Simulating the Probability of Grain Sorghum Maturity before the First Frost in Northeastern Colorado

Gregory S. McMaster 1,*, Debora A. Edmunds 1, Sally M. Jones 2, Jerry J. Johnson 2 and Merle F. Vigil 3 1 Water Management and Systems Research Unit, USDA-ARS, Fort Collins, CO 80526, USA; debbie.edmunds@ars.usda.gov 2 Department of Soil and Crop Sciences, Colorado State University, Fort Collins, CO 80523, USA; sally.jones@colostate.edu (S.M.J.); jerry.johnson@colostate.edu (J.J.J.) 3 Central Plains Resources Management Research Unit, USDA-ARS, Akron, OH 80720, USA; merle.vigil@ars.usda.gov * Correspondence: greg.mcmaster@ars.usda.gov; Tel.: +1-970-492-7404


Introduction
Grain sorghum is an important dryland crop in semiarid southeastern Colorado and may be a valuable crop to add to traditional winter wheat-based (Triticum aestivum L.) crop rotations in semiarid northeastern Colorado.Interest in growing grain sorghum in northeastern Colorado is driven partly due to its high adaptability in semiarid regions and lower production costs compared to maize (Zea mays L.) [1,2].Grain sorghum is also considered more drought tolerant than maize and higher yielding in dry years in eastern Colorado [2,3].However, successfully growing grain sorghum in northeastern Colorado is thought to be limited by shorter growing seasons and cool night temperatures in the spring and fall.Shorter growing seasons can prevent sorghum from reaching physiological maturity before the first frost occurs in the fall (which kills sorghum), thereby negatively impacting grain yield and test weight if maturity has not occurred [4].Within northeastern Colorado, generally the growing season shortens and cooler night temperatures occur as latitude increases and with proximity to the foothills of the Rocky Mountains (i.e., longitude increases), which is related to higher elevations.The mean potential growing season of a location can be estimated by the mean temperature from 1 May to the mean first frost date (Table 1).The potential growing season of locations generally followed the same pattern as observed for the mean first frost date, with the Hort Farm (17.1 °C) having the lowest mean temperature and Sterling the highest (19.1 °C), slightly higher than Stratton (19.0 °C).The three westernmost locations (Hort Farm; Fort Collins, 17.5 °C; and ARDEC, 17.6 °C) had the three shortest potential growing seasons for all locations.

Mean Maturity Date of Locations
Simulations were run for each location changing the phenological parameters (dryland, GS; irrigated, GN), maturity class (early, medium, and late), planting date (1, 8, 15, 22, 29 May and 5, 12 June), and seedbed water conditions at planting (optimum, medium, dry, and planted in dust) to The mean potential growing season of a location can be estimated by the mean temperature from 1 May to the mean first frost date (Table 1).The potential growing season of locations generally followed the same pattern as observed for the mean first frost date, with the Hort Farm (17.1 • C) having the lowest mean temperature and Sterling the highest (19.1 • C), slightly higher than Stratton (19.0 • C).The three westernmost locations (Hort Farm; Fort Collins, 17.5 • C; and ARDEC, 17.6 • C) had the three shortest potential growing seasons for all locations.

