Returning to Our Roots: Making Plant Biology Research Relevant to Future Challenges in Agriculture

Over the next 50 years, considerable stress will be placed on worldwide crop production by a combination of factors, including an increased human population, an increase in the crops used per person, and a number of environmental issues. Given current trends, it will be necessary to approximately double yields worldwide during this time period, and meeting this challenge will require a considerable effort. This article explores the nature of the challenge and the requirements for meeting it. These include novel technical advances and fundamental discoveries as well as new multidisciplinary ways of organizing research to ensure that researchers and technologists target the advances and discoveries that are most needed and effectively use them to enhance important crop traits.

THE PRINCIPAL CHALLENGE: SUSTAINABLE, WORLDWIDE DOUBLING OF CROP YIELDS

In 1798, Thomas Malthus published his essays on the potential consequences of a rapid increase in human population. He later stated that “the cause to which I allude, is the constant tendency in all animated life to increase beyond the nourishment prepared for it” (Malthus, 1826). Since 1800, the world's population has increased ∼20-fold, and there is little doubt that on average we have more to eat than did our ancestors who lived during Malthus' time. This is due to the conversion of extensive land areas to agriculture, especially in North and South America, and the introduction of numerous technological advances, including mechanization, synthetic fertilizers, chemical control of pests and diseases, and improved high-yielding crop varieties. As these advances were applied worldwide, farm yields increased so much that overproduction of crops had become a serious economic concern by the late 20th century.

Looking ahead, the human population is predicted to increase ∼50%, before reaching a maximum of ∼9 to 10 billion in ∼50 years (United Nations, 2004; U.S. Census Bureau database, http://www.census.gov/ipc/www/idb/). However, much more than a 50% increase in food production, probably a 100% increase, will be required to meet the needs and expectations of this population for two principal reasons. First, a burgeoning middle class in developing countries, such as China and India, and the concomitant increase in consumption of more animal products is increasing demand for the primary grain and oil seed crops more rapidly than population is growing. Second, the use of crops for industrial products, particularly ethanol from starch, to meet increasing energy demand is also growing at a rapid pace.

ENTERING THE UNKNOWN: DIFFICULTIES IN MEETING THE CHALLENGE

The success of agriculture in meeting humanity's needs while the population grew 20-fold is extremely impressive but not necessarily predictive of future trends. Although increasing farm yields led to overproduction in the past, in almost every year since the year 2000, more grains have been consumed than produced, leading to an initially slow and subsequently more rapid diminution of grain stocks (see www.fas.usda.gov for worldwide supply and demand statistics for all crops). During the past year, grain commodity prices increased significantly after a steady decline over the previous 30 years.

There is no longer a large land bank that could be planted with agricultural crops that is not already being used for this purpose. Rather, the quantity and quality of prime agricultural lands is now diminishing due to human habitation and degradation through poor farming practices. In addition, much marginal and environmentally sensitive land has been taken out of production in Western countries. This means that any significant increase in crop production must be met with increased crop yields. The expected population increase together with an expected increase in crop use per person implies a need to roughly double yields per acre worldwide of all important field crops, and this will need to be accomplished in an environmentally and economically sustainable fashion. While this has been done before, the next doubling will be a much greater challenge. As an example, maize yields have increased in a roughly linear fashion by ∼1.8 bu/acre/year. However, when maize yields averaged 40 bu/acre, it took only ∼20 years to double yields. Now that the average yield is 160 bu/acre, it will take 80 years to double yields if the long-term pattern is maintained.

Complicating matters, a number of large-scale changes are occurring that are expected to have a large effect on crop production for food and feed. These include the availability of water and nutrients, issues around crop land, and uses of crops or crop land for alternative purposes besides food. By far the most important environmental factors impacting crop yields are the availability of sufficient water and nutrients. Water availability varies from year to year and leads to a more-or-less predictable variation in crop production. However, in some irrigated crop production regions, it is not clear that the rate of water use is sustainable. In addition, two variables relating to climate change are unpredictable. First, we naturally tend to assume that the climate in which we have lived our lives is normal for a particular region. However, this is not always the case. As an example, it has been found that the Canadian prairies have been much wetter than usual during the time when they were settled and farmed on a massive scale (Schindler and Donahue, 2006). If this area reverted to the long-term mean, then much of one of the most important grain production areas would be unavailable for agriculture. Second, while it is difficult to predict exact rainfall pattern changes due to human-induced global warming, two effects are clear. There will be more evaporation with a hotter climate, necessitating more rainfall to stay even (Schindler and Donahue, 2006). Furthermore, with the melting of glaciers, glacial rivers like those that flow through the Canadian prairies will have much less flow during the summer when their use for irrigation would normally be greatest. The current long-term drought in Australia is a lesson on the effects of water shortages leading to a large decrease in crop production.

