March, 2016

Application of the GYGA approach for Ghana GYGA coordination team and Dr. S. Adjei-Nsiah

Contents 1. Description of agriculture, cropping systems, climate, and soils in Ghana (by Dr. S. AdjeiNsiah) ......................................................................................................................................... 3 2. Data sources and their use ...................................................................................................... 4 2.1. Harvested area and actual yields ...................................................................................... 4 2.2. Soil data ............................................................................................................................ 4 2.3. Weather data and reference weather stations ................................................................... 6 2.4. Crop and management information .................................................................................. 6 3. Crop growth simulations and model calibration .................................................................... 8 3.1. Used crop growth models ................................................................................................. 8 3.2. Data for model calibration ................................................................................................ 8 3.3. WOFOST calibration and simulation ............................................................................... 9 3.3.1. WOFOST crop parameter sets for growth simulations .............................................. 9 3.3.2. Initialization of available soil moisture for simulation with WOFOST ................... 10 3.4. HybridMaize calibration and simulation ........................................................................ 10 3.4.1. HybridMaize crop parameter sets for growth simulations ....................................... 10 3.4.2. Initialization of available soil moisture for simulation for HybridMaize ................. 11 3.5. Oryza2000 calibration and simulation............................................................................ 11 3.5.1. Oryza2000 crop parameter sets for growth simulations ........................................... 11 3.5.2. Initialization of available soil moisture for simulation for Oryza2000 .................... 12 4. Calculation of mean water limited yield level and yield gap per buffer zone...................... 13 5. References ............................................................................................................................ 13 Appendices ............................................................................................................................... 16 Appendix A Fraction of precipitation lost by surface runoff (based on literature review) ... 16 Appendix B WOFOST Crop data files for Ghana................................................................. 16 Millet medium duration ...................................................................................................... 16 Sorghum medium duration ................................................................................................. 17 Appendix C Part of the WOFOST rerun files with some of the selected soil-crop-weathersowing dates-year combinations for Ghana .......................................................................... 20 1

Millet .................................................................................................................................. 20 Sorghum.............................................................................................................................. 21 Appendix D ORYZA2000 Input data.................................................................................... 23 Crop data file ...................................................................................................................... 23 Rerun file ............................................................................................................................ 27 Appendix E Relationship between maize planting density and seasonal water deficit as observed in the US Corn Belt ................................................................................................ 28

2

1. Description of agriculture, cropping systems, climate, and soils in Ghana (by Dr. S. Adjei-Nsiah) Agriculture accounts for about 28% of Ghana’s Gross Domestic Product and employs more than half of the workforce, mainly small landholders. About 7.85 Mha is under cultivation. Out of this only 0.2% is under irrigation. Cocoa, oil palm, rubber and citrus constitute the major cash crops. The major starchy staples include maize, cassava, plantain, yam, cocoyam, rice, sorghum and millet (Table 1) and occupy a total land area of 3.40Mha, representing about 43% of total cultivated land. Table 1. Average (2006-2011) production, harvested area and yield of major food crops in Ghana Crop Cassava Yam Plantain Cocoyam Maize Sorghum Millet Rice

Yield (t ha-1) 14.01 16.35 10.6 6.55 1.68 1.12 1.13 2.30

Harvested Area (ha) 846,772 361,273 317,572 233,946 899,767 261,262 181,153 151,354

Total Production (t) 11,863,768 5,192,053 3,365,251 1,531,673 1,508,964 293,804 186,717 347,380

Source: SRID, MoFA

There are six agro-ecological zones in Ghana: Sudan Savannah, Guinea Savannah, Coastal Savannah, Forest/Savannah transitional zone, Deciduous Forest zone and the Rain Forest zone. Total annual rainfall ranges from 780 mm in the dry eastern coastal belt to 2,200 mm in the wet south-west corner of the country. The rainfall pattern is uni-modal in the Coastal, Sudan and Guinea Savannah zones, but bi-modal in the three remaining zones. Soils in the forest zone which developed over granites and phyllite, are mainly acrisols and lixisols and are deep, easily tilled and offer very little resistance to root growth. Soil total nitrogen is low, soils are acidic (pH 5.0-6.2) and their available P is low. Soils in the Guinea and Sudan savannah zones which are mainly Lixisols, Luvisols and Plinthosols, are shallow to moderately deep, are medium to light textured, and are developed over voltaian sandstones, granite, phyllite and schists. The soils are generally low in organic carbon and nitrogen. Nitrogen and P supply are less in the savannah zones than in the forest zones. A detailed description of soils in Ghana can be found in Brammer (1962) and Obeng (1975). Agriculture in Ghana is predominantly on smallholder basis with about 90% of the farmers cultivating less than one hectare. It is characterized by traditional methods of farming using hoe and cutlass. There is little mechanization, except in the forest/savannah transitional and the Guinea Savannah zones, where mechanization involving the use of tractor for land preparation is practiced. Bullock land preparation is also practiced in the Sudan savannah zone. Cereal crops (mainly maize, sorghum and millet) are produced in annual single-crop systems in the lower rainfall areas in the three northern regions. Maize is produced in annual singlecrop systems in the higher rainfall area in the southern forest zone and in annual double-crop 3

systems in the forest/savannah transitional zone. Typical double-crop systems in this zone include maize-maize, maize-cowpea and groundnut-maize. In the three northern regions, sorghum and millet are often intercropped with cowpea and/or maize and in the southern forest zone maize is often intercropped with one or more other crops such as cassava, cocoyam and plantain.

2. Data sources and their use Data that are used for the yield gap analyses for Ghana, are given in the following. More information about the applied GYGA approaches can be found at: http://www.yieldgap.org/web/guest/methods-overview

2.1. Harvested area and actual yields District-level data on annual actual yields were retrieved from the Ghana Ministry of Food and Agriculture (http://mofa.gov.gh/site/). We used all available actual yield data between 2005 till 2011 to calculate average actual yields per buffer zone (see the file with actual yields for more details). This has been done as follows: (a) determine the district(s) that best overlap with the reference weather station buffer; (b) calculate the average yield per buffer zone (via weighted averaging) based on the actual yields reported for the districts reported. Harvested areas were retrieved from the HarvestChoice SPAM crop distribution maps (You et al., 2006, 2009). 2.2. Soil data Soil data have been derived from the “AfSIS-GYGA functional soil information of SubSaharan Africa” database (RZ-PAWHC SSA v. 1.0, Leenars et al. 2015, see link). We have used effective root zone depth (ERZD, in cm), available water holding capacity of fine hearth (between field capacity – pF=2.3 – and permanent wilting point, in %v/v) and gravel content to created 28 soil classes which consist of 7 classes of available water holding capacity aggregated over ERZD (i.e., 4, 5, 6, 7, 8, 9 and 10 %v/v, adjusted by gravel content) and 4 rootable soil depth classes (i.e., 40, 75, 115 and 150 cm). These soil classes are described in Table 2, along with the soil texture that corresponds to the same plant available water holding capacity in the root zone in Hybrid Maize (as based on tropical pedo-transfer functions). We selected soil classes until achieving 50% area coverage of crop harvested area within reference weather station (RWS) buffer zones, with at least 3 dominant soil classes and at most 5 dominant soil classes. Then, water-limited yield potential was simulated for all selected soil classes and we discarded soil classes in which simulated water-limited yield potential is extremely low and highly variable, hence, unlikely to be used for long-term annual crop production. The mean water limited yield potential per buffer zone was then calculated by weighing (based on the area fractions per soil class) the simulated water limited yields for each of the selected soil classes.

4

Crop growth simulation have been done assuming the following soil and landscape characteristics: (a) no surface storage of water, (b) sufficient permeability of the soil to prevent soil saturation, (c) no ground water influence, (d) loss fraction of precipitation by surface runoff based on literature research as compiled in Appendix A (values based on the assumptions of optimal management and application of mulching) and depending on drainage class and slope angle; the drainage class per soil unit is derived from ISRIC-WISE (Batjes, 2012) data base and the slope angle is the mean angle per buffer zone (from HWSD slope map, see link) after clipping the zone by the crop harvested area mask , and (e) rooting depth is only limited by the soil in case that is indicated by ISRIC-WISE and/or the country agronomist. Table 2. Description of the 28 soil classes. Available water capacity of fine earth Effective (v%) aggregated over Soil class root zone ERZD, with FC = pF depth (cm) 2.3 with correction for gravel content 1 ≤4 40 2 5 40 3 6 40 4 7 40 5 8 40 6 9 40 7 ≥10 40 8 ≤4 75 9 5 75 10 6 75 11 7 75 12 8 75 13 9 75 14 ≥10 75 15 ≤4 115 16 5 115 17 6 115 18 7 115 19 8 115 20 9 115 21 ≥10 115 22 ≤4 150 23 5 150 24 6 150 25 7 150 26 8 150 27 9 150 28 ≥10 150