Mean Maturity Date of Locations
Simulations were run for each location changing the phenological parameters (dryland, GS; irrigated, GN), maturity class (early, medium, and late), planting date (1,8,15,22,29 May and 5, 12 June), and seedbed water conditions at planting (optimum, medium, dry, and planted in dust) to estimate mean maturity date.The mean simulated maturity date differed among locations, phenological parameters, planting dates, maturity class, and seedbed water conditions (full data not shown).Consistent patterns were found for all locations, where maturity dates were earliest for GS parameters (primarily due to shorter grain filling duration), and increasingly later as: (1) maturity class changed from early to late (requiring more thermal time to reach maturity); (2) planting dates were delayed (delaying the beginning of accumulation of thermal time); and (3) soil moisture at planting decreased (delaying seedling emergence).Some of these patterns can be seen for simulations using the GS parameters for the Early variety planted into Optimum seedbed water conditions for the seven planting dates (Table 2).For instance, the maturity date is later as planting date is delayed for all locations.Maturity dates within a planting date were later as latitude, longitude, and elevation of the location increased.For example, for the 1 May planting date, Stratton had the earliest maturity date (31 August) and the Fort Collins (26 September), ARDEC (29 September), and Hort Farm (9 October) locations had the latest maturity dates.Given the relationship of maturity date, mean first frost date, and mean temperature from 1 May to the first frost date of locations with latitude, longitude, and elevation, a relationship between maturity date and both first frost date and mean temperature would be predicted.A highly significant negative relationship between maturity date and both first frost date and mean temperature was found for all planting dates, as illustrated when using the dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions (Figures 2  and 3).There appears to be a trend of increasingly negative slope as planting date was delayed, and it is likely that this is partly due to increased instances at locations with earlier first frost dates or lower temperatures where maturity was not reached by 31 December, and therefore set to 31 December in calculating the mean maturity date.Agronomy 2016, 6, 44 6 of 17

Probability of Reaching Maturity at Locations
Simulations to estimate the probability of reaching maturity before the first frost date were run for each location changing the phenological parameters (dryland, GS; irrigated, GN), maturity class (early, medium, and late), planting date (1,8,15,22,29 May and 5, 12 June), and seedbed water conditions at planting (optimum, medium, dry, and planted in dust).The probability of reaching maturity at a location for each combination of GN/GS, planting date, maturity class, and soil water at planting scenario was calculated by determining whether the simulated maturity date for each year was before, or after, the first frost date for the year.Initial analysis of these simulations examined the dryland (GS) and irrigated (GS) phenological parameters.The GS and GN phenological parameters resulted in similar probabilities of reaching maturity before the first frost, although GS parameters usually had a slightly higher probability than GN parameters, regardless of maturity class, planting date, or seedbed water conditions at planting (data only shown for simulations using Early maturity class and Optimum seedbed soil water conditions at planting, Table 3).Because sorghum is normally grown under dryland conditions, presentation of probabilities for reaching maturity at the nine locations will be shown using the GS parameters, maturity class, seedbed water conditions at planting, and planting date (Figures 3-11).All locations show similar patterns for the probability of reaching maturity based on maturity class (later maturity classes have lower probabilities), seedbed water conditions (lower soil water reduces probabilities), and planting dates (later planting dates lower the probabilities).The highest probabilities of reaching maturity at a location were for Early maturity class, Optimum seedbed water conditions, and the earliest planting date (1 May).The lowest probabilities were for Late maturity class, Planted in Dust seedbed water conditions, and the latest planting date (12 June).The Early and Medium maturity classes had relatively similar probabilities of reaching maturity for all locations except the Hort Farm; the Late Maturity class had the lowest probabilities for most sites, the exceptions tending to be the sites with the highest probabilities such as Stratton.Within the Early and Medium maturity classes, little differences in the probability of reaching maturity for a location were noted between Optimum and Medium seedbed soil water (except for the Hort Farm).

Discussion
As expected, a general relationship was observed, that as the latitude, longitude, and elevation of a location increased, the mean first frost date occurred earlier and the mean temperature from 1 May to the first frost date (i.e., the potential growing season) was lower.Further, highly significant negative relationships between the mean maturity date for a location and both the mean first frost date and mean temperature from 1 May to the first frost date were found.Therefore, it was not surprising that simulated probabilities of reaching maturity before the first frost date differed among locations, and could be grouped based on latitude, longitude, and elevation.The scenario using the dryland phenological parameters (GS), Early maturity class, 1 May planting date, and optimum seedbed soil water conditions at maturity, which had the highest probability of reaching maturity at all locations, illustrates this point.Five locations had probabilities over 90% of reaching maturity: Stratton (100%), Sterling (98.4%),Greeley LIRF (95.5%),Akron (94.8%) and Drake Farm (91.7%), and these locations had the combination of being at lower latitudes, longitudes, and elevations.Conversely, the three locations with the highest combination of latitude, longitude, and elevations had the lowest probabilities of reaching maturity (ARDEC, 76.2%; Fort Collins, 71.7%; and Hort Farm, 60.0%).Previous work of Sauer et al., 2014 [5] calculated simulated probabilities for emergence date of 15 May for two early maturity class hybrids for three of our locations and found comparable results: Stratton (91%), Akron (89%), and Fort Collins (75%).These simulation results suggest there are many areas in northeastern Colorado where sorghum can usually reach maturity before the first frost, if agronomic practices maximizing the likelihood of sorghum reaching maturity are used.
Selecting agronomic practices maximizing the probability of reaching maturity, such as choosing an early maturity class hybrid and practices that result in beginning the accumulation of thermal time

