RootIX: An Easy and Ready to Use Tool for the Study of Plant Root Growth and Architecture.

Background: Agronomical and plant physiological studies employ a large variety of compounds that impact root development. The effect of these compounds is often evaluated by root architecture analyses using dedicated software. During the past decade, a growing number of tools proposed complex and deep analysis of the root system architecture. While these software applications are often complex and require specic set up, here, we propose a simple method based on the most common tools used by biologists: ImageJ and Excel®. Results: First, roots are measured manually with ImageJ following a succession of operations (i. choose a plant; ii. Measure: a. the primary root - b. lateral roots). Secondly, RootIX, an Excel le with dedicated macros that will automatically extract, sort and treat the data, is launched. Tables and graphics describing the data are generated. Root length, elongation rate and additional lateral roots are the main indicators provided among many others. We evaluated the effect of algae extracts on tomato roots growth with RootIX. We highlighted that these compounds negatively affect root growth while promoting generation of lateral roots. Conclusion: Ease of use and settings as well as complete data analyses without any cost are the scaffold of RootIX.


Background
Growing in sand, mud, rocks and water, the sessile plants expand their roots to nd nutrients and water. During their deployment roots will be challenged many times when microbes would try to invade their tissues (Chuberre et al. 2018) as well as dehydration or a changing salt concentration would generate osmotic stress (Robbins and Dinneny 2015).
Thus, this organ may extend and modulate its growth to ensure the best supply of water and nutrients to the whole plant (Robbins and Dinneny 2015). The best way to e ciently carry out these functions is to optimize the volume of soil explored by the root and optimize the root surface area (Satbhai et al. 2015). These parameters are signi cantly de ned by the 3D con guration of the root system architecture. Indeed the spatial organization (root sizes, number and distribution of lateral roots) and the length and density of root hairs may affect drastically the soil/root interface surface (Satbhai et al. 2015).
The root behavior is controlled by intrinsic pathways that refer to the plant internal conditions such as hormonal balance and their receptor availability, transcription factors activation; and extrinsic pathways where the similar networks are monitored by environmental stimuli such as soil moisture or pathogens aggression (Jung and McCouch, 2013). Thus the perception of a microbe associated molecular patterns (MAMP) such as the bacterial agellin peptide ( g22) affects the root growth and lateral root formation of Arabidopsis thaliana (Beck et al. 2014; Stringlis et al. 2018).
Control of root growth is mediated by a complex balance between hormones. This implies antagonism, e.g. auxin promotes lateral root initiation (Dubrovsky et al. 2008) while it is inhibited by gibberellins (Gou et al. 2010); and interaction, e.g. jasmonates promote lateral root formation through its interaction with auxin but also inhibit primary root growth (Xue and Zhang 2007;Sun et al. 2009). To gain more details on the subtle hormone control of root growth, the reader may appreciate the nice reviews written by Jung and McCouch, 2013 or Rutten and ten Tusscher, 2019.
Despite the strong interest to understand how the hormonal balance may be controlled by a given treatment, in many cases, the point to determine is just the effect on elongation and lateral root formation.
Numerous software solutions bring extensive and deep root architectural analysis, such as WinRHIZO, IJ_Rhizo (Pierret et al. 2013) or archiDART (Delory et al. 2018). However, many scientists and papers prefer to measure root lengths with their regular tools (ImageJ, ruler,…) ( We faced this context while studying the effect of algae extracts on roots. After several failures with some of the above-mentioned programs, we get back to ImageJ manual measurements and developed the RootIX solution to extract, sort and treat the huge quantity of generated data.
Finally, we propose to share this tool that may be appreciated by all of those who just need the essential details about root growth and lateral root formation rather than complete 3D architecture, and by the way who want to save time and money.

