Published in Agron J 91:940-946 (1999)
© 1999 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
Agronomy Journal 91:940-946 (1999)
© 1999 American Society of Agronomy
GRAIN AND OIL CROPS
A General Thermal Index for Maize
Lianne M. Dwyera,
Douglas W. Stewarta,
L. Carriganb,
B.L. Maa,
P. Neavea and
D. Balchina
a Agric. & Agri-Food Canada, Eastern Cereal & Oilseed Res. Ctr., Ottawa, ON, K1A 0C6, Canada
b Pioneer Hi-Bred Int., Plant Breeding Div., Willmar, MN 56201 USA
dwyerl{at}em.agr.ca
 |
ABSTRACT
|
|---|
Thermal indices predict and describe development rate more accurately than time in days and are commonly used to rate maize (Zea mays L.) for maturity. Separate temperature response functions for the vegetative and grain-filling periods predict more accurately time to maturity than a single function for the two periods combined. However, use of two functions requires a priori knowledge of the silking date, which becomes the transition date from the vegetative function to the grain-filling function. The objective of this study was to evaluate the sensitivity of estimates of silking and maturity dates to the transition date between vegetative and grain-filling functions and to develop a protocol to combine the two temperature response functions in a general thermal index (GTI) for maize. Frequency distributions of mean daily air temperatures for five 20-d periods spanning mid-June to late September at 19 locations in the northern USA and southern Ontario from 1992 to 1995 indicated few days (
12%) with mean daily air temperatures less than 15°C before late August. This was significant, as the two response functions diverged significantly at temperatures below 15°C. Standard errors in estimating maturity date using different transition dates remained small (<7.5 d) unless the transition date was delayed beyond the first week of September. Based on this analysis, a standard transition date of 1 August was proposed for the GTI. Testing on an independent data set indicated that the GTI and a transition date of 1 August provided more accurate estimates of the planting to maturity period than growing degree days (GDD) or crop heat units (CHU), with a standard error of 8.2 d (compared with 14.5 d using GDD and 12.5 d using CHU).
Abbreviations: CHU, crop heat units • GDD, growing degree days • GTI, general thermal index • MRMR, Minnesota relative maturity rating • SEE, standard error of the estimate
|
INTRODUCTION
|
|---|
ESTIMATION
of maize development rate is needed to market hybrids of specific maturity, to zone production areas, and to schedule stage-sensitive operations such as herbicide spraying. Thermal indices predict and describe development rate more accurately than time in days and are commonly used to rate maize for maturity (Shaykewich, 1995). The most common thermal indices are growing degree days (GDD; Wang, 1960) and crop heat units (CHU; Brown and Bootsma, 1993). An ideal index would estimate a constant number of heat units for a given genotype to reach a specific development stage. Existing indices provide reliable estimates of thermal time required for vegetative development (the interval between planting and silking), but estimates of thermal time required for grain filling (the interval between silking and maturity) have been found to vary with location and year (Major et al., 1983; Plett, 1992). Most frequently, the GDD concept overestimates the heat units required for grain filling (Dwyer et al., 1996). This is particularly evident in years with below-normal temperatures, when maturity estimates using GDDs may be several hundred heat units too high (Roth and Yocum, 1997).
Recently, separate temperature response functions have been developed for the vegetative and grain-filling periods of maize (Stewart et al., 1998). These equations, which were fitted to data from 28 hybrids collected at 19 locations ranging from 39° to 48° N lat over 4 years, differed from those assumed by GDD and CHU and also from each other, especially at temperatures below 15°C. Use of the fitted response functions reduced standard errors in estimating the entire period from planting to maturity by 50% compared with GDD estimates, with a reduction in the standard errors in estimating the grain-filling period alone from 13.6 d based on GDD estimates to 7.0 d using the fitted response function (Stewart et al., 1998). However, use of the fitted response functions required a priori knowledge of the silking date, which became the transition date from the vegetative function to the grain-filling function. Such a variable transition date would not be appropriate for routine application of the temperature response functions in a thermal index.
