2. ALMANAC model simulated processes

Phenological development defines the duration of various plant growth stages and determines the length of the forage growing season. The ALMANAC model simulates phenological development with a growing degree day (GDD) system, species-specific base temperature, and optimum growing temperature. The sum of GDD calculates the duration of the growing season. Anthesis date is predicted with a defined fraction of the total GDD sum to physiological maturity. Daylength and drought affect simulated forage phenology as described below.

Leaf area growth is simulated on a whole canopy basis, with potential leaf area index (LAI) defined for each species/ecotype/variety. These are hereafter referred to as just "species." The climate and soils at different sites often dictate the plant density of forages, thereby affecting the potential LAI.

The development of LAI over the growing season is simulated with a 0.0 to 1.0 "S" curve defined for each species. Thus, LAI is simulated as a function of the ratio (current summed GDD)/(GDD to maturity). This ratio typically approaches 1.0 as the plants approach anthesis and transition from forage production to reproduction. The "S<sup>00</sup> curve thus defines the potential leaf area growth over the growing season.

Daily dry matter accumulation is simulated using a radiation use efficiency (RUE) approach. The potential dry matter produced each day is a function of the amount of photosynthetically active radiation (PAR) intercepted by the leaf canopy on that day. The RUE is a species-specific value (g of dry matter per MJ of intercepted PAR).

Partitioning among plant parts is also on a whole canopy basis. The root and shoot partitioning is defined by two parameters. Plants initially partition a greater fraction of the total dry matter production into the roots. This fraction decreases as plants approach anthesis. Stresses, especially drought, reduce the above-ground dry matter production more than the root dry matter production. This causes drought stress to change the simulated root:shoot ratio.

Partitioning of plant energies to the seed is simulated with a harvest index (HI) approach. The fraction of the total plant weight in the seed at maturity relative to the total plant weight is the species-specific HI parameter. While very small for most forage species due to their relatively small fruits and seeds, the partitioning into the seed begins after anthesis and is complete by physiological maturity.

c. Plant competition for water and nutrients, including woody species competing with her-

Deriving plant parameters for a forage species and accommodating ecotypes: The steps for

Model testing against independent data: Soft calibration of the model via comparison of outputs to independent data to ensure the model is working reasonably well is described and

Knowledge gaps and areas for future improvement as a guide for additional research: Finally, the knowledge gaps and potential areas for improvement are outlined, as a guide for

Phenological development defines the duration of various plant growth stages and determines the length of the forage growing season. The ALMANAC model simulates phenological development with a growing degree day (GDD) system, species-specific base temperature, and optimum growing temperature. The sum of GDD calculates the duration of the growing season. Anthesis date is predicted with a defined fraction of the total GDD sum to physiological matu-

Leaf area growth is simulated on a whole canopy basis, with potential leaf area index (LAI) defined for each species/ecotype/variety. These are hereafter referred to as just "species." The climate and soils at different sites often dictate the plant density of forages, thereby affecting

The development of LAI over the growing season is simulated with a 0.0 to 1.0 "S" curve defined for each species. Thus, LAI is simulated as a function of the ratio (current summed GDD)/(GDD to maturity). This ratio typically approaches 1.0 as the plants approach anthesis and transition from forage production to reproduction. The "S<sup>00</sup> curve thus defines the poten-

Daily dry matter accumulation is simulated using a radiation use efficiency (RUE) approach. The potential dry matter produced each day is a function of the amount of photosynthetically active radiation (PAR) intercepted by the leaf canopy on that day. The RUE is a species-specific

Partitioning among plant parts is also on a whole canopy basis. The root and shoot partitioning is defined by two parameters. Plants initially partition a greater fraction of the total dry matter production into the roots. This fraction decreases as plants approach anthesis. Stresses, especially drought, reduce the above-ground dry matter production more than the root dry matter pro-

duction. This causes drought stress to change the simulated root:shoot ratio.

rity. Daylength and drought affect simulated forage phenology as described below.

Soils and weather data: The available and required soils and weather data are described.

deriving plant parameters for various forage species and ecotypes are outlined.

baceous forages

potential additional research.

the potential LAI.

2. ALMANAC model simulated processes

tial leaf area growth over the growing season.

value (g of dry matter per MJ of intercepted PAR).

discussed.

38 Forage Groups

Environmental stresses decrease leaf area expansion and dry matter accumulation. As described below, the model simulates the impacts of a variety of stresses each day. The most severe stress each day constrains leaf area growth and dry matter accumulation. Leaf area growth is more sensitive, especially to drought, than is dry matter growth.

