ultimately for assessing the role
of crop and livestock products among pairs of FAO countries (e.g., from
agricultural trade may play in increasing (or decreasing) sustainable P
country i to country j) from 1986 to 2011. In this study, we used data on
management (West et al., 2014). Identifying potential P scarcity and
material import only because these data have been reported to be more
water quality “winners” and “looser” in an increasingly globalized
reliable that export data (see SI). The trade data we used does not account
world is can also help make more informed P management decisions
for food aid (but this represents a very small amount of global trade, see

134
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

Fig. 2. P flows through gross export (a), gross import (b) and net budget (c) of agricultural products (in Mg P/yr) in 2011. Net agricultural product P budget was
calculated as the difference between gross agricultural product P import minus gross agricultural product P export.

SI), and does not always account for re-export of products (which is also while the data on mineral P fertilizer trade (as MAP, DAP, TSP, P acid
often reported to be minimal, see SI). The P content of the 397 agricultural and P rock) were obtained from personal communication by the In-
products considered in this study and associated data sources are given in ternational Fertilizer Association (see SI for additional details on data
Table S 1. All the data treatments were performed with R (R Development treatment and Table S 2 for the corresponding full dataset). We also
Core Team, 2009). compared the total annual global P flow values of traded crops to P in
global crop production, for years 1961 and 2011 (data on crop pro-
duction were obtained from the ‘production’ module of FAOSTAT3).
2.1. Total P flow through international trade of agricultural products

Total P flow through global trade of agricultural products was cal- 2.2. Country agricultural P flows through international trade
culated annually from 1961 to 2011 by summing the P flows for all the
imported crop and livestock products and for all the countries in each The P flow through import and export of agricultural products for
year and was then divided into categories by classifying each of the each country was calculated using the FAOSTAT data on agricultural
crop and livestock products into one of the following eight categories: product import per country for the year 2011. The year 2011 was se-
soybeans (including soybean cakes); pulses (other than soybeans); fruits lected because it corresponds to the last year for each all the informa-
and vegetables; forages; cereals (including grain maize, sorghum, ra- tion about trade were available in a consistent manner at the time of the
peseed and sunflower); feed cakes (excluding soybean cakes); animal study. We checked that this year was not too specific (from a global
products (meat, egg, milk and dairy products, honey, skins, wool, trade point of view). In particular, according to FAOSTAT, we found
hairs); other (including spices, industrial and fibre crops). that total import of agricultural products corresponded to 1357 T US$
In order to estimate the contribution of trade to P resource dis- in 2011, just 6% higher than the average 1278 T US$ for the 2010–2012
placement, we compared the total global P flow to both the amount of years. In addition, according to IFADATA, mineral P fertilizer use in
mineral P fertilizer used at the global scale (for years 1961 and 2011) 2011 was just 0.2% lower than the average 18.9 Tg P for the 2010–2012
and to the amount of mineral P fertilizer that was traded (for year years. This makes the year 2011 representative of the global trade at the
2011). Data on mineral P fertilizer use were obtained from IFADATA2 beginning of the 2010′s.

2 3
http://ifadata.fertilizer.org/ucSearch.aspx (accessed on June, 6th, 2017) http://www.fao.org/faostat/en/#data/QC (accessed on June, 6th, 2017)

135
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

Fig. 3. P inflow through import of agricultural

products expressed as a percentage of domestic
mineral P fertilizer use in 2011. For instance,
values of 800 for a given country mean that P
inflow through import of crop and livestock
products in that country is eight time bigger
than the amount of P applied to domestic soils
as mineral fertilizers.

The gross P import (and export) was calculated by summing the P 200 exporter countries). The total P flow estimated from detailed
inflows (and outflows) over all crop and livestock products imported agricultural trade among pairs of countries amounted to 2.5 Tg P
(and exported) by each country while the net P budget was calculated whereas it should have been equal to 2.8 Tg P (3.0 Tg P × 95%, while
as the gross P import minus the gross P export. The P import and export 3.0 Tg P amounted to the total P flow through global trade of agri-
were expressed both per country and per area under arable and per- cultural products, see Section 3.1). However, to our knowledge, the
manent crop of each country. The data on arable and permanent crop trade matrix provided by FAOSTAT is the most comprehensive database
area were collected from the ‘inputs’ module of FAOSTAT4 for year on agricultural product exchange. Although incomplete, it is the best
2011. Finally, the P import through trade of agricultural products was existing way to draw a global picture of P flows in trade.
compared to the domestic mineral P fertilizer use. Data on mineral P
fertilizer use by country were also extracted from IFADATA. 3. Results

2.3. P flows through trade of agricultural products among pairs of countries 3.1. Total P flow through international agricultural product trade

