Combining novel monitoring tools and precision application ...

Review Received: 9 November 2009

Revised: 5 January 2011

Accepted: 5 January 2011

Published online in Wiley Online Library: 28 March 2011

(wileyonlinelibrary.com) DOI 10.1002/ps.2134

Combining novel monitoring tools and precision application technologies for integrated high-tech crop protection in the future (a discussion document) Carolien Zijlstra,a∗ Ivar Lund,b Annemarie F Justesen,c Mogens Nicolaisen,c Peter Kryger Jensen,c Valeria Bianciotto,d Katalin Posta,e Raffaella Balestrini,d Anna Przetakiewicz,f Elzbieta Czemborf and Jan van de Zandea Abstract The possibility of combining novel monitoring techniques and precision spraying for crop protection in the future is discussed. A generic model for an innovative crop protection system has been used as a framework. This system will be able to monitor the entire cropping system and identify the presence of relevant pests, diseases and weeds online, and will be location specific. The system will offer prevention, monitoring, interpretation and action which will be performed in a continuous way. The monitoring is divided into several parts. Planting material, seeds and soil should be monitored for prevention purposes before the growing period to avoid, for example, the introduction of disease into the field and to ensure optimal growth conditions. Data from previous growing seasons, such as the location of weeds and previous diseases, should also be included. During the growing season, the crop will be monitored at a macroscale level until a location that needs special attention is identified. If relevant, this area will be monitored more intensively at a microscale level. A decision engine will analyse the data and offer advice on how to control the detected diseases, pests and weeds, using precision spray techniques or alternative measures. The goal is to provide tools that are able to produce high-quality products with the minimal use of conventional plant protection products. This review describes the technologies that can be used or that need further development in order to achieve this goal. c 2011 Society of Chemical Industry Keywords: crop protection; monitoring; detection; precision spraying

1

INTRODUCTION

616

Plant diseases, pests and weeds are major problems in crop production, where they lead to yield and quality losses. Attention to the pesticides used to control these problems has been increased, as they can have detrimental effects on the environment and human health. Therefore, an optimal utilisation that would lead to a reduction in the use of pesticides is desirable but requires detailed knowledge of the occurrence and distribution of diseases, pests and weeds in the field at the earliest possible stage of development. At present, this would involve a time-consuming manual inspection of the fields and skilled personnel, as it is difficult or even impossible to detect and distinguish diseases and pests, in particular at a sufficiently early stage. At the same time, crop production is moving towards fewer and larger farms, while pesticide use is envisaged to become even more restricted. This implies that farm managers will become increasingly dependent on automatic equipment to monitor and target the application of pesticides. In a high-tech crop protection system, where the reduction in pesticide use is an issue, it is important to focus on the optimal combination and integration of innovative monitoring tools for the early detection of weeds, pests and diseases and on precision

Pest Manag Sci 2011; 67: 616–625

spray applications to reduce the treatment area and use of pesticides. Earlier identification could also allow the grower to use non-chemical measures more efficiently. Examples are biological control,1 application of plant extracts, use of pheromones,2 application of beneficial organisms, culturing measures, use of vacuum cleaners to suck up insects,3,4 UV treatment to kill

∗

Correspondence to: Carolien Zijlstra, Wageningen UR, Plant Research International, PO Box 69, 6700AB Wageningen, The Netherlands. E-mail: [email protected]

a Wageningen UR, Plant Research International, Wageningen, The Netherlands b University of Southern Denmark, Odense M, Denmark c Aarhus University, Department of Integrated Pest Management, Slagelse, Denmark d Institute of Plant Protection IPP – CNR, Turin, Italy e Szent Istvan University, Plant Protection Institute, Gödöllo, Hungary f Plant Breeding and Acclimatization Institute, Radzikow, Blonie, Poland

www.soci.org

c 2011 Society of Chemical Industry

Integrated high-tech crop protection in the future

www.soci.org

Decision engine

Monitoring environment, planting material and soil before growing period

Interpretation of data no

Damage expected

yes

Action Monitoring Macroscale

Interpretation of data

no

Abnormalities localised

yes

Monitoring

Interpretation of data

Microscale Damage expected no

yes

Action

Figure 1. Generic model of an innovative crop protection system. The first monitoring step and the subsequent decision step are performed before the growing period; the other steps are performed during the growing period.

microorganisms,5 use of pheromones for mating disruption,6 mass trapping,7 lure and kill8 and destruction of weeds in the crop rows by removal by high-pressure air or finger hoeing. In the future, systems may be developed that could monitor for the presence of relevant organisms in the soil and planting material, or regularly monitor the field and air. When knowledge of damage thresholds and dose–response relations is available, the obtained data could be translated into spray maps. Plant protection products should then be applied as early and efficiently as possible using precision application technology. With such integrated systems, agriculture will be able to continue to deliver high-quality products and maintain high yields obtained with less chemical input. This review describes the innovative monitoring techniques and precision spray technologies that could be combined and used in the future as valuable tools to control diseases, pests and weeds in an optimal and environmentally friendly way. Assuming in the future that use of only minimal amounts of pesticides will be allowed, this review describes how this could be achieved if it were not restricted by cost or the amount of required research needed for operation of the crop protection system to be developed. Most of the required techniques are available; however, they have not yet been combined and integrated into functional crop protection systems. Environmental as well as farm management data are obviously also important to include in decision-making; however, technologies for collection of these data, as well as ways to process data in order to obtain useful information, are not covered in this review. The aim is to draw attention to already available and upcoming technologies that offer the potential of implementing future high-tech crop protection.

2 INNOVATIVE TOOLS FOR MONITORING DISEASES, PESTS AND WEEDS

Pest Manag Sci 2011; 67: 616–625

2.1 Monitoring before the growth period In order to minimise diseases, pests and weeds, the planting material or seeds must be of the best quality. The presence of pathogens on the seed or planting material must be known so that infected lots can be discarded in order to reduce chemical treatments and to avoid the introduction of new diseases. For example, several potato diseases can be introduced into the field from infected seed tubers. As the inoculum levels of pathogens are frequently low in the starting material, sensitive tests are extremely important for effective disease monitoring. Polymerase chain reaction (PCR) or real-time PCR techniques can detect minute quantities of pathogen DNA, even from a single fungal spore or single bacterial cell. However, it may be difficult to correlate the total amount of inoculum DNA with the actual number of infected seeds in order to quantify the disease because quantitative PCR cannot distinguish between a seed batch with a high amount of pathogen DNA in a few seeds from a batch with low amounts in many seeds.9,10 The probability of detecting a pathogen in the starting material depends on the sample size and the incidence of infection. The sample sizes should therefore be optimised in order to maximise the probability of detecting the organism. The soil should be checked for the occurrence of diseases and pests before the growing period. Once the diseases and pests have been detected above the threshold levels in the soil, appropriate decisions can be made, such as the choice of crop or


wileyonlinelibrary.com/journal/ps

617

In order to obtain an efficient crop protection strategy, it is important to monitor for diseases, pests and weeds before, during and after the growth period. Before the growth period, data

should be collected that characterise the field and environmental conditions. The starting material (seeds, tubers or plants) and soil should have the amounts of diseases, pests and weed seeds below the threshold levels. During the growth period, the field should be monitored at a macroscale level for the presence and progress of diseases, pests and weeds. This will help indicate the location and time where special efforts are needed. Identified ‘hotspots’ should, if relevant, be examined in detail at a microscale level (single-plant scale) for further characterisation and quantification of the causal agent. The collected data should be analysed to decide on an appropriate action plan (Fig. 1).

