Journal of Bioeconomics

, Volume 16, Issue 3, pp 223–238

Increasing cooperation among plants, symbionts, and farmers is key to past and future progress in agriculture


    • Ecology Evolution & BehaviorUniversity of Minnesota

DOI: 10.1007/s10818-014-9179-7

Cite this article as:
Denison, R.F. J Bioecon (2014) 16: 223. doi:10.1007/s10818-014-9179-7


The collective welfare of crop plants, their microbial symbionts, farmers, and society can be undermined by tragedies of the commons. A crop could increase resource allocation to grain if each plant invested less in sending roots into soil already explored by neighbors and less in stem growth. But evolutionary fitness depends on which plants capture the most soil resources and light (e.g., by growing taller than their neighbors), not just on the efficiency with which those resources are used. As for symbionts, with several strains infecting each plant, only host-imposed sanctions limit the fitness of strains that divert more resources to their own reproduction, at the expense of activities that benefit their host plant. Similarly, individual farmers do not necessarily benefit from pest- and resource-management practices that benefit farmers collectively or society as a whole. Plant breeders have increased crop yields by reversing past selection for individual fitness and they could breed for crops that would favor more-cooperative microbial symbionts. Better aligning interests among farmers and society may be more difficult.


AgricultureCooperationTragedy of the commons Symbiosis Plant breedingPest management

1 Introduction

We depend on agriculture for most of our food, for materials like cotton or leather, and increasingly for chemical feed-stocks for industrial production of various products. These demands are expected to increase with population growth, economic growth (increasing demand for meat and alcohol), and depletion of nonrenewable resources. For example, plastics, adhesives, and lubricants that are now made from petroleum will be made from crops in future decades.

Like other human activities, agriculture can cause pollution or have other negative effects on natural ecosystems, including conversion of natural ecosystems into farms. Increasing production on existing farmland could reduce the need and the economic incentives for such conversions. One recent analysis concluded that meeting our food needs through high-yield agriculture causes less environmental damage than practices whose lower yields require more land to produce the same amount of food, even if the lower-yielding practices cause less pollution per hectare (Phalan et al. 2011). This is a complex issue, however (Matson and Vitousek 2006), beyond the scope of this review. Similarly, the long-term sustainability of current and alternative practices is a major issue, which I have addressed to some extent elsewhere (Denison et al. 2004; Denison 2012).

This review focuses mainly on agricultural efficiency as a key to limiting land and water use by agriculture. Agriculture already uses 35 % of ice-free land (Foley et al. 2007) and up to 80 % of the water extracted from rivers and wells (Condon et al. 2004), so using land and water efficiently is particularly important.

I will argue that opportunities to increase efficiency often involve tradeoffs between individual and community benefits. Past evolution has almost always favored the individual fitness of plants and of their symbionts over the collective performance of plant and symbiont communities. These past “tragedies of the commons” (Hardin 1968) have left many opportunities for plant breeders to develop more “cooperative” crops, as well as crops that impose selection for more-cooperative symbionts (Denison 2012). I close with a brief discussion of analogous agriculture-related challenges involving cooperation among humans.

2 Efficiency in agriculture

“Efficiency” is the ratio of a desirable output to an input. Agricultural outputs include grain or milk, while inputs include water, sunlight, and minerals like nitrogen or phosphorus. “Land” can be considered an input that supplies all of the above, to some extent. Crop yield (kg/ha) can therefore be considered “land-use efficiency.” External inputs often include minerals, in addition to labor and fossil fuels.

Different definitions of efficiency may be useful for different purposes. For example the water-use efficiency ratio for a crop might have grain yield (kg/ha) as the numerator, but there are various options for the denominator, depending on what components of water use are of most concern. These include rain plus irrigation, irrigation only, crop water uptake plus various losses (soil evaporation, runoff, and drainage below the root zone), or crop water uptake only. Crops that have greater water-use efficiency by one definition may be less-efficient by another. For example, should a plant with a limited nitrogen supply make a few high-nitrogen leaves or many low-nitrogen leaves? Each high-nitrogen leaf photosynthesizes more, relative to its water uptake, but lower total leaf area means less shading of the soil, so more water is wasted through soil-surface evaporation (Condon et al. 2004).

