Skip to main content
Download PDF

A low-cost and open-source solution to automate imaging and analysis of cyst nematode infection assays for Arabidopsis thaliana

Plant Methods volume 18, Article number: 134 (2022) Cite this article

Abstract

Background

Cyst nematodes are one of the major groups of plant-parasitic nematode, responsible for considerable crop losses worldwide. Improving genetic resources, and therefore resistant cultivars, is an ongoing focus of many pest management strategies. One of the major bottlenecks in identifying the plant genes that impact the infection, and thus the yield, is phenotyping. The current available screening method is slow, has unidimensional quantification of infection limiting the range of scorable parameters, and does not account for phenotypic variation of the host. The ever-evolving field of computer vision may be the solution for both the above-mentioned issues. To utilise these tools, a specialised imaging platform is required to take consistent images of nematode infection in quick succession.

Results

Here, we describe an open-source, easy to adopt, imaging hardware and trait analysis software method based on a pre-existing nematode infection screening method in axenic culture. A cost-effective, easy-to-build and -use, 3D-printed imaging device was developed to acquire images of the root system of Arabidopsis thaliana infected with the cyst nematode Heterodera schachtii, replacing costly microscopy equipment. Coupling the output of this device to simple analysis scripts allowed the measurement of some key traits such as nematode number and size from collected images, in a semi-automated manner. Additionally, we used this combined solution to quantify an additional trait, root area before infection, and showed both the confounding relationship of this trait on nematode infection and a method to account for it.

Conclusion

Taken together, this manuscript provides a low-cost and open-source method for nematode phenotyping that includes the biologically relevant nematode size as a scorable parameter, and a method to account for phenotypic variation of the host. Together these tools highlight great potential in aiding our understanding of nematode parasitism.

Introduction

Plant-parasitic nematodes represent a fraction of the total number of free-living nematode species [1] but are widely studied due to their agricultural importance: their parasitism accounts for over 10% of the annual life-sustaining crop losses, costing the industry roughly 100–157 billion U.S. dollars per year [1,2,3]. In some cropping systems, plant-parasitic nematodes are the dominant plant pathogen of any kind [4].

Cyst nematodes are one of the major groups of plant-parasitic nematode, infecting a variety of crops globally. In general, the cyst nematode life cycle is as follows; a cyst, dormant in the soil for potentially decades, containing hundreds of eggs, is activated by the presence of diffusates from roots growing nearby. From these eggs, the infective second-stage juvenile (J2) hatches. The J2 migrates towards the growing root, penetrates host cells using a rapid thrusting action of their needle-like stylet, through which they secrete a variety of proteins to facilitate parasitism [5]. In the vasculature of the plant, cyst nematodes stimulate the formation of a "feeding-site" by re-differentiating existing procambial or pericyclic cells [6,7,8]. Feeding sites are unique pseudo-organs, unlike any other plant tissue, and are characterised by reduced and fragmented vacuoles, an increased smooth endoplasmic reticulum, and proliferated ribosomes, mitochondria, and plastids [9,10,11]. These organs are formed by partial cell wall dissolution and subsequent protoplast fusion of hundreds of adjacent cells, giving rise to a large, multi-nucleate, syncytial cell from which the nematode derives all external nutrition.

These obligate biotrophs will remain attached to this single feeding site for several weeks, maintaining a prolonged interaction with living host tissue. As the nematode feeds its body swells and increases in size dramatically. The sex of the nematode is at least in part determined during parasitism, and is linked to food consumption [12]. If the feeding site does not provide sufficient nutrition, the nematode terminally develops into a male, stops feeding, regains mobility, hatches out of the J3/4 body, and leaves the plant tissue for mating. Fertilised females produce eggs within their body, the body wall tans and hardens, and she falls off the plant leaving a cyst containing eggs that can live for years within the soil [13].

Current control measurements focus on the reduction of the density of plant-parasitic nematodes in the field and minimising the spread to other agricultural lands. Keeping the nematode population under a "damage threshold" is mainly achieved by crop rotation and the use of resistant cultivars and/or pesticides. However, in larger population densities, the rotation scheme may become economically unviable. Additionally, the use of the most effective nematicides is restricted due to concerns about the effects on human health and the environment. Improving genetic resources and developing impactful cultivars is therefore a current and future focus of many pest management strategies for plant-parasitic nematodes.

Plant genes that impact the outcome of nematode infection are varied. As with all plant-pathogens, classical resistance genes (NLR-type) have been identified against plant-parasitic nematodes, but in general our understanding of resistance remains limited due to the slow nature of screening for new ones. One of the major bottlenecks in identifying the plant genes that impact the outcome of infection is phenotyping. This is due in large part to the root-parasitic nature of these pathogens and the requirement of manual screening.

The most accessible infection assay is based on a well-established H. schachtii:Arabidopsis thaliana tissue culture system [14]. In brief, plants are grown in sterile tissue culture and infected about four weeks after germination. The total number of individuals infecting each plant is scored by manual assessment under the microscope. While extremely powerful, the assay has some clear limitations: (i) it requires access to microscopy equipment; (ii) due to the manual effort involved only a small number of genotypes can be screened at a time; and (iii) it does not account for the phenotypic variation of the infectable tissue of the host.