To estimate P flows among agricultural product trade partners, we We found that the total P flow through international agricultural
summed the P contained in the specific crops and livestock products trade increased from 0.4 Tg P in 1961 to 3.0 Tg P in 2011 (Fig. 1), a
imported by a country, separately for each country it traded with. The nearly eight-fold increase, with an acceleration in the last ten years.
data on agricultural product exchanges among pairs of countries were International agriculture trade played an increasing role in the global
obtained from FAOSTAT for the year 2011. The agricultural products anthropogenic P cycle; P flows in trade of agricultural products were
considered here were limited to the 67 most important products that equivalent to just 9% of all P contained in mineral fertilizer used an-
account for 95% of total P flows through global trade in 2011 (the list is nually at the global scale in 1961 but jumped to 17% by 2011. The same
given in Table S 3). The P flows through trade of agricultural products year, P flows in agricultural product trade were equivalent to 27% of
among pairs of countries were then disaggregated into three categories: the P that was traded in mineral fertilizer at the global scale (which
food, feed, and other uses (e.g., fibre, fuel, seeds, losses and probably amounted to 11.7 Tg P). Perhaps most importantly, of the P taken up by
some poorly documented uses) by using the ‘food balance sheet’ module crops globally (excluding grasslands), the 9% subsequently exported on
of FAOSTAT (see Section 1.2 in SI). Phosphorus flows among pairs of the international market in 1961 became 20% by 2011.
countries were aggregated by world regions (the aggregation key is We found that global P flows through trade of crop products were
provided in Table S 4). Finally, we compared these P flows in agri- driven by a limited set of commodities: exchanges of cereals, soybeans
cultural products to the P flows in traded mineral fertilizers in terms of and feed cakes represented more than 80% of total P flows through
magnitude and directions, and by world regions. Data were presented international trade over the whole 1961–2011 period (respectively 39,
using circular plots which were elaborated following the ‘Circlize’ R 32 and 10% of total P flows in 2011, see Fig. 1; the list of the top 10
method (Abel and Sander, 2014; Krzywinski et al., 2009). The plots products in terms of P flows is given in Table S 5).
were filtered by removing P flows lower than the 0.65 quantile. Pair
agricultural product trade data was not available for all countries from
3.2. Country-level P budgets are affected by international trade
FAOSTAT, with data gaps especially in Africa (Lassaletta et al., 2014a),
making our analysis of trade incomplete (we cover 123 importer and
We found that the structure of the P flows through international
trade of agricultural products was clearly uneven: in 2011, just 11
4
http://www.fao.org/faostat/en/#data/RL (accessed on June, 6th, 2017) countries contributed 75% of total exported P in agricultural products

136
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

Fig. 4. P flows through agricultural product trade among world regions (in Tg P/yr) in 2011. The origins and destinations of P flows are each assigned a colour and
represented by the circle’s segments. The direction of the flow is represented by the origin country’s colour. The intensity of P exchange is indicated by the width of
the flow. Because the flow width is nonlinearly adapted to the curvature, it corresponds to the flow size only at the beginning and end points. Tick marks on the circle
segments show the total inflows and outflows. Because the figure was filtered in order to remove small, insignificant flows, some flows may not appear in the Figure
(e.g., from Finland within Western Europe as a P exporter).

(Fig. 2a) while the same amount of P was imported by 29 countries gross P export through trade of agricultural products. Second, P imports
(Fig. 2b and Table S 6). Net agricultural P exports were observed in through agricultural product trade represented a significant P inflow for
major agricultural exporters such as Brazil, Argentina, USA, Canada, many countries, often much greater than domestic P fertilizer use for
France, Ukraine, Russia, India and Australia (Fig. 2c) while net agri- countries in Western and Northern Europe and in Africa (Fig. 3), which
cultural P imports were observed in countries with dense population have intensive imports of feedstuff (Western Europe) or very low mi-
(China and Japan), intensive livestock industry (Northern and Western neral P fertilizer use (Sub-Saharan Africa). For those countries, P im-
Europe) or with intense food security challenges, in particular around ports through trade of agricultural products exert a strong influence on
the Mediterranean basin (Fader et al., 2013). Such net P imports the domestic P cycle by accumulating (as animal manure and human
through agricultural trade may be affecting the accumulation of P in waste) in soils and urban areas, and by leaching to water bodies and the
countries, although it is also important to consider the potential re- environment (Grote et al., 2005).
export of P through processed products, which we did not consider
here. Those results are only slightly modified when expressed per area 3.3. Structure of the bilateral P flows in agricultural products among regions
of arable and permanent crop (Fig. S 1). In particular, they show that
China has in fact a small net P import and that Western Europe coun- We found a strong spatial pattern of bilateral agricultural product P
tries have both positive and negative net P budgets. flows through trade that resulted in unbalanced, long-distance P re-
We also found that international trade of agricultural products sig- source displacement at the global scale. North America clearly emerged
nificantly affected domestic P cycles for many countries. First, for most as a P supplier through agricultural trade to Asia (and a few other
countries, sizeable P exports through agricultural trade were associated countries), while South America contributed specifically to P inflows
with high levels of domestic mineral fertilizer use (Fig. S 2) making into Western Europe and Asia (Fig. 4). As such, Asia emerged as a large
these exporting countries also large users of mineral fertilizers. For P sink fed by imports of agricultural products from North and South
instance, we estimated that 85% of the countries that are in the upper America as well as a large number of other regions, highlighting that
quartile of domestic P fertilizer use are also in the upper quartile of this continent is highly dependent on agricultural imports. The same