Greater efficiency in use of a particular input is an important goal if the input is costly or scarce—water-use efficiency may not matter in seasons or regions where water is superabundant—and if any negative side-effects of greater efficiency are negligible. These conditions are not always met. Maximizing efficiency does not always minimize risk. For example, meat and milk contain only a fraction of the energy and protein in the feed eaten by the animals, but including animals in agriculture can reduce risk (Denison 2012). Four-year crop rotations that include alfalfa reduce the risk of crop epidemics, relative to shorter rotations (e.g., corn-soybean). Rotations with alfalfa can be as profitable as shorter ones (Davis et al. 2012), but only if there is enough demand for alfalfa hay, which mainly depends on the market for beef and milk. Forages can be grown safely on slopes where grain crops would risk erosion. Even a grain crop that fails—for example, one that runs out of water before making grain—can be grazed (Bender 1993), but the economic value of forages and grazing again depend on the market for meat or milk.

3 Evolution has increased some aspects of efficiency

The plant seeds we harvest as grain were also key to the Darwinian fitness of crops and their wild ancestors. All else being equal, therefore, past natural selection tended to increase the efficiency with which sunlight, water, and nutrients were converted into seeds. Some examples of such improvements are given below. However, various tradeoffs have limited the extent to which natural selection has improved efficiency. These include tradeoffs in adaptation to past versus present conditions and tradeoffs between individual fitness and the collective performance of plant and symbiont communities. This paper focuses on individual-versus-community tradeoffs. Crop-improvement efforts that ignore such tradeoffs are less likely to succeed (Denison 2012).

Nitrogen is often scarce in natural ecosystems, like those where the wild ancestors of our crops evolved, so there was strong selection for nitrogen-use efficiency. Nitrogen is essential for photosynthesis and an essential ingredient in grain protein. We could define nitrogen-use efficiency as the ratio of either photosynthesis rate (gCO\(_{2}\) plant\(^{-1}\) s\(^{-1}\)) or grain protein (g protein/plant) to plant nitrogen content (gN/plant). These definitions can lead to different conclusions. For example, one way that plants enhance these efficiencies is to move nitrogen within the plant. As older leaves lower in the canopy are shaded by younger leaves above them, nitrogen is moved from lower to upper leaves, increasing photosynthetic nitrogen-use efficiency. Much of the nitrogen that ends up in grain protein was transferred from leaves, which lowers photosynthetic rates of the source leaves and therefore reduces the ratio of photosynthesis rate to total plant nitrogen. But this transfer starts with shaded lower leaves, which were contributing less to photosynthesis. By the time grain matures, more than 70 % of plant nitrogen has typically been transferred to grain (Sinclair et al. 2004).

Water-use efficiency at the leaf level can be defined as the ratio of photosynthesis (\(\text {CO}_{2}\) uptake) to transpiration (water evaporation from leaves). Water vapor diffuses out of the leaf through the same stomatal pores that let \(\hbox {CO}_{2}\) in. Leaf-level water-use efficiency (WUE) can therefore be expressed as the ratio of the diffusion rates for \(\hbox {CO}_{2}\) and water vapor, each driven by its own concentration gradient:
$$\begin{aligned} \hbox {WUE} = (\hbox {Ca} - \hbox {Ci})/(\hbox {Wi} - \hbox {Wa}) \end{aligned}$$
Here, Ca and Wa are the \(\hbox {CO}_{2}\) and water-vapor concentrations in the atmosphere (the exterior end of the stomatal pore) and Ci and Wi are the corresponding concentrations at the interior end of the stomatal pore. Humans are collectively increasing Ca, but individual farmers have little influence over it, except sometimes in greenhouses.

Growing crops during humid and cool months, when Wa is greater and Wi is less, can increase WUE. If that is not possible, then the most-obvious route to improving WUE is to decrease Ci. Decreases in Ci depend on uptake of leaf-interior \(\hbox {CO}_{2}\) by the photosynthetic enzyme, rubisco. Rubisco is a protein, so it contains nitrogen. Nitrogen-deficient leaves take up less \(\hbox {CO}_{2}\), as noted above, making Ci greater and WUE less. High-nitrogen leaves have greater WUE, but photosynthesis and yield may increase less than twofold with a twofold increase in leaf nitrogen. So high-nitrogen leaves have lower nitrogen-use efficiency, resulting in a tradeoff between water-use efficiency and nitrogen-use efficiency.

Plants with the C4 photosynthetic pathway, however, have both greater water-use efficiency and greater photosynthetic nitrogen-use efficiency than plants with the older, C3 photosynthesis, pathway. C4 plants pump \(\hbox {CO}_{2}\) from the bulk interior of the leaf (decreasing Ci and thereby increasing WUE) into special compartments, the bundle-sheath cells. The resulting high \(\hbox {CO}_{2}\) concentrations in the bundle-sheath cells mean that less rubisco is needed there, so C4 plants have greater photosynthetic nitrogen-use efficiency, here defined as the ratio of photosynthesis rate to leaf nitrogen. They still need nitrogen for grain protein, however, so the ratio of grain-protein yield to fertilizer applied (another measure of nitrogen-use efficiency) is not necessarily greater for C4 crops. Water-use efficiency is probably their main advantage.