Discussion
As expected, a general relationship was observed, that as the latitude, longitude, and elevation of a location increased, the mean first frost date occurred earlier and the mean temperature from 1 May to the first frost date (i.e., the potential growing season) was lower.Further, highly significant negative relationships between the mean maturity date for a location and both the mean first frost date and mean temperature from 1 May to the first frost date were found.Therefore, it was not surprising that simulated probabilities of reaching maturity before the first frost date differed among locations, and could be grouped based on latitude, longitude, and elevation.The scenario using the dryland phenological parameters (GS), Early maturity class, 1 May planting date, and optimum seedbed soil water conditions at maturity, which had the highest probability of reaching maturity at all locations, illustrates this point.Five locations had probabilities over 90% of reaching maturity: Stratton (100%), Sterling (98.4%),Greeley LIRF (95.5%),Akron (94.8%) and Drake Farm (91.7%), and these locations had the combination of being at lower latitudes, longitudes, and elevations.Conversely, the three locations with the highest combination of latitude, longitude, and elevations had the lowest probabilities of reaching maturity (ARDEC, 76.2%; Fort Collins, 71.7%; and Hort Farm, 60.0%).Previous work of Sauer et al., 2014 [5] calculated simulated probabilities for emergence date of 15 May for two early maturity class hybrids for three of our locations and found comparable results: Stratton (91%), Akron (89%), and Fort Collins (75%).These simulation results suggest there are many areas in northeastern Colorado where sorghum can usually reach maturity before the first frost, if agronomic practices maximizing the likelihood of sorghum reaching maturity are used.
Selecting agronomic practices maximizing the probability of reaching maturity, such as choosing an early maturity class hybrid and practices that result in beginning the accumulation of thermal time as early as possible (e.g., earlier planting dates with optimum seedbed water conditions), may result in lower yield.Likely yield reductions can partly be attributed to early maturity hybrids almost always having lower yields than late maturing hybrids, less tillering, and shorter grain filling periods [9,15,16].Further, very early planting dates with cooler soil and air temperatures are not conducive to optimal sorghum emergence and early growth [17].Our simulation results indicate that usually selecting medium maturity classes and medium seedbed water conditions at planting, and often the first 2-3 planting dates, will have minimal reductions of the probability of reaching maturity.However, it would be beneficial to do additional field studies to verify these results, and also to try to quantify additional known agronomic practices (e.g., seeding rate, row spacing, nutrient availability) for inclusion in simulation models such as PhenologyMMS.Nevertheless, these simulation results provide producers with more suggestions for successfully growing sorghum in northeastern Colorado and improving yield potential.

Materials and Methods
The Phenology Modular Modeling System (PhenologyMMS V1.3), developed by the United States Department of Agriculture-Agricultural Research Service, was run using historical weather data for nine locations in northeastern Colorado to predict the date of physiological maturity and whether this occurred before the first frost in the fall.Two sets of phenological parameters estimating the thermal time (i.e., growing degree-days, GDD) between developmental stages were used in the simulations.One set provides GDD estimates for "optimum" conditions such as fully irrigated or very high rainfall, and denoted as GN (for GDD non-stressed).The other set is for extremely dry, but not lethal, conditions indicative of dryland conditions with below average rainfall in semiarid production regions, and denoted as GS (for GDD stressed).For each location, all combinations of seven planting dates (1,8,15,22,29 May and 5, 12 June), three maturity classes (early, medium, and late), and four general categories of soil water in the seedbed at planting (optimum, medium, dry, and planted in dust) were simulated for each year of historical weather data using both GN and GS parameter values.The probability of reaching maturity at a location for each combination of GN/GS, planting date, maturity class, and soil water at planting scenario was calculated by determining for each year whether the simulated maturity date was before, or after, the first frost date for the year.Additional details on PhenologyMMS, locations, and input files are provided below.

