Disentangling genotype × environment (G×E) controls of flowering time requires phenotypes that link molecular regulation, developmental physiology and environment. Here, we integrated time-resolved measurements of apical development, final leaf number (FLN), and expression of the flowering-time genes VRN1, VRN2 and VRN3 across contrasting temperature and photoperiod regimes in six wheat genotypes spanning a wide range of developmental sensitivities. By combining controlled-environment phenotyping with concurrent gene-expression profiling, we show that environmentally driven variation in FLN is coherently explained by shifts in the timing of key apical transitions and associated VRN gene-expression dynamics. These integrated datasets were used to parameterise and interrogate the Cereal Anthesis Molecular Phenology (CAMP) model, enabling direct comparison between observed foliar gene-expression time courses and modelled gene activity. While overall developmental responses were well captured by the model, systematic differences between observed and modelled gene-expression patterns highlight the importance of distinguishing foliar expression from apical regulatory activity, as well as differences in temporal scaling. Building on this framework, we present a phenotyping protocol based on FLN responses to defined temperature and photoperiod treatments that delivers unconfounded developmental phenotypes explicitly linked to underlying genetic regulation.

Crop models are vital for trait assessment and breeding, yet most lack mechanistic detail in carbon and nutrient partition and its effects on growth and yield. The generic organ arbitrator for biomass partitioning recently was developed, but it has NOT been widely validated apart from the initial testing. The study aimed to assess the organ arbitrator for simulating leaf and tiller numbers, leaf size and leaf area, biomass and nitrogen (N) partition into different organs, and the distributions of root length, root biomass and N in soil profile. We conducted the field experiment with two cultivars of winter wheat under three irrigation treatments in water-limited areas of the North China Plain from 2016 to 2018. The monitoring metrics were employed to assess the performance of APSIM next generation Wheat model (APSIM NG). The findings revealed that original APSIM NG overestimated tiller numbers, leaf area index (LAI), shoot and root biomass and nitrogen (N), but underestimated grain yield, with the Nash Sutcliffe Efficiency (NSE) ranging from -12.26-0.92. Modification to leaf, root growth parameters and temperature response curves of thermal time led to improved simulation of the development and growth of winter wheat. The simulation for individual leaf size was optimal (coefficient of determination (R2) = 0.95, NSE = 0.81). Similarly, the simulations for tiller density (except at the recovery stage) and for LAI from recovery to jointing also performed best, with Root Mean Square Error (RMSE) values of 302.91 tillers/m2 and 1.83 m2/m2, respectively. The biomass in above-ground, stems, grains of winter wheat under different water treatments, providing further confidence for the model to be used for trait evaluations (RMSE ranging from 38.04 to 123.85 g/m2, NSE ranging from 0.83 to 0.96). However, the modified model tended to overestimate partitioning to leaves and underestimate that to stems and spikes. In addition, the modifications also overestimated crop N uptake (with RMSE values of 11.09, 10.16, and 1.36 g/m2 for the N content in above-ground biomass, leaves, and spikes, respectively). The simulation of root biomass and N and their distributions in the soil profile was good, except for the underestimation of root biomass, length and N in the top soil layer. The study highlights the potential value of improved APSIM NG model to target phenotype, offering potential targets for genotype selecting in water-limited conditions. Further improvement in the model components in N uptake and root growth in the top soil layer is still required for the APSIM NG.

Process-based crop models encapsulate agro-ecosystem knowledge within modular components that have been developed across different crop modelling platforms. However, structural heterogeneity and differences in programming languages limit component reuse. The Agricultural Model Exchange Initiative recently proposed Crop2ML, a framework for crop modeling platform interoperability to address this issue. Nevertheless, transforming existing models into Crop2ML specifications still requires substantial manual curation of semantic metadata, which is error-prone and time consuming. Recent advances in large language models (LLMs) have increased their capacity to support automated code analysis through multi-agent system architectures. Here, a workflow was developed leveraging LLMs to automatically create Crop2ML model specifications by extracting and inferring metadata describing input and output variables specifications. The workflow was evaluated on source code written in three programming languages and originating from six different crop modeling platforms. Experiments conducted on eight soil temperature components from these widely used platforms showed a F1-Score exceeding 93% for input/output variables extraction and successful inference of variable specifications. The proposed approach introduces a novel consensus-based multi-agent methodology that improves the reliability of automated extraction results. Overall, this study demonstrated that LLMs-driven multi-agent systems can significantly accelerate the automatic generation of interoperable models with semantically rich model documentation and specification. This ultimately reduces the barriers to crop model reuse.