5

Corresponding texture class in Hybrid Maize (as based on tropical PTF results) Loamy sand Sandy clay loam Clay loam Sandy loam Silt loam Silt clay loam Loam Loamy sand Sandy clay loam Clay loam Sandy loam Silt loam Silt clay loam Loam Loamy sand Sandy clay loam Clay loam Sandy loam Silt loam Silt clay loam Loam Loamy sand Sandy clay loam Clay loam Sandy loam Silt loam Silt clay loam Loam

For rice, the approach was a bit different, because the ISRIC-WISE database lacked essential information on percolation rates, presence of plowpan, bunding and groundwater table depth. In rainfed lowland, there are often bunds that greatly restrain run-off. Groundwater levels are generally high and thus limiting water losses through drainage. In rainfed upland there are generally no bunds and soils are freely draining. Often due to the high water demand rice is found on the more clayey soil types, where drainage is less and water retention higher. Estimates of the fractions upland and lowland rainfed rice within each cropping system and zone were made by K. Saito, based on various sources including national experts and reports. One clayey soil type was assumed for all the upland rainfed rice systems and one clayey soil type was assumed for all the lowland rainfed. In the upland soil we assumed deep groundwater (1000 cm), no bunds, no plowsole and a high percolation rate when saturated. In the lowland we assumed shallow groundwater (20 cm), bunds (25 cm high), no plowsole and a low percolation rate when saturated. Exact parameter values can be found in appendix D in the reruns file. 2.3. Weather data and reference weather stations Historical daily weather data sets have been collected from the Ghana Meteorology Agency. Weather sets are available for 8 locations in Ghana and contain ten or more years of data. Linear interpolation or weather propagation have been used to fill missing weather data. For more information about the weather data per location, see the file weather_station_metadata.xls. Based on crop harvested area distribution and the climate zones defined for Ghana (Van Wart et al., 2013) per crop several reference weather stations (RWS) were selected (Table 3). Table 3. Selected weather stations and % coverage of total harvested area Crop

Rainfed maize

Rainfed sorghum

Rainfed millet Rainfed rice Irrigated rice

% coverage national harvested area (sum selected RWSs)

Selected RWS (#) Bolgatanga Kete-Krachie Koforidua Sefwi- Bekwai Sunyani Wa Yendi (7) Bolgatanga Wa Yendi (3) Bolgatanga Wa Yendi (3) Bolgatanga Yendi (2) Bolgatanga Yendi (2)

46%

78%

80% 41% 36%

2.4. Crop and management information Management practices for each RWS buffer zone were retrieved by the local country agronomist. Requested information included: dominant crop systems and their proportions of 6

the total harvested area, planting dates, dominant cultivar name and maturity, and actual and optimal plant population density. The crop and management information per zone is given in Table 4, 6 and 7.

Table 4. Crop and management information for maize, sorghum and millet in different RWS buffer zones of Ghana as compiled by the country agronomist (Source: Dr. S. Adjei-Nsiah). For rice, data compiled by the country agronomist were complemented with data from dr. K. Saito (AfricaRice) Water regime

% crop area under this system

Sowing window

Rainfed

33.3%

15-31 May

Rainfed

33.3%

1 -30 June

Rainfed

100%

1 June-15 July

Maize

Rainfed

37%

1 April-15 May

Double: (maize – maize)

Maize

Rainfed

37%

15-31 August

Single maize

Maize

Rainfed

63%

1-31 July

Double: (maize – maize)

Maize

Rainfed

50%

1 April-15 May

Koforidua

Double: (maize – maize)

Maize

Rainfed

50%

1 August-15 September

Koforidua SefwiBekwai SefwiBekwai Sunyani

Single maize

Maize

Rainfed

50%

15 April-15 May

Single maize

Maize

Rainfed

80%

1 April-15 May

Single maize

Maize

Rainfed

20%

1 August-15 September

Double: (maize – maize)

Maize

Rainfed

50%

1 April-15 May

Sunyani

Double: (maize – maize)

Maize

Rainfed

50%

1 August-15 September

Wa

Single: Maize

Maize

Rainfed

100%

1 June-15 July

Yendi

Single maize

Maize

Rainfed

100%

1 June-15 July

Wa

Single: Millet

Millet

Rainfed

100%

15 May-30 June

Yendi

Single: Millet

Millet

Rainfed

33.3%

15 June-15 July

Bolgatanga

Double: Rice, Dry Season

Rice

Irrigated

44%

15 January-15 February

Bolgatanga

Double: Rice, Wet Season

Rice

Irrigated

44%

15 July-15 August

Bolgatanga

Single: Rice, Dry Season

Rice

12%

?

Bolgatanga

Single: Rice, Wet Season

Rice

Irrigated Rainfed lowland

100%

15 June-15 July

Yendi

Double: Rice, Dry Season

Rice

Irrigated

33%

Yendi

Double: Rice, Wet Season

Rice

Irrigated

33%

Yendi

Single: Rice, Dry Season

Rice

Irrigated

34%

Yendi

Single, Wet Season

Rice

Bolgatanga

Single: Sorghum

Sorghum

Rainfed lowland Rainfed

Wa

Single: Sorghum

Sorghum

Yendi

Single sorghum

Sorghum

Weather station

Cropping system

Bolgatanga

Single: Millet

Bolgatanga

Single: Millet

Bolgatanga KeteKrachie KeteKrachie KeteKrachie Koforidua

Single: Maize

Early Millet Late Millet Maize

Double: (maize – maize)

Crop

7

15 January to 15 February 1 July- 7 August 15 January - 15February

100%

15 June – 15 July

100%

15 May-15 June

Rainfed

100%

15 May-15 June

Rainfed

100%

1 June-15 July

The sowing days used for the simulations are determined as the first day within the sowing window when the cumulative rainfall exceeded 20 mm (counting starts at the first day of the sowing window).

3. Crop growth simulations and model calibration 3.1. Used crop growth models The crop growth simulations for sorghum and millet in Ghana have been carried out with the crop growth simulation model WOFOST version 7.1.3 (release March 2011) (Supit et al., 1994, 2012; Wolf et al., 2011). Note that for the sowing dates, we have used either the actual date (e.g. June 15) if given, or otherwise the average date for the given period (e.g. given: June --> used date: June 15, (see Table 5). For maize the crop growth model HybridMaize version 2013.4.1 has been applied (Yang et al., 2006). For rice a subversion of the model ORYZA2000 version 2.13 (Bouman et al 2001) was specially developed in the GYGA project. The first modification in the subversion was the introduction of a new heat sterility module (van Oort et al., 2014). The reason for implementing this module was that the default model dramatically overestimated sterility in the semi-arid regions in the Sahel, where much irrigated rice is grown. The second modification was a simplified simulation of phenology, namely a fixed duration was applied for each year (but different per site and growing season). Such a simplification was necessary due to lack of data for proper phenology calibration (van Oort et al 2011). 3.2. Data for model calibration Based on experimental information reported in the literature, we have compiled data for the main crop characteristics for maize, sorghum and millet growing in Ghana (Table 5). For rice, no such data could be compiled. These characteristics can be considered representative for optimal (i.e. no water and no nutrient limitation) growing conditions in the different zones of Ghana. These crop characteristics have been used for testing and possibly calibrating the model parameters. Table 5. Crop characteristics for main crop types in Ghana to test and calibrate the crop model parameters, being representative for a high-yield variety growing under optimal conditions with respect to water and nutrient supply and optimal management1 Crop, Zones Period from in Ghana emergence to maturity (days) Grain maize, 105 - 135 all zones Sorghum, all 135 - 145 zones Millet, all 85 - 145 zones

Period frac-tions LAI-max from emergence to (m2 m-2) flowering and from flowering to maturity (%) 50% - 50% 4 to 7 55% - 45%

3 to 7

Total biomass above-ground (kg dry matter per ha) 10000 to 16000 9000 to 14000

3600 to 5600

0.35 to 0.45

62% - 38%

3 to 7

9000 to 14000 2700 to 4200

0.25 to 0.35

1

Yield (kg dry matter per ha)

5000 to 8000

Harvest index (Yield / Total biomass above ground) 0.45 to 0.55

Crop characteristics are based on crop data for Ghana in Table 4, expert knowledge and experimental information, as reported by Kpongor, 2007; Morris et al., 1999; Obeng-Bio, 2010; Quansah, 2010; Adjei-Nsiah, 2012; DTMaize bulletin, 2013 and MacCarthy et al., 2009.

