Reply to 'Temperature and drought effects on maize yield'

Lobell et al. reply —

The comments by Basso and Ritchie¹ are unconvincing for several reasons. First, the main point of our paper² (hereafter referred to as L13) was to gain insight into the mechanism behind the observed negative correlation between extreme heat and yields, a point that Basso and Ritchie do not address. Second, the Agricultural Production Systems Simulator (APSIM) uses a daily time step, and the soil–water balance of the dynamic crop model does account for the amount and timing of rainfall and incorporates algorithms to predict runoff and deep drainage. The timing of rainfall has, of course, an effect on simulated yields, and is one of the many reasons why extreme heat is not a perfect predictor of yields.

Third, they claim that estimates of potential water demand in APSIM are too high, and report values for Penman evapotranspiration (ET) that are all below 6 mm. It is unclear what their table purports to show, or why they choose 15 MJ m⁻² radiation for Penman when an assumption of 40 g m⁻² for potential growth in APSIM would correspond to much higher radiation levels of about 25 MJ m⁻². In addition, contrary to the claim made by Basso and Ritchie, many studies have reported ET in maize fields of 10 mm or more per day. Howell et al. report³ Penman–Monteith calculated values of 12 mm for high-temperature and -radiation conditions. In addition, Piccinni et al. note⁴ that based on large field lysimeter studies, crop water requirements for well-watered maize are 1.2 times greater than the Penman–Monteith estimate, even at such high values. The high Penman–Monteith values for high demand (high radiation and temperature) conditions are reinforced by further large lysimeter studies in irrigated maize⁵ that show measured ET rates of greater than 10 mm per day for a significant part of the crop cycle (for high leaf-area index).

All of these data closely reflect the estimates derived from the algorithms in APSIM. The evidence highlights that the estimate of potential water use is highly dependent on environmental conditions, and the calculations presented by Basso and Ritchie do not compare like with like. The geographical location is not of concern; the issue is the prevailing environment. While location might affect the frequency of conditions, high demand conditions occur throughout the midwest during the growing season and more so in some years.

Note that even if differences existed between the two approaches, these differences are not evidence that one approach is right and the other is wrong. For example, Monteith argued⁶ that transpiration is limited by growth (as reflected in the APSIM approach), rather than vice versa, as reflected in the Penman approach. We would like to see additional modelling groups repeat the analysis of L13; perhaps some will find that their models are less sensitive to extreme heat or more sensitive to rainfall than both APSIM and observations indicate. However, Basso and Ritchie do not provide any such analysis, they provide no evidence to convince us that our analysis overstated the role of vapour-pressure deficit, and they provide no alternative explanation for the observational fact that yields are much more strongly associated with extreme heat than is rainfall.