The global demand for food is projected to increase in the coming decades, driven by population growth, climate change, pandemics, shifts in food consumption and biofuel use (Tilman et al., 2011; Godfray and Robinson, 2015; van Dijk et al., 2021). Ensuring that crop production is sufficient to meet future goals is a challenge for plant and agricultural sciences.

In Africa, agricultural crops provide food and income for smallholder farmers and consumers. Despite the huge agricultural potential, agricultural productivity in African countries continues to remain the lowest in the world (Bjornlund et al., 2020). Many studies have indicated that yields of several important staple crops may be stagnating or even declining across the continent (Roudier et al., 2011; Knox et al., 2012; Ray et al., 2012; Parkes et al., 2018). This is the case for key staple crops such as maize, rice, wheat, millet, sorghum, cowpea, cassava and yam, which together account for a large portion of the population’s diet. Therefore, food supply systems would be negatively affected if yield gains in these crops continue to slow due to environmental stresses and production constraints.

Addressing food security in Africa is a vast challenge that needs to be tackled in many complementary directions. Infrastructure development adapted to local needs, good farming practices, management, and political will are some of the major axes of development for food security. Improving crop performance and tolerance/resistance to biotic and abiotic conditions is the challenge facing the scientific community and innovative methods are needed.

Advanced field phenotyping, e.g. using digital approaches, has developed substantially over the past decade and provides means for real-time monitoring of response to environmental stresses and nutrition, and aids unravelling the relationships between yield and complex genotypic traits. The identification of genotypes with superior traits of agricultural interest remains one of the major targets for the genetic improvement of crops (Varshney et al., 2021).

The genomes of many agricultural crops such as rice (Matsumoto et al., 2005), sorghum (Paterson et al., 2009), maize (Schnable et al., 2009), soybean (Schmutz et al., 2010) and recently wheat (Appels et al., 2018) have been sequenced. However, the advances made in genomic approaches such as maker-assisted selection and high-throughput sequencing (Crossa et al., 2017; Scheben et al., 2018) are yet to be complemented with accurate field phenotyping methods (Minervini et al., 2015). Most of the traits of agronomic relevance (e.g., yield) are complex, and quantitative, requiring tools for their phenotypic assessment in the field (Reynolds et al., 2020). Furthermore, open field rather than controlled environment measurements are more likely to be useful in identifying genotypes that will perform better in farming practice, especially when large plots that mimic real farm conditions (i.e., environmental and management conditions) are employed (Rebetzke et al., 2014).

In addition, precision agriculture (PA) is becoming increasingly important in today’s technologically advanced world (Langemeier and Boehlje, 2021; Gobezie and Biswas, 2023) and PA remains one of the cardinal principles of field phenotyping. The PA farming management concept relies on modern digital techniques to monitor and optimize agricultural production processes to improve crop performance (Hedley, 2015; Gokool et al., 2023). Despite PA’s contributions to sustainable agriculture, its use in resource-constrained smallholder farming environments, particularly in Sub-Saharan Africa (SSA), has been very limited (Gobezie and Biswas, 2023). Recent developments in sensor technologies, machine vision, and higher-resolution digital cameras, in tandem with advanced data processing power and other portable tools have paved the way for high-throughput plant phenotyping in the field to benefit crop breeding programs (Deery et al., 2014; Zhang et al., 2016; Araus et al., 2022; Ahmed et al., 2023). From the field phenotyping perspective, these emerging technologies are enabling automated intensive data collection and increasing the ability to investigate plant function and structure through non-invasive methods with high accuracy. Such field phenotyping methods will aid crop improvement efforts to meet the expected demand for food and agricultural products in the future.

The development and application of these high-throughput tools for field phenotyping are currently focused on the main staple crops grown in the most developed agricultural regions. Over the decades, breeders and agronomists in Africa have used traditional phenotyping based on manual methods either for selecting traits or for improving yields through changes in agronomic practices (Iizumi and Sakai, 2020). However, traditional phenotyping in breeding is time-consuming, laborious and data collection is insufficient to fulfil the needs of plant breeders which impedes breeding progress. Therefore, further advances in phenotyping methods and appropriate implementation are required to increase the effectiveness of selection in breeding programs, speed up genetic gains, reduce costs and enable monitoring of plant status more efficiently than is currently feasible. The sophistication and cost of current plant phenotyping equipment (Reynolds et al., 2019) have restricted them from being widely applied in the developing world and especially in Africa. Additionally, insufficient technical, operational, regulatory restrictions and conceptual capacity in the plant science community have further limited implementation. Therefore, it is timely to begin to apply these technologies more widely, both geographically and with respect to target crops in Africa. Affordable high-throughput phenotyping aims to achieve reasonably priced solutions for all the components comprising the phenotyping pipeline which will promote their adoption for the breeding of African crops (Whalen and Yuhas, 2019; Bongomin et al., 2022).