The C4 pathway has evolved independently many times (Christin et al. 2009). Of major crops, maize is C4 but rice and wheat are C3. Efforts are underway to convert rice to C4, but it is not known whether this will succeed. Although C4 rice might provide major benefits, it would be an example of humans copying one of natural selection’s innovations, rather than something original.

A more-innovative biotechnology approach to improving photosynthesis was to move some \(\hbox {CO}_{2}\)-releasing processes to the chloroplasts, facilitating the uptake of \(\hbox {CO}_{2}\) by photosynthesis (Kebeish et al. 2007). This required the introduction of several genes from bacteria. These changes apparently increased photosynthetic \(\hbox {CO}_{2}\) uptake in the model plant, Arabidopsis, but this approach has not yet been implemented in crops.

One popular biotechnology approach to increasing efficiency is to increase the expression of existing genes that are thought to contribute to efficiency (Nelson et al. 2007). However, mutants with greater expression of those genes must have arisen repeatedly in the wild ancestors of our crops. If greater expression of a gene consistently increased individual fitness in all environments, the mutants would already have displaced the ancestral genotype. Therefore, it seems likely that the levels of gene expression our crops inherited from their ancestors are close to optimal, at least in terms of individual fitness in past environments (Denison 2012). We might therefore expect increasing the expression of key genes to decrease efficiency and yield, rather than increasing it. This has been true, for example, of some genes thought to enhance drought tolerance (Lawlor 2013).

This does not mean, however, that it is impossible for plant breeding or biotechnology to improve crops in ways that natural selection cannot do or has not yet done. When there are tradeoffs in adaptation to present versus past conditions, natural selection may lag significantly behind current conditions. For example, if natural selection were allowed to operate in agricultural fields, crops would eventually evolve resistance to newly evolved or newly introduced pests. Plant breeders, however, have often greatly accelerated this process.

When there are tradeoffs between individual fitness and the collective performance of plant and symbiont communities, on the other hand, natural selection will always favor individual fitness. In such cases, increases in efficiency may require reversing the effects of past natural selection.

4 Tradeoff-based crop improvement

As an example of tradeoffs between current and past conditions, consider the implications of ongoing increases in atmospheric \(\hbox {CO}_{2}\). The optimal characteristics of rubisco depend on atmospheric \(\hbox {CO}_{2}\) concentration, due to certain tradeoffs. Current versions of the photosynthetic enzyme, rubisco in our crops are apparently adapted to lower, preindustrial \(\hbox {CO}_{2}\) (Zhu et al. 2004). Modifying rubisco to better match current or future \(\hbox {CO}_{2}\) could increase water-use efficiency and yield, but it is not clear how soon we will develop the knowledge and technology to make that possible.

Tradeoffs between individual fitness and crop/symbiont community performance may offer simpler options for improvement, relative to tradeoffs between past and present conditions (Denison 2012). For example, water-use efficiency is usually greater in the morning than in the afternoon (Kumar et al. 1999), for reasons implicit in Eq. 1. In the morning, Wa (atmospheric humidity) is usually high, while low leaf temperatures decrease Wi. The difference, Wa-Wi, in water-vapor concentration across the stomatal pore is what drives transpirational water loss from leaves. Therefore, transpiration is less, and water-use efficiency greater, under cool and humid morning conditions.

A crop could therefore increase its water-use efficiency by partly closing the stomata on hot and dry afternoons, so that plants trade water for \(\hbox {CO}_{2}\) mainly in the morning. How would this conservative approach affect yield? Photosynthesis per day would be less with morning-only photosynthesis. If the duration of crop growth is limited by soil water, however, using water more efficiently would make the soil water last longer, which could increase total seasonal photosynthesis and yield.

We know that, all else being equal, natural selection favors water-use efficiency, as in the evolution of C4 photosynthesis. But C4 photosynthesis benefits individual plants. The benefits of forgoing afternoon photosynthesis, however, would depend on the entire plant community adopting that strategy. Otherwise, water left in the soil by a more-efficient plant would be used by its less-efficient neighbors. It is understandable, therefore, that past natural selection missed this route to water-use efficiency (Denison 2012).

Plant breeders, however, could select plants that use water mostly in the mornings. In thermal-infrared images on hot afternoons, plants that are transpiring less would show up as warmer than their neighbors, whose leaves would be cooled by their greater transpiration rates (Amani et al. 1996). Greater water use in the morning would also be detectable, if it occurs.

