This Article Abstract Full Text (PDF) Supplemental Data Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (14) Citing Articles via Google Scholar Google Scholar Articles by Brewer, M. T. Articles by van der Knaap, E. Search for Related Content PubMed PubMed Citation Articles by Brewer, M. T. Articles by van der Knaap, E. Agricola Articles by Brewer, M. T. Articles by van der Knaap, E.

BIOINFORMATICS

Development of a Controlled Vocabulary and Software Application to Analyze Fruit Shape Variation in Tomato and Other Plant Species^1,[W]

Department of Horticulture and Crop Science, The Ohio State University/Ohio Agricultural Research and Development Center, Wooster, Ohio 44691 (M.T.B., E.v.d.K.); Department of Statistics (L.L.) and Department of Computer Science and Engineering (K.F.), The Ohio State University, Columbus, Ohio 43210; and Department of Mathematics and Computer Science, The College of Wooster, Wooster, Ohio 44691 (N.D., S.G.)

	ABSTRACT

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 
The domestication and improvement of fruit-bearing crops resulted in a large diversity of fruit form. To facilitate consistent terminology pertaining to shape, a controlled vocabulary focusing specifically on fruit shape traits was developed. Mathematical equations were established for the attributes so that objective, quantitative measurements of fruit shape could be conducted. The controlled vocabulary and equations were integrated into a newly developed software application, Tomato Analyzer, which conducts semiautomatic phenotypic measurements. To demonstrate the utility of Tomato Analyzer in the detection of shape variation, fruit from two F₂ populations of tomato (Solanum spp.) were analyzed. Principal components analysis was used to identify the traits that best described shape variation within as well as between the two populations. The three principal components were analyzed as traits, and several significant quantitative trait loci (QTL) were identified in both populations. The usefulness and flexibility of the software was further demonstrated by analyzing the distal fruit end angle of fruit at various user-defined settings. Results of the QTL analyses indicated that significance levels of detected QTL were greatly improved by selecting the setting that maximized phenotypic variation in a given population. Tomato Analyzer was also applied to conduct phenotypic analyses of fruit from several other species, demonstrating that many of the algorithms developed for tomato could be readily applied to other plants. The controlled vocabulary, algorithms, and software application presented herein will provide plant scientists with novel tools to consistently, accurately, and efficiently describe two-dimensional fruit shapes.

Tomato fruit is classified according to 10 shape categories such as rounded, high rounded, ellipsoid, or pyriform (International Plant Genetic Resources Institute, 1996; International Union for the Protection of New Varieties and Plants, 2001). Additionally, the distal end of the fruit is categorized as indented, flat, or pointed, whereas the proximal end of the fruit is categorized as flat or indented (International Union for the Protection of New Varieties and Plants, 2001; International Plant Genetic Resources Institute, 1996). While these classifications are useful to group tomato varieties and describe cultivars, the classification scheme cannot be utilized to conduct precise quantitative measurements in a reliable and systematic manner. In addition, the terminology of fruit shape attributes is not described in sufficient detail and tends to be taxon specific. While the taxon-specific terminology may not pose a problem for intraspecies comparisons, cross-species comparisons may be hampered by the lack of agreed-upon terms describing common attributes. Thus, querying information across databases and facilitating retrieval would require development of a structured controlled vocabulary that is arranged in ontologies. Trait ontology provides a structured framework to describe and quantify plant phenotypes (Bruskiewich et al., 2002; Yamazaki and Jaiswal, 2005). Objective phenotypic evaluation can then be conducted for comparative purposes and genetic analysis of quantitative trait loci (QTL). Therefore, a first step toward a comprehensive fruit trait ontology database requires development of agreed-upon terms and associated descriptions for each attribute or trait.

To date, most phenotypic analyses consist of time-consuming manual measurements and subjective scoring of traits that limit the detection of underlying genes. Software-aided measurements of attributes such as height, width, area, and perimeter are most commonly conducted with ImageJ, a public domain program developed at the U.S. National Institutes of Health (program created by W.S. Rasband; http://rsb.info.nih.gov/ij/). ImageJ is a versatile program that allows the user to make minor adjustments to the image. However, objective measurements such as height and width are neither automated nor exported efficiently, and extensive and detailed phenotypic analyses that describe subtle differences in shape such as degree of circular shape and the slope along the boundary of the fruit require development of novel algorithms for the attributes. In addition, many of the descriptors necessary to characterize shape cannot be scored objectively without proper software tools. For example, the degree of distal end indentation would be difficult to rate on a scale, and the scoring of this trait would be inconsistent between years, plots, and persons. Moreover, precise phenotypic measurements are necessary to sufficiently characterize loci and the underlying genes that contribute to shape variation. Thus, an accurate and objective method for conducting phenotypic analyses combined with a concise and detailed set of descriptors and terms for fruit shape attributes is necessary.