www.soci.org cultivars to grow. At the same time, the identification of beneficial organisms in agrosystems, such as arbuscular mycorrhizal fungi (AMF), fluorescent pseudomonads and others that take part in the biogeochemical cycle, may be applied in order to characterise soil biofertility. The most commonly used techniques for the identification of microorganisms in soil samples have relied upon traditional microbiological methods. The main limitations of these methods are that they are time consuming and laborious and require extensive taxonomical knowledge, which, all together, often complicates timely disease management decisions. Realtime PCR is currently one of the most commonly used techniques for accurate quantification of specific pathogens or beneficials, including bacteria,11 fungi12,13 and nematodes.14 PCR inhibitors in the DNA extracts from soil have been a major obstacle but have been minimised by optimised extraction methods which can obtain highly purified DNA from complex environmental samples. As a more exotic approach, it can be mentioned that there have been attempts to detect insects in soil through their sounds or the vibrations they generate to communicate or through noises that are produced during feeding and general movement.15 The spatial distribution in fields is of major importance when determining how samples have to be collected and which size of samples will be required to achieve a desired level of accuracy. Pooling multiple small subsamples into one sample, or processing subsamples from a homogenised bulk sample, is a desired sampling method.16 Optimal sampling strategies for plant parasitic nematodes have been studied extensively.17 Spatial patterns of nematodes (and also of other pests, pathogens and diseases) vary not only from field to field and from region to region but also both horizontally and vertically within fields. Been and Schomaker17 described different sampling approaches and scales. For instance, the dimensions of hotspots differ from nematode species to nematode species. While potato cyst nematodes cause the emergence of hotspots with a distinct shape of relatively small dimensions, other species, such as Meloidogyne chitwoodi, M. fallax and Pratylenchus penetrans, seem to cause extremely large infestations in a short time. Before the growth period, the soil should also be monitored for the occurrence of weeds. Annual weed populations are recruited from the soil seedbank, and species composition and density can to some extent be predicted from the composition of the seedbank.18,19 It seems that the relationship is stronger for species with low dormancy, such as grasses,19 than for broadleaf weeds where primary dormancy is typically stronger. Forcella et al.20 concluded that a thorough understanding of dormancy will be essential for the development of useful seedling emergence models. Such models are a prerequisite before pre-emergence herbicides can be applied according to thresholds. Perennial weeds can also be monitored in the soil. However, a very precise monitoring can be carried out much easier at harvest time or in stubble, as described later. Spectral analysis technologies can also be used to detect remaining residues of crops or weeds.21,22

618

2.2 Monitoring during the growth period at a macroscale 2.2.1 Diseases The detection of presymptomatic plant responses that are not yet apparent in visual spectrum images provides specific signatures for the diagnosis of distinct diseases. Often this is done by monitoring the changes in the physiological state of the plants. Thermal reflectance and fluorescence imaging in particular have proven their potential by detecting stress-related changes in the


C Zijlstra et al.

pattern of light emissions from plant leaves.23 Multiple imaging plant stress (MIPS)24,25 is an example of a technique that is based on fluorescence imaging of green parts of plants. This technology is non-invasive and enables the effect of pathogens on photosynthesis to be visualised and quantified. Optical and spectral analysis have been used for the indirect detection of apple scab on apple leaves.26 Specific wavelengths in the reflected light were used to discriminate between healthy leaves and spots infected with apple scab. The technology showed that apple scab could be detected 2 days after infection, whereas visual scouting could only identify the first symptoms after 10–12 days. Another approach could be the use of electronic noses (e-noses). E-noses are arrays of electronic gas sensors that can measure specific volatiles through a pattern recognition algorithm.27 E-noses can be utilised to detect pathogens directly, if they release volatiles,28 or indirectly to detect volatiles released by plants as a consequence of pathogen attack.29 Many plant diseases caused by fungi are spread by airborne spores. By monitoring air samples for fungal spores, it may be possible to predict the risk of epidemics.30 Therefore, air sampling, in combination with appropriate diagnostic methods, is highly relevant to ensure that timely action is taken. The simple form of sampling is ‘passive’ sampling, where airborne particles are deposited by gravitation onto a surface such as a microscope slide, culture plate, filter, leaf or plant.31 The ‘active’ samplers are volumetric samplers where a known amount of air is sampled. These are primarily used when a measure of the concentration of fungal spores in the air is required.31 The choice of which sampler to use depends on the biological characteristics of the organism and the environment where the sampling is going to take place. In principle, any method for the detection of pathogens could be applied to air samples. Real-time PCR has been used to quantify several plant pathogens such as Sclerotinia sclerotiorum,32 Botrytis squamosa,33 Fusarium circinatum34 and Monilinia fructicola35 in air samples. Immunoassays for the detection of airborne spores have also been developed for several fungal plant pathogens,36 – 40 but, for most of these molecular methods, the sample has to be processed in the laboratory before detection and quantification. However, the microtitre immunospore trapping device (MTIST) allows the direct trapping of spores in microtitre wells by means of a suction system and enables rapid quantification of spores by ELISA without any processing of the sample.41 Biosensors hold potential for systems where sampling and detection are integrated. Biosensors are analytical devices that integrate a biological sensing element with physical or electrochemical signalling. Most attention has been directed towards the development of sensors for foodborne bacteria and biological threat agents.42,43 There are few reports on biosensors for plant pathogen detection.44,45 Biosensors provide the possibility of quantifying pathogens in real time or near-real time, and the detection can be label free, which minimises sample preparation.46 Furthermore, some biosensors are regenerable, which would be preferable in a continuous monitoring system. However, the development of automated biosensors for ‘on-site’ use is still in its infancy, and only a few systems that integrate air sampling with a biosensor device have been described. The autonomous pathogen detection system (APDS), which is capable of continuously monitoring the environment for airborne biological threat agents, is an example of a fully integrated and autonomous system.47 The system is capable of continuously monitoring the environment for airborne biological threat agents. The system integrates continuous aerosol