The objective of the present study was to evaluate the functions described by Stewart et al. (1998) for the vegetative and grain-filling periods as the basis of a general thermal index (GTI) for the entire growing period. In particular, the sensitivity of estimates of silking and maturity dates to the transition date between the vegetative and grain-filling functions were determined and a protocol was developed for a simple method to combine the two temperature response functions into a single general thermal index for maize.
|
Materials and methods
|
|---|
Coefficients for the temperature response functions were fit to field data on 28 Pioneer maize hybrids at 19 locations in the north-central and northeastern USA and southern Ontario from 1992 to 1995 as described in Stewart et al. (1998) (Tables 1 and 2) . Hybrids in this data set were rated from 75 to 110 d relative maturity based on the Minnesota relative maturity rating system (MRMR; Peterson and Hicks, 1973). These hybrids and locations from 39 to 48° N lat represent maize production systems that depend on a thermal index for management decisions. Later-maturing hybrids grown further south are typically less limited by heat units for the season.
View this table:
[in this window]
[in a new window]
|
Table 1 Sites in the 19-location, 4-year (1992–1995) data set and in a second 5-location data set (2, 3, or 4 years, 1994–1997)
|
|
View this table:
[in this window]
[in a new window]
|
Table 2 Maize hybrids in the 19-location, 4-yr (1992–1995) data set and in the 5-location data set (2, 3, or 4 years, 1994–1997)
|
|
Dates for planting, silking (50% of plants with silk), and maturity (starch layer advanced completely to the cob on 50% of plants; i.e., 0% milk line) were recorded based on observations taken a minimum of three times weekly on the central two rows of four-row plots replicated two times. Daily maximum and minimum air temperatures were measured at each location from planting to maturity.
Temperature response functions (Fig. 1)
fitted to the vegetative (FT(veg)) and grain-filling (FT(fill)) periods by Stewart et al. (1998) were the following cubic polynomial functions:
 | (1) |
where TA is the average daily air temperature and B0, B1, and B2 are empirical coefficients. During the vegetative period, B0 was never significantly different from zero and the two-coefficient model
 | (2) |
provided the best fit. For the grain-filling period, the cubic term was never significantly different from zero and
 | (3) |
provided the best fit. The general thermal index was derived as follows:
 | (4) |
where FT is the temperature response function fitted to the vegetative or grain-filling periods, n is the number of days in a period, and 
t is a time step in days.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 1 Temperature response as a function of mean daily air temperature for the vegetative (Eq. [2]; solid line) and grain filling (Eq. [3]; dashed line) periods
|
|
Following Stewart et al. (1998), coefficients of variation were calculated for thermal time estimated using (i) actual time in days, (ii) GDD, (iii) CHU, and (iv) the two separate response functions (Eq. [2] and [3]), together termed the General Thermal Index (GTI), for the planting to maturity periods. The GDD, CHU, and GTI were used to calculate time in days from planting to silking and from silking to maturity, and these calculated days were compared with recorded days. A standard error of the estimate (SEE) was calculated between recorded time in days and time in days estimated from GDD, CHU, and GTI calculations. A mean square error (MSE) was then back-calculated for comparison of the regression models using the F-test (Steel and Torrie, 1980). Hybrid thermal averages for the entire planting to maturity period, based on all locations and years, are presented in Table 2. Note that the size of GTI values was set equivalent to that of GDD values.
It was noted that the two temperature response functions in GTI diverged at mean daily temperatures below about 18°C, but were generally quite similar at temperatures between 15°C and 30°C (Fig. 1). Thus, significant differences in estimates by the two functions occurred on days with mean daily air temperatures below 15°C. Frequency distributions of mean daily air temperatures were generated for 14-d and 20-d incremental periods from mid-June (Day 170) to late September (Day 270). Frequencies were plotted for the midpoints of 5°C increments of mean daily air temperatures for 20-d incremental periods as this period length identified differences in distribution of mean daily temperature with time in relatively few (five) periods.
The GTI was used to calculate time in days from planting to a transition date using Eq. [2] and from the transition date to maturity using Eq. [3], and then the sum of calculated days was compared with recorded days for the planting to maturity period. Transition dates ranged from Day 180 to Day 310, at 10-d intervals. Mean square errors, in days, were calculated between measured days and estimated days calculated from hybrid thermal averages required for the planting to maturity period, as a function of the transition date.