Drought stress is simulated using the potential evapotranspiration (PET), calculated as a function of daily weather variables. The available soil water in the current rooting zone is calculated each day based on rainfall, soil infiltration, and soil water-holding capacity. If available soil water in the current rooting zone is insufficient to meet the plant's demand (based on PET and leaf area index), the model simulates a drought stress response in the plant through decreased leaf expansion rates and reduced dry matter accumulation rates.

Nutrient stresses, particularly nitrogen (N) and phosphorus (P) stresses, reduce plant growth. These nutrient stresses are simulated with a supply and demand approach. Plant N and P nutrient uptake is simulated with three input parameters that define how nutrient demand changes during the growing season. For each plant species, the optimum amount of available N and P is defined for each species early in plant development, near anthesis, and at physiological maturity. These three values are used to calculate the potential nutrient uptake from the soil each day. If the N and P in the current rooting zone are insufficient to meet demand (calculated from the optimum percentage of the nutrient and the potential daily plant dry matter growth), the model simulates nutrient stress by decreasing the species' dry matter accumulation rate and leaf expansion rate.

Temperature stress can also reduce plant growth in the model. Each plant species has a defined base temperature and an optimum temperature. When daily temperature is below the base temperature, cold temperature stress occurs. When temperatures are above the optimum, high temperature stress occurs.

Aeration stress is also simulated. When soils are saturated with water, aeration stress occurs in the model. Plants have variable sensitivity to aeration stress, as defined by the species-specific value of critical aeration factor (CAF). Plants such as eastern gamagrass (Tripsacum dactyloides L.) and rice (Oryza sativa) are less sensitive to poor aeration conditions, such as flooding, while upland grasses are more sensitive.

There are components in the model developed specifically for forage simulation. Forage development is not only dependent on GDD accumulation, but also on daylength and stresses. In order to accommodate growth dynamics typical of arid ecosystems where forage species are grown without irrigation, the model was modified so that sufficient drought stress stops GDD accumulation in the model. This is in addition to the direct effects on leaf area growth and dry matter increases as discussed above. As we began simulating plant growth in more arid environments, we had to introduce the ability to halt plant development when drought stress became sufficiently intense. Thus, we introduced a function that stops GDD accumulation (thus stopping phenological development), when the zero-to-one drought stress factor is less than 0.4.

The ALMANAC model is capable of simulating growth patterns exhibited by different types of forages. Cool season forages such as tall fescue often exhibit two intervals of active growth, with a slowdown during the hottest days of the year. Actual growth patterns of tall fescue in southwestern Missouri over 3 years [12, 13] are shown in Figure 1. We incorporated a midseason dormancy function to simulate this. Thus as daylength gets sufficiently long plant growth slows and stops. The model now simulates this rapid growth in the spring, slowing and stoppage of growth near mid-season, and subsequent late summer and early fall growth. The forage-simulation functionality of ALMANAC stops plant growth and development when the maximum photoperiod of the year is reached and restarts growth when the photoperiod subsequently gets sufficiently short to trigger reinitiation of growth. This model functionality was developed based on observed tall fescue growth curves measured in Missouri [12–14].

Subsequently, we tested the model's simulation of tall fescue yields with USDA-NRCS reported yields for a number of sites and soils across the main regions of tall fescue pastures in the U.S. [14]. We used long-term measured weather and the appropriate soil parameters for these simulations. We compared the simulated yields to the reported yields for the lowyielding sites, the high-yielding sites, and for all the sites pooled (Figure 2). The model with this function did an excellent job of simulating tall fescue yields on sites with differing reported yields across the main areas of tall fescue production in the U.S.

Additionally, ALMANAC simulations accommodate winter dormancy, typically observed in forage species when daylength gets sufficiently short in the fall. This capacity has been well tested on winter wheat [15], which is planted in the fall, goes dormant during the winter, then restarts growth in the spring. A parameter (DORMNT) defines this interval by defining the hours of photoperiod near the minimum for the latitude when plants are dormant. If the value is 1.0, during the winter when the photoperiod is within 1.0 hour of the minimum for the latitude, plants remain dormant.

Simulation of grazing and hay harvest is especially important when simulating forages. The model resets development (summed GDD, LAI, and height) when the simulation includes a grazing event or the forage is cut for hay. The model simulates a daily value for plant height from the fraction of GDD relative to the physiological maturity value and a species-specific plant height parameter (CHT). When forages are grazed or cut for hay, this height is reduced. If grazing or hay cutting reduces the plant height by 90%, the summed GDD for that day is reduced by 90% and the leaf area and above-ground dry matter is reduced by 90%. The plants then begin regrowth the following day.