Pest Manag Sci 2011; 67: 616–625


www.soci.org

sampling, in-line sample preparation fluidics and antibody and PCR-based detection. 2.2.2 Pests Insects and mites can be sampled for detection purposes in a non-selective way using traps, e.g. coloured sticky traps, water traps or suction traps.48 After collection, identification can be performed by, for example, visual inspection or by DNA-based detection techniques.49 Insects and mites can be detected by pheromone traps or by using other attractants.2,50 Pheromone traps that can detect hundreds of species of insects in all sectors of agriculture, horticulture and forestry are currently available (see, for example, www.pherobank.com). However, depending of the level of specificity, a subsequent analysis is necessary. Identification of insects by visual inspection is very difficult, even for experts. Molecular methods offer the opportunity of identifying the insects, regardless of their developmental stage. Odours or volatiles produced by the insects51,52 may be used to monitor insects and mites by means of e-noses. Another potential monitoring technique for pests is based on acoustic detection, as sounds produced by insects can be species specific.53,54

Pest Manag Sci 2011; 67: 616–625

2.3 Monitoring during the growth period at a microscale 2.3.1 Diseases In many cases, the causal agent of a disease can be identified from a visual examination of the symptoms; however, in some cases this may not be possible. The sampled tissue can then be tested using a variety of methods, the most widely used being ELISA or PCR, to identify the causal agent. A range of ELISA-based tests for individual pathogens are commercially available. PCR-based detection techniques have been developed for numerous pathogens, including fungal, bacterial and virus diseases, over the last couple of decades (for reviews, see Refs 78 and 79). The different approaches for the identification of plant pathogens (although often based on a similar technology) can be divided into (i) laboratory-based tests, which require a sample to be sent to a central laboratory for testing or (ii) fieldbased tests, which can be performed directly in the field by the farmer or advisor. Direct identification in the field requires specialised test kits that have been designed to be fast, robust and easy to use. Antibody-based lateral flow devices that give an answer in a few minutes have been developed for a number of pathogens (http://www.pocketdiagnostic.com). The devices have been developed for a number of viruses and bacteria but only for a few fungal diseases. Portable PCR, which is based on the rugged Cepheid SmartCycler II (http://www.cepheid.com), has been developed as a proof of concept for Phytophthora ramorum80 and Xylella fastidiosa.81 However, the system could also be easily implemented for other pathogens. Lab-on-a-chip devices are miniaturised microfluidics systems that combine sampling, DNA extraction, amplification and real-time detection in one disposable system. The development of such devices for the point-of-care detection of pathogens is a rapidly developing area82 but has not yet been implemented for plant pathogens. If such systems could be developed into robust, user-friendly and cost-effective systems, they would also have potential applications in agriculture, especially when they can be combined with immediate spray actions at the spot of detection. 2.3.2 Pests Some insects and mites will be found all over the field, whereas others are found in spots in the field. Once a hotspot has been detected, identification at a microscale can be performed using



619

2.2.3 Weeds Weeds are generally confined to the location, and changes in species composition and population size occur slowly over time compared with diseases and pests. A distinction should be made in agricultural crops between annual seed propagated weeds and vegetative propagated perennial weeds. However, both groups of weeds occur in patches in fields, as documented in the literature.55 – 57 The population size of annual seed propagated weeds is related to the size of the seedbank, which is further determined by seed production in the preceding crops and seed persistence. The germination of annual weeds is triggered by the soil cultivation measures used during crop establishment.58,59 The emergence pattern of both annual and perennial weeds can be described by species-dependent hydrothermal time models.20,60 – 62 The emergence of weeds is therefore usually closely correlated with crop emergence. Weed monitoring should be carried out at an early growth stage in order to achieve effective control with a minimal herbicide input.63 The weed cover at this early growth stage usually ranges from 1% to a few percent. This means that, in principle, most of the foliar-acting herbicides could be wasted on the soil or crop.64,65 The scope for reducing herbicide input using new technology to sense weeds and precision application systems to apply herbicides is immense. With the use of new technologies, site-specific weed control has the aim of reducing herbicide input and retaining efficacy of weed control. Different approaches to site-specific weed monitoring and control are being explored and have recently been reviewed by Christensen et al.66 These range from patch monitoring and control, with standard broadcast techniques,67,68 patch monitoring and control of weeds in decimetre small cells with specially adapted application techniques,69 to the ultimate concept of individual plant detection and treatment.70 Weed infestations in fields have been manually recorded and mapped in a number of studies in recent years.56,71,72 Some of these studies have been conducted in the same field for several years in order to study both the spatial distribution and the temporal stability of patchiness. The potential saving for a herbicide with only foliar activity using targeted site-specific weed control can be determined from such studies with a precision that depends

on the grid size used. The studies have shown that weed patches are generally stable, especially for perennial weeds and annual seed propagated weeds with seed dispersal prior to combine harvesting.55 This means that a weed map can be used, to some extent, to predict patch locations for the following year(s). The manpower required, however, makes manual weed recording and mapping unfeasible. Hence, implementing the concept in practical agriculture requires automated weed detection. The techniques explored so far include leaf or plant shape recognition, leaf or plant texture and colour or spectral reflectance or combinations of these features or techniques, as recently discussed by Slaughter et al.73 In row crops, the pattern and plant spacing of crop plants can be a further help to discriminate between crop and weeds. Nieuwenhuizen74 described a vision system that detects volunteer potato plants using the regularity of the sugar beet plants in the row. There has been some success distinguishing crops from weeds;73,75 – 77 however, in order to exploit the full potential of site-specific weed control, weeds must be identified at a species level in order to achieve the full saving of herbicide use.

www.soci.org morphological features or, in more difficult cases, molecular techniques such as described for diseases. 2.3.3 Weeds As described in Section 2.2.3, weed density and species composition are usually heterogeneous in fields, and weed coverage is often around or below 1% when a weed control is carried out. Microscale detection and the control of single weed plants therefore constitute the ultimate site-specific weed control concept and offer the possibility of reducing the input of foliar-acting herbicides by up to 90%, and theoretically more.

3

INTERPRETATION OF DATA

A crucial element of the presented crop protection system is the decision engine. This decision engine contains a decision support system for crop protection. Before and during the cropping period, collected data (locations and amounts of diseases, pests and weeds, environmental and historical crop management data of the field) will be used as input for the decision engine. The decision engine analyses all relevant data in such a way that advice can be given on how to manage the crop in the most efficient and environmentally friendly way in order to prevent crop damage that could be caused by pests, diseases and weeds. In order to do this, it is necessary for the decision also to contain information regarding damage thresholds, dose–response relations, biology, ecology, population dynamics, etc. The use of a specific cultivar or seed treatment with (biological) control agents is an example of the actions that could be recommended before the growth period. The recommendations given during the growth period concern whether to spray or not. When spraying is advised, it is recommended that plant protection products be applied as targeted as possible using modern precision application technology. Advice, such as adapting the spray volume, will be based on input information from earlier scouting or real-time sensor data.83 The nature and amount of application is defined by means of an algorithm, i.e. a dose–effect curve for a specific disease, pest or weed. These algorithms should be specifically optimised or developed for each disease, pest, weed and pesticide combination. Guided by the spray map and the position, the sprayer will be controlled according to the recommended actions, which could concern, for example, whether to spray or not, whether to spray with one or several nozzles at the same time, producing different spray volumes, or whether to adapt the nozzle type or spray pressure in order to change the quality of the spray.

