John Libbey Eurotext

Science et changements planétaires / Sécheresse

Is crop breeding the first step to fill the yield gap? Volume 24, issue 4, Octobre-Novembre-Décembre 2013


See all figures


Auteur(s) : Philippe Monneveux1, Oscar Ortiz1, Othmane Merah2

1 International Potato Center Avenida La Molina 1895 La Molina Lima Perú

2 Université de Toulouse INP-ENSIACET Laboratoire de chimie agro-industrielle F–31030 Toulouse France

Reprints: P. Monneveux

The major challenge for the global community is to provide food for a population expected to reach 9 billion by 2050. This means that food production should increase 50-70% during the same period (Schade and Pimentel, 2010), with similar or even less resources. Food demand is mainly determined by population and income growth and the demand for biofuels (Rosegrant et al., 2008). The demand for cereals for food is projected to decrease in most regions (except sub-Saharan Africa) as people increasingly consume livestock products. Livestock in turn will drive demand for feed grain, especially maize (Fischer et al., 2009).

Our capacity to meet this challenge without affecting carbon-rich and biodiverse natural ecosystems ultimately depends on increasing production on each hectare of currently used arable land with fewer resources. Increasing farmers’ yield will consequently remain a critical component of the global strategy to achieve food security. Farmers’ yield depends on the yield potential of the crop and the gap between this and the actual yield obtained by farmers. The magnitude of the gap depends on physical, biological and socioeconomic factors and their interactions (Lobell et al., 2009).

The diverse expressions of yield

Yield potential

Yield potential (YP) is the yield of cultivar when grown in environments to which it is adapted, with non-limiting nutrients and water, and with pests, diseases and weeds effectively controlled (Evans and Fisher, 1999). YP is determined by genotype characteristics and environmental factors like solar radiation, temperature, and atmospheric concentration of carbon dioxide (figure 1). For a given location YP corresponds to an optimum planting date allowing optimal use of radiation and sums of temperature.

Simulation growth models provide the most accurate estimate of YP (De Bie, 2000; Yang et al., 2004; Lobell et al., 2005; Aggarwal et al., 2006; Lobell et al., 2009). Calibrated using the latest cultivars, most crop models simulate phenological development in relation to photo-thermal time, net assimilation, resource allocation to different organs, transpiration and soil water dynamics on a daily or hourly time step (Jones et al., 2003). Weaknesses of most models are: i) the lack of sensitivity to short-term abiotic stresses that affect yield determining steps, leading to over-estimation of yield (Fischer et al., 2009); and ii) the low resolution that do not permit to include all possible variations in environments.

Attainable yield

Attainable yield (AY) is the yield of experimental plots with the best-known management practices for a given agroecology. As it is impossible to achieve perfect growth conditions (plant density allowing rapid leaf canopy development and earliest interception of incoming solar radiation, crop management practices that eliminate nutrient deficiencies or toxicities, damage from insect pests and diseases, competition from weeds, homogeneity in soil properties) in field conditions, AY estimated from field experiments is lower than YP estimated by crop models. However, AY is sometimes used by plant breeders to find the most productive genotypes or to evaluate genetic progress, obtained by plotting AY of historic sets of varieties against variety year of release (Fischer et al., 2009).

Actual farm yield

Actual farm yield (FY) is the average yield achieved by farmers, at different spatial and temporal scales. FY is determined by the direct action of nutrient deficiencies or imbalances, poor soil quality, diseases, insect pests and weed competition that depend on crop management techniques. FY is lower than AY because of the limitations on controlling the environment and the sub-optimal agronomic management. In addition, farmers tend to maximize profits rather than just increase production. In most situations, the market price of the crop and costs of essential inputs are such that economic returns are highest at input levels that are below what is required to reach optimum production.

Maximum yield achieved by a farmer in a region of interest is exceptionally used to estimate yield potential (Sadras et al., 2002; Lobell et al., 2009). This is only appropriate in both highly favorable climatic conditions and intensively managed cropping systems.

Yield gap(s)

The gap between YP and FY is highly variable worldwide (Lobell et al., 2009). In most situations it exceeds 30%. It can be less than 20% in irrigated wheat systems in North-West India where scarcity of land and abundance of labor favor technologies that tend to minimize yield gap, but reaches 80% for tropical maize in Africa, where biophysical and management conditions result in frequent biotic and abiotic stresses (Fischer et al., 2009).

Yield gap between YP and FY can be broken down into two components (figure 1). The first component (Gap I) represents the difference between YP and AY. The second component (Gap II) is the gap between AY and FY, mainly caused by the impact of technologies and intensive practices that can be found in experimental stations but not at the farm level.