Rapid prediction and control of flowering time is essential for breeding crops resilient to changing climates. Current models often fail to predict flowering time in new cultivars because molecular models lack integration of environmental signals, while physiological models inadequately capture the interactions of vernalization, photoperiod and temperature. This leads to mischaracterized genotypes and inaccurate forecasts. A new Cereal Anthesis Molecular Phenology (CAMP) model was developed for wheat. It explicitly integrates the regulatory roles of three major 'virtual' flowering genes (Vrn1, Vrn2, and Vrn3) with environmental cues. A novel phenotyping strategy based on main stem leaf number was introduced to shorten the time required for data collection and model calibration. CAMP predicted flowering time within 4-7 d across 64 genetically diverse wheat cultivars grown under contrasting environments. The leaf-number phenotyping method reduced phenotyping time by more than 80%, offering a practical alternative to resource-intensive field trials. Together, these advances enable accurate cultivar characterization and scalable prediction of flowering behaviour. CAMP enables the ability to predict flowering time directly from genotypic data (e.g. SNPs), eliminating the need for costly controlled-environment experiments. This represents a step change in molecular-physiological modelling, supporting faster deployment of new cultivars and more effective design of wheat for future climates.

Effective parameter calibration is vital for accurate crop model simulations but is challenged by extensive parameter sets and inherent uncertainties in agroecosystem models. This study systematically evaluated three global sensitivity analysis (GSA) methods (Morris, Sobol-Martinez, and eFAST) alongside three parameter optimisation algorithms (Nelder-Mead simplex, DREAM-zs, and L-BFGS-B) to enhance parameter estimation and simulation accuracy in APSIM Next Generation, a widely used agroecosystem model. Sensitivity analysis results indicated that the Morris method identified the broadest set of influential parameters due to its inclusive parameter selection strategy, while Sobol-Martinez provided more targeted parameter identification by clearly distinguishing impactful parameters. Conversely, eFAST was highly selective, pinpointing fewer parameters of highest impact, beneficial for computational efficiency. Among the optimisation methods, the Bayesian DREAMzs algorithm consistently produced superior model predictions across evaluated output variables, including phenology, biomass, leaf area index, and grain yield, outperforming the frequentist Nelder-Mead and gradientbased L-BFGS-B methods. However, DREAM-zs required significantly higher computational resources, particularly at higher iteration settings. The uncertainty analysis revealed that interactions among sensitivity analysis methods, optimisation algorithms, and wheat genotypes dominated the sources of uncertainty in parameter estimation. This underscores the necessity of carefully selecting and integrating complementary sensitivity analysis and optimisation methods tailored to specific modelling objectives. Our findings demonstrated a robust methodological framework to improve calibration accuracy, reduce predictive uncertainty, and ultimately support more reliable agricultural decision-making.

Disentangling genotype × environment (G×E) effects is critical to understand the performance of wheat across different environments. A framework for doing this was previously presented in a model that integrated knowledge of crop physiology and the Vrn gene feedback loop to explain and predict the time of anthesis. The aims of this study were: 1) provide an updated description of the Cereal Anthesis Molecular Phenology (CAMP) model; 2) to verify the model’s assumptions regarding the relationship between Vrn gene expression and the timing of phenological stages in a set of diverse genotypes and environments; 3) to use the CAMP model to establish a phenotyping strategy for use in genetic studies and model parameterisation. Six wheat genotypes with a range of cool temperature and photoperiod sensitivities were evaluated. Apical development, final leaf number (FLN) and temporal expression of Vrn1, Vrn2 and Vrn3 were compared with model predictions. There was a clear relationship between FLN responses to cool temperature and photoperiod, the timing of phenological events and the patterns of Vrn gene expression for all genotypes. There was general agreement between the temporal patterns of foliar gene expression observed with those assumed by CAMP, but some obvious discrepancies. These may be related to differences between gene expression in foliar (observed) and apical (assumed by the model) parts of the plant, or differences in the way observed and modelled gene expression are scaled. Overall, the model described all the observed development responses to environment and provides a basis for building quantitative predictions of field-based development from genotypic and environmental data. A protocol is presented for phenotyping wheat using FLN measured in specific combinations of temperature and photoperiod. It allows easy and unconfounded measure of key developmental phenotypes that clearly relate to the genetic make-up of the plants and underlying gene expression profiles. ### Competing Interest Statement The authors have declared no competing interest. Ministry of Business, Innovation and Employment (MBIE) Strategic Science Investment Fund, Sustainable Agro-ecosystems Grains Research and Development Corporation, National Phenology Initiative (ULA00011)