8

3.3. WOFOST calibration and simulation The temperature sums required for phenological development per crop type in WOFOST are calibrated on the basis of the observed crop calendars (see Tables 3 and 5) and the climate conditions per RWS buffer zone in Ghana. We may assume that sorghum and millet are in general produced in Ghana without application of irrigation water. However, to simplify the calibration of the model parameters related to crop growth and phenological development, we have done the crop model calibration for optimal conditions (see crop characteristics in Table 5). This means that water supply and nutrient supply are optimal to attain high yield levels and that crop protection and other management activities are all optimally performed. 3.3.1. WOFOST crop parameter sets for growth simulations We used for the simulations with WOFOST the standard crop parameter sets as compiled by Van Heemst (1988). These parameter sets were later slightly adapted for western Africa. The new crop parameter sets are given in the files SORG-med-Gha-GYGA.CAB and MILL-medGha-GYGA.CAB for respectively, sorghum and millet (Appendix B). In the indicated files the following parameters are adapted for the GYGA-simulations: (a) temperature sums (TSUM1 and TSUM2) required for the modelled phenological development from crop emergence until flowering and from flowering to maturity, as calibrated for the climate conditions and the crop data per RWS buffer zone in Table 4; the derived and applied TSUM1 and TSUM2 values for the different zones are given in Table 6; (b) maximal rooting depth, which is set at 100 cm for millet and sorghum; (c) life span of leaves growing at 35°C (SPAN) for sorghum has been increased to 42, whereas SPAN for millet has kept the same and similar value (=42); (d) correction factor for evapo-transpiration (CFET) has been increased from 1.0 to 1.1 for both sorghum and millet (see Appendix B); (e) for millet the maximum leaf CO2 assimilation rate, which is dependent on development stage, has been decreased from 85 kg ha-1 hr-1 to 70 kg ha-1 hr-1 for the first development stages. Table 6. Temperature sums (TSUM1 and TSUM2) required for the modelled crop phenological development from crop emergence until flowering and from flowering to maturity as calibrated for the climate conditions and the crop data in Table 4 for sorghum and millet in the different RWS buffer zones of Ghana Weather station Bolgatanga Bolgatanga Wa Yendi Bolgatanga Wa

Cropping system

Cropping cycle

Single: Millet Single: Millet Single: Millet Single: Millet Single: Sorghum Single: Sorghum

Early Millet Late Millet Millet Millet Sorghum Sorghum

Water regime Rainfed Rainfed Rainfed Rainfed Rainfed Rainfed

Growth TSUM1 TSUM2 duration1 (0Cd) (0Cd) 150 1070 950 90 740 660 150 1070 950 150 1070 950 150 1240 1230 140 1130 1120

Yendi Single sorghum Sorghum Rainfed 120 1000 880 Growth duration is assumed to be the period from sowing to maturity, of which the period from sowing to crop emergence is estimated at about 10 days. The actual sowing date is determined as the first day within the sowing 1

9

window (for calibration of the sowing window is Table 4 used) when the cumulative rainfall exceeded 20 mm (counting starts at the first day of the sowing window)

3.3.2. Initialization of available soil moisture for simulation with WOFOST For single cropping systems (i.e. one crop grown per year) and for the first crop in double cropping systems the simulation of the soil water balance has been started 90 days before the sowing date and thus generally in the dry season. At this start of the simulation the total amount of available soil moisture is set at 3.3 cm (i.e. soil moisture content being one third of the water holding capacity between field capacity and wilting point, thus: 0.33 × (SMFCSMWP)). For double cropping systems the simulation of the soil water balance for the second crop has been started at the sowing date, and thus in the wet season and after the maturity of the previous crop. At this start of the simulation the total amount of available soil moisture is set at 6.7 cm (i.e. soil moisture content set at roughly 0.7 × (SMFC –SMWP)).

3.4. HybridMaize calibration and simulation 3.4.1. HybridMaize crop parameter sets for growth simulations In HybridMaize, a single genotype specific parameter is required for simulations (Yang et al., 2006). Phenological development stages in HybridMaize progress according to the accumulation of growing degree days (GDD), calculated as daily mean temperature minus a base temperature (10 °C for maize). Total GDD (from sowing to maturity) was calculated for each site based on the site-specific crop growth duration and the actual temperature records. Through optimization, total GDD were estimated for each site so that the long-term average simulated date of maturity matches the one reported by the country agronomist. Date of silking is calculated internally in HybridMaize based on relationships between GDD to silking versus total GDD. If the calculated total GDD of a particular site was greater than ±25% of other nearby sites, this site was further investigated to determine if any errors persisted in the weather data used to calculate GDD or country agronomist partners were contacted to determine if a misspecification of cropping season had occurred. If total GDD calculations of nearby sites were within 5% of each other, a fixed GDD (roughly the average of these GDD) for all sites within the region was used. To avoid unrealistically long crop cycle lengths, the maximum GDD was set at 1900°Cd based on maximum GDD values reported in the literature. Since our objective was to estimate attainable maize production using best available technology, we simulated yields of modern hybrid cultivars while the optimal plant population density for each location was determined based on the relationship between plant population and seasonal water deficit developed for US maize (Grassini et al. 2009) (Figure H1, Appendix H), with maximum set at 80,000 plant ha-1 (average plant density of irrigated corn in Nebraska) and a minimum set at 35,000 plant ha-1. The rationale was that the observed plant population density gradient along the east-west water deficit gradient in the US Corn Belt is a very good proxy for optimal planting density, for a given water deficit level, in a real 10

farm context where crop producers do have good access to markets, inputs, and extension services. Seasonal water deficit was estimated as the difference between total precipitation and total evaporative demand (i.e., reference grass-based evapotranspiration) between sowing and physiological maturity at each location (RWS). Table 7. Input parameters used in the HybridMaize model for simulation of rainfed maize in Ghana. Croppin g system

Water regime

% crop under this area

Bolgatanga

Single

Rainfed

100%

Sowing date used for model calibration and simulations 6/15

KeteKrachie KeteKrachie KeteKrachie Koforidua

Double

Rainfed

19%

4/1

120

1900

25

76

Double

Rainfed

19%

7/31

120

1900

50

80

Single

Rainfed

63%

7/15

120

1900

25

80

Double

Rainfed

50%

4/15

120

1900

25

80

Koforidua

Double

Rainfed

50%

8/15

120

1900

25

65

Kusi

double

Rainfed

50%

4/15

120

1900

25

40

Kusi

double

Rainfed

50%

9/1

120

1900

25

40

Kusi

single

Rainfed

100%

4/15

120

1900

25

40

SefwiBekwai SefwiBekwai Sunyani

Single

Rainfed

20%

8/15

120

1900

50

76

Single

Rainfed

80%

4/15

120

1900

50

80

Double

Rainfed

50%

5/1

120

1900

25

78

Sunyani

Double

Rainfed

50%

9/1

120

1900

50

66

Wa

Single

Rainfed

100%

6/15

120

1900

75

80

Yendi

Single

Rainfed

100%

6/25

120

1900

50

80

Weather station

Growth duration (days till maturity)

GDD (°C, base temp = 10°)

120

1900

Soil available water content at planting (mm) 50

Plant population (1,000 ha-1) 80

3.4.2. Initialization of available soil moisture for simulation for HybridMaize To account for variation in soil moisture at planting, the HybridMaize model was used to calculate available soil water available at planting for each system. This calculation was then used to inform decisions on available soil water at planting (as a percent of available water holding capacity of reported soils for each site). 3.5. Oryza2000 calibration and simulation 3.5.1. Oryza2000 crop parameter sets for growth simulations We used for rice the standard crop parameter sets as compiled by Bouman et al. (2001) for variety IR72. Two sets of parameters were modified, namely for phenology and for sterility. For phenology the crop duration was assumed to be the same in each year (but different per zone and cropping system, Table 4). Within the growing season, a fixed duration of 30 days was assumed for the period from flowering to maturity (Vergara and Chang (1981), Van Oort 11

et al 2011). Within the pre- and post-flowering phases a constant development rate was assumed (but different per zone and cropping system). A sensitivity analysis showed that this simplification had very limited effect on simulated long term average Yp and Yw when compared with a temperature sum based approach, as errors of overestimation of duration in one year cancelled out against errors of underestimation in other years. The second modification was to simulate heat induced sterility based on concepts described in van Oort et al (2014). We present the new equations plus their new parameters below. Rice is only vulnerable to heat sterility at the time when the anthers are out in the open. The time of the day in which rice is flowering was simulated as: FLHAS = 12.7 - 0.348*TMIN7 FLTIME = SUNRIS + FLHAS

Where TMIN7 is the average of the minimum temperatures in the previous 7 days and FLHAS is the time of flowering in hours after sunrise and SUNRIS is the time of sunrise. Air temperature at flowering time as calculated from minimum temperature, maximum temperature, daylength and time of sunrise as: TAIRFL = TMIN + (TMAX-TMIN)*SIN(PI*(FLTIME-SUNRIS)/(DAYL+3))

Saturated vapour pressure at the time of flowering (kPa) and vapour pressure deficit at flowering (kPa), with early morning vapour pressure (VP) as an input weather variable were calculated as: VPFL = 0.1*6.10588*EXP(17.32491* TAIRFL/(TAIRFL +238.102)) VPDFL = VPSFL – VP

With these, panicle temperature corrected for transpirational cooling was simulated as: TPANFL = TAIRFL - (1.29*VPDFL+0.01)

And then finally spikelet fertility (=1-sterility) was simulated as: SFHEAT

= EXP(14.3-0.408*TPANFL)/(1+EXP(14.3-0.408*TPANFL))

3.5.2. Initialization of available soil moisture for simulation for Oryza2000 Due to the great sensitivity of rice to water stress, we assumed farmers would only sow rice in a saturated soil profile. Therefore we assumed the soil profile was saturated at the start of every simulation.

