"Global crop exposure to critical high temperatures in the reproductive period: historical trends and future projections," by S. M. Gourdji, A. M. Sibley & D. B. Lobell, ERL 8 (2013); doi:10.1088/1748-9326/8/2/024041

Environmental Research Letters, 8 (2013) 024041; doi:10.1088/1748-9326/8/2/024041

Global crop exposure to critical high temperatures in the reproductive period: historical trends and future projections

Sharon M. Gourdji, Adam M. Sibley and David B. Lobell

Abstract

Long-term warming trends across the globe have shifted the distribution of temperature variability, such that what was once classified as extreme heat relative to local mean conditions has become more common. This is also true for agricultural regions, where exposure to extreme heat, particularly during key growth phases such as the reproductive period, can severely damage crop production in ways that are not captured by most crop models. Here, we analyze exposure of crops to physiologically critical temperatures in the reproductive stage (T_crit), across the global harvested areas of maize, rice, soybean and wheat. Trends for the 1980–2011 period show a relatively weak correspondence (r = 0.19) between mean growing season temperature and T_crit exposure trends, emphasizing the importance of separate analyses for T_crit. Increasing T_crit exposure in the past few decades is apparent for wheat in Central and South Asia and South America, and for maize in many diverse locations across the globe. Maize had the highest percentage (15%) of global harvested area exposed to at least five reproductive days over T_crit in the 2000s, although this value is somewhat sensitive to the exact temperature used for the threshold. While there was relatively little sustained exposure to reproductive days over T_crit for the other crops in the past few decades, all show increases with future warming. Using projections from climate models we estimate that by the 2030s, 31, 16, and 11% respectively of maize, rice, and wheat global harvested area will be exposed to at least five reproductive days over T_crit in a typical year, with soybean much less affected. Both maize and rice exhibit non-linear increases with time, with total area exposed for rice projected to grow from 8% in the 2000s to 27% by the 2050s, and maize from 15% to 44% over the same period. While faster development should lead to earlier flowering, which would reduce reproductive extreme heat exposure for wheat on a global basis, this would have little impact for the other crops. Therefore, regardless of the impact of other global change factors (such as increasing atmospheric CO₂), reproductive extreme heat exposure will pose risks for global crop production without adaptive measures such as changes in sowing dates, crop and variety switching, expansion of irrigation, and agricultural expansion into cooler areas.

1. Introduction

Increases in mean annual temperatures have been evident throughout the world in the past few decades [1], including in most major agricultural regions [2]. Along with mean temperature changes, there has been an increase in the occurrence of warm temperature extremes, and a simultaneous reduction in cold extremes [3–5]. The trends in occurrence of events historically considered extreme are the natural consequence of a shifting mean, but changes in the shape of the temperature distribution have also likely played a role [5–7]. As with changes in mean temperature, the trends in extremes are primarily attributable to greenhouse gas emissions [8, 9], and therefore are expected to continue in the coming decades.

Hotter average temperatures affect crop production via several mechanisms, including by speeding rates of crop development and evapotranspiration [10–12]. The effects of changing mean conditions are fairly well captured by crop simulation models frequently used to assess climate change impacts, with the notable exception of effects of mean climate on pests and diseases [13]. However, extreme temperatures (both hot and cold) can also cause physiological damage and lead to crop failure, particularly during sensitive stages of the crop cycle (e.g., the flowering or reproductive stage) [14, 15]. Although widely recognized by the crop modeling community, the effects of extremes are often not well quantified and are generally not captured in existing models [16–19].

Crops are particularly sensitive to extreme heat in the reproductive period, where it can substantially reduce grain number and final yields, and in the most negative cases lead to complete crop failure [20–22]. Teixeira et al. [23] examine projected exposure to reproductive heat stress for wheat, maize, rice and soybean in the period 2071–2100. Their analysis considered a single global circulation model (GCM) projection, without correction for potential GCM temperature bias in agricultural regions. Nonetheless, the study suggested the potential for large increases in area exposed to reproductive heat stress on a global basis, particularly for maize and rice, although adaptive measures are shown to partially mitigate impacts, especially for wheat.

Although crop heat stress is expected to increase by the end of the 21st century, there has been little work done to document changes already occurring in global cropping systems, and only then in individual regions [19, 24]. Here, we examine historical rates of exposure to crop-specific critical temperatures, henceforth referred to as T_crit, in the 30-day window around flowering, i.e. the reproductive period. We assess historical trends throughout the global area of maize, wheat, rice, and soybean, to test whether occurrence has been changing in the last three decades. We also consider sensitivity to future warming over the next few decades (until the 2050s), in order to identify areas at risk of large increases in exposure.

Open access:  http://iopscience.iop.org/1748-9326/8/2/024041/article