There are limited examples in the literature reporting nematode phenotyping and associated phenotypic analysis using sensors such as red–green–blue (RGB) cameras [15,16,17]. However, some work on open-source hardware design for programmable imaging in plant research is reported. For example, PhenoBox [18], SeedGerm [19], CropSight [20], LiDARPheno [21], and HyperScanner [22]. In all these tools, a Raspberry Pi (a small, affordable computer) was used in combination with a light sensor to capture images of the subjects of interest. For analysing acquired images, open solutions such as Leaf-GP [23] using software analytic packages based on the coding language Python, phenoSEED [24], AutoRoot [25], and PYM [26] using image analysis software ImageJ/FIJI automation scripts to perform object segmentation and trait analysis. Taken together, the above-mentioned examples clearly illustrate a scope for development of a hardware/software solution for screening plant parasitic nematode infection.

In this paper we describe a low-cost easy to build and adopt image analysis approach to measure nematode infection. We describe methods to: 3D print and construct an “imaging tower”; capture images of infected plants in the well-established tissue culture system; use image analysis scripts to isolate, count, and measure parasitic nematodes; and normalise these data to the available root surface area at time of infection.

Materials and methods

Cultivation of nematodes and setting up the infection assay

To culture Heterodera schachtii on Sinapis alba (cv albatross) and for the infection assay on Arabidopsis thaliana (either Col-0 or N804585), seeds were surface sterilised with 20% dilution of 3.6% sodium hypochlorite (ParoZone, Henkel) for 20 min and washed six times with sterile double distilled H2O. The seeds were kept at 4 °C overnight to improve and synchronise germination [27]. The seeds were sown on sterile standard KNOP-medium (Duchefa Biochemie) [28] in vitro in a 15 cm sterile Petri dish (SARSTEDT) for S. alba and 5 cm deep well petri dishes (Thermo-Fisher) for A. thaliana. For purposes of female counting the agar was dyed using various food colouring (Limino, DYL-ghm-20210126–324) as described in Table 1. The different colours were tested in two different concentrations, a darker and a lighter variant. For purposes of cyst counting the agar remained undyed.

Table 1 Amount of food colouring used in µL per 25 mL of KNOP-medium
Full size table

Plants grew on a 16 h day 21 °C and 8 h night 20 °C cycle in an MLR-352-PE growth chamber (Panasonic). Cysts were then soaked in 3 mM Zinc Chloride (SIGMA–ALDRICH) to promote hatching in a specialised hatching jar (Jane Maddern Cosmetic, 250 mL) with two 2.5 cm plastic rings (alt-intech Tube Perspex) holding a 20 µm mesh (Sigma–Aldrich). Five days after hatching, J2 nematodes that had passed through the mesh were collected by pipetting. The density of nematodes was adjusted to 1 nematode/µL using sterile double distilled H2O containing 0.01% v/v Tween (Sigma–Aldrich). Once the plants reached an age of 21–28 days, the roots were inoculated with ~ 300 J2 nematodes on S. alba and 26 days with ~ 80 J2 nematodes on A. thaliana by pipetting the suspension on the roots. On S. alba it took 10–12 weeks at 20–25 °C in darkness for the cysts to develop. Images of infected A. thaliana plates were taken depending on the need at least on two weeks post germination for quantification of root surface area, 21 days for female nematodes and 21 + days for encysted nematodes.

Building hardware for image requirements

Open hardware design

Images of petri-dishes were taken using a custom imaging tower. The imaging components of the device are assembled onto a custom 3D printed apparatus (https://github.com/OlafKranse/A_low_cost_imaging_tower). The STL files were sliced using CURA (Ultimaker) with an infill of 20%, a layer height of 0.15 mm, a wall thickness of 1.2 mm and a printing speed of 40.0 mm/s. The nozzle temperature was set to 205 °C and the build plate to 60 °C. The two components of the tower were printed in tough PLA with PVA as support (Ultimaker) in an Ultimaker S3 equipped with an 0.4 mm AA and BB core. The PVA was dissolved for 24 h in tap water. The 12 MP RPI-HQ-CAMERA (RASPBERRY-PI) was mounted in the holes of the top part of the tower using M2 bolts (2 cm), nuts, and washers (RS components) and the 16 mm Telephoto Lens (RASPBERRY-PI) was bolted onto the camera as instructed in Fig. 1. The camera was connected to the Raspberry Pi via a 30 cm Ribbon Cable (THE PI HUT). Lastly, the 60 LED RGB STRIP LIGHT 6400 K SET IP65 12 V (V-TAC) was slid through the extrusions on the side of the bottom part (Fig. 1).