12

4. Calculation of mean water limited yield level and yield gap per buffer zone Crop growth simulations for the different RWS-soil type-crop type-sowing date combinations (see some combinations in the rerun files with resp. CLFILE-SOFILE-CRFILE-IDSOWISYR (=year) given in Appendix C) in Ghana have been done for both potential (=irrigated) and water limited (=rainfed) conditions, to indicate the degree that yield levels may increase by application of irrigation water. Crop production systems in Ghana are mainly rainfed. Hence, only the water limited yields (Yw) have been used to calculate the yield gap. The mean Yw values per crop type per RWS buffer zone were calculated from the Yw values simulated for each crop type-sowing datecrop system-soil type combination per zone, weighted to their relative areas. Next, the yield gap per RWS buffer zone is calculated as the difference between the mean Yw value per zone and the mean actual yield per zone. Note that the time period of the actual yields and that of the Yw values is partly different (i.e. mean of actual yields based on yields from 2005 up to and including 2011 and mean of simulated yields based on simulations for the available weather data between mainly 1998 and 2012).

5. References Adjei-Nsiah, S., 2012. Response of maize (Zea mays L.) to different rates of palm bunch ash application in the semi-deciduous forest agro-ecological zone of Ghana. Applied and Environ-mental Soil Science, Volume 2012, Article ID 870948, 5 pages, doi:10.1155/2012/870948. Batjes, N.H., 2012. ISRIC-WISE derived soil properties on a 5 by 5 arc minutes grid. ISRIC report no. 2012-1, ISRIC, Wageningen, The Netherlands. Bouman, B.A.M., M.J. Kropff, T.P. Tuong, M.C.S. Wopereis, H.F.M. ten Berge, and H.H. Van Laar. 2001. ORYZA2000: Modeling Lowland Rice. International Rice Research Institute, Los Baños, Philippines and Wageningen University and Research Centre, Wageningen, The Netherlands. Brammer, H. (1962). Soils of Ghana. Pp. 88-126. In Brian Wills (Ed.). Agriculture and Land Use in Ghana. London, Oxford University Press. DTMaize bulletin, 2013. A Quarterly Bulletin of the Drought Tolerant Maize for Africa (DTMA) Project. Volume 2, no. 1, March 2013, CIMMYT, Narobi, Kenya Kpongor, 2007 (PhD thesis). Spatially explicit modeling of sorghum (Sorghum bicolor (L.) Moench) production on complex terrain of a semi-arid region in Ghana using APSIM J.G.B. Leenaars, T. Hengl, M. Ruiperez González, J.S. Mendes de Jesus, G.B.M. Heuvelink, J. Wolf, L.G.J. van Bussel, L. Claessens, H. Yang, K.G. Cassman, 2015. Root Zone Plant-Available Water Holding Capacity of Sub-Saharan Africa soils, version 1.0. Gridded functional soil information (dataset RZ-PAWHC SSA v. 1.0).

13

MacCarthy, D.S., Sommer, R., Vlek, P.L.G., 2009. Modeling the impacts of contrasting nutrient and residue management practices on grain yield of sorghum (Sorghum bicolor (L.) Moench) in a semiarid region of Ghana using APSIM. Field Crops Research 113, 105–115. Ministry of Food and Agriculture (2012). Agriculture in Ghana, Facts and Figures. SRID, MoFA, Accra. Morris, M.L., R. Tripp, and A.A. Dankyi. 1999. Adoption and Impacts of Improved Maize Production Technology: A Case Study of the Ghana Grains Development Project. Economics Program Paper 9901, CIMMYT, Mexico, D.F. Obeng, H.B. 1975. Soils of the Savanna Zones of Ghana – Their physico -chemical characteristic classification and management, pp. 11-23. In: H.B. Obeng and P.K. Kwakye (Eds.). Savanna soils of the sub-humid and semi-humid and semi-arid regions of Africa and their management ISSS Comm. I, IV, V and VI – State Publishing Corporation Accra. Ghana. Obeng-Bio, E., 2010. Selection and ranking of local and exotic maize (Zea mays L.) genotypes to drought stress in Ghana. Msc.-thesis, Kumasi, Ghana. Quansah, G.W., 2010; Effect of organic and inorganic fertilizers and their combinations on the growth and yield of maize in the semi-deciduous forest zone of Ghana. MSc.-thesis, Kumasi, Ghana. Supit, I., Hooijer, A.A., Van Diepen, C.A. (Eds.), 1994. System description of the WOFOST 6.0 crop simulation model implemented in CGMS. European Communities (EUR15956EN), Luxembourg. Supit, I., Van Diepen, C.A., De Wit, A.J.W., Wolf, J., Kabat, P., Baruth, B., Ludwig F., 2012. Assessing climate change effects on European crop yields using the Crop Growth Monitoring System and a weather generator. Agric. Forest Meteorol. 164, 96-111. Van Heemst, H., 1988. Plant data values required for simple and universal simulation models: review and bibliography. Simulation reports CABO-TT, Wageningen, The Netherlands. van Oort, P.A.J., Saito, K., Swart, S.J., Shrestha, S., 2014. A simple model for simulating rice heat sterility as a function of flowering time and transpirational cooling. Field Crops Research 156: pp. 303-312 van Oort, P.A.J., Zhang, T., de Vries, M.E., Heinemann, A.B., Meinke, H. 2011. Correlation between temperature and phenology prediction error in rice (Oryza sativa L.). Agricultural and Forest Meteorology 151(12): 1545-1555. Vergara, B.S., Chang, T.T., 1985. The flowering response of the rice plant to photoperiod. In: A Review of Literature, 4th ed. The International Rice Research Institute, Manila, Philippines, http://books.irri.org/9711041510content.pdf. Wolf, J., Hessel, R., Boogaard, H.L., De Wit, A., Akkermans, W., Van Diepen, C.A., 2011. Modeling winter wheat production over Europe with WOFOST – the effect of two new zonations and two newly calibrated model parameter sets. In: Ahuja, L.R., Ma, L. (Eds.), Methods of Introducing System Models into Agricultural Research. Advances in Agricultural Systems Modeling 2: Trans-disciplinary Research, Synthesis, and Applications, ASA-SSSA-CSSA book series, 297-326.

14

Yang, H.S., A. Dobermann, K.G. Cassman, and D.T. Walters. 2006. Features, applications and limitations of the Hybrid-Maize simulation model. Agron. J. 98:737-748

15

Appendices Appendix A Fraction of precipitation lost by surface runoff (based on literature review) Table 8. Surface runoff fraction of total seasonal precipitation (in %) for soils that are cultivated with cereals and are mulched, used for simulations with WOFOST Drainage class, Slope angle, in % 0-2 2-6 6-10 >10

Very poor

Insufficient

Moderate

Well drained

10 13.3 16.7 20

6.7 10 13.3 16.7

3.3 6.7 10 13.3

0 3.3 6.7 10

Appendix B WOFOST Crop data files for Ghana Millet medium duration ** ** ** ** ** ** ** ** **

File MILL-med-Gha-GYGA.CAB CROP DATA FILE for use with WOFOST Version 7.0 Reference: Heemst, H.van, 1988. Plant data values required for simple and universal simulation models: review and bibliography. Simulation reports CABO-TT. Some changes included for Millet (medium duration) for Ghana for global yield gap atlas

CRPNAM='Pearl Millet, medium duration, Ghana, Global yield gap atlas' ** emergence TBASEM = 12.0 TEFFMX = 32.0 TSUMEM = 60. ** phenology IDSL = 0 DLO DLC TSUM1 TSUM2 DTSMTB

= 1.0 = 0.0 = 1030. = 650. = 0.00, 10.00, 27.00, 35.00, 45.00,

DVSI = 0. DVSEND =

2.00

! lower threshold temp. for emergence [cel] ! max. eff. temp. for emergence [cel] ! temperature sum from sowing to emergence [cel d] ! ! ! !

indicates whether pre-anthesis development depends on temp. (=0), daylength (=1) , or both (=2) optimum daylength for development [hr] critical daylength (lower threshold) [hr] ! temperature sum from emergence to anthesis [cel d] ! temperature sum from anthesis to maturity [cel d] 0.00, ! daily increase in temp. sum 0.00, ! as function of av. temp. [cel; cel d] 17.00, 17.00, 00.00 ! initial DVS ! development stage at harvest (= 2.0 at maturity [-])

** initial TDWI = 3.00 ! initial total crop dry weight [kg ha-1] ** ! Not used as imput by WOF6_0 model LAIEM = 0.006912 ! leaf area index at emergence [ha ha-1] RGRLAI = 0.0500 ! maximum relative increase in LAI [ha ha-1 d-1] ** green area SLATB = 0.00, 0.40, 0.85, 2.00, SPA = 0.000 SSATB = 0.0, 0.0, 2.0, 0.0 SPAN = 42. TBASE = 10.0

! ! ! ! !