4

ACTIONS

Non-chemical measures are preferred. When the decision is taken to spray, it should be done with high precision in order to reduce the spread of pesticide to the surroundings while obtaining a good biological efficacy.

620

4.1 Precision spray techniques Today, the standard spray strategy is to apply the pesticide on the basis of the whole field without taking into account the variability that can be encountered within the field. Precision application technology opens the way towards operating on smaller treatment units in the field by applying pesticides according to the sitespecific demand.


C Zijlstra et al.

With the use of new technologies, site-specific weed control aims at reducing herbicide input and retaining the efficacy of the weed control. As discussed in Section 2.2.3, different approaches are being explored to monitor weeds in crops, and a corresponding precision weed control implement must be available for each level of resolution used for monitoring the weeds. The spatial resolution levels range from a uniform treatment of the whole field, a treatment of weed patches or subfields with clusters of weeds sprayed with standard broadcast techniques,67,68 patch spraying of decimetre small cells with specifically adapted application techniques,69 to the ultimate concept of individual plant treatment.70,74 Several researchers have reported a reduction in the use of pesticides by treating only the weed patches. Gerhards and Christensen,84 referring to weed control, reported a reduction in herbicide use with a map-based approach. The reduction in winter cereals was about 60% for herbicides against broadleaf weeds and 90% for grass weed herbicides. The average savings for grass weed herbicides in sugar beet and maize were 78% in maize and 36% in sugar beet. For herbicides against broadleaf weeds, 11% were saved in maize and 41% in sugar beet. Although savings have been documented, the large working load necessary to generate the spray map makes this concept unrealistic for practical purposes. Implementing site-specific weed control requires the automated detection of weeds, as discussed in Section 2.2.3. Pesticide savings could easily be achieved with robust weed sensors mounted onto conventional boom sprayers. The first step could be boom sectionwise on/off or volume rate regulation, or even single-nozzle control. The precise application techniques that have recently been developed are able to vary dose rates considerably using pulse width modulation nozzles (Weed-It) or multinozzle holders (Lechler VarioSelect) with a switchable number of nozzles that vary the flow rate.85 These methods allow a continuous change in the dose rate from 50 to 300 L ha−1 or a stepwise change from 50 to 600 L ha−1 in 12 steps. Another step was the introduction of sprayers to handle several pesticides meant for individual weed species. These kinds of sprayer are already available.67 The Cerberus patch sprayer is equipped with three boom lines and software for the automatic spray application of three different pesticides. Research on this system has been conducted by the University of Hohenheim, Germany, and is now at the commercial phase. Although the research activities look promising, there is still a need for more work before the systems can be actually implemented. Nevertheless, some sensors are already on the market. The WeedSeeker sensor (Ntech) can detect all green materials and apply pesticides to a detected area. Such systems are at present used to control weeds in stubble and below trees in orchards, etc. The challenges are to develop real-time vision-based systems that can discriminate between crops and weeds. The progress in the research and the development of faster processors will open the way towards applying pesticides according to the site-specific demand. The use of a grid system is described by Lund et al.69,86 The system is a real-time vision system that only applies herbicides to those grids in the field in which weeds have been detected. The system consists of video cameras that capture images of the soil surface in front of the spraying boom. The images are analysed, and the location of the detected weed plants is saved. The image is divided into rectangular 30 × 107 mm cells, and any cells in which weed plants have been detected are marked for spraying on a small spraying map. The theoretical potential saving in pesticide usage for a field with a weed density of 220 plants m−2 and with weeds


Pest Manag Sci 2011; 67: 616–625


www.soci.org

Pest Manag Sci 2011; 67: 616–625

The nozzle settings are defined on the basis of the canopy distance from the trunk (Fig. 2, step 2) to match the spray volume to the canopy volume. A sprayer application map is generated for the different segment heights (Fig. 2, step 3), which defines how many nozzles should be opened and where the segments are compared to the canopy contour line. This application map is transferred to the sprayer controller (Fig. 2, step 4). However, these systems97,99,100 – 102 are still at the experimental prototype stage but are expected to be on the market within 5 years. Precision spraying also helps prevent the pollution of the environment through agrochemicals. Spray drift from the fields onto neighbouring crops, residential areas and/or waterways should be minimised. Spray drift mainly depends on the wind speed and direction in the field during application. Sprayer settings can minimise drift according to the choice of spray nozzle type, air velocity settings of the sprayer and shielding of the spray process. The nozzle type, in combination with the spray pressure used, defines the amount of small drops in the spray fan103 than can be blown away, and this fraction can be classified.83 Shielded spray types, such as a tunnel sprayer or a shielded sprayer,104 have shown spray drift reductions of more than 90% compared with standard spray applications, and have reduced spray volume by 30% or more because of capturing the spray passing through the canopy and recycling it. New regulations on safety buffer zones around waterways, wells and vulnerable living areas require adaptations of spray techniques and nozzle types according to their distance from these areas. Sensors that measure the wind speed and direction are used to change the sprayer settings (spray pressure, nozzle type), depending on where the sprayer is in the field in relation to the vulnerable area based on GPS.102

5

CONCLUSIONS

The implementation and use of innovative technologies to monitor and apply pesticides could contribute to the reduction in pesticide use in future farming systems. Farms will become larger, and manual and visual inspections will be too time consuming. High-tech and cost-effective systems will therefore be needed. Growing public concern about the use of pesticides and the possibility of these chemicals being increasingly restricted to the targeted control of diseases, pests and weeds mean that it will be necessary to minimise pesticide use. Implementation of an integrated crop protection system that combines monitoring with the targeted application of pesticides would lead to less environmental impact, less use of agrochemicals, less labour-intensive farming and higher yields and products of higher quality containing less pesticide residues. The individual elements of an integrated high-tech crop protection system, such as monitoring techniques, decision support systems and precision spray techniques, are generally available but still have to be improved before they can be linked together in an innovative crop protection system like the one described in this review. Whether this should be implemented in one on-site system that includes all the steps, from detection to spraying, or several independent steps (where, for example, detection is performed in the laboratory rather than on site in the field) depends on the nature of the causal agent and on the monitoring system. Multidisciplinary research is needed to develop a fully operative system that integrates monitoring, data analysis and precision application. Monitoring systems based on molecular detection need further development to become robust, low-cost and on-site devices. Such monitoring systems have already been developed for other applications