Plant materials and growth conditions
Although this method and tool can be used with many plant and root types, we developed it on tomato (Solanum lycopersicum var. St Pierre) seedlings that were used for all the experiments.
Seeds were surface sterilized with 70% (v/v) ethanol for 1 min followed by immersion in 0.87% (v/v) bleach for 1 min. After several washes in sterile distilled water, the seeds were strati ed ve days at 4°C to promote seed germination and nally sown on half-strength MS medium (Sigma). Agar 1% (Sigma) was used as a gelling agent. Media is disposed in 24 cm x 24 cm plates (Thermo Scienti c™ Nunc™ Square BioAssay Dishes 500 cm²). Seeds were grown in a photoperiodic chamber under an 16 h daylight period at 25°C for ten days before treatment. 19 A macro is an automated program that can be launch from a simple button present in an Excel worksheet. Here, RootIX will analyze and organize the data obtained with ImageJ.
ImageJ is a public image processing program developed at the National Institute of Health. It allows to analyze images with a large variety of tools. ImageJ can be downloaded for free from http://rsbweb.nih.gov/ij/ and is available for Windows, Mac OS and Linux. Although ImageJ's potential is almost unlimited (Schindelin et al. 2015), here we only require the simplest tool of "Segmented Line" measurement (right click on the line symbol).

Images acquisition
There is no precise images properties requirement if their quality is good enough to enable precise measurement and allow to distinguish roots that cross each other.
Here, images of plant were acquired with a scanner (Epson Expression 11000XL). Since the scans were conducted on sealed plates, the focus was manually adjusted to 5.0 in order to reach the best focus. Scan images were saved in ".tif" with the following routine properties: 800 dpi, 48-bit color depth.

Root sequential measurements
Roots are measured with the ImageJ segmented line tool. To get the appropriate length values, make sure to adjust the scale.
It is required to conduct the measurements in a speci c sequence (Fig1): 1. Choose a plant 2. Measure the primary root length (See the "special cases" section below, for monocotyledons or indiscernible primary root).
3. From the primary root apex, go back up to the hypocotyl measuring each lateral root on the way. *Triple the record of a measurement for tertiary roots. *Quaternary roots or higher order of rami cation are not taken into account and should be recorded as tertiary roots. 4. Choose a second plant and follow the same sequence. 5. When the measurements of the picture are done, save the measurement results in ".tsv" (preferentially) or ".csv" le format. It is required to give a name that contains the condition (e.g. "water"), the time point (e.g. "t01h") and the replicate (e.g. "R1"). In the examples given above the le name would be: It is recommended to use at least ve decimals on the ImageJ records (on ImageJ: Analyze -set measurements… -decimal places (0-9): 5).