0.0018, ! specific leaf area 0.0027, ! as a function of DVS [-; ha kg-1] 0.0019, 0.0019 specific pod area [ha kg-1] specific stem area [ha kg-1] as function of DVS life span of leaves growing at 35 Celsius [d] lower threshold temp. for ageing of leaves [cel]

** assimilation KDIFTB = 0.0, 0.50, 2.0, 0.50 EFFTB = 0.0, 0.50, 40.0, 0.50

! ! ! !

extinction coefficient for diffuse visible light [-] as function of DVS light-use effic. single leaf [kg ha-1 hr-1 j-1 m2 s] as function of daily mean temp.

16

Extremely well drained 0 0 3.3 6.7

AMAXTB

=

TMPFTB

=

TMNFTB

=

0.00, 1.30, 2.00, 0.00, 8.00, 20.00, 35.00, 45.00, 5.00, 12.00,

70.00, 70.00, 0.00 0.00, 0.00, 1.00, 1.00, 0.00 0.00, 1.00

! max. leaf CO2 assim. rate ! function of DVS [-; kg ha-1 hr-1] ! reduction factor of AMAX ! as function of av. temp. [cel; -]

! red. factor of gross assim. rate ! as function of low min. temp. [cel; -]

** conversion of assimilates into biomass CVL = 0.720 ! efficiency of conversion CVO = 0.730 ! efficiency of conversion CVR = 0.720 ! efficiency of conversion CVS = 0.690 ! efficiency of conversion

into into into into

leaves [kg kg-1] storage org. [kg kg-1] roots [kg kg-1] stems [kg kg-1]

** maintenance respiration Q10 = 2.0 ! rel. incr. in resp. rate per 10 Cel temp. incr. [-] RML = 0.0300 ! rel. maint. resp. rate leaves [kg CH2O kg-1 d-1] RMO = 0.0100 ! rel. maint. resp. rate stor.org. [kg CH2O kg-1 d-1] RMR = 0.0100 ! rel. maint. resp. rate roots [kg CH2O kg-1 d-1] RMS = 0.0150 ! rel. maint. resp. rate stems [kg CH2O kg-1 d-1] RFSETB = 0.00, 1.00, ! red. factor for senescence 2.00, 1.00 ! as function of DVS [-; -] ** partitioning FRTB = 0.00, 0.10, 0.25, 0.40, 1.00, 2.00, FLTB = 0.00, 0.10, 1.00, 2.00, FSTB = 0.00, 0.10, 1.00, 1.60, 2.00, FOTB = 0.00, 1.00, 1.60, 2.00,

0.50, 0.50, 0.30, 0.17, 0.00 0.00 1.00, 1.00, 0.00, 0.00 0.00, 0.00, 1.00, 0.00, 0.00 0.00, 0.00, 1.00, 1.00

! fraction of total dry matter to roots ! as a function of DVS [-; kg kg-1]

** death rates PERDL = 0.030 ! RDRRTB = 0.00, 1.50, 1.5001, 2.00, RDRSTB = 0.00, 1.50, 1.5001, 2.00,

max. rel. 0.000, 0.000, 0.020, 0.020 0.000, 0.000, 0.020, 0.020

death rate of leaves due to water stress ! rel. death rate of roots ! as a function of DVS [-; kg kg-1 d-1]

** water CFET DEPNR IAIRDU

use = 1.10 = 4.5 = 0

** rooting RDI = 10. RRI = 4.0 RDMCR = 150.

! fraction of above-gr. DM to leaves ! as a function of DVS [-; kg kg-1] ! fraction of above-gr. DM to stems ! as a function of DVS [-; kg kg-1]

! fraction of above-gr. DM to stor. org. ! as a function of DVS [-; kg kg-1]

! rel. death rate of stems ! as a function of DVS [-; kg kg-1 d-1]

! correction factor transpiration rate [-] ! crop group number for soil water depletion [-] ! air ducts in roots present (=1) or not (=0) ! initial rooting depth [cm] ! maximum daily increase in rooting depth [cm d-1] ! maximum rooting depth [cm]

** nutrients ** maximum and minimum concentrations of N, P, and K ** in storage organs in vegetative organs [kg kg-1] NMINSO = 0.0100 ; NMINVE = 0.0032 NMAXSO = 0.0300 ; NMAXVE = 0.0105 PMINSO = 0.0014 ; PMINVE = 0.0005 PMAXSO = 0.0080 ; PMAXVE = 0.0025 KMINSO = 0.0030 ; KMINVE = 0.0070 KMAXSO = 0.0080 ; KMAXVE = 0.0280 YZERO = 200. ! max. amount veg. organs at zero yield [kg ha-1] NFIX = 0.00 ! fraction of N-uptake from biol. fixation [kg kg-1]

Sorghum medium duration ** File Sorg-med-Gha-GYGA.CAB ** ** CROP DATA FILE for use with WOFOST Version 7.0

17

** ** ** ** ** **

Reference: Heemst, H.van, 1988. Plant data values required for simple and universal simulation models: review and bibliography. Simulation reports CABO-TT. Some changes included for Sorghum (medium duration) for Ghana for global yield gap atlas

CRPNAM='Sorghum, medium duration, Ghana, Global yield gap atlas' ** emergence TBASEM = 10.0 TEFFMX = 30.0 TSUMEM = 70. ** phenology IDSL = 0 DLO DLC TSUM1 TSUM2 DTSMTB

= = = = =

DVSI = 0. DVSEND =

1.0 0.0 950. 850. 0.00, 10.00, 30.00, 45.00, 2.00

! lower threshold temp. for emergence [cel] ! max. eff. temp. for emergence [cel] ! temperature sum from sowing to emergence [cel d] ! ! ! !

indicates whether pre-anthesis development depends on temp. (=0), daylength (=1) , or both (=2) optimum daylength for development [hr] critical daylength (lower threshold) [hr] ! temperature sum from emergence to anthesis [cel d] ! temperature sum from anthesis to maturity [cel d] 0.00, ! daily increase in temp. sum 0.00, ! as function of av. temp. [cel; cel d] 20.00, 20.00 ! initial DVS ! development stage at harvest (= 2.0 at maturity [-])

** initial TDWI = 5.00 ! initial total crop dry weight [kg ha-1] ** ! Not used as imput by WOF6_0 model LAIEM = 0.02688 ! leaf area index at emergence [ha ha-1] RGRLAI = 0.0500 ! maximum relative increase in LAI [ha ha-1 d-1] ** green area SLATB = 0.00, 1.00, 2.00, SPA = 0.000 SSATB = 0.0, 0.0, 2.0, 0.0 SPAN = 42. TBASE = 10.0 ** assimilation KDIFTB = 0.0, 0.70, 2.0, 0.70 EFFTB = 0.0, 0.50, 40.0, 0.50 AMAXTB = 0.00, 1.00, 1.30, 1.60, 1.90, 2.00, TMPFTB = 0.00, 8.00, 20.00, 35.00, 45.00, TMNFTB = 5.00, 12.00,

0.0020, ! specific leaf area 0.0020, ! as a function of DVS [-; ha kg-1] 0.0020 ! specific pod area [ha kg-1] ! specific stem area [ha kg-1] ! as function of DVS ! life span of leaves growing at 35 Celsius [d] ! lower threshold temp. for ageing of leaves [cel] ! ! ! !

extinction coefficient for diffuse visible light [-] as function of DVS light-use effic. single leaf [kg ha-1 hr-1 j-1 m2 s] as function of daily mean temp. 70.00, ! max. leaf CO2 assim. rate 70.00, ! function of DVS [-; kg ha-1 hr-1] 61.00, 43.00, 20.00, 0.00 0.00, ! reduction factor of AMAX 0.00, ! as function of av. temp. [cel; -] 1.00, 1.00, 0.00 0.00, ! red. factor of gross assim. rate 1.00 ! as function of low min. temp. [cel; -]

** conversion of assimilates into biomass CVL = 0.720 ! efficiency of conversion CVO = 0.730 ! efficiency of conversion CVR = 0.720 ! efficiency of conversion CVS = 0.690 ! efficiency of conversion

into into into into

leaves [kg kg-1] storage org. [kg kg-1] roots [kg kg-1] stems [kg kg-1]

** maintenance respiration Q10 = 2.0 ! rel. incr. in resp. rate per 10 Cel temp. incr. [-] RML = 0.0300 ! rel. maint. resp. rate leaves [kg CH2O kg-1 d-1] RMO = 0.0100 ! rel. maint. resp. rate stor.org. [kg CH2O kg-1 d-1] RMR = 0.0100 ! rel. maint. resp. rate roots [kg CH2O kg-1 d-1] RMS = 0.0150 ! rel. maint. resp. rate stems [kg CH2O kg-1 d-1] RFSETB = 0.00, 1.00, ! red. factor for senescence 2.00, 1.00 ! as function of DVS [-; -] ** partitioning FRTB = 0.00, 0.20, 0.40, 0.60, 0.80, 1.00, 1.10, 2.00,

