Each method of measurement investigated in the current work (including the novel rind penetration method) had its attendant advantages and disadvantages. Comparisons can be drawn between the different measurement techniques in the following areas: cost of tools required, training required to perform the measurements, time required to carry out the measurements, the random error introduced by the user, and consistency of measurement values compared to the other methods studied in this paper. Each of these areas is discussed in the paragraphs below.
Cost of Tools
Each measurement method used in this study required at least one tool. The caliper method required a pair of digital calipers. A representative cost of a pair of digital calipers is $20 to $100. The image analysis technique required a computer, a scanner, and software. The cost of the computer and scanner together are estimated at $400 to $1,000. The software (i.e. ImageJ) was open source and therefore incurred no cost. The rind puncture method required an Instron universal testing frame, a computer, and a MATLAB license. Together, these items cost approximately $50,000. In summary, the most expensive method was the rind puncture method by a wide margin, while the least expensive was the caliper method.
Training for all three methods took approximately the same amount to time to carry out. For the caliper method, a 10 minute demonstration of proper caliper usage was all that was required. For the image analysis method, researchers watched a 10 minute video to familiarize themselves with the software tools they would be using. For the rind puncture method, training consisted of a 10 minute demonstration of the procedure. In other words, each method required approximately 10 minutes of training.
Time to Complete Measurements
The total time spent by all researchers in carrying out the various measurements are summarized in Table 3. The image analysis method was the most time consuming, requiring 630 minutes to complete. The most time intensive process for image analysis was sectioning the stalk samples (450 minutes). The caliper method was the next most time intensive requiring 430 minutes to complete. The least time intensive method was the rind puncture method, which required only 65 minutes. It should be noted that several automated image analysis algorithms have been presented in the literature that could reduce the time to scan and compute stalk diameter and rind thickness (5, 6, 16-21). However, the authors are unaware of any reported high throughput sectioning procedures that would reduce time to section stalk cross-sections below the reported 450 minutes required in this study. Thus, the rind puncture methods would still be significantly faster even if automated image analysis algorithms were employed.
Inter user variability was measured for the caliper and image analysis methods and was found to be small to moderate (< 2% for diameter measurements and < 9% for rind thickness measurements). Because a given measurement site can only be punctured a single time, no comparisons of inter user variability were made for the puncture method. The authors expect inter user variability for rind thickness measurements to be significantly higher when measuring pith filled plant stems as it can be difficulty to determine the boundary between pith and rind.
Agreement Between Methods
To determine the level of agreement between measurement systems a linear correlation analysis was conducted. Each system exhibited coefficients of determination (R2 values) greater than 0.84 (see Table 3). The high level of agreement between the different methods suggests that any of these methods could be used to obtain accurate measurements of rind thickness and diameter. The attendant advantages and disadvantages of each method are discussed in the sections below.
Advantages/Disadvantages of Caliper Measurements
Calipers are an inexpensive tool. They are easy to obtain and easy to use. They can be used with equal ease in a laboratory or in the field. However, their capacity for high throughput measurements of rind thickness is limited, making measurements of large sample sets impractical. Calipers would be a preferred tool in studies requiring immediate measurements of rind thickness of plant stalks/stems for a relatively small sample set (i.e., < 100 samples/user).
Advantages/Disadvantages of Image Analysis Methods
The Image analysis method to determine stalk/stem diameter and rind thickness was effective but required more sample preparation (i.e. cutting a thin cross section capable of being placed on a flatbed scanner) than the other two methods. In this study 15% of the samples (i.e. 17 internodes) were destroyed during sectioning. This method requires tools that are not easily portable to the field, limiting this method primarily to laboratory settings. Image analysis would be a preferred method for experiments requiring measurements of rind thickness of small to large sample sizes in laboratory settings, so long as the samples are easy to section. An added advantage of the image analysis method is that it does not requiring contacting the sample. When measuring soft or deformable samples caliper readings are highly dependent upon the amount of force the user applies to the sample. Image analysis techniques and other non-contact methods are often more suited to measure such samples as compared to calipers.
Advantages/Disadvantages of the Rind Puncture Method
The rind puncture method is the only non-lethal method of measuring rind thickness that could potentially be used in a field setting. For example, puncture tests are frequently used in field studies of maize (Zea mays) and other grasses to assess stalk strength without inducing plant fatality (e.g., 18). In this study a universal material testing frame / system was used to conduct the puncture tests. Materials testing frames are largely immobile and inappropriate for field work. The authors chose to use a universal testing frame to validate the puncture test methodology. However, they are currently developing a portable handheld device to conduct puncture tests of plant stems and stalks. The primary advantage of such a device would be the ability to determine stalk diameter and rind thickness in the field without inducing plant fatality. The authors are not aware of any other methods to non-destructively measure rind thickness in a field. In the meantime, laboratory-based puncture tests which utilize a universal testing system can offer a high degree of automation, allowing for high throughput measurements of rind thickness and diameter. Such tests are best suited for large sample sets (> 100 samples). Table 4. presents a quantitative summary of each methods advantages and disadvantages. The last column of the table presents the average of the R2 values between the given technique and the two other methods investigated in the study.
Complexity of Obtaining Accurate Rind Thickness Measurements
Rind thickness measurements for all methods demonstrated lower R2 values than diameter measurements. This was partly due to difficulty associated with identifying the correct plane of measurement. For example, for the image analysis and caliper measurements there was uncertainty as to what two points should be used to calculate rind thickness when a geometric irregularity in the stalk cross-section was at or near the measurement location. Caliper measurements of rind thickness were also sensitive to variations in pressure applied by the user.
High throughput measurement techniques are needed to advance many fields of plant science. The lack of high throughput phenotyping methods is a major bottleneck to plant improvement and to increasing food security worldwide (22). For example, high throughput methods of measuring rind thickness and diameter will greatly aid in addressing the problem of stalk lodging (plants breaking prior to harvest). Stalk lodging is a multi-billion dollar a year problem in grain crops and causes harvest losses that are estimated to range from 5% - 20% globally (23). Stalk geometry (diameter and rind thickness) account for as much as 70% of the variation in stalk strength (24). Because most plant stems are tapered measurements of diameter and rind thickness need to be acquired at multiple locations along the length of the stem. For example, the novel puncture technique outlined in this paper was recently used by the authors to acquire diameter and rind thickness measurements of every internode of a set of 1000 maize stalks. The collected data enabled an in-depth analysis of the taper and load bearing structure of the measured stalks and revealed insights into the structural efficiency of maize stems (25).