Few studies have covered the use of modern field phenotyping approaches employing remote sensing in Africa (e.g., Mutanga et al., 2016; Chivasa et al., 2017; Buchaillot et al., 2019; Bongomin et al., 2022; Kassim et al., 2022). For instance, Bongomin et al. (2022) recently reviewed the status of field phenotyping in Uganda with focus on the application of drones and image analytics.

In this review, we provide a background on African agriculture and cover the concept of digital field phenotyping, focused on traits that may be measured by emerging technologies and which could be applicable to African crops. The current developments of field phenotyping in Africa, including initiatives, implementation challenges and prospects are comprehensively reviewed. We observed that the lack of suitable field phenotyping infrastructures and approaches using digital technologies is limiting the development of improved crop cultivars and will negatively affect the agricultural industry and food security in Africa. We emphasize the potential for incorporating cutting-edge, low-cost phenotyping tools (i.e., portable field sensors, UAVs) into breeding schemes and for identifying agricultural crop responses to environmental constraints through field experimentation. Finally, we consider policy directions for tackling the implementation challenges (i.e., practical, financial, geographical and political) of digital field phenotyping and realizing the full potential of available field phenotyping resources (i.e., technologies, tools and know-how) appropriate for African crops.

African countries are important producers of major crops with diverse agro-climatic and ecological conditions, and cultural diversity (Leakey et al., 2022). Sub-Saharan West Africa is composed of a wide variety of ecosystems and an equally high number of production systems (https://www.fao.org/3/AC349E/ac349e04.htm). Generally, crop production is concentrated in areas with a favourable combination of agro-bioclimatic conditions. In the Sahelian zone, cereals such as millet and sorghum are the predominant crops with annual rainfall (200-600 mm), transitioning to maize, groundnuts and cowpeas farther south in the Sudanian savannah zone (the so-called “Middle Belt”). These food crops are among the top five harvested crops in the Sahelian countries – Burkina Faso, Senegal, Mauritania, Mali, Chad and Niger. According to FAOSTAT (2018a) data, maize is the major essential staple food in sub-Saharan Africa, accounting for nearly 20% of total calorie intake. The same source indicates that in Sub-Saharan West Africa, millet and sorghum account for roughly 64% of total cereal production. Across the rainy forests of the Guinean zone (1200-2200 mm of rainfall per year) crops are predominantly root and tuber crops such as cassava and yams which are mostly cultivated in Ghana, Nigeria, Côte d’Ivoire and Sierra Leone. Yam is the second most important crop in Africa in terms of production after cassava (FAOSTAT, 2018a). Rice, on the other hand, is one of the most widely harvested crops in this humid zone, ranking first in Guinea, Liberia and Sierra Leone in terms of area harvested (Soullier et al., 2020; Duvallet et al., 2021).

Crop production in West Africa is mostly rainfed and crop production is vulnerable to climate change, which manifests itself in unpredictably high temperatures and erratic rainfall patterns (Sultan and Gaetani, 2016; Affoh et al., 2022). The five principal crops in West Africa in terms of harvested area (in millions of hectares per year on average in the last decade) are cassava (81), maize (19), millet (10), sorghum (12), yam (57) (FAOSTAT, 2022). Major cash crops are cocoa, coffee and cotton. Declining soil fertility and unpredictable climate change impacts (among other factors) have made it difficult to maintain the yields of these major crops (Shimeles et al., 2018). Over the last three decades, the agricultural sector in West Africa has been characterized by strong production growth in some major staple crops culminating in increased production volumes for both domestic and export markets (Blein et al., 2008; FAO, 2015). Similarly to West Africa, Central Africa’s principal food crops include cassava, peanuts, sorghum, millet, maize, sesame and plantains. Additionally major cash crops for export include cotton, coffee and tobacco (Ochieng et al., 2020).