Much of the yield increase already achieved by plant breeders came from reversing past natural selection for individual fitness, when individual fitness conflicted with the collective performance of plant communities (Denison 2012). The first Green-Revolution rice variety, IR8, was developed by Peter Jennings at the International Rice Research Institute. His selection criteria included short stems and small, erect leaves (Jennings 1964). These criteria were based on previous research by Tsunoda showing that these traits are associated with greater fertilizer response in sweet potato, soybean, and rice.

A few years later, Colin Donald (1968) pointed out that these traits all involve a tradeoff between “competitive ability of cultivars against other genotypes on the one hand, and their capacity for yield in pure culture on the other.” Shorter plants, which allocate more resources to grain than to stem, are less competitive for light. They quickly die out in competition with taller plants (Jennings and de Jesus 1968), but that is not a problem if the whole field is of the shorter genotype and weeds are well controlled.

The optimal leaf angle depends on similar tradeoffs. When the sun is overhead, horizontal leaves intercept more light, increasing individual-plant fitness. Erect leaves, in contrast, spread the available sunlight over a larger leaf area. Because photosynthesis shows decreasing returns with photons per square centimeter of leaf, a crop with erect leaves can have greater total photosynthesis and yield (Pendleton et al. 1968), although this can require closely-spaced plants (Angus et al. 1972).

Recent suggestions for additional traits linked to individual-versus-community tradeoffs include reducing solar tracking. Like crops with erect leaves, plants with less solar tracking could have greater whole-crop photosynthesis through better distribution of sunlight. Tracking may have evolved, despite negative effects on efficiency, partly because it increased shading of nearby competitors (Denison et al. 2010).

Similar tradeoffs occur underground. A plant that sends roots into soil already explored by its neighbors uses resources that could otherwise have been used to make more seeds, without increasing total uptake of soil water and nutrients by the crop as a whole. So overall resource-use efficiency decreases. But a plant that takes a larger share of those soil resources can increase its individual fitness. Natural selection therefore favored more allocation to horizontal roots than is optimal for community level productivity and crop yield (Zhang et al. 1999). To some extent, plant breeding for yield may have reversed past natural selection for excessive root growth under neighbors (Zhu and Zhang 2013).

Darwin referred to “the astonishing waste of pollen by our fir-trees” (Darwin 1859). In maize, too, more pollen is produced than needed to fertilize all the female flowers. Pollen is produced by male flowers, tassels, which consume resources directly and also reduce photosynthesis by shading leaves (Duncan et al. 1967). The shaded leaves are often on neighboring competitors, however, so it is not surprising that individual selection favored larger tassels than was optimal for community productivity. Plant breeders have reduced tassel size greatly, apparently as an indirect result of selection for yield (Duvick and Cassman 1999).

Animal populations, too, would be more productive if fewer resources were allocated to males. However, when females outnumber males, an individual who produces mostly male offspring will typically have more descendants (Fisher 1930; Dawkins 1976). So natural selection maintains mechanisms that keep the sex ratio near 50:50. In domesticated animals, however, humans often maintain a female-biased sex ratio (through culling rather than breeding), thereby producing more milk, calves, or eggs. Similarly, individual selection favors more aggression among hens than is optimal for their collective egg production. Human-imposed group selection led to less aggression among hens and greater group productivity of eggs (Muir 1996).

Jacob Weiner has noted that past selection for individual fitness may not always have maximized competitiveness (Weiner et al. 2010). In particular, plant responses that help them escape shading by neighbors (e.g., increasing height) may not be identical to the responses that would maximize shading of those neighbors (e.g., increased leaf area). He argues that “weed suppression by a crop population is a ‘communal’ activity.” Weed suppression may therefore be the sort of “public good” not necessarily maximized by natural selection. If so, it may be possible to breed crops that suppress weeds more, preferably without increasing their mutual suppression of each other. As in the other examples above, changes in crop management may be needed to get the most benefit from such a weed-suppressing crop. Weiner proposed high crop density and more-uniform planting as key elements.

5 Individual-versus-community tradeoffs in symbionts

Crop plants typically interact with a wide range of microbes, some of which provide substantial benefits. Among these beneficial microbes are symbiotic rhizobia and mycorrhizal fungi (Denison and Kiers 2011). If we can resolve certain conflicts of interest among these symbiotic microbes and their plant hosts, we may be able to increase the benefits they provide.

Rhizobia are soil bacteria, best known for their symbiosis with legumes. Inside legume root nodules, most rhizobia convert nitrogen from the soil atmosphere into forms the plants can use. Thanks to rhizobia, legumes can be grown without nitrogen fertilizer.