Identification of major constraints responsible for yield gaps

Assessing the relative contribution of constraints causing yield gaps is essential to define optimal crop management and breeding objectives. Three types of factors are traditionally considered (Van Ittersum and Rabbinge, 1997). Growth defining factors that determine YP cannot be controlled through management, including solar radiation, temperature and plant characteristics (figure 1). Growth limiting factors comprise water and nutrients which are considered essential inputs for plant growth and determine Gap I. Growth reducing factors which determine Gap II include pests, diseases and weeds.

Comparing yields from affected and non affected areas remains the best approach to estimate yield losses due to different factors (Oerke, 2006). In the case of biotic stresses, “potential losses” correspond to losses occurring in absence of physical, biological or chemical crop protection while “actual losses” comprise losses occurring despite the crop protection practices. The efficacy of crop protection practices is calculated as a percentage of potential losses prevented. Actual and potential losses vary with the considered crop (figure 2). Estimated efficacy of control is around 51.4, 43.4, 54.5, 46.2 and 56.2% for rice, wheat, maize, potato and soya, respectively (Oerke, 2006). A similar approach is also used in the case of abiotic stresses. In the case of tropical maize, losses due to drought in lowland tropics averaged 17% of AY (Edmeades et al., 1992) and can reach up to 60% in severely drought-affected regions (Rosen and Scott, 1992).

How to increase yield in farmers’ fields?

By increasing yield potential

Yield potential increased substantially in the past decades. It is expected to largely drive FY growth, particularly in intensive cropping systems, when FY approaches 70 to 80% of YP (Fischer et al., 2009). Crop physiologists consider YP as the product of three factors: the photosynthetically active radiation (PAR) intercepted by green tissue over the life of the crop (PARi, in megajoule [MJ]), the radiation use efficiency (RUE) or efficiency with which PARi is converted into above-ground biomass (RUE, in g/MJ), and the harvest index of the crop (HI, ratio of the harvested part of the crop to total biomass) (Passioura, 1977).

In many crops yield progress during the past 50 years has been due to a higher proportion of carbon partitioned to the harvested organs (higher HI) (Jain, 1986). In most modern cereal cultivars, HI seems to be close to its biological maximum (i.e., 60%) suggesting that further genetic gains in YP may rather come from biomass increases (Shearman et al., 2005). In cereals, PARi intercepted by green tissue over the life of the crop can be increased by selecting for greater “stay green” (Borrell et al., 2001) or for erected leaves (Beadle and Long, 1985). This approach has been particularly effective in sorghum and rice, respectively.

Radiation use efficiency is the ratio of gross photosynthesis minus crop respiration to radiation intercepted. Its values during vegetative growth under optimal conditions are about 4.0 for sugar cane, 3.3 for maize, 3.3 for potato, 3.1 for sunflower, 2.9 for wheat, 2.8 for rice, 2.6 for sorghum, 2.6 for barley, 2.1 for soybean and peanut and less than 2.0 for grain legumes (Mitchell et al., 1998). Theoretical maximum limit to RUE is 5.8 g/MJ in C3 crops and 6.9 g/MJ in C4 crops, which are much higher than the values reported above (Long et al., 2006). Different ways have been proposed to genetically increase RUE (table 1).

Table 1 Potential breeding objectives to enhance radiation use efficiency (RUE), justifications and perspectives.

Breeding objectives for enhancing RUE Justification Perspectives
Reduced crop respiration The reduction of crop respiration increases RUE Only modest prospects of improvement (Loomis and Amthor, 1999)
Reduced photorespiration Varietal differences reported in wheat (Monneveux et al., 2003) Mainly in C3 crops
Uppermost leaves nearly vertical, so that they are not light-saturated, while lower leaves are almost horizontal to ensure that almost all the light is intercepted. Increasing RUE under radiation levels of 10-50 percent full radiation Potential to increase RUE
by as much as 40% at midday in full sunlight (Long et al., 2006)
Increasing Pmax There is an association between RUE and Pmax Higher Pmax observed in modern cultivars suggesting scope for improvement (Fischer et al., 1998)
Increasing, by genetic engineering, Rubisco efficiency to capture CO2, or increasing the supply of CO2 or other limiting substrates to the enzyme Engineering of the Rubisco, may increase efficiency to capture CO2, or the supply of CO2 or other limiting substrates to the enzyme Predicted RUE increase at annual rates of 1-4% over the next 10 to 20 years (Long et al., 2006)
Modify the photosynthetic way Idem Complex, difficult to deliver in short term (Hibberd et al., 2008)
Reducing the thermal sensitivity of Rubisco activase (Salvucci, 2008) Higher Rubisco activity at high temperatures Pertinent in hot areas and to face climate change effects