In Northern Africa, particularly Morocco, crop production is regionally diverse owing to different climatic conditions, agro-ecological zones, land-crop tenure and farming systems (Ouraich and Tyner, 2018). This geographical diversity results in varied agriculture, with crops ranging from cereals and vegetables to fruits and nuts, grains, legumes, etc., that contribute significantly towards the country’s agricultural sustainability and food security. Cereal production accounts for 65% of cultivable agricultural areas (Ouraich and Tyner, 2018). Most cereal production occurs under rainfed conditions. As a result, productivity performance is influenced by precipitation levels. For instance, 7.3 million tonnes of wheat were produced in 2018 making it the 20th largest producer in the world and 2.8 million tonnes of barley being the 15th largest producer in the world (FAOSTAT, 2018b). However, drought is a persistent threat to crop production especially the lowlands where cereals are grown are particularly at risk because of the wide variations in annual precipitation (Verner et al., 2018; Meliho et al., 2020). In recent years, quinoa has sparked particular attention in Morocco (Choukr-Allah et al., 2016; Hirich et al., 2021). It remains one of the most nutrient-dense crops and is recognized as a ‘Superf Food’ due to its nutritional benefits. Thus, Morocco is one of the few North African countries capable of achieving self-sufficiency in food production (Saidi and Diouri, 2017).

Grains and cereals (e.g., maize, wheat, barley, oats and sorghum) are South Africa’s most important crops occupying more than 60% of the acreage under cultivation (FAO, 2022). Together, these crops account for one of the largest agricultural industries contributing more than 30% to the total gross value of agricultural production (FAO, 2022). Maize, the country’s most important crop and largest locally produced field crop, is a dietary staple supplying most of the carbohydrate needs, a source of livestock feed and is an export crop (Epule et al., 2022).

The country has emerged as the largest maize producer and exporter in the Southern African Development Community (SADC) region and Africa as a whole (Fisher et al., 2015; FAO, 2022). According to the FAO, 2022, in 2021 South Africa produced 17 million metric tonnes of maize, making it the 9th largest producer in the world. Moreover, it produced 2.6 million metric tonnes of potato and 2.3 million metric tonnes of wheat. Largely, South Africa has a semi-arid climate characterized by summer and winter rainfall seasons. Unpredictable weather conditions due to climate change have a severe impact on maize and wheat production which accounts for more than 36% of the total value of field crops (Bradshaw et al., 2022).

Smallholder farmers dominate agriculture in East African countries, contributing up to 90% of total agricultural production (Salami et al., 2010; Livingston et al., 2011). A cereal‐legume mixed cropping pattern is the dominant system that includes maize, millet, sorghum and wheat (Van Duivenbooden et al., 2000). Over 40% of the region is covered by the maize mixed cropping system, which is followed by the pastoral (14%), root crop (12%) and cereal-root crop mixed system (11%) (Adhikari et al., 2015). Teff is a significant crop in the Ethiopian highlands, while other significant crops in the area include cassava, bananas and rice. The mixed cropping system in East Africa is based on millet in the drier regions and on maize and cassava in the humid regions (Adhikari et al., 2015). The main cash crops in most of the East African countries in SSA are coffee, tea, cotton, tobacco and sugarcane. Rainfall variability negatively impact on crop production in East African countries (Palmer et al., 2023). Generally, the major challenges to crop production in Africa are unproductive soils, pests and diseases, drought, and poor crop management (Tadele, 2017). The distribution of major crops in each sub-region except Northern Africa is summarized in Figure 1 .

Experiments with repeated trials in diverse environments are often necessary to screen plants for desirable traits. This becomes problematic when there is the need to screen a large panel of genotypes for valuable traits (i.e., yield potential or abiotic and biotic stress tolerance) to assess genotype, environment, and management (G × E× M) interactions (Araus and Cairns, 2014). Over the years, the measurement of individual plants in controlled conditions has dominated most of the phenotyping research. However, controlled environments often do not accurately mimic plant growth and development in field conditions (White et al., 2012). Field phenotyping is becoming more widely recognized as the approach that gives the most accurate representation of traits in real-world cropping systems (Tariq et al., 2020). Thus, field phenotyping is an important component of crop improvement to assess how the genotype, the environment, and their interaction (G × E) influence quantitative traits in a complex and dynamic manner (Fiorani and Schurr, 2013; Araus and Cairns, 2014; Neilson et al., 2015). Furthermore, field phenotyping is employed to discover novel traits, identify new germplasm carrying relevant but complex traits for breeding, and for testing proof of concept to validate traits (Watt et al., 2020). Traditionally, destructive sampling has been used to quantify certain observable plant traits, including laboratory analysis to characterize phenotypes based on their genetic and physiological functions. Digital phenotyping approaches seek to reduce this need (Tripodi et al., 2022; Virlet et al., 2022).