Land use issues also present a considerable future challenge. First, there is the issue of using prime farm land for other human uses, an inexorable and somewhat predictable process. The second issue is that we have been using industrial agricultural practices for approximately the past half century. During that time a considerable portion of the topsoil has been lost. Current practices of crop production, such as the use of no- and low-till, are a considerable improvement, but losses still occur, albeit at a lower rate. Although average yields have shown a steady increase during this period of time, it is unknown how long this trend can continue. By relying on the status quo, we are wagering that these practices will be sustainable on a global scale for 50 more years.

Finally, there is the issue of the use of grain for ethanol production and the potential use of other biomass feedstocks produced from agricultural land for this purpose. Last year, ∼20% of the U.S. maize crop was used for ethanol, which increased the demand enough to cause a doubling in prices. This in turn has led to a significant increase in the maize acreage planted this year, causing decreased planting of other, less profitable crops. However, use of the entire U.S. maize crop for ethanol would satisfy only a small part of the nation's liquid energy needs, creating a potentially unlimited market for this use.

A sensible alternative to the use of grain for ethanol is to use cellulose, which can come from two potential sources using crop waste and marginal lands. While using crop waste, such as maize stalks left over after harvest, would at first glance appear to be an ideal use, several preliminary studies have shown that this practice would lead to significant yield losses in subsequent years due to degradation of soil quality (for example, see Johnson et al., 2006). While growing plants such as switchgrass and poplar on marginal agricultural lands may be feasible, there would be no way to prevent farmers with prime land from growing these energy crops if the economic gain was greater than that for food crops. For these reasons, it seems certain that the use of agricultural land for energy production creates another source of future uncertainty for food production.

In addition, the growing biofuel industry is having a strong impact on current agricultural research. For example, in Canada, almost all new agricultural research support is in the area of bioproducts, of which biofuels is a major component. This is occurring in other countries as well. This is not all bad, of course, as it is always worth developing alternative ways of using agricultural products, and much of the work must involve studies on how to improve plant productivity that may have application to food crop production. Nevertheless, it means that many other important areas of research are not being funded, as will be discussed in more detail.

CHALLENGING ISSUES IN PRIVATE AND PUBLIC PLANT GENETICS RESEARCH

Plant genetics companies spend somewhere in the neighborhood of $2.0 to $2.5 billion per year on research to develop better seed products. While this is a substantial sum, the largest pharmaceutical company, Pfizer, alone spends $7.5 billion dollars per year on research and development. Furthermore, due to the economics of the seed business, only maize generates enough of a profit from seed sales to justify bringing the whole panoply of modern techniques to bear on developing better seed products. By contrast, a relatively minor sum is spent on research into, for example, wheat genetics, despite its importance as a food crop. Even with regards to research on maize, there is no way that any one company or set of companies has the resources to undertake all worthwhile experiments. In addition, there is no economic incentive to do research for poor farmers in developing countries where income from seed sales cannot support significant research spending. This means that not all of the potential capabilities for genetic improvement are being used for crops that feed most of the world's population.

This situation has important consequences for considering how the results from basic plant biology research could be applied to important problems. Here is one scenario to exemplify this problem. Let us assume that a researcher studying tolerance to abiotic stress does excellent research on a model system like Arabidopsis that gives leads on important genes that modulate drought tolerance. Applying this information to important crop plants requires at the very least the following capabilities: access to information on advanced breeding lines, the ability to produce transgenic plants and/or to use a large marker set to check for the presence of allelic variation for these genes, a way of testing for phenotype for this trait, and the ability to test whether any genetic improvement has any negative pleiotropic effects and works under a variety of different environmental conditions. Developing a valid way to test for a drought-tolerant phenotype in a realistic fashion requires developing testing sites where irrigation is the only water source, since variable amounts of precipitation will make it very difficult to do controlled experiments. Industrial research organizations can do these things well but will only do so for crops and geographies where there is a reasonable chance for an economic return on their research investment. Thus, much of this costly foundational applied research for these crops and geographies must be done in the public sector.

Low commodity prices decrease the political will to support public sector research on enhancing agronomic traits for crop production. After all, why support research whose goal will lead to increased overproduction of field crops, decreased commodity prices, and thus decreased farmer profitability? There has been an enormous amount of brilliant research done to understand the genetic basis of plant growth and development, particularly in Arabidopsis, over the past 25 years. However, during this period there has been a concomitant shrinkage in the public sector capability for using this knowledge to improve important crop plants. This is no doubt in part due to the change in public perception of biotechnology and partly due to the sense that industry would supply all of the applied innovation.