Locations and Weather Data
General details of the nine locations used in the study are listed in Table 1.Location names in Table 1 are shortened for the manuscript:

•
Akron is located at the USDA-ARS Central Great Plains Research Station, about 185 km southeast of Fort Collins.Additional location details provided in [18].

•
Ault is about 24 km east of Fort Collins.

•
Drake Farm is located on a producer's farm about 15 km east of Fort Collins.Additional location details provided in [20].

•
Fort Collins is located at the foothills of the Rocky Mountains.Three weather stations located within the city were used to get the entire historical weather records and fill in missing data.

•
Greeley LIRF is located at the USDA-ARS Limited Irrigation Research Farm immediately north of Greeley and about 45 km southeast of Fort Collins.The airport weather station immediately to the east of the site was used for the initial years, and in 2008 a weather station was installed at the research site and used for later years.Additional location details provided in [21].

•
Hort Farm is located at the Colorado State University Horticultural Farm on the northeast edge of Fort Collins.Additional location details provided in [22,23].

•
Sterling is located about 160 km east of Fort Collins.

•
Stratton is located about 330 km southeast of Fort Collins.
All weather data were either directly collected from previous experiments, CoAgMet (COlorado AGricultural Meteorological nETwork [24]), NOAA records, formerly the National Climatic Data Center (NCDC) [25], or Colorado Climate Center [26].Occasional missing daily maximum or minimum temperature data were estimated with several different techniques (e.g., mean of previous and subsequent day, using values from another nearby weather station, etc.), otherwise larger gaps of missing data resulted in not using the year in the simulations.The first frost event in a year was estimated when the daily minimum temperature was <−2 • C [2,27].