Different measurement approaches including novel technologies such as non-invasive imaging, robotics and sensor positioning systems have been incorporated in well-designed field phenotyping installations for high-throughput phenotyping (e.g., Araus and Cairns, 2014; Kirchgessner et al., 2017; Shakoor et al., 2017; Virlet et al., 2017; Pieruschka and Schurr, 2022). These significant strides in field phenotyping have fostered a major international collaborative effort directed toward data and protocol standardization (Pieruschka and Schurr, 2019; Lorence and Jimenez, 2022). The appeal of these platforms is the increased throughput and objectivity in data collection compared to traditional field approaches.

Non-invasive portable devices, ground-wheeled, motorized gantry scanalyzer systems, agricultural robots and aerial vehicles that deploy a wide range of cameras and sensors, together with high-performance computing are currently required to conduct field phenotyping in a timely and economical manner ( Figure 2 ). Together, these platforms are able to phenotype plant characteristics throughout the season in field environments (White et al., 2012; Fritsche-Neto and Borém, 2015; Jimenez-Berni et al., 2018; Furbank et al., 2019; Li et al., 2021).

In recent years, manned and unmanned aerial vehicle (UAV) remote sensing platforms have emerged as convenient high-throughput tools for field phenotyping (Pajares, 2015; Shi et al., 2016; Feng et al., 2021). These remote sensing approaches, particularly UAVs enable quick and non-destructive high throughput phenotyping, with the benefit of adaptable and convenient operation (Yang et al., 2017a). These phenotyping platforms can combine multiple sensors such as digital cameras, infrared thermal imagers, light detection and ranging (LiDAR), multispectral cameras and hyperspectral sensors for various assessments of morphological and physiological plant traits (Gonzalez-Dugo et al., 2015; Yang et al., 2017a; Camino et al., 2018; Roitsch et al., 2019).

Alternatively, field phenotyping can be accomplished on the ground utilizing a fully automated fixed-site phenotyping platform (e.g., Kirchgessner et al., 2017; Virlet et al., 2017; Bai et al., 2019), hand-held sensors, portable spectroradiometers, hand-pushed carts or high-clearance tractors carrying multiple high-resolution sensors to measure phenotypic features non-destructively (Comar et al., 2012; Andrade-Sanchez et al., 2014; Crain et al., 2016). The use of rapid non-invasive portable devices that carry sensors for crop status monitoring has advanced field data collection due to their applicability and ease of operation (Parks et al., 2012; Yang et al., 2014; Condorelli et al., 2018). Recently, field phenotyping has become more flexible by integrating ground-based and aerial platforms (Potgieter et al., 2018; Furbank et al., 2019; Ninomiya, 2022). Table 1 summarizes the diverse ground-based and aerial field phenotyping platforms, their applications, advantages, and limitations.

For field phenotyping, traits that have been evaluated by sensors in the field have been reviewed recently by Watt et al. (2020) and include for example; (a) plant morphological development (i.e., including seed establishment and growth of the crop, the timing, and dynamics of flower and fruit development); (b) functional traits that are related to the photosynthetic capacity and carbon uptake during the phenological growth phase; (c) traits related to biotic and abiotic stress resistance/tolerance; (d) traits that determine crop water status (e.g., water uptake and transpiration and water-use efficiency) of plants; (e) yield-related traits and harvest quality of crops (i.e., biomass yield) and (f) the structural and functional root traits (i.e., root architecture). These traits have been previously classified into morphometric and physiological parameters (Qiu et al., 2018). Traits such as plant height, stem diameter, leaf area or leaf area index, leaf angle, stalk length and in-plant space are morphometric parameters. Physiological parameters include traits such as photosynthetic rate, chlorophyll content, water stress, leaf water content, biomass, and salt resistance, which together can impact plant growth. It should be emphasized that different phenotypic traits have specific time frames within the phenological cycle of the plant when they are relevant for the breeder and farmer. Currently, the most researched crops in field phenotyping are economic crops, such as wheat, maize, barley, sorghum, tomato, bean and grape because they have significant economic value for agricultural development. A challenge is to extend phenotyping into the vast range of African crops, some of which may be of only local importance.