CONTEXT Over the last 26 years, researchers globally have successfully applied the soil nitrogen (N) model in the Agricultural Production Systems sIMulator (APSIM) to simulate N cycling and its effects on crop production across a range of agricultural systems and environments. As the modelling community further expands its focus to include environmental impacts of farming, it needs the model to be fit for this broader purpose. OBJECTIVE Accurately modelling N loss via different pathways demands more of the model and so, to inform and prioritise future development needs, we embarked on a detailed review of APSIM's soil N modelling capability. METHODS We conducted a comprehensive search of APSIM Soil N model verification studies and found 131 relevant publications across a wide range of systems, applications, and processes. We examined their approaches and findings, and distilled out the lessons learnt. RESULTS AND CONCLUSIONS The model-data comparisons showed strong performance across all modelled processes, despite limited changes to the core of the soil N model since its inception. The model's relatively simple conceptual pool approach to modelling carbon (C) dynamics with N cycling linked via C:N ratios, has proven remarkably versatile. However, these conceptual pools have posed challenges relating to initialisation methods and the resulting sensitivity of predictions at different time scales, e.g. long-term C trajectories vs. short-term seasonal N dynamics. Correctly predicting timing of N loss on a daily timestep also proved challenging, but this level of resolution may not always be required. APSIM's adaptable code structure facilitated the creation of model prototypes (e.g., ammonia volatilisation and N in runoff) allowing testing of different conceptualisations ahead of formal release. SIGNIFICANCE APSIM is one of the most widely used agricultural systems models. This review, which covers model documentation, model-data comparisons, various approaches to parameterisation, and prototypes for additional processes, consolidates decades of research into insights about the model and its functioning. The review highlights the importance of model evaluations across a wide range of applications to ensure model robustness, to identify issues that may be masked in single studies, and to allow the emergence of solutions with broad applicability.

Increasing global food demand will require more food production1 without further exceeding the planetary boundaries2 while simultaneously adapting to climate change3. We used an ensemble of wheat simulation models with improved sink and source traits from the highest-yielding wheat genotypes4 to quantify potential yield gains and associated nitrogen requirements. This was explored for current and climate change scenarios across representative sites of major world wheat producing regions. The improved sink and source traits increased yield by 16% with current nitrogen fertilizer applications under both current climate and mid-century climate change scenarios. To achieve the full yield potential—a 52% increase in global average yield under a mid-century high warming climate scenario (RCP8.5), fertilizer use would need to increase fourfold over current use, which would unavoidably lead to higher environmental impacts from wheat production. Our results show the need to improve soil nitrogen availability and nitrogen use efficiency, along with yield potential. Martre et al. found that to achieve the full yield potential of improved wheat varieties, nitrogen fertilizer use would need to increase fourfold over current use, which would unavoidably increase the environmental impacts of wheat production.