137
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

Fig. 5. P flows through trade of mineral P fertilizers among world regions (in Tg P/yr) in 2011. See Fig. 4 for reading details.

holds true, but to a lesser extent, for Central America which appeared Morocco) appeared as a large P exporter (mostly as P rock and P acid)
highly connected to North America through imports. Our results also together with Central Asia (mostly China and Russia). South Asia
showed that while Western Europe imported significant amount of P (mostly India), and to a lesser extent Europe and South America, ap-
through trade of agricultural products, important flows remained in- peared as large mineral P sinks, highligthing their dependance towards
ternal to this region, probably due to intense exchanges among a high P fertilizer exporting countries. The traded P flows through fertlizer
number of small countries. Finally, our results confirmed that Africa were also larger (11.1 Tg) than those traded through agricultural pro-
played only a minor role in agricultural products P flows globally by ducts (3.0 Tg).
being mostly an importer of P in agricultural products of moderate
intensity (the detailed trade among individual countries is given in Fig. 4. Discussion
S 3).
The pattern of P flows through trade of agricultural products among 4.1. P flows in agricultural product international trade are an increasingly
regions differed when disaggregated into food, feed and other uses. P important component of global nutrient cycling
flows in food trade were dominated by exports from North America to
Asia and internal cycling within Europe (Fig. S 4a) whereas feed flows Our results demonstrate that global P flows through international
were mostly due to export from South America to Western Europe as trade of agricultural products have become an important feature of the
well as internal cycling within Europe (Fig. S 4b). P flows through trade global P cycle, accounting for 20% of the P in global crop production,
for other uses (mostly fibre, fuel and seeds; see SI) were mostly driven 17% of the P globally used as mineral fertilizer, and 27% of the P that
by imports from Western and Central Asia (Fig. S 4c). In total, 0.7 Tg P was traded as mineral fertilizers in 2011. The international trade of
(28% of agricultural global P flows) were exchanged through trade of agricultural products therefore contributes strongly to P resource dis-
food products while trade for feed and other uses amounted to 1.1 placements among world regions, for instance, by removing 0.7 Tg P
(44%) and 0.7 Tg P (28%) in 2011, respectively. from North America through agricultural exports and adding 0.8 Tg P to
In contrast with P flows through trade of agricultural products, we the food systems in Central Asia through imports. Trade of agricultural
found a substantially different –and in fact less diverse- structure of P products is making more P available for recycling in some regions,
flows through trade of mineral P fertilizers (Fig. 5). Africa (mostly while in other places, these opportunities are reduced. Perhaps more