In previous work, algorithms for some of the shape attributes were defined and shown to have a genetic basis (Van der Knaap and Tanksley, 2003). Here, additional trait terms and mathematical descriptors of shape attributes were developed to further improve phenotypic analyses. The terms were proposed for broader acceptance in the community and future incorporation into a trait ontology database. The mathematical descriptors were implemented in a software application, Tomato Analyzer. This program performs semiautomatic, objective, and quantitative measurements of attributes that will accelerate phenotypic analyses and eliminate subjective scoring of many fruit shape traits. Consequently, Tomato Analyzer can provide more accurate, consistent, and objective results and measure shape attributes that are impossible or impractical to determine manually. To validate that Tomato Analyzer functions properly and to show its use for fruit other than tomato, we used the software to collect phenotypic data on tomato fruit from segregating F₂ populations as well as on a sample of fruit from other species.

	RESULTS

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 

Trait Ontology Terms and Mathematical Descriptors

An accurate description of fruit shape requires the development of a common vocabulary that encompasses a range of plant species. The structure of the terms needs to follow the True Path Rule that states that the pathway from a child term all the way up to the parent term must be accurate (Bruskiewich et al., 2002). As an example, Figure 1 displays the child term distal end fruit shape angle as an instance of distal end fruit shape. In turn, distal end fruit shape is an instance of fruit end shape and follows the path fruit trait, fruit shape, and fruit end shape. The trait terms and path relationships that were developed are listed in Figure 1. Instances of fruit trait are fruit shape and fruit size. Fruit shape is a parent term of fruit shape index, fruit shape triangle, fruit shape eccentric, and fruit end shape. Fruit shape eccentric and fruit end shape are parent terms of additional child terms (Fig. 1). The remaining fruit shape terms, fruit shape circular, ellipsoid, heart, and rectangular, describe how well the fruit surface depicts a circle, ellipse, etc. and determine the uniformity and homogeneity of objects. Instances of fruit size are fruit mass, fruit area, fruit height, fruit perimeter, and fruit width (Fig. 1). Tomato-specific synonyms for some of the terms are listed as well as the acronyms for each term. The acronyms will be used for detailed QTL analyses pertaining to fruit shape.

View larger version (13K): [in this window] [in a new window]   Figure 1. Trait ontology terms. Generic terms and their relationship as instance of other terms are indicated by the arrows. The synonyms are tomato-specific terms of the generic terms. The acronyms are used in QTL analyses. The Image column lists the Figure 2 section that displays how the mathematical descriptor for each term or acronym is measured. na, Not applicable.

View larger version (107K): [in this window] [in a new window]   Figure 2. Descriptors of fruit morphology traits. A, Fruit shape index I: ratio of maximum height to width, H/W. Fruit shape index II: ratio of mid height to mid width, Hm/Wm. B, Fruit shape triangle: the ratio of proximal width to distal width, w1/w2. C, Fruit shape eccentric obovoid: the position of the widest width of the fruit, 4 x (y – 0.5) if y > 0.5; 0 otherwise. D, Fruit shape eccentric horizontal asymmetry, (|n – ni|)/number of columns L. E, Fruit shape eccentric vertical asymmetry, (|m – mi|)/number of rows L. F, Distal fruit end shape angle at position 5% above the tip from the fruit. G, Distal fruit end shape angle at position 5% above the tip from the fruit. H, Distal fruit end blockiness: ratio of fruit width at the distal end to mid width, w2/Wm. I, Distal end indentation area relative to total fruit area. J, Proximal fruit end shape angle. K, Proximal fruit end blockiness: ratio of fruit width at the proximal end to mid width, w1/Wm. L, Proximal fruit end indentation: shoulder height, (h1 + h2)/2H. M, Proximal fruit end indentation: area, indentation area relative to total fruit area. N, Fruit shape circular, fitting precision R2. O, Fruit shape ellipsoid, fitting precision R2. P, Fruit shape heart: taperness function, 1 – w2/W + w1/W. w2 = average width below widest width W; w1 = average width above widest width W. Q, Fruit shape rectangular: the ratio of maximum area inscribing the rectangle to the minimum area of the enclosing rectangle, Sin/Sout.