621

growing in patches was estimated to be at least 80% compared with a normal broadcast application. The treatment of individual plants using an ultraprecise chemical application system has an even higher potential for herbicide savings. Lee et al.87 have developed a real-time robotic weed control system for single-plant application of herbicides, and they tested the system by conducting field tests. Giles et al.88 investigated the biological performance of a robotic weed control system. Søgaard and Lund70 developed a microboom with a linear array of 20 evenly spaced tubings covering a 100 mm spray boom. Nieuwenhuizen74 described a system for the automated detection and control of volunteer potato plants in sugar beet fields. The system works with a single droplet application and it controls 95% of the volunteer potato plants without significant damage to the sugar beet plants at a forward speed of 0.8 m s−1 . Single-plant herbicide application will also be effective in reducing losses on the soil surface. Jensen and Spliid89 have quantified this loss for broadcast spraying in four different arable crops. They measured losses of up to 99%. Each weed seedling covers the soil surface, with a mean area of about 100 mm2 , and the density of weeds is usually in the range of 100–400 plants m−2 , which corresponds to a surface weed seedling area of about 1–4%. This means that more than 96% of a foliar-acting herbicide applied by broadcast spraying will be lost on the soil surface and potentially cause an environmental impact. With this figure in mind, single-plant application could be a valuable method to develop, as it potentially could lead to huge herbicide savings. Sensors that quantify crop parameters, such as the quantity of the biomass and photosynthesis activity, are already available on the market. Sensors used to evaluate plant stress90,91 or spectral analysis of the crop canopy parameters92 – 94 would open the way towards more target-oriented spraying for crop protection purposes. Smaller treatment units could be achieved in the field95,96 using these methods. The first experiments with these systems, which apply pesticides based on remote crop canopy sensing, show a reduction of more than 90% in the use of treatments early in the growing season, with an average 25% reduction in use over the whole growing season (potato, flower bulbs).95,96 The development of precision crop protection for fruit and vine growing has the aim of reducing chemical residues in these crops and minimising the environmental impact of agrochemical applications through precision dosing and croporiented applications of the chemicals, thereby contributing to the sustainable production of safe and healthy fruit. The agrochemical application tools and methods in sustainable fruit production have to be precisely oriented towards the crop itself and the surrounding crop. The first steps to quantify canopy structure and variations have already been made and involve driving alongside the rows of trees and bushes and adapting the spray quantity according to the measured canopy parameters.97,98 In this way, the variation in crop growth within a field on a certain date, as well as the variation in canopy amount and structure during the growing season, can be quantified and used to vary the dose. However, there is a lack of algorithms to transfer the sensor information to the spray activity. The basic steps of such a system are presented in Fig. 2. In order to control the sprayer, the controller needs to know which trees are in front of the sections and how they are shaped (Fig. 2, step 1). A tree shape and volume map is generated according to the three-dimensional orchard maps that are supplied by an aerial photogrammetric system.99

www.soci.org

C Zijlstra et al.

Figure 2. Schematic layout of the steps made from a target definition in the orchard to actual precise spraying. 1: a tree placement and volume map; 2: a nozzle setting definition by canopy contour distance; 3: a sprayer application map; 4: a working segmented cross-flow spraying with a canopy-adapted spray volume.90 .

622

such as medical point of care and as warning devices to civilians concerning bioterrorism. These could relatively easily be transferred to agricultural settings. After monitoring, data should be transformed into spray maps and recommendations for the farmers. However, this requires knowledge of the damage thresholds and dose–response relations and also basic information on biology, ecology, population dynamics, etc., for each of the relevant diseases, pests and weeds. Some novel detection techniques are more sensitive, and efficient two-step monitoring, from a macroscale level to a microscale level, will enable earlier detection of diseases, pests and weeds. This will, for instance, enable more successful use of biological control agents, as it will be possible to apply them earlier. Similarly, the innovative spray techniques will require less agrochemical control agents. This means that the use of innovative diagnostic and precision spray techniques will require new action thresholds and dose–response relationships, which will require a substantial research effort. The costs of some of the proposed innovative methods, such as real-time PCR, vision technology, precision spraying, ICT for decision support systems, etc., are relatively high at the moment. However, the price of the recently developed equipment will decrease in time. Some immunological devices, such as lateral flow devices, do not require expensive equipment, whereas others can be regenerable, again reducing costs. The economic benefits of the system result from the reduction in manpower and pesticides. It would be advisable firstly to work out a system for one crop, and, once the proof of principle has been achieved for this crop, it could be worked out for other crops in a much faster way. It


is foreseen that the application of the crop protection system could result in a reduction of 90% in the currently used pesticides, depending on the type of crop such as row crop or orchards. The innovative crop protection system discussed in this review will obviously not be implemented in the near future but is meant as an inspiration to solve challenges for future farming, if the use of pesticides is to become even more restricted. Whether the described model for crop protection will be implemented in the future depends on several factors, such as the context in which the farmer is going to operate, i.e. the development of markets, public concern about pesticide use and policy-making in general.

ACKNOWLEDGEMENTS Funding was provided by the European Community [Network of Excellence (European Network for the Durable Exploitation of Crop Protection Strategies (ENDURE)]. The authors would like to thank Dr Maria Teresa Della Beffa and Kirsten Jensen for editing the reference list.

REFERENCES 1 Hajek AE, McManus ML and Junior ID, A review of introduction of pathogens and nematodes for classical biological control of insects and mites. Biol Control 41:1–13 (2007). 2 Jutsum AR and Gordon RFS, Insect Pheromones in Plant Protection. John Wiley & Sons, Inc., New York, NY (1989). 3 Boiteau G, Misener GC, Singh RP and Bernard G, Evaluation collector for insect pest control in potato. Am J Potato Res 69:157–166 (1992).