0.55, 0.45, 0.35, 0.20, 0.15, 0.05, 0.00, 0.00

! fraction of total dry matter to roots ! as a function of DVS [-; kg kg-1]

18

FLTB

=

FSTB

=

FOTB

=

0.00, 0.40, 0.60, 0.80, 1.00, 1.20, 2.00, 0.00, 0.40, 0.60, 0.80, 1.00, 1.20, 1.30, 2.00, 0.00, 1.00, 1.20, 1.30, 2.00,

** death rates PERDL = 0.030 RDRRTB = 0.00, 1.50, 1.5001, 2.00, RDRSTB = 0.00, 1.50, 1.5001, 2.00, ** water CFET DEPNR IAIRDU

use = 1.10 = 5.0 = 0

** rooting RDI = 10. RRI = 4.0 RDMCR = 150.

1.00, 1.00, 0.87, 0.59, 0.26, 0.00, 0.00 0.00, 0.00, 0.13, 0.41, 0.74, 0.85, 0.00, 0.00 0.00, 0.00, 0.15, 1.00, 1.00

! fraction of above-gr. DM to leaves ! as a function of DVS [-; kg kg-1]

! fraction of above-gr. DM to stems ! as a function of DVS [-; kg kg-1]

! fraction of above-gr. DM to stor. org. ! as a function of DVS [-; kg kg-1]

! max. rel. death rate of leaves due to water stress 0.000, ! rel. death rate of roots 0.000, ! as a function of DVS [-; kg kg-1 d-1] 0.020, 0.020 0.000, ! rel. death rate of stems 0.000, ! as a function of DVS [-; kg kg-1 d-1] 0.020, 0.020 ! correction factor transpiration rate [-] ! crop group number for soil water depletion [-] ! air ducts in roots present (=1) or not (=0) ! initial rooting depth [cm] ! maximum daily increase in rooting depth [cm d-1] ! maximum rooting depth [cm]

** nutrients ** maximum and minimum concentrations of N, P, and K ** in storage organs in vegetative organs [kg kg-1] NMINSO = 0.0100 ; NMINVE = 0.0035 NMAXSO = 0.0320 ; NMAXVE = 0.0120 PMINSO = 0.0014 ; PMINVE = 0.0005 PMAXSO = 0.0060 ; PMAXVE = 0.0025 KMINSO = 0.0025 ; KMINVE = 0.0070 KMAXSO = 0.0075 ; KMAXVE = 0.0280 YZERO = 200. ! max. amount veg. organs at zero yield [kg ha-1] NFIX = 0.00 ! fraction of N-uptake from biol. fixation [kg kg-1]

19

Appendix C Part of the WOFOST rerun files with some of the selected soil-crop-weathersowing dates-year combinations for Ghana Millet RUNNAM='1'; CRPNAM='Millet for YIELD GAP calcul.'; CRFILE='MILL-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=152; ISDAY=92; TSUM1=1070.; TSUM2=950.; IDURMX=200; SMW=0.100; SMFCF=0.140; RDMSOL=40.; NOTINF=0.033; WAV=0.5; RUNNAM='2'; CRPNAM='Millet for YIELD GAP calcul.'; CRFILE='MILL-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=152; ISDAY=92; TSUM1=1070.; TSUM2=950.; IDURMX=200; SMW=0.100; SMFCF=0.150; RDMSOL=40.; NOTINF=0.033; WAV=0.7; RUNNAM='3'; CRPNAM='Millet for YIELD GAP calcul.'; CRFILE='MILL-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=152; ISDAY=92; TSUM1=1070.; TSUM2=950.; IDURMX=200; SMW=0.100; SMFCF=0.160; RDMSOL=40.; NOTINF=0.033; WAV=0.8; RUNNAM='4'; CRPNAM='Millet for YIELD GAP calcul.'; CRFILE='MILL-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=152; ISDAY=92; TSUM1=1070.; TSUM2=950.; IDURMX=200; SMW=0.100; SMFCF=0.170; RDMSOL=40.; NOTINF=0.033; WAV=0.9; RUNNAM='5'; CRPNAM='Millet for YIELD GAP calcul.'; CRFILE='MILL-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998;

20

INYEAR =1; IDSOW=152; ISDAY=92; TSUM1=1070.; TSUM2=950.; IDURMX=200; SMW=0.100; SMFCF=0.180; RDMSOL=40.; NOTINF=0.033; WAV=1.1; Etc.

Sorghum RUNNAM='1'; CRPNAM='Sorghum for YIELD GAP calcul.'; CRFILE='SORG-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=137; ISDAY=77; TSUM1=1240.; TSUM2=1230.; IDURMX=200; SMW=0.100; SMFCF=0.140; RDMSOL=40.; NOTINF=0.033; WAV=0.5; RUNNAM='2'; CRPNAM='Sorghum for YIELD GAP calcul.'; CRFILE='SORG-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=137; ISDAY=77; TSUM1=1240.; TSUM2=1230.; IDURMX=200; SMW=0.100; SMFCF=0.150; RDMSOL=40.; NOTINF=0.033; WAV=0.7; RUNNAM='3'; CRPNAM='Sorghum for YIELD GAP calcul.'; CRFILE='SORG-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=137; ISDAY=77; TSUM1=1240.; TSUM2=1230.; IDURMX=200; SMW=0.100; SMFCF=0.160; RDMSOL=40.; NOTINF=0.033; WAV=0.8; RUNNAM='4'; CRPNAM='Sorghum for YIELD GAP calcul.'; CRFILE='SORG-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=137; ISDAY=77; TSUM1=1240.; TSUM2=1230.; IDURMX=200; SMW=0.100; SMFCF=0.170;

21

RDMSOL=40.; NOTINF=0.033; WAV=0.9; RUNNAM='5'; CRPNAM='Sorghum for YIELD GAP calcul.'; CRFILE='SORG-med-Gha-GYGA.CAB'; CLFILE='Gha0.'; ISYR =1998; INYEAR =1; IDSOW=137; ISDAY=77; TSUM1=1240.; TSUM2=1230.; IDURMX=200; SMW=0.100; SMFCF=0.180; RDMSOL=40.; NOTINF=0.033; WAV=1.1; Etc.

22

Appendix D ORYZA2000 Input data Crop data file ********************************************************************** * Template Crop data file for ORYZA2000 rice growth model * * File name : IR64OPT.CRP * * Crop : Oryza sativa cv. IR72 * * Experiment : Parameter values derived from various experiments * * at IRRI, Los Banos, Philippines. * * Information : Bouman BAM, Kropff MJ, Tuong TP, Wopereis, MCS, ten * * Berge HFM, van Laar, HH. ORYZA2000. IRRI, Los Banos. * * Phenology : Pepijn: Phenology parameters calibrated for IR64 * * with pheno_opt_rice calibration program and calling * * SUBDDN.f90 in ORYZA1.f90. See van Oort et al 2011 * * for details about the calibration * * Other : Pepijn this file is input for an adapted version of * * the ORYZA2000 version 2 number 13 model. New * * subroutines have been added for grainfilling and * * sterility. * ********************************************************************** * 1. Phenological development parameters * Note: not used in GYGA project, in GYGA project these parameters are provided through the reruns file TBD = 14. ! Base temperature for development (oC) TBLV = 8. ! Base temperature for juvenile leaf area growth (oC) ! not calibrated by pheno_opt, keep to default! TMD = 999. ! Maximum temperature for development (oC) TOD = 31. ! Optimum temperature for development (oC) DVRJ = 0.011404 ! Development rate in juvenile phase (oCd-1) DVRI = 0.011404 ! Development rate in photoperiod-sensitive phase (oCd-1) DVRP = 0.016667 ! Development rate in panicle development (oCd-1) DVRR = 0.045455 ! Development rate in reproductive phase (oCd-1) MOPP = 10.0 ! Maximum optimum photoperiod (h) PPSE = 0.0 ! Photoperiod sensitivity (h-1) SHCKD = 0.0 ! Relation between seedling age and delay in phenological ! development (oCd oCd-1) COLDMIN = 12. ! Lower air temperature treshhold for growth (oC) COLDEAD = 3. ! Consecutive number of days below COLDMIN that crop dies (-) * 2. Leaf and stem growth parameters RGRLMX = 0.0085 ! Maximum relative growth rate of leaf area (oCd-1) RGRLMN = 0.0040 ! Minimum relative growth rate of leaf area (oCd-1) SHCKL = 0.25 ! Relation between seedling age and delay in leaf area ! development (oCd oCd-1) *SHCKL = 0.00 ! Relation between seedling age and delay in leaf area * ! development (oCd oCd-1) *RGRLMX = 0.011 ! Maximum relative growth rate of leaf area (oCd-1) ! RGRL = (30-8)/(31-14)*0.0085 = 0.011 * Switch to use SLA as table (give values below) or as fixed function SWISLA = 'FUNCTION' ! Give function parameters ASLA, BSLA, CSLA, DSLA, SLAMAX *SWISLA = 'TABLE' ! Give SLA as a function of DVS in the table SLATB * If SWISLA='FUNCTION', supply SLA function parameters: * SLA = ASLA + BSLA*EXP(CSLA*(DVS-DSLA)), and SLAMAX ASLA = 0.0024 ! (-) BSLA = 0.0025 ! (-) CSLA = -4.5 ! (-) DSLA = 0.14 ! (-) SLAMAX = 0.0045 ! maximum value of SLA (ha/kg) * If SWISLA='TABLE', supply table of specific leaf area (ha kg-1; Y value) * as a function of development stage (-; X value): *SLATB = 0.00, 0.0045, * 0.16, 0.0045, * 0.33, 0.0033, * 0.65, 0.0028, * 0.79, 0.0024, * 2.10, 0.0023, * 2.50, 0.0023 * * D:\africarice\ORYZASAHEL\ORYZAS1S\RICE.DAT SLATB = 0.0,0.00333 ,0.65,0.0024 ,1.00,0.00205 2.10,0.00177