This paper describes the data set that was used to test the accuracy of twenty-nine crop models in simulating the effect of changing sowing dates and sowing densities on wheat productivity for a high-yielding environment in New Zealand. The data includes one winter wheat cultivar (Wakanui) grown during six consecutive years, from 2012-2013 to 2017-2018, at two farms located in Leeston and Wakanui in Canterbury, New Zealand. The simulations were carried out in the framework of the Agricultural Model Intercomparison and Improvement Project for wheat (AgMIP-Wheat). Data include local daily weather data, soil profile characteristics and initial conditions, crop measurements at maturity (grain, stem, chaff and leaf dry weight, ear number and grain number, grain unit dry weight), and at stem elongation and anthesis (total above ground dry biomass, leaf number per stem and leaf area index). Several in-season measurements of the normalized difference vegetation index (NDVI) and the fraction of intercepted photosynthetically active radiation (FIPAR) are also available. The crop model simulations include both daily in-season and end-of-season results from twenty-nine wheat models.

The relative time to maturity of grain crops is an important consideration for producers, yet there are no universally accepted classification schemes for cultivar phenology to guide decisions on variety selection and time of sowing. A first edition of an industry guide for wheat variety maturity was recently developed for use across Australia, representing a significant step forward for the grains industry. The aim of this paper was to revise and extend this industry guide to make it more robust, agronomically functional and meaningful to industry. The Australian Cereal Phenology Classification (ACPC) presented herein was developed using an unprecedented phenological data set with a diverse array of genotypes, environments and management. Field experiments were carried out with 70 wheat and 30 barley cultivars at 15 sites across Australia between 2017 and 2020. Thermal time to anthesis data were used to rank cultivars according to their relative phenology and divide them into classes, and then boundary cultivars of both species were selected to separate these classes. The resulting classification scheme divides wheat and barley into phenology classes ranging from ‘quick’ to ‘mid’ to ‘slow’. New cultivars to market can be assigned a phenology classification based on their thermal time to anthesis relative to the boundary cultivars. The ACPC will help growers, agronomists, breeders and researchers make informed decisions regarding cultivar comparison and selection while reducing misclassification and confusion across regions. The same methodology used to derive and validate the ACPC can be applied internationally to standardise descriptions of crop phenology.

Increasing global food demand will require more food production without further exceeding the planetary boundaries, while at the same time adapting to climate change. We used an ensemble of wheat simulation models, with sink-source improved traits from the highest-yielding wheat genotypes to quantify potential yield gains and associated N requirements. This was explored for current and climate change scenarios across representative sites of major world wheat producing regions. The sink-source traits emerged as climate neutral with 16% yield increase with current N fertilizer applications under both current climate and mid-century climate change scenarios. To achieve the full yield potential, a 52% increase in global average yield under a mid-century RCP8.5 climate scenario, fertilizer use would need to increase fourfold over current use, which would unavoidably lead to higher environmental impacts from wheat production. Our results show the need to improve soil N availability and N use efficiency, along with yield potential.

Early-vigour is a plant trait phenotypically characterised by a rapid expansion of leaf area in early stages of crop growth, before canopy closure. Although early-vigour is already used as a selection criteria in breeding programmes for grain cereals, its genetic variability and potential benefits for oat cover crops is unknown. In this study, we screened 231 oat lines from a commercial forage breeding programme using high-throughput field phenotyping to quantify canopy development rates through aerial photographs. Results showed a wide genetic variability for early-vigour in oats, with canopy cover differences of up to ∼20% during the approximate 30 days period from emergence to full canopy cover. Two oat genotypes with contrasting canopy cover development rates (opposite quartile ranges of the population) were subsequently selected for a detailed investigation of underlying mechanisms explaining early-vigour during two field trials. For these genotypes, the size of individual leaves was found to be the main factor driving differences, with 50% larger leaf area before the sixth leaf from the base of the main tiller in the high early-vigour genotype. A process-based biophysical model (APSIM-NextGen oats) was then parameterised to quantify variability on potential early-vigour benefits across four representative agricultural target environments. This was done by simulating the two selected oat genotypes across 30-years of historical weather data from each of the four distinct climatic zones, considering two contrasting soil types and three possible cover crop sowing dates. Results showed early-vigour to provide overall positive ecosystems services with pooled medians of 7–18% increase in above ground biomass and 5–13% increase in nitrogen uptake which caused a consequent 4–9% reduction in nitrogen leaching losses depending on the location/soil/management combination. Such decline in relative plant-trait effects across different components of the production system (above-ground biomass, above-ground N and N leaching reduction) was also accompanied by increasing variability in responses, with pooled coefficients of variation around 31%, 69% and 80% respectively. Similarly, our results also highlight the relative dilution of trait effects across scales. For instance, a 50–70% difference in basal leaf size at the plant-organ scale caused a 6–10% potential reduction of N leaching at the agricultural system scale. These results highlight the importance of multi-metric evaluations of spatial and temporal variability in trait effects to better inform breeding and selection programmes. Finally, our practical implementation of previously conceptualised methods illustrates the increased depth of understanding about plant-trait benefits when combining interdisciplinary approaches such as high throughput phenotyping, classic crop physiology field experimentation and biophysical modelling. The principles of this approach can be extended to assess the relative value of other traits across different species, managements and environments.