Forage plant communities often have mixtures of species, due to the diversity typical of a native prairie, due to intercropping of legumes and grasses to better accommodate nutrient demands, or due to invasion of the forage site by undesirable herbaceous or woody plants. The ALMANAC model is capable of simulating both nitrogen fixation benefits to non–nitrogenfixing species and competition between plant species. The ALMANAC model was initially

Figure 1. Measured plant growth rates (kg ha<sup>1</sup> day<sup>1</sup>

growth rates within each year at α = 0.05. Source: adapted from Kiniry et al. [14].

) for "Kentucky 31" and "Bar Optima" tall fescue in 2011, 2012, and

Forage Yield Estimation with a Process-Based Simulation Model

http://dx.doi.org/10.5772/intechopen.79987

41

2013 at Mt. Vernon, MO. The Wilcoxon Rank Sum test was performed to compare "Bar Optima" and "Kentucky 31"

became sufficiently intense. Thus, we introduced a function that stops GDD accumulation (thus stopping phenological development), when the zero-to-one drought stress factor is less

The ALMANAC model is capable of simulating growth patterns exhibited by different types of forages. Cool season forages such as tall fescue often exhibit two intervals of active growth, with a slowdown during the hottest days of the year. Actual growth patterns of tall fescue in southwestern Missouri over 3 years [12, 13] are shown in Figure 1. We incorporated a midseason dormancy function to simulate this. Thus as daylength gets sufficiently long plant growth slows and stops. The model now simulates this rapid growth in the spring, slowing and stoppage of growth near mid-season, and subsequent late summer and early fall growth. The forage-simulation functionality of ALMANAC stops plant growth and development when the maximum photoperiod of the year is reached and restarts growth when the photoperiod subsequently gets sufficiently short to trigger reinitiation of growth. This model functionality was developed based on observed tall fescue growth curves measured in Missouri [12–14].

Subsequently, we tested the model's simulation of tall fescue yields with USDA-NRCS reported yields for a number of sites and soils across the main regions of tall fescue pastures in the U.S. [14]. We used long-term measured weather and the appropriate soil parameters for these simulations. We compared the simulated yields to the reported yields for the lowyielding sites, the high-yielding sites, and for all the sites pooled (Figure 2). The model with this function did an excellent job of simulating tall fescue yields on sites with differing reported

Additionally, ALMANAC simulations accommodate winter dormancy, typically observed in forage species when daylength gets sufficiently short in the fall. This capacity has been well tested on winter wheat [15], which is planted in the fall, goes dormant during the winter, then restarts growth in the spring. A parameter (DORMNT) defines this interval by defining the hours of photoperiod near the minimum for the latitude when plants are dormant. If the value is 1.0, during the winter when the photoperiod is within 1.0 hour of the minimum for the

Simulation of grazing and hay harvest is especially important when simulating forages. The model resets development (summed GDD, LAI, and height) when the simulation includes a grazing event or the forage is cut for hay. The model simulates a daily value for plant height from the fraction of GDD relative to the physiological maturity value and a species-specific plant height parameter (CHT). When forages are grazed or cut for hay, this height is reduced. If grazing or hay cutting reduces the plant height by 90%, the summed GDD for that day is reduced by 90% and the leaf area and above-ground dry matter is reduced by 90%. The plants

Forage plant communities often have mixtures of species, due to the diversity typical of a native prairie, due to intercropping of legumes and grasses to better accommodate nutrient demands, or due to invasion of the forage site by undesirable herbaceous or woody plants. The ALMANAC model is capable of simulating both nitrogen fixation benefits to non–nitrogenfixing species and competition between plant species. The ALMANAC model was initially

yields across the main areas of tall fescue production in the U.S.

latitude, plants remain dormant.

then begin regrowth the following day.

than 0.4.

40 Forage Groups

Figure 1. Measured plant growth rates (kg ha<sup>1</sup> day<sup>1</sup> ) for "Kentucky 31" and "Bar Optima" tall fescue in 2011, 2012, and 2013 at Mt. Vernon, MO. The Wilcoxon Rank Sum test was performed to compare "Bar Optima" and "Kentucky 31" growth rates within each year at α = 0.05. Source: adapted from Kiniry et al. [14].

developed to simulate competition between crops and weeds and has been applied to communities of plants such as native range sites and woody plants competing with forages. Aspects of competition simulated in ALMANAC include competition for light, water, and nutrients.