,1.25,0.00177,

* * D:\africarice\ORYZASAHEL\ORYZAS1S\RICE.DAT * RGRMAX parameters * Equation used in RGRMAX in ORYZA1.f90: * RGRMAX = RGRMX / (1.+EXP(RGRA+RGRB*TMIN)) * this continuous function produces same shape as linear interpolation from table * RGRMTB = -10.,0. ,10.0,0.0 ,15.,2. ,20.,14. ,24.,17. ,40.,17. *RGRMX = 0.17 *RGRA = 12.8 *RGRB = -0.72

23

* this continuous function produces same shape as linear interpolation from table * RGRMTB = 0.,0. ,12.,0. ,18.,11. ,20.,14. ,23.,16. ,30.,16. RGRMX = 0.16 RGRA = 12.4 RGRB = -0.73 * D:\africarice\ORYZASAHEL\ORYZAS1S\RICE.DAT *======================================= * valeurs … lire pour la variete : IR64 *====================================== *SLATB = 0.0,0.00333 ,0.65,0.0024 ,1.00,0.00205 * 2.10,0.00177

,1.25,0.00177,

* Table of specific green stem area (ha kg-1; Y value) as a function of * development stage (-; X value): SSGATB = 0.0, 0.0003, 0.9, 0.0003, 2.1, 0.0000, 2.5, 0.0000 * 3. Photosynthesis parameters FRPAR = 0.5 ! fraction of sunlight energy that is ! photosynthetically active (-) SCP = 0.2 ! Scattering coefficient of leaves for PAR (-) CO2REF = 340. ! Reference level of atmospheric CO2 (ppm) ! don't touch this parameter CO2 = 382. ! Ambient CO2 concentration (ppm) - pvo20131223: 382 ppm is around year 2006/7, years in which Michiel's experiments conducted * source CO2: ftp://ftp.cmdl.noaa.gov/ccg/co2/trends/co2_annmean_mlo.txt * Table of light extinction coefficient for leaves (-; Y-value) as a function * of development stage (-; X value): KDFTB = 0.00, 0.4, 0.65, 0.4, 1.00, 0.6, 2.50, 0.6 * Table of extinction coefficient of N profile in the canopy (-; Y-value) * as a functionof development stage (-; X value): KNFTB = 0.0, 0.4, 2.5, 0.4 * Table of light use effiency (-; Y-value) as a function of * temperature (oC; X value): EFFTB = 0.,0.54, 10.,0.54, 40.,0.36, 100.,0.0 * Table of effect of temperature on AMAX (-; Y-value) as a function of * temperature (oC; X value): REDFTT = -10., 0., 10., 0., 20., 1., 37., 1., 43., 0., 99., 0. * Table of N fraction in leaves on leaf area basis (g N m-2 leaf; Y-value) * as a function of development stage (-; X value): NFLVTB = 0.00, 0.54, 0.16, 0.54, 0.33, 1.53, 0.65, 1.22, 0.79, 1.56, 1.00, 1.29, 1.46, 1.37, 2.02, 0.83, 2.50, 0.83 * 4. Maintenance parameters * Maintenance respiration coefficient (kg CH2O kg-1 DM d-1) of: MAINLV = 0.02 ! Leaves MAINST = 0.015 ! Stems MAINSO = 0.003 ! Storage organs (panicles) MAINRT = 0.01 ! Roots TREF *TREF Q10

= 25. = 20. = 2.

! Reference temperature (oC) (ORYZA2000) ! Reference temperature (oC) (GECROS) ! Factor accounting for increase in maintenance ! respiration with a 10 oC rise in temperature (-)

* 5. Growth respiration parameters * Carbohydrate requirement for dry matter production (kg CH2O kg-1 DM leaf) of: CRGLV = 1.326 ! Leaves CRGST = 1.326 ! Stems CRGSO = 1.462 ! Storage organs (panicles) CRGRT = 1.326 ! Roots

24

CRGSTR = 1.11

! Stem reserves

LRSTR

! Fraction of allocated stem reserves that is ! available for growth (-)

= 0.947

* 6. Growth parameters FSTR = 0.20 ! Fraction of carbohydrates allocated to stems that ! is stored as reserves (-) TCLSTR = 10. ! Time coefficient for loss of stem reserves (1 d-1) SPGF = 64900. ! Spikelet growth factor (no kg-1) ! value for IR72 !SPGF = 52000. ! Spikelet growth factor (no kg-1) ! ~0.8*64900, for sensitivity analysis !SPGF = 88060. ! Spikelet growth factor (no kg-1) ! average from Cotonou experiment, varieties IR64, FARO 35 (ITA212), Kogoni 91-1 NSPM2X = 100000. ! Maximum number of spikelets per m2 ! average from Cotonou experiment, varieties IR64, FARO 35 (ITA212), Kogoni 91-1 NSJA = 2000. ! Juvenile spikelets per g-1 N at 14 days before heading. Range: 1000.-3000. WGRMX

= 0.0000249 ! Maximum individual grain weight (kg grain-1)

NSP_FRC = 0 ! simulate NSP !NSP_FRC = 2 ! force NSP to observed value NSP_OBS = 400000000. ! Observed number of spikelets per hectare * pvo20130507: new input parameters for subroutine SUBGREC * (1) Under source limitation some grains will be only partially filled * A crop with GFRCP = 1 tries to fill completely as much as possible of its grains; the remaining are partially filled at very low weight (almost empty) * A crop with GFRCP = 0 produces 0 completely filled grains but many, relatively heavy, partially filled grains * GFRCP = 1 is most realistic and conform simulations with original ORYZA2000v2n13 model * (2) All grains (partially and completely filled) heavier than WGREC contribute to the yield that can be commercially sold *GFRCP = 0.9 ! Grain Filling Ratio Complete / Partial (0 to 1). GFRCP = 1.0 ! Grain Filling Ratio Complete / Partial (0 to 1). WGREC = 0.0000200 ! Weight GRain EConomic (kg grain-1) * Partitioning tables * Table of fraction total dry matter partitioned to the shoot (-; Y-value) * as a function of development stage (-; X value): FSHTB = 0.00, 0.50, 0.43, 0.75, 1.00, 1.00, 2.50, 1.00 * Table of fraction shoot dry matter partitioned to the leaves (-; Y-value) * as a function of development stage (-; X value): FLVTB = 0.000, 0.60, 0.500, 0.60, 0.750, 0.30, 1.000, 0.00, 1.200, 0.00, 2.5 , 0. * Table of fraction shoot dry matter partitioned to the stems (-; Y-value) * as a function of development stage (-; X value): FSTTB = 0.000, 0.40, 0.500, 0.40, 0.750, 0.70, 1.000, 0.40, 1.200, 0.00, 2.5 , 0. * Table of fraction shoot dry matter partitioned to the panicles (-; Y-value) * as a function of development stage (-; X value): FSOTB = 0.000, 0.000, 0.500, 0.000, 0.750, 0.000, 1.000, 0.600, 1.200, 1.000, 2.5 , 1. * Table of leaf death coefficient (d-1; Y-value) as a function of development * stage (-; X value): DRLVT = 0.00, 0.000, 0.60, 0.000, 1.00, 0.015, 1.60, 0.025, 2.10, 0.050, 2.50, 0.050 * 7. Carbon balance parameters * Mass fraction carbon (kg C kg-1 DM) in the: FCLV = 0.419 ! Leaves FCST = 0.431 ! Stems FCSO = 0.487 ! Storage organs (panicles) FCRT = 0.431 ! Roots FCSTR = 0.444 ! Stem reserves