Fig. 1
figure 1

Explosion diagram of the imaging tower. A The separate components labelled with their corresponding end position. The item numbers indicate a position in the table below. B The final product after assembly. C A table of components. The item number corresponds with the numbers in A

Full size image

Data collection and calibration

Before inoculation (four weeks post germination), images were taken using a custom script controlling the tower described above (https://github.com/OlafKranse/A_low_cost_imaging_tower). Petri dishes were wiped clean using disposable wipes (Kimwipes) before capture. The images were separated into folders corresponding to the mutant line. Using ImageJ commands [29], the script, in order: converted the image to 8bit, subtracted the background, enhanced the contrast, ran a threshold, converted the output to a mask, performed a watershed, defined an area of interest, analysed and counted the resulting overlay using the parameters as described in the script (https://github.com/OlafKranse/A_low_cost_imaging_tower). A slightly adjusted script was used for plates containing dye (https://github.com/OlafKranse/A_low_cost_imaging_tower). The root surface area for all the images in the folder were exported as a CSV file. By adding colour thresholding to the above mentioned scripts, leaf surface area was isolated from images (https://github.com/OlafKranse/A_low_cost_imaging_tower).

Traditional counting and automatic counting

Ground truth by manual scoring

The number of females and males in manual counting was scored under a S9D Stereomicroscope (Leica). For manual size quantification an image was taken using the above-described imaging machine and an outline was drawn using the polygon selections tool in ImageJ (Fig. 2). Using the measure command, the area within the outline was quantified in number of pixels.

Fig. 2
figure 2

An example of a manual outline drawn around a female cyst nematode using ImageJ. (Left) before drawing an outline. (Right) the outline drawn

Full size image

Script-based counting and quantifiable traits

Automatic counting was performed on images taken as described above. Depending on the treatment a different script was used to calculate the number and size of females. Before isolation, the colour histogram for all images was normalised to the first image in the dataset using a custom python script (https://github.com/OlafKranse/A_low_cost_imaging_tower). The images were then processed in ImageJ for two different nematode life stages: (i) tanned cyst nematodes; (ii) female nematodes.

  1. (i)

    Automatic counting of cysts. Using ImageJ commands, the script, in brief: thresholded colour, converted the image to 8bit, removed pixel outliers, performed a watershed, waited for the user to define an area of interest, analysed and counted the resulting overlay using the parameters as described in the script (https://github.com/OlafKranse/A_low_cost_imaging_tower).

  2. (ii)

    Semi-automatic counting of females. The well-established watershedding addon MorphoLibJ [30] was used for segmentation of female nematode from images of coloured agar. The tool parameters were set to a morphological gradient type and a watershed segmentation threshold both with a radius of 5. The display type was set to catchment basins and the number and size of segmentation was scored using the analyse regions function.

Results

Cyst nematodes are naturally root parasitic obligate biotrophs, and notoriously difficult to phenotype during infection. Their ability to infect depends, although not exclusively, on a variety of host factors (including genotype and physiology), for example, root surface area. To account for this, custom hardware (low-cost, 3D printed imaging platform) and associated custom software (simple image analysis pipelines) were developed to standardise/semi-automate root and nematode phenotyping in the model Heterodera schachtii:Arabidopsis thaliana pathosystem.

A cost-effective, open-source 3D printable imaging device

Based on in vitro 5 cm petri dish cultures of individual A. thaliana plants on transparent medium, a custom 3D printed imaging platform was designed to capture consistent images of the root systems. Consistency between images is essential for downstream image analysis (object segmentation using thresholding). The apparatus comprises of two halves (Fig. 3A): the upper half housing a high-quality 12-megapixel camera and lens (Fig. 3B): the bottom half housing an LED strip wrapping around the base and passes through the tower (Fig. 3D). Combined with a Raspberry Pi computer, the setup can consistently capture high quality images of A. thaliana root systems (Fig. 3C).

Fig. 3
figure 3

The imaging tower. A An explosion diagram of the assembly and final product of the imaging tower on scale of 1:3. B a look inside the top half of the imaging tower. C An example image taken using the imaging tower. D The assembled imaging tower. The lens (blue arrow) is mounted to the camera (red arrow), which is connected to the Raspberry Pi (not shown). Consistent all around lighting is provided by an LED strip (yellow arrow)

Full size image

Printing the imaging tower takes 30 h (on a Ultimaker S3, Ultimaker PLA with 20% infill), with 30 min hands on time for assembly of the device. Imaging efficiency may vary depending on the speed of the Raspberry Pi and if cooling is provided. In our hands, imaging takes approximately 7 s per capture. Using these images, the extent and nature of the root system pre-infection can be readily and non-destructively analysed (Fig. 4A).