Pest Manag Sci 2011; 67: 616–625


www.soci.org

Pest Manag Sci 2011; 67: 616–625

26 van de Zande JC, Wenneker M, Meuleman J and Achten VTJM, Development of a Crop Health Sensor (CHS) to minimise spray applications in apple, in Book of Abstracts, ed. by Bjugstad N, Andersen PG, Jørgensen M, Svensson SA and Servin D. SuProFruit 2007, 9th Workshop on Application Techniques in Fruit Growing, 12–14 September 2007, Swedish University of Agricultural Sciences, Alnarp, Sweden, pp. 13–14 (2007). 27 Gardner JW and Bartlett PN, A brief history of electronic noses. Sens Actuators B 18:210–211 (1994). 28 Pavlou A, Turner APF and Magan N, Recognition of anaerobic bacterial isolates in vitro using electronic nose technology. Lett Appl Microbiol 35:366–369 (2002). 29 Jansen RMC, Miebach M, Kleist E, van Henten EJ and Wildt J, Release of lipoxygenase products and monoterpenes by tomato plants as an indicator of Botrytis cinerea-induced stress. Plant Biol 11:859–868 (2009). 30 West JS, Atkins SD, Simon D, Emberlin J and Fitt BDL, PCR to predict risk of airborne disease. Trends Microbiol 16:380–387 (2008). 31 McCartney HA, Fitt BDL and Schmechel D, Sampling bioaerosols in plant pathology. J Aerosol Sci 28:349–364 (1997). 32 Rogers SL, Atkins SD and West JS, Detection and quantification of airborne inoculum of Sclerotinia sclerotiorum using quantitative PCR. Plant Pathol 58:324–331 (2009). 33 Carisse O, Tremblay DM, Levesque CA, Gindro K, Ward P and Houde A, Development of a TaqMan real-time PCR assay for quantification of airborne conidia of Botrytis squamosa and management of Botrytis leaf blight of onion. Phytopathology 99:1273–1280 (2009). 34 Schweigkofler W, O’Donnell K and Garbelotto M, Detection and quantification of airborne conidia of Fusarium circinatum, the causal agent of pine pitch canker, from two California sites by using a real-time PCR approach combined with a simple spore trapping method. Appl Environ Microbiol 70:3512–3520 (2004). 35 Luo Y, Ma Z, Reyes HC, Morgan D and Michailides TJ, Quantification of airborne spores of Monilinia fructicola in stone fruit orchards of California using real-time PCR. Eur J Plant Pathol 118:145–154 (2007). 36 Ribbert M, Wolters A, Barth S, Stoecker M, Schaeffer A, Fischer R, et al, Immunodetection of Venturia inaequalis ascospores with phage antibodies. J Phytopathol 155:170–177 (2007). 37 Skottrup P, Frøkiær H, Hearty S, O’Kennedy R, Hejgaard J, Nicolaisen M, et al, Monoclonal antibodies for the detection of Puccinia striiformis urediniospores. Mycol Res 111:332–338 (2007). 38 Jamaux I and Spire D, Development of a polyclonal antibody-based immunoassay for the early detection of Sclerotinia sclerotiorum in rapeseed petals. Plant Pathol 43:847–862 (1994). 39 Kennedy R, Wakeham AJ and Cullington J, Production and immunodetection of ascospores of Mycosphaerella brassicicola: ringspot of vegetable crucifers. Plant Pathol 48:297–307 (1999). 40 Bossi R and Dewey FM, Development of a monoclonal antibodybased immunodetection assay for Botrytis cinerea. Plant Pathol 41:472–482 (1992). 41 Kennedy R, Wakeham AJ, Byrne KG, Meyer UM and Dewey FM, A new method to monitor airborne inoculum of the fungal plant pathogens Mycosphaerella brassicicola and Botrytis cinerea. Appl Environ Microbiol 66:2996–3000 (2000). 42 Leonard P, Hearty S, Brennan J, Dunne L, Quinn J, Chakraborty T, et al, Advances in biosensors for detection of pathogens in food and water. Enzyme Microb Technol 32:3–13 (2003). 43 Hall RH, Biosensor technologies for detecting microbiological foodborne hazards. Microbes Infect 4:425–432 (2002). 44 Skottrup P, Hearty S, Frøkiær H, Leonard P, Hejgaard J, O’Kennedy R, et al, Detection of fungal spores using a generic surface plasmon resonance immunoassay. Biosens Bioelectron 22:2724–2729 (2007). 45 Skottrup P, Nicolaisen M and Justesen AF, Rapid determination of Phytophthora infestans sporangia using a surface plasmon resonance immunosensor. J Microbiol Methods 68:507–515 (2007). 46 Skottrup PD, Nicolaisen M and Justesen AF. Towards on-site pathogen detection using antibody-based sensors. Biosens Bioelectron 24:339–348 (2008). 47 Hindson BJ, Makarewicz J, Setlur UJ, Henderer BD, McBride MT and Dzenitis JM, APDS: the autonomous pathogen detection system. Biosens Bioelectron 20:1925–1931 (2005). 48 De Bokx JA and Piron PGM, Aphid trapping in potato fields and transmission of potato virus YN. Mededelingen van



623

4 Vincent V and Boiteau G, Pneumatic control of agricultural insect pests, in Physical Control Methods in Plant Protection, ed. by Vincent C, Panneton B and Fleurat-Lessard F. Springer, Berlin, Germany, pp. 270–281 (2002). 5 Ranganna B, Kushalappa AC and Raghavan GSV, Utraviolet irradiance to control dry rot and soft rot of potato in storage. Can J Plant Pathol 19:30–35 (1997). 6 Welter SC, Pickel C, Millar J, Cave F, Van Steenwyk RA and Dunley J, Pheromone mating disruption offers selective management options for key pests. Calif Agric 59:16–22 (2005). 7 El-Sayed AM, Suckling DM, Wearing CH and Byers JA, Potential of mass trapping for long-term pest management and eradication of invasive species. J Econ Entomol 99:1550–1564 (2006). 8 El-Sayed AM, Suckling DM, Byers JA, Jang EB and Wearing CH, Potential of ‘lure and kill’ in long-term pest management and eradication of invasive species. J Econ Entomol 102:815–835 (2009). 9 Justesen AF, Hansen HJ and Pinnschmidt HO, Quantification of Pyrenophora graminea in barley seed using real-time PCR. Eur J Plant Pathol 122:253–263 (2008). 10 Chilvers MI, du Toit LJ, Akamatsu H and Peever TL, A real-time, quantitative PCR seed assay for Botrytis spp. that cause neck rot of onion. Plant Dis 91:599–608 (2007). 11 Huang J, Wu J, Li C, Xiao C and Wang G, Specific and sensitive detection of Ralstonia solanacearum in soil with quantitative, real-time PCR assays. J Appl Microbiol 107:1729–1739 (2009). 12 Lievens B, Brouwer M, Vanachter AC, Cammue B and Thomma BPHJ, Real-time PCR for detection and quantification of fungal and oomycete tomato pathogens in plant and soil samples. Plant Sci 171:155–165 (2006). 13 Schroeder KL, Okubara PA, Tambong JT, Le´ vesque CA and Paulitz TC, Identification and quantification of pathogenic Pythium spp. from soils in Eastern Washington using real-time polymerase chain reaction. Phytopathology 96:637–647 (2006). 14 Zijlstra C and van Hoof RA, A multiplex real-time PCR (TaqMan) assay for the simultaneous detection of Meloidogyne chitwoodi and M. fallax. Phytopathology 96:1255–1262 (2006). 15 Mankin RW, McCoy CW, Flanders KL, Brandhorst J, Crocker RL and Shapiro JP, Methods for acoustic detection of insect pests in soil. Proc Soil Science Society of America Conference on Agroacoustics, 3rd Symposium, 3–6 November 1998, Buoyoucos, MS (1998). 16 Schroth G and Kolbe D, A method of processing soil core samples for root studies by subsampling. Biol Fertil Soils 19:60–62 (1994). 17 Been TH and Schomaker CH, Distribution patterns and sampling, in Plant Nematology, ed. by Perry RN and Moens M. CABI Publishing, Wallingford, Oxon, UK, pp. 302–326 (2006). 18 Dessaint F, Chadoeuf R and Barralis G, Nine years’ soil seed bank and weed vegetation relationships in an arable field without weed control. J Appl Ecol 34:123–130 (1997). 19 Rahman A, James TK and Grbavac N, Correlation between the soil seed bank and weed populations in maize fields. Weed Biol Manag 6:228–234 (2006). 20 Forcella F, Arnold RLB, Sanchez R and Ghersa CM, Modelling seedling emergence. Field Crops Res 67:123–139 (2000). 21 Roy SK, Shibusawa S and Okayama T, Site-specific soil properties prediction using hyperspectral signatures of topsoil coverage and underground image by realtime soil spectrophotometer, in Precision Agriculture ‘05, ed. by Stafford JV. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 529–536 (2005). 22 Gée C, Bonvarlet L, Magnin-Robert J and Guillemin J, Weed discrimination by reflectance measurements using neural networks. Proc XIIth Internat Conf on Weed Biology, Dijon, France, pp. 487–494 (2004). 23 Chaerle L, Leinonen I, Jones HG and van der Straeten D, Monitoring and screening plant populations with combined thermal and chlorophyll fluorescence imaging. J Exp Bot 58:773–784 (2006). 24 Jalink H, van der Schoor R and Schapendonk A, A method and a device for making images of the quantum efficiency of the photosynthetic system with the purpose of determining the quality of plant material and a method and a device for measuring, classifiying and sorting plant material. Dutch Patent 1 021 800 (2004). 25 Polder G, van der Heijden GWAM, Jalink H and Snel JFH, Correcting and matching time sequence images of plant leaves using penalized likelihood warping and robust point matching. Computers and Electronics in Agriculture 55:1–15 (2007).