138
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

importantly, P flows through agricultural trade exhibit a different nutrient management laws, P imported as fertilizer and food has been
structure than P flows in fertilizer trade, implying that agricultural associated with local eutrophication (Asmala et al., 2011). Con-
product trade has a unique imprint on the global P cycle. This imprint centrated animal production comes with similar risks. For example, by
amplifies the strong anthropogenic signature of the global P cycle that specialising in animal production, Western Europe imported 0.4 Tg P/
has emerged over the last few decades (Liu et al., 2016;Ringeval et al., yr from abroad, amounting to a surprisingly large amount (16%) of
2014). For instance, the sharp differences among countries in terms of global P flows through agricultural trade (Fig. S 4b). As an animal
net agricultural trade P budget that we found (Fig. 2c), if prolonged feeding region which has experienced historically massive manure
over multiple years, could contribute to imbalanced P budgets across spreading on agricultural soils, continued import of P as feed sup-
world croplands (MacDonald et al., 2011; Vitousek et al., 2009) and to porting livestock production, if prolonged over years, could exacerbate
the build-up of soil P stocks in import-intensive regions (Gerber et al., already high soil P (Lemercier et al., 2008; Matos-Moreira et al., 2017)
2014; Sattari et al., 2012). and thus possibly contributing to losses through runoff and aquatic
The effect of trade on other nutrient and elemental cycles is also ecosystem eutrophication.
large. A quarter of reactive nitrogen emissions, for example, are asso- Growing food and feed for export has mixed and variable impacts on
ciated with international trade, and consumption habits (in terms of eutrophication risk, depending on other aspects of local context. On the
crops, animals, and textiles) in developed countries can be linked to one hand, in the US, a quarter of P fertilizer use is associated with
nitrogen related pollution in developing countries (Oita et al., 2016). export production, especially of crops used as feed such as corn and soy
Similarly, 23% of global carbon dioxide emissions from fossil fuel (MacDonald et al., 2012). The Mississippi River Basin, where most of
burning is associated with international trade; almost a quarter of these feed crops are grown, continues to contribute nutrients to the Gulf
China’s emissions are a result of producing goods for other nations of Mexico, the second largest hypoxic zone in the world (Diaz and
(Davis and Caldeira, 2010). Meat produced in Brazil and Argentina for Rosenberg, 2008; Jacobson et al., 2011). In this case agricultural pro-
export to Russia is associated with large embodied methane and nitrous duction and trade could be linked to increased local pollution. In con-
oxide emissions, powerful greenhouse gases (Caro et al., 2014). These trast, South American production of feed crops for export, including to
examples, together with our results, highlight that the spatial decou- Western Europe, may have limited water quality implications in the
pling of consumption from production of agricultural (and non-agri- feed production region. For example, P fertilizer application to soybean
cultural) products has had large effects on global biogeochemistry. fields in Mato Grosso Brazil has not resulted into local losses to wa-
One should note that our study has focussed on international trade terways as P accumulates in stable soil P pools (Riskin et al., 2013a).
only and, thus, has not fully addressed the effect of total (i.e. including Finally, in some cases international trade of food crops can over-
also sub-national) trade on the global P cycle. Although the number of come local agricultural P scarcity. For example, P flows through trade
studies that considered sub-national trade of agricultural products is contribute to P inflows that are much greater in magnitude than the
limited, they have shown that specialization of agricultural systems at amount of mineral P fertilizer used domestically in many poorer
the sub-national scale drives large amounts of feed and food transpor- countries (Fig. 3). If prolonged, this type of trade may help compensate
tation, with strong consequences on the national nutrient cycles and for local low soil fertility. For example, the annual 0.18 Tg P inflow
recycling impediments (Le Noé et al., 2016; Senthilkumar et al., 2012). through trade of agricultural products to Africa –a continent that often
However, the characterization of sub-national trade in consistent, exhibits large negative soil P budgets (Cobo et al., 2010; Sanchez,
complete, and homogeneous databases is rare, making assessing this 2002)- is significant compared to the 0.66 Tg P/yr soil deficit reported
information at the global scale virtually impossible. One should also by previous studies (Sheldrick et al., 2002).
note that we might have over-estimated the effect of trade by con-
sidering international trade between individual countries within 4.3. Implications of international trade for the sustainability of the global P
Europe. Those countries are geographically small but belong to the cycle
same supra-national federation. Removing intra-regional exchanges
within Western and Eastern Europe would lead to a 0.47 Tg P (i.e. 19%) Global agricultural trade shapes our capacity to manage P sustain-
reduction in P flows through agricultural trade. However, similar intra- ably. Our results indicate that nearly half of traded agricultural product
regional exchanges probably occur between states or provinces within P flows occur in feed crops, highlighting the key role of livestock pro-
big countries such as the US, India or China but we did not account for duction and agricultural specialization on the global P cycle.
them in this study because of a lack of standardized data. Agricultural specialization permitted by –and caused by– international
trade has in fact mixed effects on the sustainability of the global P cycle.
4.2. Potential local P implications of international agricultural trade On the one hand, it may increase overall P fertilizer use efficiency if
traded products flow from countries with high P fertilizer use efficiency
An increasingly interconnected global food system, with large flows towards countries with lower P use efficiency (Martinez-Melendez and
of both food and feed, has resulted in part from increasingly spatially Bennett, 2016; Schipanski and Bennett, 2012). On the other hand, such
decoupled food production and food consumption. Several studies have specialization may also reduce our capacity to recycle P back onto
shown that this decoupling is linked to global urbanisation (Cumming agricultural land because it physically removes P in agricultural pro-
et al., 2014), locally limited availability of land and water to support ducts (and resulting waste) from food production. Recoupling crop and
agricultural production (Fader et al., 2013), and the emergence of livestock production at the local scale, that is making sure that feed
landless animal production systems and highly specialised crop farming crops are grown in proximity to the livestock that eat them, is of course
regions (Metson et al., 2016; Naylor et al., 2005; Roe et al., 2002; a central issue if we want to maximize manure recycling, and as such
Senthilkumar et al., 2012). For global agricultural P flows, this has decrease dependence on P rock and on long distance international trade
translated into P in food flowing towards densely populated areas, and of fertilizers and crops (Garnier et al., 2016; Nowak et al., 2015).
P in feed flowing to livestock intensive regions (Fig. 2b). Technologies, economic incentives and public policies are required to
This concentration of P waste (urban wastewater and manure), in move towards such recoupling, also accounting for side-environmental
part made possible by trade, is associated with increased risk of eu- and economic effects (Metson et al., 2016; Regan et al., 2017).
trophication. Urban centres are the most important contributor to losses Finally, we identified a strong inter-connection of P flows through
of dissolved P to the oceans globally (Harrison et al., 2010). And al- the trade of agricultural products and mineral fertilizers among world
though the risk of P from wastewater reaching water bodies is higher regions (Figs. 4 and 5) reinforcing the need to consider the planet as an
where there are limited local waste treatment capacities, even in re- inter-connected system when it comes to nutrient flows (MacDonald,
gions such as the Baltic Sea, with high sewer connectivity and strict 2013). Although traded P flows through agricultural products are