Fig. 4
figure 4

Extraction of root surface area from images and correlation with nematode counts. A A visual representation of the pipeline used for calculating the root surface area. From left to right the features are extracted from the image and a region of interest is defined. Artefacts picked up in the pipeline: A1 Damage to the agar (purple arrow) and a leaf touching the agar (brown arrow). A2 A small particle on the surface of the plastic (pink arrow), clustered roots interpreted as a singular root (blue arrow). B An example of the ability to extract the leaf surface area using colour thresholding. Artifacts are introduced by difference in leaf colour (pink arrow) and damage to the agar (blue arrow). C Two models describing the relationship between number of nematodes and root surface area for two A. thaliana genotypes. The models are described as follows; Columbia 0, Males: y = 5.4 + 5.7 × 10−6x, Females: y = 17.5 + 6.33 × 10−6x. N804585, Males: y = 10 + 2.34 × 10−5x, Females: y = 11 + 1.26 × 10−5x

Full size image

By means of thresholding, a technique that groups pixels into distinct classes (typically a foreground and a background), the root surface area for A. thaliana grown in tissue culture was estimated from an image given in number of pixels (Fig. 4A). Using this technique most roots can be isolated and quantified. Minor artefacts were introduced (leaves touching the agar, damage to the agar (Fig. 4-A1), and dust on the outside of the plates (Fig. 4-A2) but do not contribute substantially to the total root surface area and, importantly, are consistently present on all plates, and therefore of minimal concern. A similar method of colour thresholding was used to extract the leaf surface area from an image (Fig. 4B), with similar constraints.

The impact of root surface area on infection

Two genotypes were compared; Columbia 0 and a T-DNA knockout mutant for AT1G07540 (N804585), a plant gene highly expressed during nematode infection [31]. Using these estimates of root area from the images, the relationship between the number of nematodes and the total root surface area for the two genotypes has been summarised by the linear models shown in Fig. 4C. The number of manually counted nematodes for both male and female increased with the total root surface area at inoculation density of 80 infective stage nematodes/plant. Approximately 4% of female and 3% of male variation in nematode number on Colombia wild-type and 10% of female and 54% of the variation in males on mutant line N804585 can be explained by root area pre infection (Fig. 4C). Together, the positive correlation in Fig. 4C demonstrates the importance of normalising nematode counts by available root are pre-infection, when comparing between but even within genotypes.

Validation of image analyses method of cyst nematode infection

The image quality from the imaging tower not only allows for quantification of roots, but also clearly shows female life stage nematode on roots from 21 days post infection (dpi) onwards (16-h day at 21 °C and 8-h night at 20 °C). There is, therefore, scope to employ similar image analysis scripts for the detection of nematodes directly from images of infected plants.

Counting cysts

Like detecting roots from images, high contrast between the entity of interest and the background is crucial for threshold-mediated object isolation. Once extracted, a feature can be scored for various parameters. In images, the encysted H. schachtii often has excellent contrast with the background, due to its tanned cuticle, and was the initial life stage of interest for automated analyses. For a series of images, each containing a single infected plant, the colour histograms were normalised to an arbitrary reference (any one of the series) using a custom python script [32]. The optimal colour thresholding for the reference image was then empirically determined to isolate the cysts, and then applied to all other images in the series. Using this technique, 147/230 cysts were identified, with 3 false positives (1.3%) introduced by artefacts on the petri-dish. Of those cysts that were identified, their area was quantified automatically. Taken together, cysts were surprisingly difficult to identify and quantify using threshold-mediated object isolation.

Counting females

In order to identify females of H. schachtii (i.e. before encystment) using threshold-mediated object isolation, it was anticipated that some additional steps would be required to increase contrast: at this life stage, they are very similar in colour to roots and senesced A. thaliana leaves. In standard KNOP media, we were unable to identify parameters that would distinguish females from surrounding roots (data not shown). To address this, we tested various contrasts by dyeing the agar with food colouring, followed by illumination with white, red, green or blue lights (Fig. 5A, 5B). By visual observation, the translucency of dyed agar does not appear any different to the undyed counterparts. The best contrast was found using a red food colouring (Limino, DYL-ghm-20210126-324) illuminated with blue light (380–500 nm, Fig. 5B). Green light provides a similarly good contrast but was more prone to highlighting artifacts (data not shown).

Fig. 5
figure 5

The spectrum of coloured agar under various lighting conditions. A The range of colours tested for contrast with nematodes and filtration potential of the background. B The colour red agar under four different lighting conditions. From left to right: White, Red, Green and Blue light. The blue lighting condition with the red agar were the most ideal combination for quantifying nematodes using the segmentation tool. The bottom image row has the contrast and brightness adjusted for visibility. Examples of female nematodes are highlighted with yellow arrows

Full size image

Using a well-established ImageJ library (morphological segmentation [30]) and manual verification, we extracted nematodes from images with 93.3% (70/75) accuracy. We were also able to measure nematode area from the images, with an R2 of 0.83 with manual area quantification compared to only an R2 of 0.44 when counting cysts (Fig. 6D). The automated segmentation tool was not able to reliably extract cysts from these conditions. For validation of these methods, the outcomes of automated and manual area were compared against each other (Fig. 6D).