PhenologyMMS Decision Support Tool Overview And Default Parameters Used in Simulations
PhenologyMMS is a decision support tool with a Java graphical user interface and a FORTRAN 90/95 science simulation model for simulating the phenological responses of different crops to varying levels of water deficits.The objectives of PhenologyMMS are to (1) provide a relatively easy tool to producers, consultants, and scientists to predict crop developmental stages and provide information about crop phenology; and (2) develop seedling emergence and crop phenology science simulation components that could be inserted into other crop simulation models.Additional details on PhenologyMMS not covered below can be found in [19,[28][29][30], and PhenologyMMS can be downloaded at ARS Agricultural Software Download and Applications website [31].
Several different temperature response functions are used for calculating GDD in PhenologyMMS, and the method used for sorghum is calculated from [32]: where T max,i and T min,i are the daily maximum and minimum temperature for day i ( • C), respectively, and T base is the base temperature ( • C) and T upper ( • C) is the crop-specific upper temperature threshold above which additional GDD are not accumulated.T base and T upper are also used in the manipulation of the equation, where if T max,i and/or T min,i < T base , T max,i and/or T min,i = T base and if T max,i and/or T min,i > T upper , then T max,i and/or T min,i = T upper .Daily values greater than zero are summed over a period of n days.Grain sorghum T base and T upper were set to 10 • C and 40 • C, respectively.The crop-specific default thermal time required between successive stages is adjusted between GN (i.e., non-water stressed conditions such as fully irrigated) and GS (i.e., non-terminal water stressed conditions such as semiarid, dryland conditions) for varying levels of water deficits.GN and GS phenological parameters for early, medium, and late maturity classes are provided for each crop, and grain sorghum parameters are given in Table 4.The seedling emergence sub-model in PhenologyMMS is a simplified version of the SHOOTGRO model [33,34].Three factors control the time of seedling emergence: soil moisture near the seed, temperature, and planting depth.It is assumed that soil moisture controls the beginning of imbibition and germination (Germ): where the daily growing degree-days (GDDG i ), which are currently calculated using Equation ( 1), are summed from planting day (Pdate) until the required number of growing degree-days (GDDG req ) for germination is reached.Gday is the day that germination occurs.GDDG req is based on the soil moisture conditions of the seedbed zone.Table 5 presents the default sorghum values for GDDG req .Once germination has occurred, temperature drives the rate of shoot growth (ElongRate i ) from the seed leading to emergence.The thermal time required for emergence (GDDE req ) is calculated by: where ElongRate i is the shoot elongation rate (mm/ • C•day) for day i based on the soil water content (see Table 5 for default sorghum values) and Pdepth is the planting depth (cm).Seedling emergence (Emerge) is then determined by multiplying the daily elongation rate by the daily GDD (GDD i ) until the required GDD (GDDE req ) have been accumulated: Seedling emergence occurs the day (Eday) that Emerge equals GDDE rep .Crop-specific parameters for germination and elongation rate in Table 5 are based on four general categories of soil moisture in the seedbed layer: Optimum (>45% water-filled pore space), Medium (35%-45%), Dry (25%-35%), and Planted in Dust (<25%).These values do not need to be precisely estimated; rather, the user can choose the category based on general conditions.PhenologyMMS lacks a soil water balance module.Therefore, a surrogate approach was to use precipitation during this time period to vary the soil moisture conditions for simulating seedling emergence.In the original seedling emergence sub-model in PhenologyMMS, daily rainfall amounts from 5-7 mm incremented the soil moisture category to the next higher level of soil moisture.Rainfall events from 7-12 mm incremented the soil moisture category two levels.Preliminary evaluation of the seedling emergence sub-model uncovered instances of emergence occurring too early when selecting Medium or Dry soil water conditions.One solution tested was to create intermediate categories between Dry and Medium and Medium and Optimum levels.Germination and elongation rate values for the intermediate categories are the mid-points between the initial soil water values for Optimum, Medium, or Dry as appropriate, and this is done internally so the user is not required to provide additional input values.The first rainfall event ≥7 mm increases the initial soil water level to the intermediate level between Dry and Medium or between Medium and Optimum.The second rainfall event ≥7 mm increments it to the Medium or Optimum level.However, if the original condition was Planted in Dust, then the model was modified so that if rainfall is from ≥7 to 12 mm the soil moisture level is advanced to Dry.Rainfall from ≥12 to 20 mm results in Medium level, and if rainfall ≥20 mm, the soil water condition is Optimum. 1 Accumulated growing degree-days (GDD) required to initiate germination.Equation ( 1) is used for calculating GDD; 2 Seedbed water conditions are based on % water-filled pore space: optimum (>45%), medium (35%-45%), dry (25%-35%), and dust (<25%); 3 Soil moisture in this category is below the minimum threshold to initiate the imbibition process.

Conclusions
To avoid yield reductions due to a frost occurring prior to physiological maturity, selecting agronomic practices that maximize the probability of reaching maturity before the first frost.Our simulation results suggest choosing an early maturity class hybrid and practices that result in beginning the accumulation of thermal time as early as possible (e.g., earlier planting dates with optimum seedbed water conditions) result in the highest probability of reaching maturity before the first frost.However, planting as early as possible expands producer options of choosing hybrids in the medium maturity class and planting when less water is available in the seedbed without significantly reducing the probability of reaching maturity.However, it would be beneficial to do additional field studies to verify these results, and also to try to quantify additional known agronomic practices (e.g., seeding rate, row spacing, nutrient availability) for inclusion in simulation models such as PhenologyMMS.

Figure 1 .
Figure 1.Frequency distribution of first frost date in a year for nine locations.

Figure 1 .
Figure 1.Frequency distribution of first frost date in a year for nine locations.