By filling the gaps

Gap I

This gap can be reduced mainly by addressing limiting factors like water and nutrition. A first way of filling this gap is by fitting the phenological development to the particular environments or by selecting for earliness to bring the growth of the crop into a moister period. Better crop nutrition, especially N nutrition, can also contribute to fill this gap by greater leaf area of longer duration and increased PARi (Bange et al., 1997). Gap I can finally addressed by improving tolerance to abiotic stresses. Significant progress has been made in improving drought tolerance in cereals, particularly in maize (Monneveux et al., 2006).

Gap II

Gap II is caused mainly by environmental conditions and technologies that are available at research stations but are not applied in farmers’ fields. In most situations gap II exceeds 30% (Fischer et al., 2009). Yield gap appears to be large in maize, especially in sub-Saharan Africa where it easily exceeds 100%. There is a space to reduce gap II because of the wide variation among farmers. Within African farming systems, lower yields are often less than half of highest yields and variation mainly relate to soil properties (Van Asten, 2003), agronomic practices (Samake et al., 2006) and farmers’ resource allocation decisions (Nkonya et al., 2005).

Contribution of breeding and agronomy in improving farmers’ yield

Past contributions

In most countries, wheat, rice and maize yield increase has been predominantly linear over the last decades (Hafner, 2003). The rate of increase for wheat and rice in developing countries however declined from the mid-1980s to about 1% annually in the most recent decade (Fischer et al., 2009).

Globally, the proportion of yield increase attributable to plant breeding during the last decades was considered to reach 50% (McLaren, 2000). It was 47% in wheat and 55% in barley in UK (Silvey, 1994), 50% for maize in the USA and 47% for sugar beet in the UK (Scott and Jaggard, 2000). The biggest contribution of plant breeding is in closing gap II and maintaining resistance levels in the face of evolving pest agents, as recently documented in the case of wheat rusts (Dubin and Brennan, 2009). Conversely, modern varieties tend to be more susceptible to weed competition, so breeding has not helped so efficiently (Fischer et al., 2009).

The contribution of crop management resides in crop protection, crop nutrition, soil preparation and development of irrigation. Efficacy of chemical control is around 55% for weeds, 31% for pests and 23% for diseases (Oerke and Dehne, 1997). Increased use of fertilizer is considered to explain one third to one half of FY growth in developing countries since the Green Revolution (Heisey and Norton, 2007). This has been obtained by increasing the doses (3.6% per year over the past decade) and applying in a more efficient way to reduce waste of resource to the farmer and leaching. Low FY and high gaps in SSA (Sub-Saharan Africa) are largely due to low use of N (5 kg/ha), related to the high price of N to grain fertilizer price ratio (Morris et al., 2007).

Over the past decade irrigated area has expanded steadily at 0.6% per year in developing countries. Given a productivity differential between irrigated and rainfed areas of 130% (Fuglie, 2008), irrigation alone accounted for about 0.2% in overall annual yield growth of 1.1% for cereal yields from 1991-2007 (Fischer et al., 2009). The adoption of the bed-planting method in the Yaqui Valley (Mexico) paralleled by the use of pre-seeding irrigation and the use of mechanical cultivation for weed control allowed an increase of yields, a reduction of application of costly herbicides and a reduction of seed rates (Sayre and Moreno Ramos, 1997).

Can the variety of tomorrow reduce agronomic management needs?

New varieties often represent a less expensive option for farmers and extension organizations and are generally adopted more readily than new management techniques, although there are cases where varietal change has been difficult or slow (Walker, 1994; Walker, 2008). Targeted breeding can help both raising YP and closing yield gaps, essentially by making varieties more resilient. However, potential losses often increase when more productive varieties are grown (eg, losses due to diseases in wheat in Germany increased from 11 to 20% when AY rose from 4 t/ha to 11 t/ha [Jaggard et al., 2010]). In the future, more efficient protection will be required, also because of the threat of new resistance in the pathogens. The pressure to reduce the use of crop protection chemicals could lead to a limitation in their use. In face of the decrease of market values of chemicals (-18% between 1998 and 2003, [Clough, 2005]), multinational companies involved in seeds and agrochemicals could be tempted to reduce their investments in research and development of new products and rather invest in GMOs (genetically modified organisms) (Fischer et al., 2009).