139
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

smaller than those associated with P fertilizers, they are still key to with R modelling. We also thank Bruno Ringeval and Sylvain Pellerin
better understanding the vulnerability and resilience of regions to P their comments on an earlier version of this manuscript and Graham
scarcity. In fact, these agricultural product trade flows add to the global MacDonald for fruitful discussions about trade. This work was funded
interconnectedness associated with P fertilizer flows as they represent by grants from Bordeaux Sciences Agro (Univ. Bordeaux) and the
the next step in the food production chain which ranges from soil fer- McGill School of Environment during TN's sabbatical, as well as by an
tilization to food product delivery. That is, a country which is not im- NSERC Discovery Grant to EMB.
porting large amounts of P fertilizers may still be importing large
amount of P in agricultural products from a country that is itself de- Appendix A. Supplementary data
pendent on imported P fertilizers, making both countries ultimately
vulnerable to changes in the country exporting the original P fertilizer Supplementary material related to this article can be found, in the
or P-rich rock that will be made into fertilizer. Such tele-connections –in online version, at doi:https://doi.org/10.1016/j.gloenvcha.2018.04.
addition to differences in spatial patterns of P flows through agri- 004.
cultural products vs P fertilizer trade– can make it difficult to identify
the impact of food and feed use on global rock phosphate depletion References
risks (Galloway et al., 2007; MacDonald et al., 2012; Matsubae et al.,
2011), but point to increased vulnerabilities in the global system Abel, G.J., Sander, N., 2014. Quantifying global international migration flows. Science
(Helbing 2013). While inter-connections among countries may provide 343, 1520–1522.
Asmala, E., Saikku, L., Vienonen, S., 2011. Import–export balance of nitrogen and
some resilience in the global P cycle (e.g., by providing different outlets phosphorus in food, fodder and fertilizers in the Baltic Sea drainage area. Sci. Total
or sources for exports and imports, respectively), the unbalanced nature Environ. 409, 4917–4922.
of the global food system (as shown in Fig. 2a and b) may also increase Bennett, E.M., Carpenter, S.R., Caraco, N.F., 2001. Human impact on erodable phos-
phorus and eutrophication: a global perspective. BioScience 51, 227–234.
risks of systemic vulnerability (Helbing, 2013; Puma et al., 2015). For Bouwman, A.F., Booij, H., 1998. Global use and trade of feedstuffs and consequences for
instance, some countries import large amounts of P as food (e.g., in the nitrogen cycle. Nutr. Cycl. Agroecosyst. 52, 261–267.
North Africa and Central Asia) or feed (e.g., in Western Europe), re- Caro, D., et al., 2014. CH4 and N2O emissions embodied in international trade of meat.
Environ. Res. Lett. 9 (11), 114005. Available at: http://stacks.iop.org/1748-9326/9/
sulting in strong dependence towards their partners and towards the
i=11/a=114005?key=crossref.c93488ee9a16fd241ae31ac3a66c1182.
global market for their food and feed supply (Fader et al., 2013). Si- Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, V.H.,
milar risks exist for major exporting countries with limited numbers of 1998. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl.
8, 559–568.
importing partners. In addition, these exporting countries are often
Cobo, J.G., Dercon, G., Cadisch, G., 2010. Nutrient balances in African land use systems
strongly dependent on P fertilizer use (Fig. S 2) and on P fertilizer across different spatial scales: a review of approaches, challenges and progress. Agric.
imports from foreign partners (Fig. 5). Such patterns highlight the Ecosyst. Environ. 136, 1–15.
vulnerability of both exporting and importing countries towards the Cordell, D., Drangert, J.-O., White, S., 2009. The story of phosphorus: global food security
and food for thought. Glob. Environ. Change 19, 292–305.
global market and P cycle, in particular in a context of high volatility of Cordell, D., Neset, T.-S.S., Prior, T., 2012. The phosphorus mass balance: identifying
P fertilizer and agricultural products prices (Daviron et al., 2011). Mi- ‘hotspots' in the food system as a roadmap to phosphorus security. Curr. Opin.
tigating such vulnerability would require enhanced food sovereignty Biotechnol. 23, 839–845.
Cumming, G.S., Buerkert, A., Hoffmann, E.M., Schlecht, E., von Cramon-Taubadel, S.,
(Puma et al., 2015), although such practices can lower domestic P use Tscharntke, T., 2014. Implications of agricultural transitions and urbanization for