www.soci.org

49

50 51 52

53 54

55 56

57

58 59 60 61 62 63

64 65 66 67

68 69

70

624

de Faculteit Landbouwwetenschappen, Rijksuniversiteit, Gent, Belgium, 50/2b, pp. 483–492 (1985). Schweigkofler W, Otrosina WJ, Smith SL, Cluck DR, Maeda K, Kabir G, et al, Detection and quantification of Leptographium wageneri, the cause of black-stain root disease, from bark beetles (Coleoptera: Scolytidae) in Northern California using regular and real-time PCR. Can J Forest Res 35:1798–1808 (2005). Carde RT and Minks AK, Insect Pheromone Research: New Directions. Chapman and Hall, New York, NY (1997). Baldwin IT, Kessler A and Halitschke R, Volatile signaling in plant–plant–herbivore interactions: what is real? Curr Opin Plant Biol 5:351–354 (2002). Kappers IF, Aharoni A, van Herpen TWJM, Luckerhoff LLP, Dicke M and Bouwmeester HJ, Genetic engineering of terpenoid metabolism attracts bodygards to arabidopsis. Science 309:2070–2072 (2005). Virant-Doberlet M and Cokl A, Vibrational communication in insects. Neotrop Entomol 33:121–134 (2004). Johnson SN, Crawford JW, Gregory PJ, Grinev DV, Mankin RW, Masters GJ, et al, Non invasive techniques for investigating and modelling root-feeding insects in managed and natural systems. Agr Forest Entomol 9:39–46 (2007). Colbach N, Forcella F and Johnson GA, Spatial and temporal stability of weed populations over five years. Weed Sci 48:366–377 (2000). Clay SA, Kreutner B, Clay E, Reese DE, Kleinjan J and Forcella F, Spatial distribution, temporal stability, and yield loss estimates for annual grasses and common ragweed (Ambrosia artemisiifolia) in a corn/soybean production field over nine years. Weed Sci 54:380–390 (2006). Izquierdo J, Blanco-Moreno JM, Chamorro L, Recasens J and Sans FX, Spatial distribution and temporal stability of prostrate knotweed (Polygonum aviculare) and corn poppy (Papaver rhoeas) seed bank in a cereal field. Weed Sci 57:505–511 (2009). Scopel AL, Ballare CL and Radosevich R, Photostimulation of seed germination during soil tillage. New Phytol 126:145–152 (1994). Jensen PK, Effect of light environment during soil disturbance on germination and emergence pattern of weeds. Ann Appl Biol 127:561–571 (1995). Donald WW, A degree-day model of Cirsium arvense shoot emergence from adventitious root buds in spring. Weed Sci 48:333–341 (2000). Calado J, Basch G and Carvalho M, Weed emergence as influenced by soil moisture and air temperature. J Pestic Sci 82:81–88 (2009). Dorado J, Sousa E, Calha IM, Gonzalez-Andujar JL and FernandezQuintanilla C, Predicting weed emergence in maize crops under two contrasting climatic conditions. Weed Res 49:251–260 (2009). Rydahl P, A Danish decision support system for integrated management of weeds, in Aspects 72: Advances in Applied Biology: Providing New Opportunities for Consumers and Producers in the 21st Century, ed. by Andrews M, Bancroft RD, Boatman ND, Bush MN, Dale MFB, Foulkes J, et alThe Association of Applied Biologists, Warwick, UK, pp. 43–53 (2004). Jensen PK and Spliid NH, Deposition of spray liquid on the soil below cereal crops after applications during the growing season. Weed Res 43:362–370 (2003). Jensen PK and Spliid NH, Deposition of spray liquid on the soil below sugar beet and potatoes after applications during the growing season. Acta Agric Scand B 53:49–55 (2003). Christensen S, Søgaard HT, Kudsk P, Nørremark M, Lund I, Nadimi ES, et al, Site-specific weed control technologies. Weed Res 49:233–241 (2009). Gerhards R and Oebel H, Practical experiences with a system for site specific weed control in arable crops using real-time image analysis and GPS-controlled patch spraying. Weed Res 46:55–70 (2006). Paice MER, Day W, Rew LJ and Howard A, A stochastic simulation model for evaluating the concept of patch spraying. Weed Res 38:373–388 (1998). Lund I, Jensen PK and Olsen HJ, Aspects of spraying small geometric cells, in Aspects 84: International Advances in Pesticide Application 2008, ed. by Alexander LS, Carpenter PI, Cooper SE, Glass CR, Gummer Andersen P, Magri B, et al. The Association of Applied Biologists, Warwick, UK, pp. 273–278 (2008). Søgaard HT and Lund I, Application accuracy of a machine visioncontrolled robotic micro-dosing system. BiosystEngng 96:315–322 (2007).


C Zijlstra et al.