Mycorrhizal fungi form symbioses with a wider range of crop plants than rhizobia do. These fungi typically provide their hosts with phosphorus. This phosphorus comes from the limited supply in the soil, so the fungi do not eliminate the need for external inputs of phosphorus. They do reduce the photosynthate cost of phosphorus uptake, however, and may provide other benefits, including protection against pathogens.

Why do these symbionts provide these benefits to their hosts? That is, why does infecting legume roots and supplying their hosts with nitrogen increase rhizobial fitness? Why does infecting roots and supplying their hosts with phosphorus increase the fitness of mycorrhizal fungi?

The benefits to the symbionts from infecting roots are fairly obvious and not entirely different from the benefits obtained by root pathogens. A root nodule founded by a single rhizobial cell may eventually contain millions of rhizobia, depending on the species (Denison and Kiers 2011). If even a tiny fraction of these rhizobia eventually escape back into the soil—the actual fraction is not yet known—then infecting roots probably increases fitness much more than remaining in the soil or on the root surface. For mycorrhizal fungi, the benefits of symbiosis are apparently even greater, as they are thought to be entirely dependent on their hosts for carbon and energy.

But, once these symbionts have infected a root, why do they provide their host with nitrogen or phosphorus? The more nitrogen the rhizobia provide, the more their host can photosynthesize (Bethlenfalvay et al. 1978). Greater host photosynthesis probably results in more photosynthate for the rhizobia. So there could be collective benefits to rhizobia infecting a plant from fixing nitrogen. Similar arguments apply to phosphorus supply by mycorrhizal fungi.

Those collective benefits have an individual cost, however. Carbon that rhizobia respire in support of nitrogen fixation could instead have been used for additional rhizobial reproduction, inside the nodule. Or the carbon could have been hoarded as energy-rich polyhydroxybutyrate, to support rhizobial survival and reproduction after they return to the soil (Ratcliff et al. 2008). Similarly, resources that mycorrhizal fungi consume in supplying their hosts with phosphorus could instead have been used for reproduction via spores.

Given these collective benefits but individual costs, what level of investment in nitrogen fixation or phosphorus transfer will result from natural selection? This partly depends on how many different strains of rhizobia or mycorrhizal fungi infect the average plant. If there were only one symbiont strain per plant, the additional photosynthate that symbionts would receive from a healthier plant would exceed the cost of providing the host with nitrogen or phosphorus. As the number of strains per individual host plant increases, however, increased sharing of benefits among competing symbiont strains would favor strains with lower investment in costly, host-benefiting activities. The fitness-optimizing symbiont investment with even two strains per plant can be as low as zero, especially if nitrogen and phosphorus are abundant in the soil (West et al. 2002). Models based on field estimates of rhizobial strains per plant (4–18) predict zero investment in nitrogen fixation, even if soil nitrogen is so scarce that plants depend on rhizobia for most of their nitrogen (West et al. 2002). Because multiple infection is also typical of mycorrhizal fungi, endophytes, and rhizosphere microbes, similar arguments apply (Denison et al. 2003). All of these interactions typically involve multiple strains per plant, creating the potential for a “tragedy of the commons” (Hardin 1968).

The above analysis assumed that host plants do not discriminate among strains, based on their contributions. For example, it assumes that the only benefit a rhizobial strain gets from fixing more nitrogen is its share of the increased photosynthate supply that results from improved host-plant nitrogen supply. That hypothesis leads to the prediction that rhizobia and mycorrhizal fungi will rapidly evolve to provide little or no benefit to their plant hosts. This prediction is incorrect, so the hypothesis that generated the prediction must be wrong (Kinraide and Denison 2003). But what if plants monitor the contributions from each strain and allocate resources accordingly? A model that makes that assumption predicts that symbionts will evolve much greater investment in activities that benefit the host (West et al. 2002).

Experiments have shown that plants do indeed respond to differences among strains in benefits provided by their symbionts. Some soybean root nodules were exposed to argon-oxygen mixtures containing almost no nitrogen gas, to prevent rhizobia in those nodules from supplying their hosts with nitrogen, while control nodules on the same plant were exposed to air, which has 80 % nitrogen. Plants reduced the oxygen supply to the interior of the nodules that fixed little or no nitrogen and, perhaps as a result, the nodules grew less and the rhizobia inside reproduced less (Kiers et al. 2003). Similar reductions in nodule growth and rhizobial reproduction have been seen in alfalfa and pea (Oono et al. 2011), although decreased growth of nodules may not always reduce the reproduction of rhizobia inside (Gubry-Rangin et al. 2010).