efficiency (Schipanski and Bennett, 2012). ecosystem services. Nature 515, 50–57.
Daviron, B., Nango Dembele, N., Murphy, S., Rashid, S., 2011. Price Volatility and Food
Security. The High Level Panel of Experts on Food Security and Nutrition of the
5. Conclusion Committee on World Food Security, Rome p. 83.
Davis, S.J., Caldeira, K., 2010. Consumption-based accounting of CO2 emissions. Proc.
Global P flow through international trade of agricultural products Natl. Acad. Sci. 107, 5687–5692.
Diaz, R.J., Rosenberg, R., 2008. Spreading dead zones and consequences for marine
has become an increasingly important component of the global P cycle, ecosystems. Science 321, 926–929.
increasing by more than 750% between 1961 and 2011, mostly due to Fader, M., Gerten, D., Krause, M., Lucht, W., Cramer, W., 2013. Spatial decoupling of
trade in cereals, soybeans and soybean cakes. Nearly half of this trade agricultural production and consumption: quantifying dependences of countries on
food imports due to domestic land and water constraints. Environ. Res. Lett. 8,
occurred in animal feed, a quarter was in human food, and another
014046.
quarter in all trade of all other agricultural products that contain P. Food Standards Agency, 2002. McCance and Widdowson’S The Composition of Foods, 6th
Food and feed imports are major P inflows for some countries, some- summary edition. The Royal Society of Chemistry, Cambridge (UK).
times in much greater quantities than mineral P fertilizer used do- Frankel, J.A., Rose, A.K., 2005. Is trade good or bad for the environment? Sorting out the
causality. Rev. Econ. Stat. 87, 85–91.
mestically. We found a strong spatial pattern in P flows through trade of Galloway, J.N., Burke, M., Bradford, G.E., Naylor, R., Falcon, W., Chapagain, A.K.,
agricultural products, with most flows originating from North and Gaskell, J.C., McCullough, E., Mooney, H.A., Oleson, K.L.L., Steinfeld, H., Wassenaar,
South America and ending in Western Europe and Asia, in addition to T., Smil, V., 2007. International trade in meat: the tip of the pork chop. Ambio 36,
622–629.
large amounts of P moving through agricultural trade within Western Garnier, J., et al., 2016. Reconnecting crop and cattle farming to reduce nitrogen losses to
Europe. This spatial pattern is in strong contrast with that observed for river water of an intensive agricultural catchment (Seine basin, France): past, present
mineral P fertilizer, implying that agricultural product trade has a un- and future. Environ. Sci. Policy 63, 76–90. Available at: https://doi.org/10.1016/j.
envsci.2016.04.019.
ique imprint on the global P cycle. Together, these results suggest that Gerber, P.J., Uwizeye, A., Schulte, R.P.O., Opio, C.I., de Boer, I.J.M., 2014. Nutrient use
large scale P movements may make it increasingly difficult to recycle P efficiency: a valuable approach to benchmark the sustainability of nutrient use in
resources from human, animal, and food waste streams back to the global livestock production? Curr. Opin. Environ. Sustain. 9–10, 122–130.
Grote, U., Craswell, E., Vlek, P., 2005. Nutrient flows in international trade: ecology and
agricultural lands than need it, and that trade may increase the vul- policy issues. Environ. Sci. Policy 8, 439–451.
nerability of both exporting and importing countries to geopolitically Harrison, J.A., et al., 2010. Magnitudes and sources of dissolved inorganic phosphorus
concentrated mineral P resources. Trade of agricultural products is a inputs to surface fresh waters and the coastal zone: a new global model. Glob.
Biogeochem. Cycles 24 (1), GB1003. http://dx.doi.org/10.1029/2009GB003590.
key piece in the puzzle of the modern P cycle, and policies to improve P
Helbing, D., 2013. Globally networked risks and how to respond. Nature 497, 51–59.
management must consider trade to achieve full success. Jacobson, L.M., David, M.B., Drinkwater, L.E., 2011. A spatial analysis of phosphorus in
the Mississippi River Basin. J. Environ. Qual. 40 (3), 931–941.
Acknowledgments Jasinski, S.M., USGS, 2011. Phosphate Rock. USGS. Available at:http://minerals.usgs.
gov/minerals/pubs/commodity/phosphate_rock/mcs-2011-phosp.pdf.
Kastner, T., Erb, K.-H., Haberl, H., 2014. Rapid growth in agricultural trade: effects on
We thank Michel Prud’Homme and the International Fertilizer global area efficiency and the role of management. Environ. Res. Lett. 9, 034015
Association for providing detailed data about mineral P fertilizer trade. (034010p).
Krausmann, F., Erb, K.-H., Gingrich, S., Lauk, C., Haberl, H., 2008. Global patterns of
We are grateful to Adrien Lepoutre and Delphine Renard for their help

