Agricultural legacies, food production and its environmental consequences

Near the end of Proust’s famous novel Remembrance of Things Past, the protagonist at last comes to understand that a person’s present can only be understood in the context of their past. In PNAS, Sattari et al. (1) remind us the same is true for ecosystems. Focusing on the present and future of agriculture, the authors argue that a legacy of excess fertilizer use has caused soil phosphorus (P) to accumulate in the world’s traditional breadbaskets and enhance recent crop yields. Further, they suggest those past P deposits can delay a potential crisis in global P supply (2). Their work is the latest contribution to a debate about one of society’s fundamental challenges: how do we feed a growing population while minimizing agriculture’s collateral damage on finite resources and the environment (3, 4)?

It is a daunting question. In a single human lifetime, agriculture has more than doubled the amount of P and nitrogen (N) cycling in terrestrial systems (5–7). Past fertilizer excesses may subsidize future crop growth but are also responsible for a litany of environmental ills: P runoff has caused widespread freshwater eutrophication (6), whereas N’s “rap sheet” includes climate change, marine eutrophication, biodiversity loss, and air pollution (7). From a food security perspective, however, there is a critical difference between N and P. Although inequitable global distribution of mineral fertilizers perpetuates malnourishment in some world regions (8), our reliance on atmospheric N₂ for fertilizer means we will never run out of N.

The same cannot be said for P. Currently, world agricultural output relies on P fertilizers extracted from a few mineral deposits that—much like fossil fuels—took millions of years to form (9). Although the extent of Earth’s P capital remains debated, absent the discovery of new reserves and better recycling, P supply may eventually limit food production (9, 10). Hence the importance of Sattari et al.’s “residual” soil P pools (1). As early as 1942, Coleman (11) argued that residual P—that missed by simple soil extractions—might be more available to plants than originally thought, a hypothesis supported by multiple recent studies (e.g., ref. 12). Sattari et al.’s analysis creates optimism that present and future crops can recoup some benefits from the mistakes of our past. Yet while information on accumulated soil P ought to inform decisions about fertilizer needs, there are several reasons to temper our optimism. History also teaches us that, in addition to crop requirements, a web of biophysical and socioeconomic factors drives agricultural decisions and outcomes (Fig. 1).

Fig. 1.

Expansion of P-intensive soy agriculture in Mato Grosso, Brazil has led to soil P storage, a legacy that may lower future P fertilizer requirements (1). However, agricultural P demand and efficiency may depend as much on climatic variability and socioeconomic drivers as on soil biogeochemical conditions. Ideally, P inputs mirror crop outputs, creating high PUE, which reduces pressure on finite P fertilizer reserves and minimizes P loss to the surrounding environment. However, high PUE is rare because suboptimal climate variation and other disturbances typically suppress crop yields below their maximum potential. Society’s overall P use efficiency is also low for reasons that range from P-intensive dietary choices to minimal P recycling. Reducing P-driven environmental damages and world fertilizer demand, while simultaneously meeting food security challenges, will require management and policy changes that increase PUE along the entire path from crop growth to human consumption. Photo credit: Chris Neill.

For example, even if residual P analyses inform fertilizer requirements, those needs are often targeted to maximum potential yields. In some ways, that makes intuitive sense: farmers do not want nutrient shortages to cause lower economic returns, thus fertilizer is often viewed as cheap insurance. However, getting the most out of a crop is a rare event. Climate variability alone causes most years to fall below maximum potential yields (13), and disturbances like pest outbreaks or ground-level ozone damage drive additional gaps between realized and maximum potential harvests (14). Thus, although overall balances are improving in some world regions (5), poor nutrient use efficiency remains common.

In the case of N, a gap between supply and demand typically means a proportionate loss of N to the environment (15). The consequences of low P use efficiency (PUE; Fig. 1) are less clear. Excess P may stick around and help crop production in the future (1), but as widespread problems with freshwater eutrophication demonstrate (6), substantial amounts also leave agricultural fields. For the most part, losses are determined by the rate of soil erosion. Where soil management effectively minimizes erosion, P losses can be small even under high application rates, and much of the excess may be available for future use. Unfortunately, soil erosion remains a significant problem in agroecosystems worldwide, one that may worsen as high-input agriculture spreads throughout equatorial regions (10).

Erosion potential depends on soil type, climate, and topography, but management also plays an enormous role. Practices that leave soils bare for extended periods increase the risk of water- or wind-driven loss, which reduces long-term P use efficiency and enhances the potential for downstream eutrophication (10). Significant erosion potential can also exist in row crops, where areas of exposed soil remain year-round and are vulnerable to extreme climate events. If the frequency and intensity of climate extremes increases (16), the potential for P erosion is also likely to rise unless management practices strongly target the dual goals of soil preservation and high PUE.

As with most aspects of agriculture, predicting the balance between P losses and storage is a challenge that must integrate biophysical and socioeconomic drivers (Fig. 1). Although still in its infancy, such integration is a focus of some Earth system modeling goals, but those efforts rarely include treatment of the P cycle. In part that is because historic hotspots of disturbance to N and P cycling have been in temperate regions where P limitation is less of a concern. However, low soil P availability is a common feature of tropical and subtropical soils, and these regions are witnessing explosive growth in high-input agriculture, particularly of P-demanding crops like soy (Fig. 1) (3–5). Tropical ecosystems also have an

Sattari et al.'s analysis creates optimism that present and future crops can recoup some benefits from the mistakes of our past.

enormous influence on the carbon cycle and climate, one that depends on interactions with P (17). As such, the need to integrate P into modern Earth system models is paramount.

Moreover, P does not act in isolation. Biogeochemical cycles are coupled, such that the behavior of one element can have substantial effects on others, and the feedbacks between elements can be vital to predicting ecosystem responses to environmental change (18). For example, projections of plant responses to rising CO₂ were substantially revised once N constraints were considered. In unmanaged systems, N limitation may curtail plant responses (19); for crops, N-fueled ozone creation can reduce yields well below those predicted from CO₂ alone (14). The ways in which P affects and responds to variations in C and N cycling are not as well known, but history tells us they may be central to predicting the future.

Models that effectively couple the C, N, and P cycles could help, especially in understanding feedbacks that play out at multiple time scales. Society has dramatically reorganized all major biogeochemical cycles (17), but our disturbances to those cycles have occurred in the blink of a geologic eye, and their legacy will be with us for generations. That fact is widely recognized in climate science: we could turn off anthropogenic CO₂ today but would be committed to future climate effects because of momentum from long-acting portions of the carbon cycle (16).

Similarly, both the N and P cycles operate on time scales from minutes to millennia, thus predicting their future will depend on the legacy of past actions. However, with N and P, that challenge comes with an additional twist. Ending fossil fuel dependence is no small task, but it is possible (and ultimately essential) to envision a carbon-free future for energy supply. By contrast, feeding the planet demands a continued reliance on significant disruption of the N and P cycles (5). There are no substitutes for these essential plant nutrients. Our challenge, then, is an optimization problem: how do we grow the food we need while minimizing environmental damage and not running out of readily accessible P reserves?

Happily, it is a challenge with some answers. Multiple solutions exist that can improve nutrient use efficiency along the path from crop growth to human consumption. These include high- and low-tech management options, policy instruments, dietary changes, and opportunities for N and P recycling (5, 7, 15, 20). Most critically, many solutions are not inventions yet to be discovered. For political, economic, and cultural reasons, they can be difficult to achieve (15), but the choices are in front of us. As Sattari et al. (1) note, every once in a while our past mistakes do not only haunt us.