A second data set was obtained on 14 Pioneer maize hybrids at five locations in the north-central USA and southern Ontario and Quebec from 1994 to 1997 (Tables 1 and 2). Hybrids in this data set, rated from 76 to 95 d relative maturity based on MRMR, were arranged in randomized complete block designs with three or four replications. Each plot was a minimum of four 0.76-m rows, 8 m long. Planting, silking, and maturity dates were noted as in the main data set. Daily maximum and minimum air temperatures were determined from recorded hourly air temperatures. This data set served as an independent test of thermal unit estimation based on GTI coefficients developed from the main data set. Locations represented in this data set were also clustered in more northern latitudes, where the probability of experiencing mean daily air temperatures 15°C near silking (transition from vegetative to grain-filling functions) was higher than in the main data set.
|
Results and discussion
|
|---|
Comparison of coefficients of variation and mean square errors indicated that GTI improved estimation of development rate over GDD and CHU. Although estimates based on CHU were better than those based on GDD and all thermal indices provided better estimates of development rate than days alone, GTI provided the best estimates of the duration of both the vegetative and grain-filling periods (Table 3)
.
View this table:
[in this window]
[in a new window]
|
Table 3 Goodness of estimate of days required from planting to silking and from silking to maturity using actual days, growing degree days (GDD), crop heat units (CHU), and the general thermal index (GTI) to estimate development in maize, based on coefficient of variation (CV) and on standard error of the estimate (SEE) for both data sets (Tables 1 and 2)
|
|
Frequency distributions of mean daily air temperatures for five 20-d periods spanning mid-June to late September indicated that temperatures for the 19 locations in the main data set began to drop rapidly beginning only in the first to third week in September (i.e., Day 250 to 270). Periods prior to this had approximately 12% or fewer of their days with mean daily air temperatures in the range 10 to 15°C (midpoint of 12.5°C; Fig. 2)
. Frequency distributions of mean daily air temperatures of the four most northern sites (Table 1) indicated that temperatures below 15°C were more common, but there were still only 14% of site-days with mean daily air temperatures below 15°C in the last three weeks of July (Day 190–210) and 18% in the first three weeks of August (Day 210–230) (Fig. 3) . Frequency distributions of mean daily air temperatures of the 17 site-years in the second data set (Table 2) do not show air temperatures of less than 15°C making up >10% until the second week of September (Day 250–270) (Fig. 4)
.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 2 Frequency distribution of mean daily air temperatures at 19 locations ranging from 39° to 48° N lat over five years (1992–1995) for Day of Year (a) 170–190, (b) 190–210, (c) 210–230, (d) 230–250, and (e) 250–270
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 3 Frequency distribution of mean daily air temperatures at four locations in the main data set (Mankato, MN; Willmar, MN; Moorhead, MN; Grand Forks, ND) with latitudes >44° N for Day of Year (a) 170–190, (b) 190–210, (c) 210–230, (d) 230–250, and (e) 250–270
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 4 Frequency distribution of mean daily air temperatures at the five locations in the second data set ranging from 41° to 45° N lat over four years (1994–1997) for Day of Year (a) 170–190, (b) 190–210, (c) 210–230, (d) 230–250, and (e) 250–270
|
|
The SEE in estimating planting to silking, silking to maturity, and planting to maturity using GTI but with different transition dates from Eq. [2] to Eq. [3] remained small (i.e., <7.5 d) until Day 250 (second week in September). Beyond this date, SEE rose rapidly (Fig. 5)
. Thus, accuracy in estimating development rate did not require a precise date for transition from Eq. [2] to [3], but it did require that Eq. [3] be used later in the grain-filling period when mean daily air temperatures were consistently below 15°C. In reality, silking occurs from mid-July to mid-August in almost all commercial hybrids at these locations. We therefore propose that a new improved thermal index can be developed using Eq. [2] and [3] with a set transition date. Day 213 (1 Aug.), a date that approximates 50% silking for many of the hybrids grown at the latitudes represented in this study, could be set as a conservative transition date. The relative insensitivity of estimates of development rate to transition dates a month on either side of 1 August suggested that a new GTI based on two temperature response functions and a set transition date of 1 August would accurately estimate maize development rate.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 5 Standard error (SEE) in estimating periods from planting to maturity in main data set when transition from Eq. [2] to Eq. [3) occurs on different days of the year (end of June to early November)
|
|
Maturity dates were estimated for the second data set (Table 2) using GTI (with a transition date of 1 August), GDD, and CHU. All three thermal indices estimated silking dates well. Estimates of maturity date based on GTI for all hybrids were more accurate than those based on GDD or CHU (Table 3). There were seven site-years in which GDD or CHU provided more accurate maturity estimates than GTI, but (with the exception of Ottawa 1997) the difference between actual maturity date and that estimated using GTI was 5 d (Fig. 6)
. Estimates based on GDD and CHU deviated from the actual maturity date by 20 d or more for five site-years. Thus, GTI provided more consistent maturity estimates for the independent data set than the other thermal indices tested.