140
T. Nesme et al. Global Environmental Change 50 (2018) 133–141

socioeconomic biomass flows in the year 2000: a comprehensive assessment of Godderis, Y., Hinsinger, P., Llusia, J., Nardin, E., Vicca, S., Obersteiner, M., Janssens,
supply, consumption and constraints. Ecol. Econ. 65, 471–487. I.A., 2013. Human-induced nitrogen–phosphorus imbalances alter natural and man-
Krzywinski, M., Schein, J., Birol, I., Connors, J., Gascoyne, R., Horsman, D., Jones, S.J., aged ecosystems across the globe. Nat. Commun. 4, 2934.
Marra, M.A., 2009. Circos: an information aesthetic for comparative genomics. Puma, M.J., Bose, S., Chon, S.Y., Cook, B.I., 2015. Assessing the evolving fragility of the
Genome Res. 19, 1639–1645. global food system. Environ. Res. Lett. 10, 024007.
Lassaletta, L., Billen, G., Grizzetti, B., Garnier, J., Leach, A.M., Galloway, J., 2014a. Food Development, R., Team, Core, 2009. R: a Language and Environment for Statistical
and feed trade as a driver in the global nitrogen cycle: 50-year trends. Computing. R Foundation for Statistical Computing, Vienna.
Biogeochemistry 118, 225–241. Regan, J.T., et al., 2017. Does the recoupling of dairy and crop production via coopera-
Lassaletta, L., Billen, G., Romero, E., Garnier, J., Aguilera, E., 2014b. How changes in diet tion between farms generate environmental benefits? A case-study approach in
and trade patterns have shaped the n cycle at the national scale: Spain (1961–2009). Europe. Eur. J. Agron. 82, 342–356.
Reg. Environ. Change 14, 785–797. Ringeval, B., Nowak, B., Nesme, T., Delmas, M., Pellerin, S., 2014. Contribution of an-
Le Noé, J., et al., 2016. La place du transport de denrées agricoles dans le cycle thropogenic phosphorus to agricultural soil fertility and food production. Glob.
biogéochimique de l'azote en France: un aspect de la spécialisation des territoires. Biogeochem. Cycles 28.
Cahiers Agric. 25 (1), 15004. Riskin, S.H., Porder, S., Neill, C., Figueira, A.Me.S., Tubbesing, C., Mahowald, N., 2013a.
Lemercier, B., Gaudin, L., Walter, C., Aurousseau, P., Arrouays, D., Schvartz, C., Saby, The fate of phosphorus fertilizer in Amazon soya bean fields. Philos. Trans. R. Soc. B
N.P.A., Follain, S., Abrassart, J., 2008. Soil phosphorus monitoring at the regional 368.
level by means of a soil test database. Soil Use Manage. 24, 131–138. Roe, B., Irwin, E.G., Sharp, J.S., 2002. Pigs in space: modeling the spatial structure of hog
Liu, X., Sheng, H., Jiang, S., Yuan, Z., Zhang, C., Elser, J.J., 2016. Intensification of production in traditional and nontraditional production regions. Am. J. Agric. Econ.
phosphorus cycling in China since the 1600s. Proc. Natl. Acad. Sci. 113, 2609–2614. 84, 259–278.
Maavara, T., Parsons, C.T., Ridenour, C., Stojanovic, S., Dürr, H.H., Powley, H.R., Van Sanchez, P.A., 2002. Soil fertility and hunger in Africa. Science 295, 2019–2020.
Cappellen, P., 2015. Global phosphorus retention by river damming. Proc. Natl. Acad. Sattari, S.Z., Bouwman, A.F., Giller, K.E., Van Ittersum, M.K., 2012. Residual soil phos-
Sci. 112, 15603–15608. phorus as the missing piece in the global phosphorus crisis puzzle. Proc. Nat. Acad.
MacDonald, G.K., 2013. Eating on an interconnected planet. Environ. Res. Lett. 8, Sci. 109, 6348–6353.
021002. Schindler, D.W., Hecky, R.E., Findlay, D.L., Stainton, M.P., Parker, B.R., Paterson, M.J.,
MacDonald, G.K., Bennett, E.M., Carpenter, S.R., 2012. Embodied phosphorus and the Beaty, K.G., Lyng, M., Kasian, S.E.M., 2008. Eutrophication of lakes cannot be con-
global connections of United States agriculture. Environ. Res. Lett. 7, 044024. trolled by reducing nitrogen input: results of a 37-year whole-ecosystem experiment.
MacDonald, G.K., Bennett, E.M., Potter, P.A., Ramankutty, N., 2011. Agronomic phos- Proc. Natl. Acad. Sci. 105, 11254–11258.