71 Heijting S, van der Werf W, Stein A and Kropff MJ, Are weed patches stable in location? Application of an explicitly two-dimensional methodology. Weed Res 47:381–395 (2007). 72 Blanco-Moreno JM, Chamorro L and Sans FX, Spatial and temporal patterns of Lolium rigidum–Avena sterilis mixed populations in a cereal field. Weed Res 46:207–218 (2006). 73 Slaughter DC, Giles DK, Fennimore SA and Smith RF, Multispectral machine vision identification of lettuce and weed seedlings for automated weed control. Weed Tech 22:378–384 (2008). 74 Nieuwenhuizen AT, Automated detection and control of volunteer potato plants. PhD Thesis, Wageningen University, Wageningen, The Netherlands, 177 pp. (2009). 75 Borregaard TH, Nielsen H, Norgaard L and Have H, Crop-weed discrimination by line imaging spectroscopy. J Agr Engng Res 75:389–400 (2000). 76 Vrindts EJ, De Baerdemaker J and Ramon H, Weed detection using canopy reflection. Precis Agr 3:63–80 (2002). 77 Longchamps L, Panneton B, Samson G, Leroux GD and Theriault R, Discrimination of corn, grasses and dicot weeds by their UVinduced fluorescence spectral signature. Precis Agr 11:181–197 (2010). 78 Ward E, Foster SJ, Fraaije BA and McMartney HA, Plant pathogen diagnostics: immunological and nucleic acid-based approaches. Ann Appl Biol 145:1–16 (2004). 79 Schaad NW, Frederick RD, Shaw J, Schneider WL, Hickson R, Petrillo MD, et al, Advances in molecular-based diagnostics in meeting crop biosecurity and phytosanitary issues. Annu Rev Phytopathol 41:305–324 (2003). 80 Tomlinson JA, Boonham N, Hughes KJD, Griffin RL and Barker I, On-site DNA extraction and real-time PCR for detection of Phytophthora ramorum in the field. Appl Environ Microbiol 71:6702–6710 (2005). 81 Schaad NW, Opgenorth D and Gaush P, Real-time polymerase chain reaction for one-hour on-site diagnosis of Pierce’s disease of grape in early season asymptomatic vines. Phytopathology 92:721–728 (2002). 82 Mairhofer J, Roppert K and Ertl P, Microfluidic systems for pathogen sensing: a review. Sensors 9:4804–4823 (2009). 83 van de Zande JC, Achten VTJM, Michielsen JMGP, Wenneker M and Koster ATJ, Towards more target oriented crop protection, in Aspects 84: International Advances in Pesticide Application 2008, ed. by Alexander LS, Carpenter PI, Cooper SE, Glass CR, Gummer Andersen P, Magri B, et alThe Association of Applied Biologists, Warwick, UK, pp. 245–252 (2008). 84 Gerhards R and Christensen S, Real-time weed detection, decision making and patch spraying in maize, sugarbeet, winter wheat and winter barley. Weed Res 43:345–392 (2003). 85 Dammer KH and Ehlert D, Variable-rate fungicide spraying in cereals using a plant cover sensor. Precis Agr 7:137–148 (2006). 86 Lund I, Christensen S, Jensen LA, Jensen PK, Olsen HJ and Søgaard HT, Cell spraying of weeds in maize (in Danish). Bekæmpelsesmiddelforskning fra Miljøstyrelsen, No. 123, 90 pp. (2008). 87 Lee WS, Slaughter DC and Giles DK, Robotic weed control system for tomatoes. Precis Agr 1:95–113 (1999). 88 Giles DK, Downey D, Slaughter DC, Brevis-Acuna JC and Lanini WT, Herbicide micro-dosing for weed control in field-grown processing tomatoes. Appl Engng Agr 20:735–743 (2004). 89 Jensen PK and Spliid NH, Deposition of pesticides on the soil surface. Pesticides Research, 65, Danish Environmental Protection Agency, 59 pp. (2003). 90 MLHD Online Manual for the Herbicide Dose Calculation Module V2.0. [Online]. Plant Research International, Wageningen, The Netherlands (2004). Available: http://mlhd.opticrop.nl/[February 2011]. 91 Polder G, Spectral imaging for measuring biochemicals in plant material. PhD Thesis, Wageningen Agricultural University, Wageningen, The Netherlands, 150 pp. (2004). 92 Bravo C, Moshou D, West J, McCartney A and Ramon H, Early disease detection in wheat fields using spectral reflectance. Biosyst Engng 84:137–145 (2003). 93 Vrindts E, Reyniers M, Darius P, DeBaerdemaeker J, Gilot M, Sadaoui Y, et al, Analysis of soil and crop properties for precision agriculture for winter wheat. Biosyst Engng 85:141–152 (2003).


Pest Manag Sci 2011; 67: 616–625


www.soci.org

94 Scotford IM and Miller PCH, Combination of spectral reflectance and ultrasonic sensing to monitor the growth of winter wheat. Biosyst Engng 87:27–38 (2004). 95 van de Zande JC, Achten VTJM, Kempenaar C, Michielsen JMGP, van der Schans D, de Boer J, et al, SensiSpray: site-specific precise dosing of pesticides by on-line sensing, in Precision Agriculture ‘09, ed. by van Henten EJ, Goense D and Lokhorst C. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 785–792 (2009). 96 van de Zande JC, Achten VTJM, Schepers HTAM, van der Lans A, Michielsen JMGP, Stallinga H, et al, Plant-specific and canopy density spraying to control fungal diseases in bed-grown crops, in Precision Agriculture ‘09, ed. by van Henten EJ, Goense D and Lokhorst C. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 715–722 (2009). 97 Balsari P, Doruchowski G, Marucco P, Tamagnone M, Van de Zande J and Wenneker M, A system for adjusting the spray application to the target characteristics. Agricultural Engineering International: the CIGR Ejournal 10:(May):Manuscript ALNARP 08 002 (2008). 98 Costa-Ferreira AM, Germain C, Homayouni S, Da Costa J-P, Grenier G, Marguerit E, et al, Transformation of high resolution aerial images in vine vigour maps at intra-block scale by semi-automatic image processing, in Proc XV Internat Symp GESCO, Juin, Porec, Croatia, Vol. 2, pp. 1372–1381 (2007). 99 Meron M, van de Zande JC, van Zuydam RP, Heijne B, Shragai M, Liberman J, et al, Tree shape and foliage volume guided precision orchard sprayer – the PreciSpray FP5 project, in Precision

100 101 102

103

104

Agriculture, ed. by Stafford J and Werner A. Wageningen Academic Publishers, Wageningen, The Netherlands, pp. 411–416 (2003). Molto M, Martin B and Gutierrez A, Pesticide loss reduction by automatic adaptation of spraying on globular trees. J Agric Engng Res 78:35–41 (2001). Solanelles F, Planas S, Escola A and Rosell JR, Spray application efficiency of an electronic control system for proportional application to the canopy. Aspect Appl Biol 66:139–146 (2002). Doruchowski G, Holownicki R, Godyn A, Swiechowski W, Jedrachowicz T, Balsari P, et al, Environmentally Dependent Application System (EDAS) for adjusting the spray application parameters according to the environmental circumstances in orchards – ISAFRUIT project, in Proceedings, ed. by Bjugstad N, Andersen PG, Jørgensen M, Svensson SA and Servin D. SuProFruit 2007, 9th Workshop on Application Techniques in Fruit Growing, 12–14 September 2007, Swedish University of Agricultural Sciences, Alnarp, Sweden, pp. 15–16 (2007). Southcombe ESE, Miller PCH, Ganzelmeier H, van de Zande JC, Miralles A and Hewitt AJ, The international (BCPC) spray classification system including a drift potential factor, in Proc Brighton Crop Prot Conf – Weeds, Brighton, UK, pp. 371–380 (1997). Wenneker M, Heijne B and van de Zande JC, Drift reduction in orchard spraying with a sensor-equipped cross-flow sprayer, in Proc VII Workshop on Spray Application Techniques in Fruit Growing, ed. by Balsari P, Durochowski G and Cross J. University of Turin, Turin, Italy, pp. 247–256 (2003).

625

Pest Manag Sci 2011; 67: 616–625