Assuming it becomes more difficult to improve productivity enough to meet crop demand, this will likely be a slow-moving crisis, and it will take a considerable effort to impress upon the public and government the need for action. One problem is that it takes time to enhance research capability in this area. Even given the necessary research capacity, previous experience suggests that it takes 10 to 15 years to use any piece of basic knowledge to improve crop genetics. Clearly, it would be best not to wait for this type of crisis to occur before trying to develop solutions.

FINDING SOLUTIONS

Some Important Traits Requiring Additional Fundamental Genetic Knowledge

A considerable amount is known about the genes controlling plant growth and development, and the question is whether this knowledge can be used in a predictive sense to help develop improved crop cultivars. The following are two examples of the possibilities of this approach. The semidwarf varieties of wheat and later of rice were fundamentally important to the green revolution (Borlaug, 1983). In addition, a high percentage of the yield increase due to breeding was due to the development of crop genetics that allowed crops to be planted at high densities, thus increasing the per acre biomass produced (Duvick, 1992). Therefore, can plant architecture and development be modulated in a controlled fashion to develop new lines that can be grown at a much higher planting density? A second example is that harvest index, the biomass partitioned into the grain relative to total plant mass, has remained around 50% for maize during the last 60 years of breeding. By contrast, the harvest index in wheat was increased because of the development of the semidwarf varieties (Borlaug, 1983). Enhanced partitioning into the grain, assuming no pleiotropic negative effects, would substantially increase yields without increasing plant biomass produced. Again, we can use fundamental knowledge of the genetic factors controlling partitioning to optimize for this trait.

The use of nitrogen fertilizer is essential to maintain current productivity and will be crucial for doubling yields. However, its production is typically the single largest energy input for crop production, and it is increasingly expensive due to escalating energy costs and thus is less affordable for poorer farmers. Furthermore, nitrogen fertilizers cause significant environmental damage, leading to considerable air and water pollution. One example of these effects is that wasted nitrogen fertilizer leads to the production of the potent greenhouse gas nitrous oxide. At the same time the one environmental change that undoubtedly will occur is an increase in atmospheric CO₂ concentration, and under the right circumstances, this could enhance plant growth. Ideally, yield doubling would be achieved without markedly augmenting the amount of nitrogen fertilizer used, as this would lead to a large increase in pollution and of economic cost. There is a linkage between the regulation of carbon and nitrogen metabolism, although our knowledge of these processes is still imperfect (Coruzzi and Bush, 2001). Therefore, the challenge is to use our knowledge of these processes to develop crop genetics with enhanced nitrogen use efficiency in a high-CO₂ world.

Finally, there are a number of important findings described by our colleagues in agronomy and the related sciences that delineate important fundamental problems. The following example is particularly intriguing. For the sake of simplicity, most research focuses on the genetic analysis of individual species, although in the real world they grow in the presence of others. Maize or soybean seedlings can be grown in proximity to weeds but under conditions where there is no competition for nutrients, water, or sunlight (Rajcan et al., 2004; C. Swanton, personal communication). However, even in the absence of this competition, growth of either in the presence of the weeds leads to a decrease of 2%/d in final yield. This implies that these crop plants detect the presence of the weed species and react to this potential competition in a way that must involve a stress response that decreases plant productivity (Rajcan et al., 2004; C. Swanton, personal communication). There must be a genetic component to this phenomenon, and discovering the fundamental processes behind this would allow us to understand and use this knowledge given the importance of inter- and intra-species competition on yield.

Developing Gene Rules for Agricultural Crops: Understanding Genotype by Phenotype as Complicated by Genotype by Environment Interactions

Most important agronomic traits that impact crop productivity, like those described above, are complex multigene traits. Crucially, the level and type of environmental variability is enormous, differing from year to year and from location to location. In a traditional breeding program, large numbers of genetic variants are tested first on a small scale. Next, a select few promising variants are tested on a larger scale, involving multiple locations and multiple years, to deal with the problem of environmental variability in a statistically significant fashion. The potential role of using fundamental genetic knowledge to improve on this process is deceptively simple: can this knowledge be used to improve the efficiency of breeding selection or, alternatively, used to generate new improved genetic variation? There are two intertwined steps involved in this process. First, can a paradigm be developed where gene knowledge can be used to predict phenotype from genotype? Second, can predictive models be developed on how different genotypes will respond to an array of different abiotic and biotic environmental stress conditions? This becomes a multifactor problem involving multiple genes important for the trait of interest, multiple potential alleles for these genes, and multiple stress conditions.