Plants can also respond to less-beneficial mycorrhizal fungi in ways that presumably reduce their fitness. When plants were simultaneously infected by a more- and a less-beneficial strain, the more-beneficial strain received more plant-derived carbon (Kiers et al. 2011).

We have called plant responses to less-beneficial symbionts “host sanctions” when they reduce symbiont fitness (Denison 2000; Kiers and Denison 2008). In contrast to sanctions against rogue states, there is no implication that a change in symbiont behavior is expected, except via an evolutionary decrease in the frequency of less-beneficial strains over years. It is not clear how, or whether, plants impose sanctions on root-surface or endophytic (tissue-infecting) microbes. These, too, are often beneficial, despite the apparent potential for tragedies of the commons (Denison et al. 2003).

Although host sanctions presumably limit the spread of the least-beneficial symbionts, current crop varieties may impose weaker sanctions than would be optimal for long-term productivity and sustainability. Intermediate rates of nitrogen fixation (as low as 50 % of potential) may not always trigger sanctions (Kiers et al. 2006). In the long run, a plant species might benefit if each individual plant cut off resources to all but its most-efficient symbionts. Future generations of plants would then acquire more-efficient symbionts from the soil, increasing their nitrogen or phosphorus supply or decreasing their cost to obtain these nutrients. But natural selection is blind to future consequences—most species go extinct. A plant that needs more nitrogen for its seeds may obtain an individual benefit from supporting nodules containing even mediocre rhizobia, whatever the future consequences for its species (Denison 2000).

So the legacies of past natural selection may include crops with too-weak sanctions, in addition to too-tall stems and too-big tassels. Plant breeders have been improving the latter two traits, but what about sanctions? Soybean sanctions against less-beneficial rhizobia may have been getting weaker over decades, perhaps due to inadvertent selection in high-nitrogen soils. Three soybean cultivars developed recently had lower yield when grown in soil containing a mixture of fixing and nonfixing rhizobia, relative to growth with only the nitrogen-fixing strain (Kiers et al. 2007). Three earlier cultivars did not show this yield loss, perhaps because they were better at allocating resources only to nodules that were fixing nitrogen. This difference between older and newer cultivars was large enough to erase the yield gains of the latter.

Could we breed for stronger host sanctions? We don’t know what plant genes are involved, but a phenotypic screen could be possible. One way to select among crop genotypes, based on their sanctions responses, would be to grow the crop genotypes separately (in pots or single-genotype field plots) in soil containing symbionts ranging from excellent to mediocre. The effects of each genotype on the composition of the symbiont community in soil would be assessed by planting a second, genetically-uniform, test-crop in the same soil, and measuring its yield (Denison 2012). For example, some soybean genotypes might enhance the growth of subsequent soybeans planted in the same soil, due to stronger sanctions this year improving the composition of soil rhizobial populations available to future crops.

Effects on subsequent crops would not necessarily be due to rhizobia, however. Some soybean cultivars may increase or decrease the abundance of soil pathogens more than others. There could also be effects on mycorrhizal fungi. Even current soybean cultivars apparently increase the abundance of mycorrhizal fungus strains that are more beneficial to maize (Johnson et al. 1992), which is often the next crop in a 2-year rotation. This trait could perhaps be enhanced further by selecting those soybean genotypes that most enhance the growth of a subsequent, genetically-uniform test-crop of maize. The same screen could select for other beneficial effects on soil microbes. For example, root exudates from some plants apparently suppress microbial conversion of ammonium to nitrate (Subbarao et al. 2013), thereby reducing nitrogen losses from soil. To the extent that soil nitrogen is a “public good”, plant production of these exudates may not have been optimized by past natural selection.

The key assumptions behind all of these potential microbe-linked improvements are that: (1) plants affect the composition of soil-microbe communities in ways that affect the subsequent growth of plants of the same or different species, and (2) past natural selection has not already optimized plant genotypes for these traits. The apparent success of human-imposed group selection in increasing benefits to plants from soil microbial communities (Swenson et al. 2000) confirms the theoretical prediction that natural selection for group benefits was weak in natural ecosystems (Gardner and Grafen 2009), leaving room for improvement by humans.

6 Individual-versus-community tradeoffs among farmers

The main point of Garret Hardin’s much-cited paper, “The Tragedy of the Commons” (Hardin 1968) was his claim that a reliance on voluntary methods to limit human population growth will be undermined by evolution: “Those who have more children will produce a larger fraction of the next generation than those with more susceptible consciences.” Differences in family size may be due more to cultural rather than genetic differences, undermining Hardin’s argument, but population growth is one of the major threats to global food security (Denison 2012).