Fruit Shape Index
Shape index is defined as the ratio of height to width (Fig. 2A ). The software contains two acronyms for fruit shape index. The first acronym, fs I, is the ratio of maximum height to maximum width (H/W). The second acronym, fs II, is the ratio of the height at mid width to the width at mid height (H_m/W_m; Fig. 2A). For both descriptors, a value greater than 1 indicates an elongated fruit, equal to 1 indicates a round fruit, and less than 1 indicates a squat fruit. Typically, the results of these two measurements are very similar and the user can select which one describes the shape of a particular object most appropriately.

Fruit Shape Triangle
The software defines triangle as the ratio of the proximal end width to distal end width, w₁/w₂ (Fig. 2B). The width is measured at user-defined distances from the proximal end. A fruit shape triangle value greater than 1 indicates that the proximal end of the fruit is wider than the distal end of the fruit, while a value less than 1 indicates that the distal end of the fruit is wider.

Fruit Shape Eccentric
The ovoid and obovoid functions describe how top or bottom heavy a fruit is, respectively. Thus, the fruit displayed in Figure 2C is considered obovoid shaped. The degree of ovoid or obovoid is calculated by identifying the position of the widest section, y, of the fruit (Fig. 2C). The following formula is then used to describe obovoid: 4 x (y – 0.5) if y > 0.5; 0 otherwise, indicating the fruit is not obovoid. The following formula describes ovoid: –4 x (y – 0.5) if y < 0.5; 0 otherwise, indicating the fruit is not ovoid.

Horizontal asymmetry and vertical asymmetry describe how asymmetric a fruit is when divided along a horizontal or vertical axis, respectively. The horizontal or vertical axes that divide the fruit are termed n and m, respectively (Fig. 2, D and E). The position of the horizontal axis, n, is determined by finding the topmost and bottommost points of the fruit and dividing by two to find the center. Likewise, the position of the vertical axis, m, is determined by finding the leftmost and rightmost points of the fruit and dividing in half to find the center. To compute horizontal asymmetry, each column of pixels, termed L_i, is determined, and the midpoint of the column, n_i, is found (Fig. 2D). Next, the difference between n_i and n is calculated and recorded. Once every column is examined, the sum of the differences is determined and divided by the number columns. Thus, the general formula for horizontal asymmetry is (|n – n_i|)/number of columns. For vertical asymmetry, each row of pixels, termed L_i, is determined, and the midpoint of the column, m_i, is found (Fig. 2E). Next, the difference between m_i and m is calculated and recorded. Once every row is examined, the sum of the differences is determined and divided by the number rows. Thus, the formula for vertical asymmetry is (|m – m_i|)/number of rows. Vertical and horizontal asymmetry values of 0 signify a perfectly symmetric shape. Horizontal asymmetry ovoid is defined by the general formula for horizontal asymmetry if there is more area above the horizontal axis n than below it; otherwise, horizontal asymmetry ovoid equals 0. Similarly, horizontal asymmetry obovoid is defined by the general formula for horizontal asymmetry if there is more area below the horizontal axis n than below it; otherwise, horizontal asymmetry obovoid equals 0.

Distal Fruit End Shape
The angle of the distal fruit tip refers to the intersection of two lines where the slope is measured via regression along the boundary of the fruit on both sides at a user-defined distance from the distal end of the fruit (Fig. 2, F and G). The slope is determined by the regression measured at ±5% when the user-defined positions are between 5% and 50% from the distal end (macro setting), whereas the slope is determined by the regression measured at ±2% when the user-defined positions are between 2% and 10% from the distal end (micro setting). The angle is measured at the point where the lines intersect and is expressed in degrees, where 180° is flat, greater than 180° is indented (Fig. 2F), and less than 180° is pointed (Fig. 2G). The term blockiness is referred to as squared or box-like shapes. Blockiness is calculated as the ratio of the width at a user-selected proportion of the height closest to the distal end of the fruit to the mid width, w₂/W_m (Fig. 2H). Indentation refers to the area of indentation at the distal end of the fruit (Fig. 2I). The distal indentation area is determined by finding the most indented distal end position, the lowest distal end on both sides of the indented position, and by the boundary of the fruit along the indented area. The computational method is described in more detail below (shoulder height; Fig. 2L). The distal end indentation area is the ratio of indentation area to total fruit area.