Fig. 6
figure 6

Isolation and quantification of nematodes from images. A A schematic workflow highlighting the steps required for manual and automated nematode counting, and the differences in output. B Isolation of female cyst nematodes from red coloured agar under blue light illumination, and the extracted overlay from the script. C Isolation of encysted nematodes under white light illumination, and the extracted overlay from the script. D Correlation of area of the nematode between manual counts and automatic counts using the script for females and cysts

Full size image

Discussion

Our knowledge on both resistance genes and susceptibility genes against plant parasitic nematodes remains scarce. The most widely studied, Gpa2, recognises the nematode effector RBP-1 once it is delivered into the syncytial feeding site: triggering an immune response and ultimately resistance [33]. For Heterodera schachtii, a commonly used model species for cyst nematodes, resistance associated with HS1pro1 was introduced in sugar beet through introgressive hybridization [34]. The presence of this gene diminishes the growth of the feeding organ, impacting the reproduction of the nematode [35]. Despite the positive effects on nematode population control in the field, the introgressed gene reduces yield [36], and is therefore not widely used [37]. However, control is not limited to resistance, the absence of susceptibility genes can also determine the outcome of infection illustrated by the deletion of AtHIPP27 and the more well understood vitamin B5 pathway gene AtPANB1 [14, 38]. A broad focus of the field is to identify genes that have direct impact on nematode parasitism.

The Heterodera schachtii:Arabidopsis thaliana infection assay is broadly used for screening of genes due to the availability of the genetic resources on the plant side of the interaction. One of the major bottlenecks that remains is the phenotyping assay because it: (i) requires access to costly microscopy equipment; (ii) lacks the ability to readily account for phenotypic variation of the infectable tissue of the host; and (iii) requires considerable time investment of a trained operator. Taken together, only a relatively small number of genotypes can be screened.

To address the first constraint, a cheap, easy to build and use, 3D-printed imaging tower was created. The construction consists of few parts ensuring ease of assembly and, with the growing number of 3D printing services [39], is available at low cost (printing service ~ £67.56 with delivery [40], or £10.87 in house at time of writing). The assembly can take images of petri-dishes every seven seconds. At two weeks post germination, the root system of A. thaliana can be isolated from the photographs using a threshold-based image analyses method and quantified in a measure of pixels.

It is hypothesised that not only the genotypic, but also the phenotypic variation of the host impacts its infectability. For example, some loss of function mutants vary in their susceptibility or resistance to nematodes [41], but in some cases also vary in their root physiology, complicating the assessment. It is hypothesised that the available root area before infection acts as a direct physical constraint to cyst nematode parasitism, which can affect the outcome of a susceptibility assay.

Due to the high variability in nematode infection a good model remains difficult to construct. The linear model in Fig. 4C, however, illustrates an interaction between the root surface area and the number of nematodes. Overall, the number of nematodes increases with the amount of root surface area for a given inoculum. Importantly, between the two tested genotypes there is a clear difference in magnitude of the effect.

On images of the same dishes at a later stage (21 + days post infection), both female and cyst life stage are clearly visible. A fully automated script for counting cysts using colour thresholding from petri-dishes was created. This method, while quick, has a high error rate (63.9% accuracy, 1.3% false positive and 34.8% false negative). Adjusting the threshold can reduce the number of false negatives, however, this typically increased the number of false positives. The technique is largely limited by the complexity of the background around nematodes and the variability in colour and contrasts between plates. To mitigate against this, various colour dyes were added to the agar which, under blue illuminated light, increase the contrast of female nematodes, and weaken the visibility of the background. The greatest distinction between nematode and background was found using Limino red food colouring illuminated with blue light (380–500 nm). Importantly, the translucency of the agar does not appear to change significantly after addition of the dye, excluding the possibility of visual obstruction of nematodes. Using both a program which isolates features from an image by means of morphological operations, and manual verification, the frequency and size of the nematode could be determined with a 93.3% accuracy and 6.7% false negatives. In addition to count, this method provides a good size correlation between fully manual and semi-automatic measurements (R2 = 0.83). Importantly, from the available data, it is unclear whether the dye is entirely innocuous (some dyes did inhibit germination, or root growth, and were not included in the experiment—others may have an impact on parasitism); regardless, the impact of dye can be disregarded with consistent use between the control and the treatment. Furthermore, the effect on parasitism appears minimal as all life stages can fully develop. Some coloured dyes do contribute to the false positives by highlighting particles on the outside of the petri-dish which must be filtered out manually. Through modifications to the traditional assay, a semi-automatic and high accuracy measurement of infection can be achieved. This, however, does little to reduce the effort involved for screening females as manual verification of the isolated mask is required. Nevertheless, at no extra effort, the biologically relevant measurement of nematode size is gained, which makes the semi-automated screening superior over manual screening for females.

A major limitation of this new screening method is that it does not allow for inclusion of male nematodes. Whilst visible on images of non-coloured agar, the current pixel density does not create a sharp enough image to reliably distinguish them from the background. Male nematodes are much smaller than females (easily obscured by roots and take up fewer pixels), transient (males regain motility to find a female), and transparent (so difficult to detect). It may or it may not be possible to address this limitation in the future with a higher resolution camera and/or a more appropriate lens allowing for reduction of undesired imaging of the area around the petri dish, optimising the use of available pixels. Regardless, accurate and semi-automated measurement of females and cysts hold considerable value for fundamental research and applied variety testing. Furthermore, the new technique is limited by its throughput as it does not allow for a decrease in time required for screening. An issue that may be addressed in the future by further automation on the hardware side.