View larger version (36K):
[in this window]
[in a new window]
|
Fig. 6 Estimated number of days from planting to maturity, using GTI with a constant transition date of 1 Aug., GDD, and CHU, minus the actual number of days for 17 site-years in the second data set. Years are designated by the last two digits. Sites are designated by the letters O for Ottawa, ON; P for Ste Polycarpe, QC; Wo for Woodstock, ON; Wi for Willmar, MN; and J for Johnston, IA
|
|
Work is ongoing to further characterize physical conditions for those site-years in which all thermal indices performed poorly (e.g., Johnston, 1996). Poor performance indicates that factors other than temperature are significantly affecting development rate. Identification of conditions in years in which thermal indices performed poorly would be an advantage in many thermal index applications. However, for the majority of site-years studied here, air temperature was the dominant environmental factor affecting development and GTI based on temperature response functions for the vegetative and grain-filling periods with a set transition date of Day 213 (1 Aug.) provided better estimates of time to maturity than either GDD or CHU.
|
ACKNOWLEDGMENTS
|
|---|
The contributions of Vlado Puskaric at Woodstock, Jean Marc Montpetit at Ste Polycarpe, Chris Zinselmeier at Johnston Pioneer Research Stations, and Lynne Evenson and Dave Meredith at Ottawa are gratefully acknowledged. Financial support was provided by a Pioneer Hi-Bred competitive grant.
|
NOTES
|
|---|
ECORC contribution No. 991411.
Received for publication October 1, 1998.
|
REFERENCES
|
|---|
- Brown, D.M., and A. Bootsma. 1993. Crop heat units for corn and other warm-season crops in Ontario. OMAF Factsheet, Agdex 111/31. Ontario Ministry of Agric. & Food, Toronto.
- Dwyer L.M., Stewart D.W., Ma B.L., Carrigan L. Performance of a revised thermal index. Can. J. Soil Sci. 1996;76:559 [Abstract from Can. Soc. Agrometeorol. at Agric. Inst. of Canada Annu. Meet., Lethbridge, AB.].
- Major D.J., Brown D.M., Bootsma A., Dupuis G., Fairey N.A., Grant et al E.A. An evaluation of the corn heat unit system for the short-season growing regions across Canada. Can. J. Plant Sci. 1983;63:121-130.
- Peterson R.H., Hicks D.R. Minnesota relative maturity rating of corn hybrids. St. Paul: Agron. No. 27. Univ. of Minn. Agric. Ext. Serv, 1973.
- Plett S. Comparison of seasonal thermal indices for measurement of corn maturity in a prairie environment. Can. J. Plant Sci. 1992;72:1157-1162.
- Roth G.W., Yocum J.O. Use of hybrid growing degree day ratings for corn in the northeastern USA. J. Prod. Agric. 1997;10:283-288.
- Shaykewich C.F. An appraisal of cereal crop phenology modelling. Can. J. Plant Sci. 1995;75:329-341.
- Steel R.G.D., Torrie J.H. Principles and procedures of statistics: A biometrical approach, 2nd ed New York: McGraw-Hill, 1980.
- Stewart D.W., Dwyer L.M., Carrigan L. Phenological temperature response of maize. Agron. J. 1998;90:73-79.[Abstract/Free Full Text]
- Wang J.Y. A critique of the heat unit approach to plant response studies. Ecology 1960;4:785-790.
This article has been cited by other articles:
|
|
|
|
 
I. C. Madakadze, K. A. Stewart, R. M. Madakadze, and D. L. Smith
Base Temperatures for Seedling Growth and Their Correlation with Chilling Sensitivity for Warm-Season Grasses
Crop Sci.,
May 1, 2003;
43(3):
874 - 878.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. L. Nielsen, P. R. Thomison, G. A. Brown, A. L. Halter, J. Wells, and K. L. Wuethrich
Delayed Planting Effects on Flowering and Grain Maturation of Dent Corn
Agron. J.,
May 1, 2002;
94(3):
549 - 558.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
L. M. Dwyer, D. W. Stewart, L. Carrigan, B.L. Ma, P. Neave, and D. Balchin
Guidelines for Comparisons among Different Maize Maturity Rating Systems
Agron. J.,
November 1, 1999;
91(6):
946 - 949.
[Abstract]
[Full Text]
|
|
|
TOP
ABSTRACT
INTRODUCTION