Accurate assessment of plant development is essential for agronomic management and scientific research, and crop development scales with repeatable and reproducible protocols are required to achieve this. Development scales currently in use are ambiguous, subjective and qualitative, they do not describe all stages in a crop's lifecycle, they do not explicitly distinguish between culm-level and crop-level development, and they are incompatible with modern analytical and computational technologies. Here we propose two new scales of wheat and barley development: the Single Culm Development Scale (SCDS) to define progression through the lifecycle of an individual culm, and the Population of Culms Development Scale (PCDS) to identify the timing of stages and duration of phases within a crop canopy. These development scales merge and fill gaps within existing scales currently in use, describe development in terms that are unambiguous, objective and quantitative, and better interface with crop simulation models, automated image analysis and other computational tools and analytical methods. The SCDS and PCDS are paired with definitive protocols to measure each stage that were developed and tested using different operators in controlled environment and field experiments.

Two datasets from red clover monoculture pastures grown in Lincoln, New Zealand, were analysed to generate coefficients to predict red clover yield. The mean annual production of established red clover was 17.0±0.48 t DM/ha, with a maximum mean growth rate of 125±9.36 kg DM/ha/day (spring Year 2). In the establishment year irrigated red clover grew at a constant rate of 7.30±0.14 kg DM/ha/°Cd (Tb = 3 °C) throughout the year. In contrast, there was a splitline linear response in Years 2 and 3, which differed between years and decreased after the second week of January. Specifically, the growth rate in Phase 1 was 7.70±0.38 kg DM/ha/°Cd in Year 2, which was 16% higher than the 6.60±0.28 kg DM/ha/oCd in Year 3. The difference probably reflected increased competition from weed grasses as red clover content declined from >95% to ~75% of total annual yield. After January, red clover grew at 3.05±0.35 kg DM/ha/°Cd, in both years. This lower rate occurred in the mid-January-July period, and probably reflected a change in partitioning of assimilate to red clover roots in response to a decreasing photoperiod. The coefficients reported here for red clover need to be validated from other datasets. However, they provide easily transferable coefficients that can be used to estimate red clover yield under nonlimiting conditions for other locations. These could be integrated into feed budgeting software to assist onfarm decision making.

Context Wheat (Triticum aestivum L.) adaptation is highly dependent on crop lifecycle duration, particularly the time at which flowering occurs in a specific environment. Frost, low solar radiation, heat and drought can significantly reduce yield if a crop flowers too early or late. Wheat genotypes have different lifecycle durations determined by plant responses to temperature (thermal time accumulation and vernalisation) and photoperiod. These responses are largely controlled by five phenology genes (two PPD1 and three VRN1 genes). Advances in crop phenology modelling suggest that flowering time under field conditions could be accurately predicted with parameters derived from photoperiod and vernalisation responses obtained in controlled environments. Aims This study quantified photoperiod and vernalisation responses of 69 Australian wheat genotypes selected for diversity at the PPD1 and VRN1 loci. Methods Spring and winter genotypes were grown in four controlled environments at a constant temperature of 22°C with photoperiod (17 or 8 h) and vernalisation (0 or 8 weeks) treatments as factors. Key results Thermal time from coleoptile emergence to flowering in spring genotypes was typically decreased more by long photoperiod than by vernalisation; the opposite was true for winter genotypes. Spring genotypes that were sensitive to vernalisation contained a sensitive allele at the Vrn-A1 locus. Conclusions There is large diversity in phenological responses of wheat genotypes to photoperiod and vernalisation, including among those with matching multi-locus genotype. Implications Data from this study will be used to parameterise and test a wheat phenology model in a future study.