Field phenotyping makes use of a variety of sensors due to the large number of phenotypic traits that must be measured. Several conventional and novel sensors such as digital cameras, range cameras, depth cameras, spectral sensors, lidar or laser sensors, thermal sensors, fluorescence sensors, multispectral cameras, hyperspectral cameras and others are employed and integrated for plant trait measurement in field phenotyping (Qiu et al., 2018; Roitsch et al., 2019; Xie and Yang, 2020).

Since plants develop rapidly during their early growth stages, frequent measurements during their establishment are a prerequisite for the quantitative selection of vigour phenotypes. Drones fitted with conventional RGB (red-green-blue) cameras, in combination with advanced image processing pipelines, can automatically detect crop stands (single plants) and determine seed emergence, germination rates and timing under extreme climatic events in the field (Liu et al., 2017).

Most plants display strong morphological changes during their phenological development, which is greatly influenced by the availability of resources and changes in abiotic and biotic factors. Therefore, the development of robust, automated, and precise methods to measure morphological plant traits in field conditions is still required (Gibbs et al., 2017).

The leaf is one of the important components of a plant. It plays a major role in plant growth given that its growing status influences the efficiency of the direct solar energy utilization by plants. Hence, it is a significant parameter in plant phenotyping. Measurements of morphometric parameters of the leaf and other canopy features (i.e., leaf area, stem height, number of tillers, and inflorescence architecture) have been evaluated using non-destructive multi-sensor approaches (Busemeyer et al., 2013; Fiorani and Schurr, 2013; Rahaman et al., 2015). However, the most frequently used geometric measure of plant canopy is the green leaf area index (GLAI), which relates the one-sided green leaf area per unit projected ground area (Chen and Black, 1992). For instance, UAV multispectral imagery has been used to characterize GLAI dynamics of a large maize panel under contrasted environmental conditions and thus holds great potential for yield predictions in breeding programs (Blancon et al., 2019). LAI can also be evaluated, indicating plant coverage, from spectral images (Dammer et al., 2016; Schirrmann et al., 2016).

Plant canopy architecture and other morphological traits of plant organs have been measured concurrently with 3D proximal sensing techniques. A body of recent reviews has compared the performances of the most common 3D sensors for high-throughput plant phenotyping (Li et al., 2014; Qiu et al., 2019). The 3D acquisition devices and approaches commonly used are LiDAR time-of-flight cameras, mono, and multi-view stereo vision and structure-from-motion. The LiDAR sensors can scan and extract morphological traits of plant organs from 3D point clouds. For example, LiDAR was used to estimate plant height, ground cover and above-ground biomass in wheat (Jimenez-Berni et al., 2018). However, LiDAR sensors are expensive (Li et al., 2014), take significant time and there is a need to increase scanning time to increase the spatial resolution. Deploying a UAV-based system may reduce this challenge.

Plant height is a key indicator for canopy structure, yield, carbohydrate storage capacity and lodging occurrence (Holman et al., 2016; Hassan et al., 2019). Additionally, it has significant applications in predicting biomass, identifying plant cultivars, plant stress and phenological stages (Aasen et al., 2015). The traditional method of measuring height using a metre rule is labor-intensive, cumbersome and low throughput. In recent years, the development of drones and imaging sensors that capture high-resolution images has enabled high-throughput plant height estimation. For instance, Holman et al. (2016) estimated wheat height using UAV-based RGB images and terrestrial LiDAR.

Chlorophyll is a vital plant trait because it is strongly related to crop physiological status and may be indicative of photosynthetic rate, crop stress, nutrition status, yield, and plant productivity (Peng et al., 2011; Maimaitijiang et al., 2017). The most popular tools for evaluating vegetation health using visible and near-infrared light are spectral sensors. Chlorophyll meters such as the SPAD-502 are frequently used instruments to measure the relative chlorophyll content. Handheld chlorophyll meters and fluorescence meters have been used to assess plant nitrogen status, photosynthesis, yield and its components in crops (Yang et al., 2014; Andrianto et al., 2017; Fernández-Calleja et al., 2020). Additionally, chlorophyll can be measured using NDVI sensors and portable spectrometers in the field (Bai et al., 2016).