These first steps towards digitizing screening are crucial in the development of a fully automated system for screening of plant parasitic nematode infection. The main limitation of the protocol to date is speed, and the requirement of modification of the agar using coloured dyes for accuracy of automated screening. Classical computer vision techniques are restricted in their capabilities to isolate objects from complex backgrounds. These traditional methods, however, can be accompanied with deep learning approaches to lift these constrains [42], and have proven already valuable in phenotyping of plant diseases [43]. Ideally, this would lift the requirement of modified agar, and manual verification of nematode count, essentially limiting the speed of screening to the speed of imaging.

Conclusion

Overall, this paper describes an accessible open-source image-based method for better insight in cyst nematode infection on Arabidopsis thaliana by including the size of the nematode as a parameter in infection screening, normalised for available infectable tissue. Further research is needed to elucidate the exact interaction between root surface area and the success on nematode infection. While the new method adds a new dimension to nematode infection screening, it is still limited in speed. Beyond the scope of this paper, machine learning approaches may lift the constraints of manual verification in the assay, and thus further increase screening speed.

Availability of data and materials

All scripts used in this experiment are available under the following github repository: https://github.com/OlafKranse/A_low_cost_imaging_tower.

References

  1. Abad P, Gouzy J, Aury JM, Castagnone-Sereno P, Danchin EGJ, Deleury E, et al. Genome sequence of the metazoan plant-parasitic nematode Meloidogyne incognita. Nat Biotechnol. 2008;26:909–15.

    Article  CAS  Google Scholar 

  2. Sasser JN. A world perspective on nematology: the role of the society. Vistas Nematol. 1987;7–14.

  3. Chitwood DJ. Research on plant-parasitic nematode biology conducted by the United States Department of Agriculture-Agricultural Research Service. Pest Manag Sci. 2003;59:748–53. https://doi.org/10.1002/ps.684.

    Article  CAS  Google Scholar 

  4. Savary S, Willocquet L, Pethybridge SJ, Esker P, McRoberts N, Nelson A. The global burden of pathogens and pests on major food crops. Nat Ecol Evol. 2019;3:430–9.

    Article  Google Scholar 

  5. Atkinson HJ, Urwin PE, Hansen E, McPherson MJ. Designs for engineered resistance to root-parasitic nematodes. Trends Biotechnol. 1995;13:369–74.

    Article  CAS  Google Scholar 

  6. Bleve-Zacheo T, Zacheo G. Cytological studies of the susceptible reaction of sugarbeet roots to Heterodera schachtii. Physiol Mol Plant Pathol. 1987;30:13–25.

    Article  Google Scholar 

  7. Wyss U. Observations on the feeding behaviour of Heterodera schachtii throughout development, including events during moulting. Fundam Appl Nematol. 1992;15(1):75–89.

    Google Scholar 

  8. Wyss U, Zunke U. Observations on the behaviour of second stage juveniles of Hetero inside host roots. Rev Nematol. 1986;9:153–65.

    Google Scholar 

  9. Bleve-Zacheo T, Rubino L, Melillo MT, Russo AM. The 33K protein encoded by cymbidium ringspot tombusvirus localizes to modified peroxisomes of infected cells and of uninfected transgenic plants. J Plant Pathol. 1997;197–202.

  10. Gheysen G, Fenoll C. Gene expression in nematode feeding sites. Annu Rev Phytopathol. 2002;40:191–219.

    Article  CAS  Google Scholar 

  11. Gray JE, Picton S, Giovannoni JJ, Grierson D. The use of transgenic and naturally occurring mutants to understand and manipulate tomato fruit ripening. Plant Cell Environ. 1994;17:557–71.

    Article  CAS  Google Scholar 

  12. Anjam MS, et al. Host factors influence the sex of nematodes parasitizing roots of Arabidopsis thaliana. Plant Cell Environ. 2020;43(5):1160–74.

    Article  CAS  Google Scholar 

  13. Hu W, Strom N, Haarith D, Chen S, Bushley KE. Mycobiome of cysts of the soybean cyst nematode under long term crop rotation. Front Microbiol. 2018;9:386.

    Article  Google Scholar 

  14. Radakovic ZS, Anjam MS, Escobar E, Chopra D, Cabrera J, Silva AC, et al. Arabidopsis HIPP27 is a host susceptibility gene for the beet cyst nematode Heterodera schachtii. Mol Plant Pathol. 2018. https://doi.org/10.1111/mpp.12668.

    Article  Google Scholar 

  15. Sohrabi S, Mor DE, Kaletsky R, Keyes W, Murphy CT. High-throughput behavioral screen in C. elegans reveals Parkinson’s disease drug candidates. Commun Biol 2021; 4:1–9. Available from: https://www.nature.com/articles/s42003-021-01731-z

  16. Bates K, Le K, Lu H. Deep learning for robust and flexible tracking in behavioral studies for C. elegans. PLOS Comput Biol. 2022;18:e1009942. https://doi.org/10.1371/journal.pcbi.1009942.