Lucerne (Medicago sativa L.) is a widely grown perennial legume worldwide which can provide high biomass and protein yields, biological N fixation, deep soil water extraction and a range of ecosystems services relevant to current and future agricultural systems. The potential to expand lucerne beyond its current cultivated areas in New Zealand, and its potential productivity across the country's contrasting climate zones, are currently un-known. To gain such insights, we estimated land suitability and spatial distribution of lucerne above-ground biomass across New Zealand lands considering contrasting growth conditions (rain-fed or irrigated for different soils types) and two simulation methods of different complexity (process-and GIS-based approaches). This aimed to assess yield-estimate spatial patterns and sensitivity to model selection for a wide range of com-binations of water supply (i.e. irrigation and soil water storage) across New Zealand climate zones. For example, highly suitable areas for lucerne cultivation, were estimated in-21 thousand km2 when considering the exclusion of steep slopes, poor soil drainage and excess annual rainfall. The two crop-yield models were applied in response to 30 years of daily historical (1971-2000) weather data downscaled at 5 km resolution on suitable areas. Simulated average lucerne yields ranged from -4.5-28 t dry matter/ha per year. Simulations showed a distinct spatial pattern of yield decline from north to south, mainly in response to decreasing temperatures. Temporally, water limited yields were up to 4-fold more variable than under irrigation, depending on the degree of drought stress across different years. Results also unveiled systematic spatial patterns of model uncertainty quantified as yield sensitivity to model selection. For instance, simulated yields were most sensitive to model selection (6-31% of total variability, Ti) within high abiotic-stress environments (e.g. low temperature and limited water supply). Overall, soil type selection accounted for most of yield variability (58-78% Ti), being particularly important in warmer environments with variable seasonal rainfall regimes (e.g. northern regions). As expected, water supply (i.e. rain-fed or irrigated systems) was relatively more impactful on yield (8-20% Ti) for limited rainfall areas, where crops are most drought prone (e.g. east coast and central southern regions). Long-term regional scale comparisons of annual lucerne yield, between 30-year simulated distributions and point-based observations from the AgYields database, helped identify hotspots of yield overestimation. Such insights are useful to guide future research on high yield gap areas (e.g. southern colder and drier locations) and highlight key areas for model improvement (e.g. representation of multiple biotic stresses). Overall, our results provide a first gridded-model assessment of lucerne suitability and yield at national scale and quantify the share of variability explained by key climatic, management and methodological components in spatial analysis studies. These insights can inform future modelling efforts and support agricultural planning that considers the expansion of lucerne and other perennial legumes.

Modelling lucerne growth requires response functions that represent seasonal partitioning of biomass into above-ground and below-ground organs. An additional challenge is to parameterize perennial organ responses across contrasting fall dormancy (FD) genotypes. Current models use empirical approaches to simulate biomass accu-mulation and partitioning. This research integrated knowledge of lucerne biomass accumulation and partitioning into the Agricultural Production Systems sIMulator (APSIM) next generation (APSIM NextGen) model frame-work. Biomass supply was calculated from light interception and total radiation use efficiency (RUEtotal), and then allocated based on the relative demand of each organ. Leaf biomass demand was parameterized as a function of specific leaf area (SLA). Stem biomass demand was parameterized as a positive power function of shoot biomass. Root biomass (taproots and crowns) showed a strong seasonal pattern. The observed decrease of root biomass in periods of an increasing photoperiod (mid-winter to mid-summer) was assumed as remobiliza-tion to shoots and carbon loss from maintenance respiration. Periods of decreasing photoperiod showed increased biomass of root caused by greater carbon partitioning to this organ. To capture this, a model opti-mization approach was used to fit required parameters. Fitted parameters included a remobilization coefficient (percentage of storage biomass per day) of 0.01common to all FD cultivars tested (FD5, FD2 and FD10). The regrowth coefficient (remobilization duration) remained constant at 0.01 post-defoliation until 250 degrees Cd for FD5, 200 degrees Cd for FD2 and 300 degrees Cd FD 10, and then declined to 0 after another 50 degrees Cd. The model was parameterized to have maximal root demand in a decreasing photoperiod to capture carbon partitioning. The model had good prediction of shoot biomass (NSE=0.70) and fair prediction (NSE=0.60) of root biomass for 42 day defoliation treatments. It was less accurate for predictions of shoot biomass under a frequent (28 day) defoliation regime. This highlights the importance to include the response to limitations caused by depleted root N reserves in future model versions. The APSIM NextGen lucerne model provided a mechanistic framework to model perennial organ biomass dynamics with structural and storage components, root maintenance respiration, remobilization in spring, partitioning in autumn and the regrowth effect. This framework accounted for differences in fall dormancy of genotypes and provided a methodology that can be integrated into models of other perennial crops.