Hardin’s paper is best-remembered for an agricultural example: over-grazing of pasture land held in common. People may graze too many cattle on public land because each individual’s benefit from grazing additional cattle on the commons can be greater than that individual’s share of the collective cost from over-grazing. There are many other examples of agriculture-related tragedies of the commons, where individually rational decisions undermine the collective well-being of farmers and societies.

Consider irrigation from underground aquifers. A farmer who pumps more water can grow more crops. If farmers collectively pump water from aquifers faster than it is replenished, then ground-water levels will drop. Deeper, more expensive wells will then be needed. Eventually, even those deeper wells can run dry. Farmers might benefit, collectively, by limiting pumping to rates that don’t drop water levels below what their current wells can reach. But over-pumping by a given farmer will only slightly increase the risk of all the wells running dry. The individual benefit of over-pumping may outweigh the individual risk. Very large farms represent a possible exception, as the depth to groundwater beneath those farms may depend mainly on their own rate of pumping.

Groundwater in some parts of Australia is salty. There, over-pumping is not the problem. In fact, the greater water consumption of perennial plants provides a public good, by keeping salty groundwater deeper in the soil, away from the soil surface. Salt water seepage damaging one farm might have been prevented if other farms, some distance away, had grown water-hungry alfalfa rather than wheat. But, if wheat is more profitable, that’s what most farmers will grow.

For an evolutionary example, consider insecticide resistance. Transgenic crops that make an insecticidal protein from Bacillus thuringiensis (Bt) are widely grown in the US, but their value depends on the continued susceptibility of pests to the Bt toxin. Bt crops select for resistance, but evolution of resistance can be slowed by keeping some land in “refuges”, where insect pests susceptible to Bt can survive (Tabashnik 2008). If susceptible pests from refuges fly into Bt fields, outnumber the few survivors there, and dominate the local “mating market”, then essentially all insects will have at least one susceptible parent. If insects need two resistant parents to have a significant fitness advantage, then the frequency of resistance genes will remain low.

How do the interests of individual farmers accord with their collective interest in slowing the evolution of Bt resistance in pests? Since insects are mobile, Bt-free refuges on one farm can help slow the evolution of resistance on another. It might seem, therefore, that individual farmers would be tempted to allocate little or no land to Bt-free refuges, to minimize their crop losses to pests. But non-Bt seed is cheaper and losses to pests within the refuges have sometimes been low. This is partly because the overall prevalence of Bt crops suppresses the landscape-level abundance of pests (Hutchison et al. 2010). So minimizing non-Bt crop area does not necessarily maximize individual-farmer profits. The profit-maximizing strategy depends on what other farmers are doing. Such “frequency-dependent selection” is also common in evolution. Individual self interest may sometimes coincide with socially optimal behavior, but this could change with prices, overall pest abundance, etc.

Australian farmers have cooperated to slow the evolution of Bt resistance, but some of their other efforts to coordinate pest management have been less successful (Zalucki et al. 2009). To control pests of cabbage-family crops, Australian farmers agreed not to grow these crops for 3 months each year, from November through January. Without their host plants, pest populations decreased across the region. Soon, however, many farmers switched to a 1-month break, which they found sufficient to significantly reduce pests on their individual farms. Even a 1-month break might have reduced pest levels across the region, if breaks were synchronized among farms. But farmers failed to synchronize their breaks, so pest problems increased.

Some area-wide pest-management strategies depend on farmers in a region all doing the same thing at the same time. However, greater crop diversity at the landscape scale can often decrease disease or pest problems. From an individual-versus-community perspective, however, one problem is that the spatial scale at which crop diversity is most beneficial for pest control—20 km, in one recent study (O’Rourke et al. 2011)—can exceed the size of individual farms. To the extent that crop diversity is a public good, individual-farmer decisions may result in less diversity than would be optimal for the agricultural community as a whole.

At regional to global scales, food shortages from crop-disease epidemics are less likely if different farmers grow different crops. For an individual farmer, however, the expectation of higher profits from alternative crops in the rare years when major crops fail may not make up for lower profits in most years. Individual-farmer decisions about which crops to grow may therefore lead to lower levels of regional or global crop diversity than would be needed to ensure food security via “portfolio insurance” (Denison 2012).

Finally, consider nitrogen fertilizer use. Up to a point, increasing fertilizer use benefits individual farmers, but also consumers and the environment. Fertilizer increases yields and farm income. Higher yields lead to lower food prices, benefiting consumers. Lower crop prices also reduce the incentive to turn drain wetlands or clear forests for farmland. But with each increment in fertilizer, a smaller fraction of the nitrogen goes into grain protein and a larger fraction ends up polluting lakes or rivers. Depending on the prices of crops and fertilizers, the socially optimal fertilizer rate may be less (or perhaps sometimes greater) than the rate that maximizes individual profit.