    Article  CAS  Google Scholar 

  17. Hebert L, Ahamed T, Costa AC, O’Shaughnessy L, Stephens GJ. WormPose: Image synthesis and convolutional networks for pose estimation in C. elegans. PLOS Comput Biol. 2021;17:e1008914. https://doi.org/10.1371/journal.pcbi.1008914.

    Article  CAS  Google Scholar 

  18. Czedik-Eysenberg A, Seitner S, Güldener U, Koemeda S, Jez J, Colombini M, et al. The ‘PhenoBox’, a flexible, automated, open-source plant phenotyping solution. New Phytol. 2018;219:808–23. https://doi.org/10.1111/nph.15129.

    Article  Google Scholar 

  19. Colmer J, O’Neill CM, Wells R, Bostrom A, Reynolds D, Websdale D, et al. SeedGerm: a cost-effective phenotyping platform for automated seed imaging and machine-learning based phenotypic analysis of crop seed germination. New Phytol. 2020;228:778–93. https://doi.org/10.1111/nph.16736.

    Article  CAS  Google Scholar 

  20. Reynolds D, Ball J, Bauer A, Davey R, Griffiths S, Zhou J. CropSight: a scalable and open-source information management system for distributed plant phenotyping and IoT-based crop management. Gigascience 2019; 8:1–11. https://academic.oup.com/gigascience/article/8/3/giz009/5304887

  21. Panjvani K, Dinh AV, Wahid KA. LiDARPheno—a low-cost LiDAR-based 3D scanning system for leaf morphological trait extraction. Front Plant Sci. 2019;10:147.

    Article  Google Scholar 

  22. Lien MR, Barker RJ, Ye Z, Westphall MH, Gao R, Singh A, et al. A low-cost and open-source platform for automated imaging. Plant Methods. 2019;15:1–14. https://doi.org/10.1186/s13007-019-0392-1.

    Article  Google Scholar 

  23. Zhou J, Applegate C, Alonso AD, Reynolds D, Orford S, Mackiewicz M, et al. Leaf-GP: An open and automated software application for measuring growth phenotypes for arabidopsis and wheat. Plant Methods. 2017;13:1–17. https://doi.org/10.1186/s13007-017-0266-3.

    Article  CAS  Google Scholar 

  24. Halcro K, McNabb K, Lockinger A, Socquet-Juglard D, Bett KE, Noble SD. The BELT and phenoSEED platforms: Shape and colour phenotyping of seed samples. Plant Methods. 2020;16:1–13. https://doi.org/10.1186/s13007-020-00591-8.

    Article  Google Scholar 

  25. Pound MP, Fozard S, Torres Torres M, Forde BG, French AP. AutoRoot: open-source software employing a novel image analysis approach to support fully-automated plant phenotyping. Plant Methods. 2017;13:1–10. https://doi.org/10.1186/s13007-017-0161-y.

    Article  Google Scholar 

  26. Valle B, Simonneau T, Boulord R, Sourd F, Frisson T, Ryckewaert M, et al. PYM: A new, affordable, image-based method using a Raspberry Pi to phenotype plant leaf area in a wide diversity of environments. Plant Methods. 2017;13:1–17. https://doi.org/10.1186/s13007-017-0248-5.

    Article  Google Scholar 

  27. Roberts EH. Temperature and seed germination. In: Symposium of the Society for Experimental Biology. 1988. p. 109–32.

  28. Hoagiand DR. Nutrition of strawberry plant under controlled conditions.(a) Effects of deficiencies of boron and certain other elements,(b) susceptibility to injury from sodium salts. In: Proceedings of the American Society for Horticultural Science. 1933. p. 288–94.

  29. W.S Rasband, ImageJ, U. S. National Institutes of Health, Bethesda. ImageJ [Internet]. Maryland, USA. https://imagej.nih.gov/ij/

  30. Legland D, Arganda-Carreras I, Andrey P. MorphoLibJ: Integrated library and plugins for mathematical morphology with ImageJ. Bioinformatics. 2016;32:3532–4.

    CAS  Google Scholar 

  31. Siddique S, Radakovic ZS, Hiltl C, Pellegrin C, Baum TJ, Beasley H, et al. The genome and lifestage-specific transcriptomes of a plant-parasitic nematode and its host reveal susceptibility genes involved in trans-kingdom synthesis of vitamin B5. Nat Commun 2021 [cited 2022 May 19]; in press.

  32. Reinhard E, Ashikhmin M, Gooch B, Shirley P. Color transfer between images. IEEE Comput Graph Appl. 2001;21:34–41.

    Article  Google Scholar 

  33. Sacco MA, Koropacka K, Grenier E, Jaubert MJ, Blanchard A, Goverse A, et al. The cyst nematode SPRYSEC protein RBP-1 elicits Gpa2- and RanGAP2-dependent plant cell death. Opperman C, editor. PLoS Pathog 2009;5:e1000564. https://doi.org/10.1371/journal.ppat.1000564

  34. Panella L, Lewellen RT. Broadening the genetic base of sugar beet: introgression from wild relatives. Euphytica. 2006;154:383–400. https://doi.org/10.1007/s10681-006-9209-1.