Data analysis
a. How to launch the analysis: Data analysis can be launched in 3 simple steps.
1. Fill in the "Parameters" sheet required cells (pale green cells) with their appropriate information following the instructions in red cells: a. File path in "B1"to access the folder that contain the data les. b. A giving name of the experiment in "B2".
c. Primary and lateral roots discrimination size in "E2" (primary root (RI) will be considered as higher and lateral root as lower than this size). See the "special cases" section below for size lower than 1cm and lateral roots longer than RI. d. Data name information of each le to analyze ("conditions" row 7 and when it is suitable "replicate", "starting" and "ending time point" respectively rows 8, 9 and 10). This batch of information is required to nd and work with the source data les.
2. Launch the macro 1: "2. Macro 1 -Collect data" that collects and gathers the data present in the indicated folder in a new Excel workbook named with the "experiment name".
b. Calibrate parameters: Although RootIX is supplied already calibrated for tomato-seedling root analysis, it is obvious that all plant species have a different root system with different shapes and sizes. Thus, it is possible to personalize this calibration.
1. It is important to adjust the primary and lateral roots discrimination size in cell "E2". Since RootIX assumes that lateral roots are shorter than primary roots, this whole number is used to discriminate between primary and lateral roots. In case the studied plants present shorter primary roots see "special cases" section below.
2. RootIX determines different classes of plants based on two parameters: the number of lateral roots per plant and the size of lateral roots. To get the most suitable ranking, values used to generate it are adjustable in the "Optional" panel: Cells C14 to C16 determine the classes of root sizes (smaller size, middle size and large sizes). While the classes are automatically de ned based on the discrimination size, they can be modi ed manually.
Cells C18 to C20 determine the classes of lateral-root number per plant.
c. Several types of calculation: RootIX collects and analyzes data from the folder whose path is lled in the worksheet (cell B1). Thus, it is required to store the les to compare in a same folder to make a comparison analysis.
The macro extracts, deduces and calculates several types of information from the data les and it presents them in up to six tables (named Table 1, Table 2, Table 6) that are disposed in a new sheet called "Analyse". The following data are grouped in Table 1 Table 2 below the previous one. Five parameters are evaluated: 1 st , it indicates the elongation rate of the whole root system (primary and lateral roots taken together).
2 nd , it estimates the part of elongation attributed to the primary root.
3 rd , the elongation rate of the primary root is indicated.
4 th , the rate of additional lateral roots is calculated. 5 th it calculates the total lateral-root length's increase.
6 th it indicates the changes of rami cation if new tertiary roots were formed.
A graph presenting these data is generated below Table 2.
When several replicates of a same condition are analyzed, average and standard deviation of these replicates are calculated for each condition in Table 3 below the second one. In this case, the graph generated present these averaged data below Table 3.
All the calculations in the "Analyse" sheet are linked to the source cells so the user can easily nd how the results are calculated.
Furthermore, RootIX also gathered in the It also offers the possibility to shift these tables in horizontal tables that can be used for ANOVA statistical analyses (i.e.: horizontal disposition required for GraphPad Prism one-way ANOVA analyses).
d. Pool the data of multiple replicates If the experiment was conducted several times or if there are several les/replicates for a same condition, it could be really useful to pool all the data of each same condition together (without affecting individual les) with the "Optional 4. Pool data" macro. It combines all the data of the different replicates in a new sheet for each condition and then reconducts the analyses on the pooled data in the sheet named "Analyse Pooled Data". e. Special cases: i. Monocotyledons or indiscernible primary root measurements If the analyzed plant is a monocotyledon with a brous root system, primary root may not be easy to reach. In that case, the rst measured root of the plant may be considered as a primary root and the others as lateral roots. The discrimination size may be set at 0 and each new plant (and so primary root) may be recorded with at least four consecutive records of the measurement (eg. 2.43566 cm record like this "1) 2.43566; 2) 2.43566; 3) 2.43566; 4) 2.43566" on the result le of ImageJ). It will trigger the manual determination of the root type.
ii. One time point measurements (no growth calculations) It may occur to compare root shapes between different conditions but regardless to time or growth evolution. In that case it is only required to complete the "conditions" information (row 7 in the "Parameters" sheet) regardless to the starting and ending time points (rows 9 and 10) that can be empty.
iii. Lateral roots longer than primary roots In certain conditions, lateral roots could be longer than primary ones. In this particular case it is required to quadruple the record of a measurement thus RootIX will consider the root as a secondary root.
iv. Most of the roots are ≤ 1 cm or units lengths need to be modi ed When studied roots are very small, the discrimination size could be smaller than one centimeter however RootIX requires a whole number. Then, the simplest solution is to convert the data in millimeters. A button named "Convert length unit of all data", reachable below the "optional" parameters, launches a macro that convert all the data in the wanted unit. Notice that it also allows to convert the unit of the data from and into all the main length units used in the West (international units (m, dm, cm, mm, µm), American units (yd, ft, In)). For example, it can convert inches to cm or cm into mm and vice versa.