Crop nitrogen content can serve as a proxy for soil fertilizer availability, assisting farmers in precision nitrogen application to the soil. UAV-based hyperspectral imaging and ground-level optical sensors (SPAD-502, Duplex, and Multiplex) have been employed to estimate nitrogen fertilization status in maize (Quemada et al., 2014). In another study, Zaman-Allah et al. (2015) used a UAV equipped with a multispectral sensor (Green, Red, and NIR) to assess low nitrogen stress tolerance in corn. Additionally, vegetation indices (VIs) derived from spectral reflectance data captured by sensors devices such as the CropScan multispectral radiometer (Zhu et al., 2008), handheld spectroradiometers and the FieldSpec (Fitzgerald et al., 2006; Tilling et al., 2007; Feng et al., 2008), Tec5 (Erdle et al., 2013) can accurately measure nitrogen status in wheat and rice.

The above-ground biomass reflects light use efficiency and growth and is vital for carbon stock accumulation and monitoring (Swinfield et al., 2019). Brocks and Bareth (2018) estimated the biomass in barley using RGB images collected by UAV. Thermal infrared sensors are mostly used to detect crop water stress since they can provide temperature information for the crop (Park et al., 2017; Poblete et al., 2018; Bian et al., 2019). Thermal infrared sensors enable the estimation of canopy temperature which is a reflection of plant transpiration and plant water status. Kumar et al. (2020) used a proximal phenotyping cart (phenocart) mounted with low-cost consumer-grade digital cameras to characterize wheat germplasm for drought tolerance under field conditions. Plant yield has been considered an important agronomic trait for field phenotyping. Bascon et al. (2022) estimated rice yield using multispectral images.

The features of the sensors (e.g., spectral resolution, spatial resolution, specificity, and cost) should be considered according to the specific applications, phenotyping needs and context. In the African context, low-cost sensors and analysis pipelines which are not complex would benefit a broader user base for plant phenotypic trait assessments. The most successful trait assessment approach incorporates in time (throughout the crop cycle) and space (at the canopy level) the performance of the crop with respect to capturing resources (e.g., radiation, water and nutrients) and the efficiency of resource utilization (Araus et al., 2008). The aforementioned traits are discussed here with specific examples of sensors and automated measurement approaches used for their evaluation in the field (see Table 2 ). The advantages and limitations of each type of sensor are indicated.

Despite the recent advances in high-throughput field phenotyping based on the non-destructive analysis of plant traits, Africa has yet to consolidate the gains of these cutting-edge technologies for research into agricultural productivity. In terms of the deployment of high-end field phenotyping tools and approaches, Africa cannot keep pace with many regions, even in the era of artificial intelligence (AI), ‘internet-of-things’ (IoT) and technological advancements, although more affordable and lean phenotyping systems are now becoming available. Community-wide surveys and exchanges conducted by the International Plant Phenotyping Network (IPPN) and European Infrastructure for Multi-Scale Plant Phenomics and Simulation (EMPHASIS) within the growing phenotyping community in recent years have identified focus areas to assess the status of global plant phenotyping and crucial bottlenecks in the emerging field.

The major bottlenecks for developing field phenotyping in Africa were non-invasive phenotyping approaches, data management and cost among others (IPPN, 2016; Rosenqvist et al., 2019). This survey further reveals that in terms of using high-intensity field approaches (e.g., automation, robotics, image analysis and data storage management) for field phenotyping, Africa ranks lowest around the world. A recent survey conducted in the framework of the IPPN and EMPHASIS projects in 2020 (IPPN, 2020) which is reported by Yang et al. (2020) and Fahrner et al. (2021), indicated that Africa is still behind in the implementation of high-throughput field phenotyping. This highlights the need for a broader deployment of high-throughput field phenotyping techniques, which are essential enablers or resources for agricultural sciences and breeding to address upcoming crop production challenges.

The IPPN over the years has been promoting the idea of strengthening modern plant phenotyping in African countries by giving travel grants to Africa and inviting students and researchers for International Plant Phenotyping symposia and internships. However, only a few institutional members are identified for collaboration in the region. In recent times, there have been some high-throughput field phenotyping research and initiatives in African countries such as South Africa, Ghana, Senegal, Morocco, Nigeria, Ethiopia, Kenya, Egypt, and Zimbabwe which is encouraging for the emerging field and will be highlighted in this review (see section 5.2 and Table 3 ).