To address this problem, it is important to understand gene function, and much of this research involves the phenotypic analysis of strong mutations in each gene or set of genes in model systems like Arabidopsis. This is a crucial first step but is not sufficient for its application to important agronomic traits. We can assume that there will be a range of variation for important traits in the base genetic lines due to allelic variation in a number of genes, with most of this variation being subtle in nature. One goal is to be able to predict the effect on phenotype of altering selected genes (either through altered expression or protein produced) in any of the base genetic lines and, ideally, to be able to predict whether this change would have a negative or positive effect under different environmental stress conditions. This will require the following things. First, we need to understand how to manipulate genetic variation to alter phenotype using all of our familiar tools. These will include the ability to produce and test genetic variants for a range of phenotypes, including the use of a range of biochemical and physiological parameters and the ability to dissect biochemical and developmental pathways to understand which genetic variants are rate-limiting for each process. It will also involve the testing of these under an array of environmental stress conditions. Second, we need to have a comprehensive understanding of the genetic variation present in each crop plant of interest, since this must be the starting point for any serious program to improve crop genetics. This would almost certainly involve knowing the allelic variability via sequencing of all of the important genetic lines, developing a wide set of test lines that vary only in particular regions, and testing these for their extremes of phenotypic variation under a variety of environmental conditions. Third, we will need to develop open systems to process and make available the enormous data sets that will accompany this research.

While it will be possible to develop the genetic variants and to do the genotypic studies of these, the most difficult aspect will no doubt be the phenotypic analysis under varied environmental conditions. The larger the set of genotypes to be tested, the more difficult it is to accomplish this in an in-depth and accurate fashion. To achieve this efficiently, better measurement tools for various phenotypic traits and environmental conditions would be useful. For example, the ability to collect data via remote sensing will be feasible for many plant traits. Finally, it will be necessary to integrate this type of data together with the above genetic/biochemical/genomic data as well as with environmental data for each test site. All of this will require a wide range of expertise.

Funding and Organizational Issues: What Is Required for Success?

There is no question that it will require a considerable increase in funding to develop the fundamental understanding required to use gene knowledge in a predictive fashion to alter important crop traits. However, two other issues need to be solved to achieve success. First, the funding constraints in the crop genetics industrial research sector means that for most crops, simply defining important genetic effects will not be sufficient and it will be necessary for public sector researchers to go considerably further to apply this knowledge than would otherwise be required. Second, success will require a combination of a wide range of expertise, including the areas of genetics and biotechnology, breeding, agronomy, and agricultural engineering. The private sector companies can develop products due to their ability to organize these required capabilities, and this is not easily achieved in the public sector. The current method of dispersed funding for academic research has been extremely successful in generating fundamental knowledge in the plant sciences and will continue to be important. Furthermore, in those institutions that still support plant breeding research, new lines can be developed in the traditional fashion, which does not require knowledge of gene function. However, in the absence of an organization capable of using all the available tools, it is difficult to envision how even the most brilliant fundamental work being done in plant cell and molecular biology can lead to useful outcomes for crops in which the industry cannot afford significant research expenditures.

Industrial research organizations can be judged by the success of their products in the marketplace. Public sector researchers doing fundamental research can be judged by their ability to publish and compete for grants. However, it has proven to be difficult to set up a public research organization that has the staying power and performance metrics to fulfill this role of shepherding basic research results into products. Furthermore, there is no question that success will require cross-national programs, if only to be able to develop the requisite testing sites. What is certain is that this will cost a considerable sum to accomplish, will need to have a long time horizon, and will need to be organized with clear metrics for success.

CONCLUSIONS ON GOALS AND ROLES

Being able to feed the world population in an environmentally sustainable fashion as it reaches its apex over the next 40 to 50 years will be difficult to achieve, but it is a vitally important enterprise. Failure will lead to extremely challenging future socio-economic conditions. Furthermore, the goal of doubling yields is clearly an international challenge, with much of the improvement needing to occur in developing nations. Improved crop genetics is not the only thing required, but it is nevertheless an important component for success.

Great strides have been made in understanding the role of different genes in plant growth and development. In addition, the advent of genomics and other similar technologies have allowed for the generation of enormous quantities of data related to these genes and their effect on phenotype. The need is clear: feeding the world's population in 50 years without increasing land for farming and in an environmentally responsible way will take considerable drive, commitment, and a substantial increase in research funding. For this to be successful, it also will require a significant change in how fundamental knowledge is translated into applications. Research must go beyond data acquisition and deeper understanding of fundamental biology into the realm of successful implementation. Otherwise, we will have failed to meet the essential needs of the future world population.