WinRhizo -ImageJ measurements comparison
As part of a study that use algae extracts as biocontrol solutions, we needed to estimate the effect of them on the tomato root architecture. Before launching the RootIX creation, WinRhizo© was used to analyze root lengths. However, regarding the nature of the images and the media where roots grown, WinRhizo© overestimated the major parts of results we were looking for (number of lateral roots, root lengths) (table 1). Indeed, the number of tips detected by WinRhizo© was 175% ± 31 % higher than measures conducted with ImageJ manual measurement (table 1). The total length of roots on a plate was lightly overvalued of 30 ± 10% (table 1). These differences can be mostly explained by the di culty of the algorithm to detect the crossing roots, thus a root crossing another one may be counted twice (on both sides of the crossed root).
Since manual ImageJ measurements gave a strongly better view of the real root structure, we choose to pursue the lengths analysis with ImageJ.

Measurement
First, each ten-day-old plant of a culture plate has been measured and saved in a same le (one data le = one condition/one plate of 19.2 ±2.4 plants). Plates were scanned immediately after treatment (T0h) and four days later (T72h). Three independent experiments were conducted. Roots were measured with ImageJ following the indicated sequence for each plant: primary root (RI) followed by its lateral roots: secondary roots (RII) and tertiary roots (RIII).
Secondly, the large amount of data generated needed to be extracted, sorted and analyzed. RootIX was developed for this purpose.
RootIX was ran and gave a high number of information that were then analyzed with GraphPad Prism6 to determine their statistically reproducibility (Fig2). This analysis was conducted on the sum of the primary root length with its lateral roots' length (RI + RII length).
The rst time point presented a satisfying reproducibility with a 95% con dence intervals (CIs) (Fig2A). The second time point display a somewhat similar reproducibility except for the AlgEx1-8000 condition where the difference between the rst (R1) and second (R2) replicates is signi cant (P: 0,0175) while R1 vs third (R3) replicate and R2 vs R3 are not signi cantly different (P respectively: 0,4284 and 0,2524) (Fig2B). Keeping in mind this slight biological variability we will assume the reproducibility of the three experiments. Thus, to get a stronger analysis, data of the three experiments were pooled together with the "Optional 4. Pool data" module (see table 2 to get an overview of the generated data).

Root development
All roots lengths were compared before and 72h after treatment in order to determine how treatments affect root elongation. We evaluated the mean of RI and RII development under the different treatments in uence with the standard error of the mean (SEM) as error value. Data were also submitted to Dunnett's multiple comparison test against the water condition.
H2O treated plants presented a RI elongation of 7.98 ±0.51mm when g22 treatment slightly reduced the elongation with slight difference to 6.02 ±0.58mm (P value = 0.0360). Notice that these proportions of adventitious roots were not calculated by RootIX.
Besides lateral root formation, the size of the RII may also be affected. Indeed, 77.1% of the tested conditions have a major range of RII comprised between 0.5 and 1 mm when most of the RII from the other 22.1% are under 0.5mm (Fig4). However, all the conditions allowed the presence of RII longer than 2cm 72h after treatment, except AlgEx2 at the higher dose (Fig4).

Discussion
Root architecture gives an overview of the plant health and ability to colonize its environment. It is mainly represented by root's length and number along with the evolution of these parameters through time. Root architecture may also be conditioned by root diameter and growth direction (Satbhai et al. 2015). RootIX was created to answer quickly this kind of question: "Does the treatment affect root growth and length?".

Advantages
Our RootIX method does not require to install speci c software nor deep con gurations. It only needs the two most common biologist software ImageJ and Excel. To ensure the RootIX operation, measurement must follow this simple sequence: 1 st a plant is chosen, 2 nd its primary roots (RI) is measured, 3 rd its lateral roots (RIIs) are measured, 4 th a second plant is chosen and so forth. Finally, RootIX provides data in a few clicks: elongation rates; number of additional RIIs; RI, RII and total roots (RI+RII of a same plant) sizes (Table 2). Graphs (Fig3 and 4) are ready to be interpreted as well as tables are ready for statistical analysis. For the complete explanation of the tool functioning, please refer to the material and methods chapter.
This tool is wanted to be totally free and thus enables low-cost experiments that may be useful for researchers with tight budget or who suffers underinvestment such as in emerging and developing countries (UNESCO and Schlegel 2015). Furthermore, RootIX is totally coded with Visual Basic and thus may be upgraded by users to offer new options or to adapt its functionality to unforeseen cases.