Like in many developing countries, field phenotyping in African countries is mostly based on conventional and traditional methodologies which rely heavily on manually recorded measurements of phenotypic data or visual assessment of plant parameters. It entails manually inspecting crops and measuring several crop characteristics that affect yield traits, including plant height, number of tillers, leaf color, leaf shape, leaf area index (LAI), chlorophyll content, growth stages, above-ground biomass and stress tolerance (Gedil and Menkir, 2019; Bongomin et al., 2022; Badu-Apraku et al., 2023). In practice, in traditional field phenotyping, breeders or research evaluators inspect the trial fields and rate the plots according to how they feel, taste, smell, and appear (Kim, 2020). Such phenotyping methods have several disadvantages such as being low-throughput, time-consuming, laborious, expensive and error-prone (Chapu et al., 2022; Xiao et al., 2022). Although these methods have been beneficial in developing new crop cultivars and improved yields, it is crucial that more effective phenotyping methods be used to increase the accuracy of data collection.

In parallel, field phenotyping is undertaken to evaluate the agronomic performance of crops in breeding programs, germplasm collections and in biotechnology programs to deliver improved cultivars that can cope with environmental stresses (e.g., Asare-Bediako et al., 2019; Gedil and Menkir, 2019; Rezende et al., 2020; Kavhiza et al., 2022). These phenotyping research targets are focused on key crops for food security but are predominantly low-throughput phenotyping based on field trials. In sub-Saharan Africa, breeding programs championed by the Alliance for a Green Revolution in Africa (AGRA) have been dedicated to priority crops such as rice, maize, cassava, yam, beans, cowpea and vegetables under various regional breeding networks for improved varieties and seed systems (FAO, 2011; AGRA, 2019).

Previous studies have used a variety of calibration data, including ground-based survey methods and crop model simulations, to predict yield in smallholder systems (Burke and Lobell, 2017; Ogutu et al., 2018). However, there has been emerging evidence in SSA suggesting inaccurate farmer-reported crop production estimates in smallholder production systems (World Bank, 2010; Gourlay et al., 2017; Abay et al., 2019; Wahab, 2020). These anomalies in crop yield estimation at smallholder, country and regional levels can cause price fluctuations (i.e., inflation), wrong national policy decisions and food insecurity among others. High-throughput and/or digital phenotyping might offer a better estimation of regional and national crop production.

Recent advances in sensor technology and the availability of free high-resolution (spatial and temporal) multispectral satellite images have also presented an opportunity to predict the yield of maize (Chivasa et al., 2017) and detect leaf spot diseases in groundnut (Sie et al., 2022), adaptation responses to early drought stress in sorghum (Gano et al., 2021) as well as mapping spatial distribution on a near real-time basis for a region, which hitherto was not feasible.

The application of emerging field phenotyping technologies has the potential to accelerate plant breeding efforts and crop production in Africa. On the other hand, most of these approaches reviewed here are at best relatively new or unknown to some of the plant science community in Africa. Field phenotyping is a critical component of crop improvement but remains a major bottleneck in African agriculture, as is the case globally. Some of the key challenges limiting the application of high-throughput field phenotyping in Africa are highlighted below.

This review provides an overview of high-throughput field phenotyping and its implications for African crops. It highlights the prospects of emerging high-throughput phenotyping techniques and integrated sensor platforms for plant trait assessment for field phenotyping that could apply to African crops. High-throughput field phenotyping has superior advantages that facilitate quick, non-destructive, and high-throughput detection, thus overcoming the shortcomings of conventional approaches. The readiness and the potential adoption of high-throughput field phenotyping for practical implementation in Africa are of paramount interest and should be demonstrated.

Field phenotyping solutions of immediate to long-term feasibility for African crops will likely rely on a combination of available techniques or prototypes of low-cost sensors and imaging approaches to study crop performance. Manual methods dominate the field phenotyping ecosystem with only a few countries beginning to explore high-throughput techniques through digital and precision agriculture. Notably, high-throughput phenotyping cannot yet completely replace manual measurements but should be promoted. The implementation of high-throughput phenotyping in general, and low-cost methods for field evaluation, is still fraught with challenges in Africa. Challenges identified by this present review include the high upfront cost of the prototypical platforms, huge funding gap, lack of conceptual and technical capacity, lack of technology transfer infrastructure and methodological approaches, lack of phenotyping network on the continent and the needed legislation in some cases, amongst others.

Lack of financial resources, a problem pervasive in African countries needs to be tackled holistically. Public-private partnerships could support resolving these financial and investment challenges to foster political will. Although in some countries, this public-private drive is already being implemented through close collaboration between universities and agricultural research organizations, these efforts need to be stepped up. In parallel, African governments should dedicate enough funding, incentives, and tools to breeders to advance research and innovations regarding high-end plant breeding. We suggest that donor support to Africa for agriculture and food security research should also consider projects in modern plant phenotyping to cope with current and projected climate change.