phorus imbalances across the world’s croplands. Proc. Natl. Acad. Sci. 108, Schipanski, M.E., Bennett, E.M., 2012. The influence of agricultural trade and livestock
3086–3091. production on the global phosphorus cycle. Ecosystems 15, 256–268.
MacDonald, G.K., Brauman, K.A., Sun, S., Carlson, K.M., Cassidy, E.S., Gerber, J.S., West, Senthilkumar, K., Nesme, T., Mollier, A., Pellerin, S., 2012. Regional-scale phosphorus
P.C., 2015. Rethinking agricultural trade relationships in an era of globalization. flows and budgets within France: the importance of agricultural production systems.
BioScience 65, 275–289. Nutr. Cycl. Agroecosyst. 92, 225–236.
Matos-Moreira, M., et al., 2017. High-resolution mapping of soil phosphorus concentra- Sheldrick, W.F., Syers, J.K., Lingard, J., 2002. A conceptual model for conducting nutrient
tion in agricultural landscapes with readily available or detailed survey data. Eur. J. audits at national, regional, and global scales. Nutr. Cycl. Agroecosyst. 62, 61–72.
Soil Sci. 68, 281–294. Tilman, D., Clark, M., 2014. Global diets link environmental sustainability and human
Martinez-Melendez, L.A., Bennett, E.M., 2016. Trade in the US and Mexico helps reduce health. Nature 515, 518–522.
environmental costs of agriculture. Environ. Res. Lett. 11, 55004. Available at: Tilman, D., Gassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agricultural
https://doi.org/10.1088/1748-9326/11/5/055004. sustainability and intensive production practices. Nature 418, 671–677.
Matsubae, K., Kajiyama, J., Hiraki, T., Nagasaka, T., 2011. Virtual phosphorus ore re- Van Vuuren, D.P., Bouwman, A.F., Beusen, A.H.W., 2010. Phosphorus demand for the
quirement of Japanese economy. Chemom. Intell. Lab. Syst. 84, 767–772. 1970-2100 period: a scenario analysis of resource depletion. Glob. Environ. Change
Metson, G.S., Bennett, E.M., Elser, J., 2012. The role of diet in phosphorus demand. 20, 428–439.
Environ. Res. Lett. 7, 044043 (044010pp). Vitousek, P.M., Naylor, R., Crews, T., David, M.B., Drinkwater, L.E., Holland, E., Johnes,
Metson, G.S., MacDonald, G.K., Haberman, D., Nesme, T., Bennett, E.M., 2016. Feeding P.J., Katzenberger, J., Martinelli, L.A., Matson, P.A., Nziguheba, G., Ojima, D., Palm,
the corn belt: opportunities for phosphorus recycling in U.S. agriculture. Sci. Total C.A., Robertson, G.P., Sanchez, P.A., Townsend, A.R., Zhang, F.S., 2009. Nutrient
Environ. 542, 1117–1126. imbalance in agricultural development. Science 324, 1519–1520.
Mueller, N.D., Gerber, J.S., Johnston, M., Ray, D.K., Ramankutty, N., Foley, J.A., 2012. Wang, J., et al., 2017. International trade of animal feed: its relationships with livestock
Closing yield-gaps through nutrient and water management. Nature 490, 254–257. density and N and P balances at country level. Nutr. Cycling Agroecosyst. 1–15 In
Nowak, B., et al., 2015. Nutrient recycling in organic farming is related to diversity in press.
farm types at the local level. Agri. Ecosyst. Environ. 204, 17–26. West, P.C., Gerber, J.S., Engstrom, P.M., Mueller, N.D., Brauman, K.A., Carlson, K.M.,
Naylor, R., Steinfeld, H., Falcon, W., Galloway, J., Smil, V., Bradford, E., Adler, J., Cassidy, E.S., Johnston, M., MacDonald, G.K., Ray, D.K., Siebert, S., 2014. Leverage
Mooney, H., 2005. Losing the links between livestock and land. Science 310, points for improving global food security and the environment. Science 345,
1621–1622. 325–328.
Oita, A., Malik, A., Kanemoto, K., Geschke, A., Nishijima, S., Lenzen, M., 2016. Withers, P.J.A., Van Dijk, K.C., Neset, T.-S.S., Nesme, T., Oenema, O., Rubaeke, G.H.,
Substantial nitrogen pollution embedded in international trade. Nat. Geosci. 9. Schoumans, O.F., Smit, B., Pellerin, S., 2015. Stewardship to tackle global phos-
Peñuelas, J., Poulter, B., Sardans, J., Ciais, P., Van der Velde, M., Bopp, L., Boucher, O., phorus inefficiency: the case of Europe. Ambio 44, S193–S206.

141