The light extinction coefficient (k) for Beer's law [16] is calculated for each harvest date as:

The value of k has been determined for a number of forages in the U.S. [17–20]. Realistic simulation of LAI is critical for these equations describing light interception. This is true for both the increase of LAI during active growth and the decline as leaves senesce. The model uses an S-curve to simulate the accumulation of leaf area increase as a function of GDD.

Similarly, as described above, biomass growth is simulated with a radiation use efficiency (RUE) approach [17, 21]. The RUE is calculated as the rate of increase in dry matter (g per m<sup>2</sup> ground area) per unit of intercepted photosynthetically active radiation (IPAR) (MJ per m<sup>2</sup> ground area). Regressions are fit with the treatment means of plant dry weight and summed IPAR for each sampling point. The RUE is the slope of the regression for this plant weight

This regression is ideally based on multiple harvest dates during the active growth period of the forage. Occasionally, when only two harvest dates are usable, RUE is calculated from differences. Only data from dates showing increases in dry matter (actively growing) are included. This constrains RUE values to periods of active growth. Data from sites experiencing drought stress are avoided. Values for FIPAR are calculated on a daily basis, with values for

Simulated light competition uses functions of [22], whereby the light interception of each plant species in the mixture is computed with the following formula: LAI\*k (k being the light extinction coefficient). These products (LAI\*k) of each species are summed and the sum used in Beer's law to compute the fraction of light interception by the whole plant community. This fraction of light intercepted for the whole plant community is then divided among the competing species by weighted fractions. The weights account for differences in species heights and LAI\*k of the species. Thus, taller species and those with higher LAI values and higher k values intercept a greater fraction of the total light intercepted by the plant community.

Water and nutrient competition are simulated with a balance sheet approach. Once intercepted light for each plant species is computed as described above, the potential daily biomass growth is calculated for each species with the total daily incident solar radiation, assuming 45% of that is PAR [23, 24]. The RUE for the species multiplied by the intercepted PAR is the potential biomass growth on any given day. Using the optimum nutrient concentrations for N and P at

).

FI <sup>¼</sup> <sup>1</sup>:<sup>0</sup> � exp �k<sup>∗</sup> ð Þ LAI (1)

Forage Yield Estimation with a Process-Based Simulation Model

http://dx.doi.org/10.5772/intechopen.79987

43

k ¼ ½ � logn ð Þ 1 FIPAR =LAI (2)

The fraction of incoming solar radiation intercepted by the leaf canopy (FI) is:

where logn = natural log of the number, and FIPAR = fraction of IPAR.

) as a function of the summed IPAR (MJ m�<sup>2</sup>

dates between measurement dates calculated by linear interpolation.

(g m�<sup>2</sup>

Figure 2. Reported (USDA-NRCS) and simulated tall fescue yields for (a) high-yielding soils, (b) low-yielding soils, and (c) high- and low-yielding soils at diverse sites in the U.S. Source: adapted from Kiniry et al. [14].

developed to simulate competition between crops and weeds and has been applied to communities of plants such as native range sites and woody plants competing with forages. Aspects of competition simulated in ALMANAC include competition for light, water, and nutrients.

The fraction of incoming solar radiation intercepted by the leaf canopy (FI) is:

$$\text{FI} = 1.0 - \exp\left(-\mathbf{k}^\* \mathbf{L} \mathbf{A} \mathbf{I}\right) \tag{1}$$

The light extinction coefficient (k) for Beer's law [16] is calculated for each harvest date as:

$$\mathbf{k} = [\log n \ (1 \text{ FIPAR})] / \text{LAI} \tag{2}$$

where logn = natural log of the number, and FIPAR = fraction of IPAR.

The value of k has been determined for a number of forages in the U.S. [17–20]. Realistic simulation of LAI is critical for these equations describing light interception. This is true for both the increase of LAI during active growth and the decline as leaves senesce. The model uses an S-curve to simulate the accumulation of leaf area increase as a function of GDD.

Similarly, as described above, biomass growth is simulated with a radiation use efficiency (RUE) approach [17, 21]. The RUE is calculated as the rate of increase in dry matter (g per m<sup>2</sup> ground area) per unit of intercepted photosynthetically active radiation (IPAR) (MJ per m<sup>2</sup> ground area). Regressions are fit with the treatment means of plant dry weight and summed IPAR for each sampling point. The RUE is the slope of the regression for this plant weight (g m�<sup>2</sup> ) as a function of the summed IPAR (MJ m�<sup>2</sup> ).