There is also a scope to genetically improve nutrient use efficiency. As production and application of N fertilizers consume huge amounts of energy, and excess is detrimental to the environment, any progress in nitrogen use efficiency (NUE) will be essential for a more economically and environmentally friendly use of valuable N resources (Xu et al., 2012).

In summary, future yield gains are probably going to depend more on breeding than on new developments in crop agronomy, at least under favourable environments, but efforts to optimize the use of nutrients and control pests, diseases and weeds should be maintained to ensure that the new varieties will express their potential (Jaggard et al., 2010).

Obstacles to the dissemination and impact of new varieties

Seeds are the vehicle of novel varieties. Understanding the seed supply system and the factors limiting the production, multiplication, and marketing is essential for promoting dissemination and diffusion of improved varieties. There are generally three reasons for low rates of adoption of new varieties: i) new varieties are not superior to the current varieties, or are inferior in some important characteristics ; ii) farmers do not have information about new varieties; or iii) farmers do not have access to seeds of new varieties.

The potential impact of new varieties depends on the biological quality of the seed. In the case of root and tuber crops, because of asexual propagation, biological quality is mainly determined by the level of disease infection or “degeneration” due to viruses and virus-like organisms (Okorogri et al., 2010; Ling et al., 2010).

Which future for crop management?

Future success of crop management will depend on the capacity to capture and address local specificities (Tittonell et al., 2007). On-farm experiments with rice in Asia (Dobermann et al., 2003) and potato in the Andes (Ortiz et al., 2004) already pointed to the importance of adjusting crop management according to field to field variability. On-farm survey, facilitated by the use of remote sensing and communication technologies advances, will continue to have a central role to explain differences in farmers’ fields yield and idenify responsible factors (Paroda, 2004). In the Yaqui Valley (Mexico), latest high-resolution satellite imagery and farm surveys revealed that yield of highly productive wheat varieties was mainly constrained by late planting (Ortiz-Monasterio and Lobell, 2007), delays in the first post-plant irrigation (Lobell and Ortiz-Monasterio, 2008) and summer fallow weeds (Ortiz-Monasterio and Lobell, 2007). These evidences suggest that even with the best possible genetic material, the full expression of the potential would depend on management decisions.

The diffusion of new technologies will highly depend on the capacity of extension services and participation of farmers (Phillips, 2010). The development of conservation agriculture (CA) for wheat, maize and soybeans in southern South America has been mainly driven by the farmers faced with the threat of soil degradation and the opportunity provided by knock-down herbicides (Ekboir, 2002).

Additional solutions to improve production should be searched at the cropping system level. Multiple cropping systems can convert a greater proportion of the potential land productivity in crop yield. Examples are the rice–wheat (Iijima et al., 2005), maize–wheat (Jin et al., 2011) and wheat–soybean (Caviglia et al., 2004) systems. A science-based crop management support is required to optimize spacing and timing of planting and application of external inputs.

An important question is to which extent crop management will be able in the future to reduce or mitigate degradation of soil structure, erosion and salinization (Bruinsma, 2003). Another burning issue is the capacity to reduce fertility losses in conditions of high exportation of minerals and poor restitutions. Finally the last issue is water shortages in irrigated areas. Climate change projections indicate that there will be less reliability on water sources. Therefore, either the genetic material has the adaptation to a wide range of levels of soil humidity, or crop management for increased water use efficiency needs to be put in place.


In technically advanced cropping systems the contribution of agronomic innovation is becoming smaller, although agronomic innovation remains very important for input use efficiency. In less technically developed systems the lack of adoption of modern agronomy is still the major cause of the yield gap. Research will have to develop more integrated approaches and new tools (Kropff et al., 1993; Van Ittersum et al., 2008). Combining experimental research and explorative studies based on biophysical and/or economic models will permit to more accurately evaluate the impact for different agro-ecological conditions and a longer time-frame (Chen et al., 2010). Modeling studies on the effects of climate change on yield are essential, particularly in developing countries and for non-cereal crops. In developing countries, bad quality of infrastructure, weak institutions and bad farm policy by leading to price disincentives at the farm gate, expensive credit, and increased risk in general can create huge obstacles to the adoption of improved technologies. Improved varieties which can cope with variable and unpredictable environmental conditions will be needed. However, it is not possible to expect that genetic improvements will replace agronomic management in full, because of the high heterogeneity and the need to provide the best possible agronomy for new genotypes to express their potential.