    Article  CAS  Google Scholar 

  35. Cai D, Kleine M, Kifle S, Harloff HJ, Sandal NN, Marcker KA, et al. Positional cloning of a gene for nematode resistance in sugar beet. Science (80-). 1997;275:832–4. https://doi.org/10.1126/science.275.5301.832.

    Article  CAS  Google Scholar 

  36. (Germany) AD-P-NB, 1992 undefined. The effects of imidacloprid on aphids and virus yellows in sugar beet. agris.fao.org [Internet]. [cited 2022 May 22]; https://agris.fao.org/agris-search/search.do?recordID=DE93U0269

  37. Märländer B, Hoffmann C, Koch HJ, Ladewig E, Merkes R, Petersen J, et al. Environmental situation and yield performance of the sugar beet crop in Germany: Heading for sustainable development. J Agron Crop Sci [Internet]. 2003 [cited 2022 May 22];189:201–26. www.blackwell.de/synergy

  38. Radakovic ZS. Identification and characterisation of Heterodera schachtii susceptibility genes AtPANB1 and HIPP27 in Arabidopsis thaliana [Internet]. PHD thesis. Rheinische Friedrich-Wilhelms-Universität Bonn; 2018 [cited 2022 May 23]. https://bonndoc.ulb.uni-bonn.de/xmlui/handle/20.500.11811/7377

  39. Rogers H, Baricz N, Pawar KS. 3D printing services: classification, supply chain implications and research agenda. Int J Phys Distrib Logist Manag. 2016;46:886–907.

    Article  Google Scholar 

  40. 3D People UK | 3D Printing Service | Order Online [Internet]. [cited 2022 May 31]. https://www.3dpeople.uk/

  41. Zhang Q, Van Wijk R, Zarza X, Shahbaz M, Van Hooren M, Guardia A, et al. Knock-down of arabidopsis PLC5 reduces primary root growth and secondary root formation while overexpression improves drought tolerance and causes stunted root hair growth. Plant Cell Physiol. 2018;59:2004–19.

    Article  CAS  Google Scholar 

  42. Mahony NO, Campbell S, Carvalho A, Harapanahalli S, Velasco-Hernandez G, Krpalkova L, et al. Deep learning vs. traditional computer vision. Arai K, Kapoor S, editors. Cham: Springer International Publishing; 2019 [cited 2022 May 17]. P. 943. http://arxiv.org/abs/1910.13796

  43. Ferentinos KP. Deep learning models for plant disease detection and diagnosis. Comput Electron Agric. 2018;145:311–8.

    Article  Google Scholar 

Download references

Funding

Work on plant-parasitic nematodes at the University of Cambridge is supported by DEFRA licence 125034/359149/3, and funded by BBSRC grants BB/R011311/1, BB/N021908/1, and BB/S006397/1. This work is also supported by an overseas foreign specialist project from the Ministry of Science and Technology of the People's Republic of China: G2021145005L.

Author information

Author notes

  1. Itsuhiro Ko

    Present address: Plant Pathology Department, Washington State University, Pullman, WA, 99164, USA

Authors and Affiliations

  1. Department of Plant Sciences, The Crop Science Centre, University of Cambridge, Cambridge, CB2 3EA, UK

    Olaf Prosper Kranse, Itsuhiro Ko, Roberta Healey, Unnati Sonawala, Siyuan Wei, Beatrice Senatori, Francesco De Batté & Sebastian Eves-van den Akker

  2. Jiangsu Collaborative Innovation Center for Modern Crop Production Co-Sponsored By Province and Ministry, Academy for Advanced Interdisciplinary Studies, Nanjing Agricultural University, Nanjing, 210095, China

    Ji Zhou

  3. Cambridge Crop Research, National Institute of Agricultural Botany (NIAB), Cambridge, CB3 0LE, UK

    Ji Zhou

Authors
  1. Olaf Prosper Kranse
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Itsuhiro Ko
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Roberta Healey
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Unnati Sonawala
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Siyuan Wei
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Beatrice Senatori
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Francesco De Batté
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Ji Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Sebastian Eves-van den Akker
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

OPK, designed the hardware and software. OPK, IK, RH, US, SW, BS, and FDB collected the data. OPK, JZ, and SEVdA wrote the main manuscript. OPK prepared the figures. All authors reviewed the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Sebastian Eves-van den Akker.

Ethics declarations

Ethical approval and consent to participate

Not applicable.

Competing interests

The authors are not aware of competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kranse, O.P., Ko, I., Healey, R. et al. A low-cost and open-source solution to automate imaging and analysis of cyst nematode infection assays for Arabidopsis thaliana. Plant Methods 18, 134 (2022). https://doi.org/10.1186/s13007-022-00963-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13007-022-00963-2

Plant Methods

ISSN: 1746-4811