25

* 8. Root parameters GZRT = 0.01 ! Growth rate of roots (m d-1) ZRTMCW = 0.25 ! Maximum depth of roots if no drought stress (m) ZRTMCD = 0.40 ! Maximum depth of roots if drought (m) * 9. Drought stress parameters * Upper and lower limits for drought stress effects ULLS = 74.13 ! Upper limit leaf rolling (kPa) LLLS = 794.33 ! Lower limit leaf rolling (kPa) ULDL = 630.95 ! Upper limit death of leaves (kPa) LLDL = 1584.89 ! Lower limit death of leaves (kPa) !ULLE = 1.45 ! Upper limit leaf expansion (kPa) !PVO20131104: default value in IR72.CRP !LLLE = 1404. ! Lower limit leaf expansion (kPa) !PVO20131104: default value in IR72.CRP ULLE = 50. ! Upper limit leaf expansion (kPa) !PVO20131104: oryza2000 book, page 88 LLLE = 260. ! Lower limit leaf expansion (kPa) !PVO20131104: oryza2000 book, page 88 ULRT = 74.13 ! Upper limit relative transpiration reduction (kPa) LLRT = 1584.89 ! Lower limit relative transpiration reduction (kPa) * Switch to use ULTR and LLTR as given above or function built in ORYZA * for the reduction in relative transpiration: SWIRTR = 'DATA' ! Use data *SWIRTR = 'FUNCTION' ! Use function *===================================================================* * Drought stress effect parameters for aerobic rice * * Values are for wheat, taken from SUCROS2 model * *===================================================================* * characteristic potential transpiration rate at a soil water * content halfway wilting point and field capacity (mm.d-1) TRANSC = 6. * Root activity coefficient (-) EDPTFT = 0.,0.15, 0.15,0.6, 0.3,0.8, 0.5,1., 1.1,1. *===================================================================* * Pvo20140316: switch for yes (1) or no (0) delay in flowering * * C.f. Oryza2000 book page 89 * *===================================================================* SWIDVEW = 0 *===================================================================* * 10. Nitrogen parameters NMAXUP = 8. ! Maximum daily N uptake (kg N ha-1 d-1) RFNLV = 0.004 ! Residual N fraction of leaves (kg N kg-1 leaves) FNTRT = 0.15 ! Fraction N translocation from roots, as (additonal) ! fraction of total N translocation from stems and leaves (-) RFNST = 0.0015 ! Residual N fraction of stems (kg N kg-1 stems) TCNTRF = 10. ! Time coefficient for N translocation to grains (d) NFLVI = 0.5 ! Initial leaf N fraction (on area basis: g N m-2 leaf) FNLVI = 0.025 ! Initial leaf N fraction (on weight basis: kg N kg-1 leaf) NMAXSO = 0.0175 ! Maximum N concentration in storage organs (kg N kg-1) * Table of minimum N concentration in storage organs (kg N kg-1 DM; Y value) * as a function of the amount of N in the crop till flowering (kg N ha-1; X value): NMINSOT = 0., .006, 50., .0008, 150., .0125, 250., .015, 400., .017, 1000., .017 * Table of maximum leaf N fraction on weight basis (kg N kg-1 leaves; Y value) * as a function of development stage (-; X value): NMAXLT = 0.0, .053, 0.4, .053, 0.75, .040, 1.0, .028, 2.0, .022, 2.5, .015 * Table of minimum leaf N fraction on weight basis (kg N kg-1 leaves; Y value) * as a function of development stage (-; X value): NMINLT = 0.0, 0.025, 1.0, 0.012, 2.1, 0.007, 2.5, 0.007 *--- Table of effect of N stress on leaf death rate (-; Y value) * as a function of N stress level (-; X value): NSLLVT = 0., 1.0, 1.1, 1.0, 1.5, 1.4, 2.0, 1.5, 2.5, 1.5

26

Rerun file The rerun file contains crop parameters (phenology), management parameters (sowing date, plant density, planting method (direct seeding/transplanting)) and soil parameters. Below, a part of the rerun file is shown and comments have been added for explanation. Descriptions of the meaning of parameters listed below can be found in Bouman et al. (2001). *RFR_ACTSOWDATE_YW_RERUNS.RER *Rerun file for rainfed rice (RFR), with actual sowing dates * First cropping system, in this case rainfed lowland * year (IYEAR) 1990, Country (CNTR) Burkina Faso, station (ISTN) 4, emergenge julian day of year (STTIME) 168, seedbed duration (SBDUR) 18 days, transplanted rice (ESTAB), 16 hills per m2 (NH), 3 plant per hill (NPLH), development rates (day-1) for pre-flowering period (DVRJ, DVRI and DVRP) and for post-flowering period (DVRR), soil parameters WCLI, WL0MX, SWITPD, SWITVP, WCSTRP, PERTB, ZWTB, KST, WCST, VGA, VGL, VGN and VGR. Atmospheric CO2 in 1990 (CO2): 333 ppm *For rainfed lowland we use parameters of WOFOST soil EC4-fine (clayey soil). Additional assumptions: Bundheight 250 mm. Groundwater at 20 cm depth. NonPuddled. No plowpan. Low percolation rate. IYEAR = 1990; CNTR = 'Bur'; ISTN = 4; STTIME = 168.; SBDUR = 18; ESTAB = 'TRANSPLANT'; NH = 16.; NPLH = 3.; NPLDS = 48.; DVRJ = 0.011111; DVRI = 0.011111; DVRP = 0.011111; DVRR = 0.033333; WCLI = 10*0.57; WL0MX = 250.0; SWITPD = 0; SWITVP = 0; WCSTRP = 10*0.56; PERTB = 0.,0.,1000.,3.7; ZWTB = 1.,20., 366.,20.; KST = 10*10.789; WCST = 10*0.57; VGA = 10*0.0321; VGL = 10*40.7; VGN = 10*1.2848; VGR = 10*0.27; C02 = 333. IYEAR = 1991; CNTR = 'Bur'; ISTN = 4; STTIME = 168.; SBDUR = 18; ESTAB = 'TRANSPLANT'; NH = 16.; NPLH = 3.; NPLDS = 48.; DVRJ = 0.011111; DVRI = 0.011111; DVRP = 0.011111; DVRR = 0.033333; WCLI = 10*0.57; WL0MX = 250.0; SWITPD = 0; SWITVP = 0; WCSTRP = 10*0.56; PERTB = 0.,0.,1000.,3.7; ZWTB = 1.,20., 366.,20.; KST = 10*10.789; WCST = 10*0.57; VGA = 10*0.0321; VGL = 10*40.7; VGN = 10*1.2848; VGR = 10*0.27; C02 = 334. … IYEAR = 2005; CNTR = 'Bur'; ISTN = 4; STTIME = 168.; SBDUR = 18; ESTAB = 'TRANSPLANT'; NH = 16.; NPLH = 3.; NPLDS = 48.; DVRJ = 0.011111; DVRI = 0.011111; DVRP = 0.011111; DVRR = 0.033333; WCLI = 10*0.57; WL0MX = 250.0; SWITPD = 0; SWITVP = 0; WCSTRP = 10*0.56; PERTB = 0.,0.,1000.,3.7; ZWTB = 1.,20., 366.,20.; KST = 10*10.789; WCST = 10*0.57; VGA = 10*0.0321; VGL = 10*40.7; VGN = 10*1.2848; VGR = 10*0.27; C02 = 354. * Next cropping system, in this case rainfed upland in the same site * year (IYEAR) 1990, Country (CNTR) Burkina Faso, station (ISTN) 4, emergenge julian day of year (STTIME) 168, seedbed duration (SBDUR) 18 days, transplanted rice (ESTAB), 16 hills per m2 (NH), 3 plant per hill (NPLH), development rates (day-1) for pre-flowering period (DVRJ, DVRI and DVRP) and for post-flowering period (DVRR), soil parameters WCLI, WL0MX, SWITPD, SWITVP, WCSTRP, PERTB, ZWTB, KST, WCST, VGA, VGL, VGN and VGR. Atmospheric CO2 in 1990 (CO2): 333 ppm *For rainfed upland we use parameters of WOFOST soil EC4-fine (clayey soil). Additional assumptions: Bundheight 0 mm. Groundwater at 1000 cm depth. NonPuddled. No plowpan. High percolation rate. IYEAR = 1990; CNTR = 'Bur'; ISTN = 4; STTIME = 170.; SBDUR = 0; ESTAB = 'DIRECT-SEED'; NH = 0.; NPLH = 0.; NPLDS = 80.; DVRJ = 0.011111; DVRI = 0.011111; DVRP = 0.011111; DVRR = 0.033333; WCLI = 10*0.39; WL0MX = 0.0; SWITPD = 0; SWITVP = 0; WCSTRP = 10*0.38; PERTB = 0.,0.,1000.,240.03; ZWTB = 1.,1000., 366.,1000.; KST = 10*99.77; WCST = 10*0.39; VGA = 10*0.0321; VGL = 10*60.9; VGN = 10*2.163; VGR = 10*0.04; C02 = 333.

27

Appendix E Relationship between maize planting density and seasonal water deficit as observed in the US Corn Belt

Figure E1. Relationship between maize planting density and seasonal water deficit as observed in the US Corn Belt. Seasonal water deficit was calculated as ETo (reference evapotranspiration) minus precipitation between sowing and physiological maturity. The grey shaded area represents the 95 % confidence interval of prediction of the linear regression. Source: Grassini et al. (2009).

28