Limits
RootIX was not designed for deep root architecture analysis but to get a rapid and precise overview of a condition effect on the root system focusing on length and number of primary and lateral roots and the evolution of these parameters through time. Since our purpose was to answer to simple question such as "Does a treatment affects root elongation or lateral root formation?", many architecture parameters were not taken in account, such as root diameter, root color or angles. For a deeper and complete architecture analysis, the user may appreciate the free powerful R packages: DART or archiDART (Le Bot et al. RootIX offers a fast analysis of data. Nevertheless, the time-consuming part of our analyses was to carry out the measurement manually to avoid errors generated by computers algorithms. Indeed, data obtained with WinRhizo© were strongly overvalued due to the di cult a liation of roots that cross each other (Table 2).
However, since it uses ImageJ, the manual measurement could be bypassed, while it is the surest method to avoid artefacts. To this end, it is possible to create/use ImageJ macro that would measure root lengths in the required sequence (RI then RIIs for each plant) to allow RootIX analyzes.
It may be frequent to record root length in a large variety of scienti c projects but the use of strong software or ImageJ packages such as evocated above are more dedicated to scientists with signi cant computing backgrounds. For example, IJ_Rhizo is an ImageJ macro and archiDART a R package, both require relevant computing capacities in ImageJ macro or R language to fully perform their operation that may be discouraging for someone with less digital experiences. Finally, soon come the time to choose to invest time in understanding how to use the software, or to simply invest the time on taking the measures and process the data with traditional methods (ruler, ImageJ manual measure, spreadsheet data treatment). Indeed some scientists prefer to use

Conclusion
RootIX gave a strong and quick access to the essence of simple measurements conducted on roots. Indeed, this tool and method pointed out the dose effect of our treatments on root elongation.
RootIX and this method were developed to give a free to use and easy to reach tool to quickly answer the question of a giving condition effect's on root growth.
Furthermore, since it uses common scientists' tools, ImageJ© and Excel©, users may be able to improve and adapt RootIX to their own speci cities. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
We gratefully acknowledge support of our research on Root Health and Protection by the ADEME (Agence de l'environnement et de la maîtrise de l'énergie), the University of Rouen, the "Structure fédérative de recherche SFR-Normandie Végétal" from Normandie (France), and the FEDER.
Authors' contributions MG managed the experiments, analyzed the plant root images and data, created the RootIX method and tool, wrote the major part of the manuscript. QA and GD prepared and generated the plant cultures, treatments and images. YL participated to the software conception and paper design. AD was a major contributor in writing the paper and supervised the scienti c project. MLFG was involved in the data analysis and interpretation. ENO supervised the research project, was involved in the paper organization and corrected the manuscript. MV coordinated the research project, was strongly involved in the paper organization and redaction and corrected the manuscript.
All authors read and approved the nal manuscript.   Figure 1 Required measurement steps. Each plant is measured following a simple sequence of steps that allows RootIX to identify the different root types. A. table will be saved as a "csv" le. One le must be related to one condition, one time point and n plants .

Figure 2
Differences between replicates root length means before treatment (A) and 72h post treatment (B). Each difference is represented as a central dote with 95% con dence intervals. Sum of each primary root length with their lateral roots' length (RI + RII length) was calculated by RootIX and provided in the " Table 4" of the "Analyse" sheet. All replicates from this table were compared with each other using GraphPad Prism 6.0 Tukey's multiple comparison test to evaluate their