Figure 2 .
Figure 2.Relationship between mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Figure 3 .
Figure 3. Relationship between mean temperature from 1 May to the mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Figure
Figure 2.Relationship between mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Figure 2 .
Figure 2.Relationship between mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Figure 3 .Figure 3 .
Figure 3. Relationship between mean temperature from 1 May to the mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Figure 4 .
Figure 4. Probability of sorghum reaching physiological maturity at Akron, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 5 .
Figure 5. Probability of sorghum reaching physiological maturity at ARDEC, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 4 .
Figure 4. Probability of sorghum reaching physiological maturity at Akron, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 4 .
Figure 4. Probability of sorghum reaching physiological maturity at Akron, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 5 .
Figure 5. Probability of sorghum reaching physiological maturity at ARDEC, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 5 .
Figure 5. Probability of sorghum reaching physiological maturity at ARDEC, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 6 .
Figure 6.Probability of sorghum reaching physiological maturity at Ault, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 7 .
Figure7.Probability of sorghum reaching physiological maturity at Drake Farm, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 6 . 17 Figure 6 .
Figure 6.Probability of sorghum reaching physiological maturity at Ault, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 7 .Figure 7 .
Figure7.Probability of sorghum reaching physiological maturity at Drake Farm, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 8 .
Figure 8. Probability of sorghum reaching physiological maturity at Fort Collins, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 9 .
Figure9.Probability of sorghum reaching physiological maturity at Greeley LIRF, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 8 . 17 Figure 8 .
Figure 8. Probability of sorghum reaching physiological maturity at Fort Collins, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 9 .
Figure9.Probability of sorghum reaching physiological maturity at Greeley LIRF, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 9 .
Figure9.Probability of sorghum reaching physiological maturity at Greeley LIRF, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 10 .
Figure10.Probability of sorghum reaching physiological maturity at Hort Farm, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 11 .
Figure 11.Probability of sorghum reaching physiological maturity at Sterling, Colorado for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 10 .
Figure10.Probability of sorghum reaching physiological maturity at Hort Farm, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 11 .
Figure 11.Probability of sorghum reaching physiological maturity at Sterling, Colorado for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 12 .
Figure12.Probability of sorghum reaching physiological maturity at Stratton, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Figure 12 .
Figure 12.Probability of sorghum reaching physiological maturity at Stratton, Colorado, for seven planting dates using non-stressed (GS) phenological parameters, three maturity classes, and four seedbed water conditions at planting.

Table 1 .
General weather information for nine locations in northeastern Colorado.

Table 2 .
Mean maturity dates for seven planting dates at nine locations in northeastern Colorado.
Each planting date used GS parameters, Early maturity class, and Optimum seedbed water.Values in Bold within parenthesis indicate mean maturity date is after the mean first frost date.1MeanDOY of maturity.If maturity was not predicted by the end of the year, maturity date was set to 31 December because the plant would be dead by 31 December, and due to lack of thermal time accumulation during the winter and early spring would result in simulated maturity many days later (e.g., DOY 100 or later).Using maturity dates in the following year would result in an earlier estimated mean maturity DOY than realistic for this table;2The first frost event in a year was estimated when the daily minimum temperature was <−2 • C.
2.Relationship between mean first frost date and mean maturity date of nine locations in northeastern Colorado for seven planting dates.Dryland (GS) phenological parameters, Early maturity class, and Optimum seedbed water conditions at planting were used for simulating maturity date.If maturity was not predicted by the end of the year, maturity date was set to 31 December.Linear regression lines are given with associated r 2 and probability of significance.

Table 3 .
Probability of reaching maturity for seven planting dates at nine locations in northeastern Colorado using either non-water stressed phenological parameters (GN) or water-stressed phenological parameters (GS).Each planting date used Early maturity class, and Optimum seedbed water conditions.Values in Bold within parenthesis indicate the probability of reaching maturity date is less than 80% (a general risk level unacceptable to producers based on personal communication).
[18,19] is the Colorado State University Agricultural, Research, Education, and Development Center about 7 km northeast of Fort Collins.Additional location details provided in[18,19].

Table 4 .
Sorghum growing degree-days (GDD) for no water stress (GN) and maximum non-terminal water stress (GS) phenological parameters for early, medium, and late maturity classes.

Table 5 .
Sorghum default parameters for seedling emergence.