In our group we use crop models mainly to i) gain an insight into the crop growth dynamics in complex agricultural systems and fluctuating environments, ii) characterize the main constraints of production environments, iii) assess potential of new technologies (e.g. success of new cultivars, optimization and construction of specific “cultivar x management packages”) in given environment and, iv) “what if analysis” and future foresight (e.g. in context of climatic changes).
APSIM model requires several inputs to simulate the crop within the system: i) daily weather information (Tmax, Tmin, rainfall, solar radiation), ii) soil information (soil type, water &nutrient status), iii) crop and genotype specific coefficient.Calculation of genotypic coefficients for APSIM is generally the most effective if done in this particular order;
Phenology (~60% fit with observed data)
Photoperiod sensitivity (wherever relevant)
Canopy (~30% fit with observed data)
Grain number/grain size (~10% fit with observed data)
Detailed description of particular crop parameters can be found at www.apsim.com. In the case of sorghum, reasonable minimum parameterization requirement can be achieved by:
Calculating the thermal time requirement from “end of juvenile” to “panicle initiation” phase (tt_endjuv_to_init, APSIM sorghum module annotation) which is generally the most variable phenological phase determining the whole crop duration.To determine the crop duration we need to observe thermal time to flowering in the field at minimum. Then tt_endjuv_to_initpheno-phase can be calculated as;
tt_endjuv_to_init= [tt to flowering (observed) – tt_germ_to_emerg (conserved) –tt_emerg_endjuv (conserved) – tt_init to flag (conserved) - tt_flag_to_flower (conserved)];
(Note: We can improve the phenological coefficients estimation if we record tt to panicle initiation, tt to flag leaf,and tt to maturity.)
Duration of the crop is sensitive to photoperiod (calculated as a function of day of year and latitude; can be outputted from APSIM) if the cultivar is photoperiod sensitive. The fotoperiod affects sorghum tt_endjuv_to_init phase and is generally stable [pp (11.5;13.5h)]. The most variable coefficient is the slope of extension in duration of tt_endjuv_to_init phase (pp_slope [tt_endjuv_to_init extension per h of pp) which has to be calculated. The minimum observation requirement is tt to flowering of the crops sown in different time of the year within photoperiod sensitivewindow11.5-13.5h. For these tt_endjuv_to_init (above)and consequently the slope of extension in duration of tt_endjuv_to_init phase can be calculated.
Released APSIM version canopy module simulates the canopy based on the total plant leaf area (TPLA) approach which requires calculation of i) TPLAmax, ii)TPLA inflection ratio, iii) TPLA production cf.
TPLA approach of canopy prediction integrates the number of fully expanded leaves, their individual size, and tiller number, and includes an adjustment for the area of expanding leaves (Hammer et al., 1993). To parameterize the TPLA coefficients, we need several observed data points on LA and corresponding TT, tiller number, thermal time to flag leaf (can be generalized by TT_to_flower-100oCdays) during the plant growth. Then TPLAmax defines the maximum total plant leaf area as:
tplaMax = (pow(FTN + 1.0, tillerCoef) * pow(finalLeafNo,mainStemCoef)) * scm2smm;
TillerCoef = 0.66 (is generally considered constant); FTN- fertile tiller No,
We optimize the mainStemCoef (using solver add-in in excel [hyperlink to xls])
TPLAinflection ratio, TPLAproductioncf define skewness and breath of the TPLAmax function (“early vigour” related parameters); then every day TPLA is calculated:
TPLA = TPLAmax / (1 + exp(-tpla_prod_coef * (TTemerg_to_now - tpla_inflection)));
TPLA inflection = (TTemerg_to_flag * tpla_inflection_ratio)
TTemerg_to_flag = Flowering time - 100 TT
Here we optimizec_tpla_inflection_ratio ; -p_tpla_prod_coef (using solver add-in in excel [hyperlink to xls])
Note: leaf_app_rate1 andleaf_app_rate2 [tt requirement per leaf-1and tt requirement per leaf-1of last 4 leaves] is considered rather constant, but can be calculated and parameters changed;
For most of the cultivated genotypes grain number [dm_per_seed] and grain size [maxGFRate] coefficients are rather common. However, these coefficients can be calculated as: Maximum grain number is a function of the plant growth rate between panicle initiation and the start of grain filling;
[maxGFRate = (deltaDMPlant)/nDays]
and the final grain number is calculated by dividing the growth rate by a genotypic value.
Grain number = growthRate/dm_per_seed
Grain size is determined by grain growth rate, the effective grain-filling period, and the redistribution of assimilates post-anthesis. Grain size is determined by calculating the potential daily grain filling rate which should not exceed the set maximum grain filling rate (maxGFRate, genotype specific value). Grain demand or potential grain filling rate (maxGFRate) is a function of the increase in plant biomass per grain per degree day (totDMCaryopsis, above). Then:
For us gems means GEMS, or G*E*M*S (genotype by environment by management by society) interactions, i.e. the fact that crop yields results from complex biophysical interactions while acceptance depends on farmer/consumer preferences. This complexity becomes an opportunity when it is cracked into components that can be analysed, understood, predicted, and then used to prioritise research investments to maximise return. This is what we do, and this is when GEMS become gems!
For us gems means GEMS, or G*E*M*S (genotype by environment by management by society) interactions, i.e. the fact that crop yields results from complex biophysical interactions while acceptance depends on farmer/consumer preferences. This complexity becomes an opportunity when it is cracked into components that can be analyzed, understood, predicted, and then used to prioritize research investments to maximise return. This is what we do, and this is when GEMS become gems!
A crop performs in different ways in different sites, years and agronomic managements. These are called genotype-by-environment-by management(G*E*M) interactions, and they are a main challenge for breeders and agronomists. There is one more layer of interaction, even more complex: the society (S). Farmers and consumers have different desires, needs, expectations, and a cultivar that fits one may not fit the other (G*E*M*S interactions). The puzzle is complex and challenging but if its components are understood, specific interventions can be undertaken.For instance, breeding for a particular genotype (G, with particular physiological characteristics), for a particular environment (E, with a particular kind of drought pattern that requires a specific adaptive trait), in a particular management practice (M, for instance a sowing density, or a N fertilizer treatment), and targeted to particular farmer/consumer (S, for instance a genotype that produces a lot of rich stover for cattle ranchers) is the need of the hour.