Lucerne (Medicago sativa L.) canopy expansion, as quantified by leaf area index (LAI), is the crop process that determines the amount of intercepted total radiation during each regrowth cycle. A challenge is to capture seasonal changes of canopy expansion rate in response to the environment. This research integrates parameters and functions of lucerne canopy expansion into the Agricultural Production Systems sIMulator (APSIM) next generation (APSIM NextGen) model (LeafArea module) to simulate canopy expansion and light interception. Over 20 years of detailed field experimental datasets, with multiple treatments, from Lincoln University were used for model development. Functions derived from a fall dormancy (FD) 5 rated genotype were grown under an industry standard defoliation treatment to parametrize the model. These functions were tested further using genotypes with an FD2 or FD10 rating under longer and shorter defoliation regimes, all under irrigated conditions. The APSIM NextGen lucerne model predicted the LAI expansion pattern in each growth cycle as a double sigmoid curve requiring functions that define the lag phase, basal bud initiation, the linear leaf area expansion rate (LAER; m2 m-2 degrees Cd), and canopy senescence which represents the loss of LAI over time. LAI was well predicted for experiments under the standard (42-day) and long (84 day) defoliation treatments for FD5, with NashSutcliffe efficiency (NSE) of 0.61 and 0.55. However, the derived parameters and functions overestimated LAI under an extreme short defoliation treatment (28-day), NSE values ranged from 0.38 to 0.78. LAER was lower for the short-defoliation intervals (28-day), probably due to a depletion of carbon and nitrogen reverses in perennial organs. For FD2 and FD10, different LAER functions were generated from field observed data and used to improve simulation agreement. There was fair agreement for the 84-day treatment (NSE of 0.32) and the 42-day treatment (NSE of 0.38), but poor agreement for the 28-day treatment for FD10 (NSE = -0.88). The estimated extinction coefficient (k) was the same for seedling and regrowth crops, and consistent across defoliation treatments and FD classes. With the LeafArea module and k value, the APSIM NextGen lucerne model can now estimate daily LAI and intercepted radiation. Future model development includes validating the LeafArea module in different environments. However, a more mechanistic model approach is required to link canopy expansion to carbon and nitrogen reserves in lucerne plants that experience intense defoliation.

Crop multi-model ensembles (MME) have proven to be effective in increasing the accuracy of simulations in modelling experiments. However, the ability of MME to capture crop responses to changes in sowing dates and densities has not yet been investigated. These management interventions are some of the main levers for adapting cropping systems to climate change. Here, we explore the performance of a MME of 29 wheat crop models to predict the effect of changing sowing dates and rates on yield and yield components, on two sites located in a high-yielding environment in New Zealand. The experiment was conducted for 6 years and provided 50 combinations of sowing date, sowing density and growing season. We show that the MME simulates seasonal growth of wheat well under standard sowing conditions, but fails under early sowing and high sowing rates. The comparison between observed and simulated in-season fraction of intercepted photosynthetically active radiation (FIPAR) for early sown wheat shows that the MME does not capture the decrease of crop above ground biomass during winter months due to senescence. Models need to better account for tiller competition for light, nutrients, and water during vegetative growth, and early tiller senescence and tiller mortality, which are exacerbated by early sowing, high sowing densities, and warmer winter temperatures.