Is the problem “how to keep the majority of people acting towards the public (i.e. their own) long-term good” (Zalucki et al. 2009)? Not always. A farmer who grows cabbage crops when other farmers refrain may benefit from higher prices, year after year. That farmer’s contribution to increased pest abundance will often be small, relative to his individual benefit. A farmer who applies nitrogen up to the point where her marginal cost equals her marginal benefit may pollute more than those who fertilize less, perhaps contributing to fish deaths downstream, but her individual contribution to pollution will have only a tiny effect on the price she pays for fish.

It has been claimed that Hardin ’s (1968) classic paper advocated “centralized government” (Dietz et al. 2003) to prevent tragedies of the commons. Centralized government could be one mechanism for Hardin’s proposed “mutual coercion, mutually agreed upon” (Hardin 1968), although he did discuss the problem of government corruption. But “mutual coercion” is also a good description of effective local self-government, such as “forest patrols, which promote cooperation by sanctioning free riders” (Vollan and Ostrom 2010) Often, local self-government (mutual coercion, locally agreed upon) depends on incentives or sanctions from central government. Examples include protection of local fishing grounds from outside groups, or government assessments of the “annual allowable timber quota and the rent each group has to pay to the local forest administration” (Rustagi et al. 2010).

7 Conclusions

Tragedies of the commons undermine the resource-use efficiency of crop plants, decrease benefits from microbial symbionts, and create perverse incentives for farmers. These situations represent opportunities for improvement, but some potential improvements may be difficult to implement.

Reversing past selection for individual-plant competitiveness, when it conflicts with the overall productivity of the plant community, has been responsible for most past increases in yield potential (Denison 2012). Shorter, higher-yielding Green Revolution wheat and rice varieties are the best-known example. Much of this progress has been a side-effect of selection for yield (Duvick and Cassman 1999), but a greater focus on tradeoffs could accelerate progress. This might be particularly true for plant traits that benefit future crops (such as enriching the soil with the most-cooperative indigenous symbionts), a potential benefit that has not yet been a plant-breeding objective.

Improving cooperation among crop plants and their symbionts may be easier than resolving conflicts among farmers and larger groups. Many opportunities to improve crop and symbiont cooperation could be achieved by a single plant breeder, independent of more-widespread acceptance of these ideas. If the resulting cultivars have higher yield or require fewer inputs, they are likely to be widely adopted, even without broader understanding of the reasons for their better performance.

One caveat is that more cooperative crops may require somewhat different management practices. For example, crops with more-erect leaves typically need closer spacing to fully intercept the available solar radiation. A soybean cultivar that imposes stricter sanctions might sacrifice some nitrogen fixation the first year it is grown, with longer-term benefits coming from improved rhizobial populations in future years. Still, taking these factors into consideration would represent only a small increase in the overall cognitive load faced by modern farmers, who already need expertise in crops, soils, pests, machinery, personnel management, and marketing.

Improving the alignment of individual-farmer interests with those of farmers collectively, and society as a whole, will be more difficult. Correctly identifying such conflicts and possible solutions is difficult enough. For example, pollution from over-fertilization is typically addressed in one of several ways. A set of “best-management practices” may be proposed. A governmental agency may ask for voluntary compliance, perhaps threatening to impose sanctions on all farmers in a watershed, if they fail to reduce pollution, collectively. But this is exactly analogous to the original problem: a farmer who follows less-profitable “best-management practices” pays an individual cost, for the uncertain chance of sharing in a collective benefit.

Some government-imposed practices or incentives may not be optimal. For example, assuming steady-state conditions, total nitrogen pollution from a farm is not the amount of nitrogen applied, but rather the amount applied minus the amount exported in products sold (e.g., in grain or milk protein), by conservation of matter. Taxing that difference would promote innovative ways of increasing the fraction of applied nitrogen that ends up as food rather than as pollution, in ways that a simple tax on fertilizer does not (Denison 2012). But government officials who promoted a flawed policy have a strong incentive to exaggerate its success (Campbell 1969).

Discussions among farmers, agricultural scientists, and others can identify workable solutions, as in the case of resistance management for Bt crops in Australia (Zalucki et al. 2009). But so long as agricultural problems (from pollution to food security) attract little attention from the nonagricultural public, politicians may not invest enough time in these issues to distinguish workable solutions from simplistic ones.

Copyright information

© Springer Science+Business Media New York 2014