This regression is ideally based on multiple harvest dates during the active growth period of the forage. Occasionally, when only two harvest dates are usable, RUE is calculated from differences. Only data from dates showing increases in dry matter (actively growing) are included. This constrains RUE values to periods of active growth. Data from sites experiencing drought stress are avoided. Values for FIPAR are calculated on a daily basis, with values for dates between measurement dates calculated by linear interpolation.

Simulated light competition uses functions of [22], whereby the light interception of each plant species in the mixture is computed with the following formula: LAI\*k (k being the light extinction coefficient). These products (LAI\*k) of each species are summed and the sum used in Beer's law to compute the fraction of light interception by the whole plant community. This fraction of light intercepted for the whole plant community is then divided among the competing species by weighted fractions. The weights account for differences in species heights and LAI\*k of the species. Thus, taller species and those with higher LAI values and higher k values intercept a greater fraction of the total light intercepted by the plant community.

Water and nutrient competition are simulated with a balance sheet approach. Once intercepted light for each plant species is computed as described above, the potential daily biomass growth is calculated for each species with the total daily incident solar radiation, assuming 45% of that is PAR [23, 24]. The RUE for the species multiplied by the intercepted PAR is the potential biomass growth on any given day. Using the optimum nutrient concentrations for N and P at

Figure 2. Reported (USDA-NRCS) and simulated tall fescue yields for (a) high-yielding soils, (b) low-yielding soils, and

(c) high- and low-yielding soils at diverse sites in the U.S. Source: adapted from Kiniry et al. [14].

42 Forage Groups

the current growth stage, the demands for N and P are calculated. If insufficient N and/or P is present in the current rooting zone, the model reduces simulate growth rates to account for N and/or P stress. This simulation of the balance of nutrients is done for each species within a mixture. The model accounts for variability in root scavenging capacities between species only through differences in the current rooting depth of each species. Potential rooting depths of various plant species are derived from measurements reported in the literature for forages grown on soils with no restrictive soil layers (such as in [17]).

For each soil layer, the values for saturation, drained upper limit, and lower limit are used by the model. Soil organic matter is another input that impacts plant-available water and soil carbon balances in the model. The amount of runoff from rainfall events is calculated with the traditional runoff curve number system. The runoff is simulated with the slope and type of

Forage Yield Estimation with a Process-Based Simulation Model

http://dx.doi.org/10.5772/intechopen.79987

45

4. Deriving plant parameters for a forage species and accommodating

The group of readily derived plant parameters includes the potential leaf area index (LAI), the development curve for LAI over the growing season, the light extinction coefficient for Beer's law (k), the radiation use efficiency (RUE), the duration of the season in degree days, the harvest index for seeds (HI), and the N and P concentrations for each species over the growing season. All of these should be derived from measurements of a plant stand grown in a relatively stress-free environment to establish potential values for these for each forage species and ecotype. This means that ideally species being measured in field conditions should not

Details on taking field measurements for deriving plant parameters are outlined in detail under the headings: "Gathering Field Data, How to Use Ceptometer: AccuPAR LP-80 Basics Standard "[31] and "Taking measurements for ALMANAC: Sampling Protocol Standard with Photos" (https://www.ars.usda.gov/plains-area/temple-tx/grassland-soil-and-water-research-labo-

Field-derived values for the critical species-specific parameters have been described previously [17–21]. The model simulates light interception by the leaf canopy with Beer's law [16]

Figure 3. Intercepted photosynthetically active radiation (IPAR) measurements using an AccuPAR LP-80 Ceptometer at

ground cover.

ecotypes

4.1. Field plant species measurements

ratory/docs/193226/) (Figure 3).

Bishop, California, and Bryan, Texas.

have stresses due to drought or nutrient deficiency.

Likewise, potential plant transpiration is calculated from the potential evapotranspiration and the total community LAI. If soil water in the current rooting zone is insufficient to meet the species' demand, simulated drought stress occurs and limits growth. This occurs for all plant species present. However, it should be noted that a deeper rooted plant species may have access to soil water (and nutrients) not available to any competing shallower rooted species; ALMANAC accommodates different rooting depths of species. The deeper rooted plant species may have adequate soil water and nutrients to avoid drought and nutrient stresses when a shallower rooted species is stressed. The ALMANAC model does not currently simulate hydraulic lift dynamics and the potential impacts of lift on water and nutrient redistribution.