This will open the possibility of investing more in current sensor and imaging technologies for field data collection and the use of cost-effective phenotyping technologies that are already available to increase the throughput, quantity and quality of phenotypic data. The wide range of applications for these phenotyping technologies makes them good candidates and feasible choices for adoption in Africa which hitherto were prohibitive in terms of cost and deployment. The advantages of improved sensor-platform integration have facilitated the development of complete phenotyping systems that can gather, integrate and store data for many subsystems concurrently in a structured, efficient and cost-effective way. Such platforms have been widely adopted by research groups in developed countries and are gradually adopted by plant breeders in Africa as the technology develops and the benefits are proven.

In addition to the adoption of high throughput field phenotyping approaches in African countries, precision agriculture will also greatly benefit and revitalize the establishment of closer interaction between breeders and farmers to develop protocols mutually for the optimal use of improved crop varieties. The tools and knowledge exchange are expected to spur a second green revolution to meet the agricultural challenges to feed the ever-increasing African population. In terms of advancing field crop phenotyping in Africa for agricultural crop sustainability, we propose that breeding priority should be given to the most important staple crops such as maize, wheat, yam, cassava, cowpea, sorghum, etc. These crops form the backbone for food security and hence their improvement is crucial in the wake of prevailing climate change and production constraints. We suggest that each country selects traits that are of high demand and relevance by farmers and consumers when designing breeding strategies. In parallel, high-throughput phenotyping should be incorporated into national agricultural research policies and prioritize the practical implementation of field phenotyping. By and large, these could be achieved when governmental and private sector participation, as well as financial support, is readily available.

To overcome the challenges with the deployment of phenotyping tools and the integration of software to deliver accurate data acquisition, processing, analysis and management, a multidisciplinary team of expert-level skills and competencies may be required. This will necessitate deliberate training and capacity improvement of African plant scientists and students in software engineering and computer science domains, including AI, demanding true interdisciplinary partnerships to provide meaningful results and inform decision-making, while addressing the issue of training cost and related risks. In this instance, we recommend technological adoption rather than complete technological development considering the financial constraints and the low-level expertise in software and equipment development. However, as the plant phenotyping industry develops the development of new technologies from scratch may be feasible in Africa.

Furthermore, we propose encouraging collaborations between the African plant science community with their international counterparts to foster collaborative research, effective technological transfer and adoption. This review recommends close collaboration with the IPPN and similar phenotyping networks to benefit from the unprecedented investments made in field phenotyping infrastructures globally. Consequently, crop scientists may leverage ground-breaking advancements in high-throughput field data collection, image analysis and data management. Efforts should be made to foster synergies among different African countries by establishing transnational interdisciplinary networks that incorporate expertise in all aspects of plant breeding.

To address the limited investments in science, technology and innovation (STI), a commitment for expanded and long-term funding of agricultural research and development is essential. At the policy and operational levels, barriers must be overcome to allow the smooth establishment of public-private partnerships for transformational change in research and demand-driven technologies for breeders and farmers. There is renewed interest both from private and public institutions in developed countries to support African agriculture. Hence, African agricultural institutions need to develop strategies and synergies that include building partnerships that must be implemented to tackle the challenges, especially in the face of climate change and food insecurity.

The widespread adoption of high-throughput field phenotyping techniques in African countries could only be made possible in plant breeding programs if it can be proven as something worthwhile in terms of genetic gains attained with resources invested. Hence, costs must be reasoned in relation to the precision, repeatability, heritability, cost per unit plot or trait, prevailing climatic and economic condition, etc., required in a particular phenotyping activity. Given what has been said, to ensure that such implementation of field phenotyping can be translated into yield gains, low-cost phenotyping tools must be adopted. On this basis, affordable, easy-to-handle, reliable tools, and phenotyping infrastructures for small to large-scale field phenotyping may become a strategic choice and pave the way for practical implementation. Such technologies applicable to phenotyping methodologies should be available soon due to the high demands and efforts by the phenotyping community in Africa.

DC, NV, TW, FM, and MH provided the conceptualization of the manuscript. DC drafted, wrote, and edited the manuscript. NV, MC, AR, MM, TW, FM, and MH substantially reviewed the manuscript. TW, FM, and MH provided supervision. All authors read and approved the final manuscript. All authors contributed to the article.