Proximal Fruit End Shape
The angle of the proximal fruit end refers to the angle from the shoulder points to the site of pedicel attachment or the proximal end, where 180° is flat and greater than 180° is concave (Fig. 2J). Blockiness is calculated as the ratio of the width at a user-selected proportion of the height from the top of the fruit to the mid width, w₁/W_m (Fig. 2K). Indentation is measured by two methods, either as the "shoulder height, psh" (Fig. 2L) or as the "indentation area, piar" (Fig. 2M). Shoulder height is calculated by first locating the most indented point (P) at the proximal end. A straight vertical line from point P to the center of gravity is drawn. Next, a line perpendicular to the vertical line is drawn through the boundary, selecting the intersection points A and B. Finally, shoulder height points h₁ and h₂ are defined to be the maximal distance from the arc A-P and B-P to the line AB, respectively. Shoulder height is defined as (h₁ + h₂)/2H (Fig. 2L). The larger the shoulder height value is, the more indented the fruit at the proximal end. The indentation area is calculated by finding the area between the two shoulder points, the lowest position P, and the boundary. The "proximal end indentation area, piar" is the ratio of the indentation area to the total fruit area.

Fruit Shapes Circular, Ellipsoid, Heart, and Rectangular
These functions are related to homogeneity and uniformity, that is, similarity of the object to the common shapes circle, ellipse, heart, and rectangle. The method developed for circular and ellipsoid determines the fitting precision R². This value represents the coefficient of determination and reflects how well the actual shape fits a circle or ellipse based on regression (Fig. 2, N and O). The closer the value to 1, the more similar the fruit is to a circle or ellipse. Heart shape is a function of three object characteristics: the location of the maximum width y (Fig. 2C), the shoulder height (Fig. 2L), and the taperness (Fig. 2P). To calculate heart, the location of the maximum width is described as 1 – y, where y is the widest point. The shoulder height is described as (h₁ + h₂)/2H (see above). The taperness is described as 1 – w₂/W + w₁/W. w₁ and w₂ are the average width above and below the widest width, respectively. The weight of these individual components is expressed as 0.25 x [(1 – y) x (1 – w₂/W + w₁/W)] + 20 x [(h₁ + h₂)/2H] and is returned as the heart shape value. Rectangular is calculated as the ratio of S_in/S_out where S_out is the minimum area of an enclosing rectangle and S_in is the maximum area of the inscribing rectangle (Fig. 2Q). Thus, the closer the value is to 1, the more rectangular the shape of the object.

Fruit Size
Fruit size measurements calculated by Tomato Analyzer include area, width, height, and perimeter. For width, two descriptors are available, the widest width (fd I, largest horizontal cross section) and the width measured at the midpoint of the height (fd II; Fig. 2A). There are also two descriptors for height available: the highest height (fl I, largest vertical cross section) and the height measured at the midpoint of the width (fl II; Fig. 2A).

Tomato Analyzer Application

The Tomato Analyzer application requires digital images of cut fruit saved in jpeg format. When loaded into the application, the entire image appears in the left viewing window (Fig. 3 ). The software is designed to recognize objects of a certain size and image resolution, measured in dots (pixels) per inch (dpi). Generally, the smaller the object, the higher the resolution required to provide accurate analysis. The implementation of the equations with the software relies on obtaining the x and y coordinates of a pixel in a jpeg image of the fruit objects. The software automatically determines the boundaries of fruit in a scanned image. The object boundary is extracted through contour tracing, which results in a list of adjacent points describing the border of an object in an image. All phenotypic measurements are calculated based on the boundaries.

View larger version (66K): [in this window] [in a new window]   Figure 3. Tomato Analyzer application.

The units used for the attribute values can be selected as pixels, centimeters, millimeters, or inches. As most of the measurements are ratios, selecting the appropriate dpi setting consistent with the resolution of original jpeg image is only required for the size measurements including perimeter, area, height, and width.

The measurements saved setting allows the user to select which attribute values to compute for display and export. Individual attributes or an entire measurement cluster can be selected or deselected. Shape attributes are divided into several clusters in Tomato Analyzer application and largely follow the grouping listed in Figure 1. Basic Measurements group comprises all the fruit size traits; Fruit Shape Index comprises fruit shape index I and II; Homogeneity comprises circular, ellipsoid, and rectangular; Distal Fruit End Shape comprises macro and micro angle, indentation, and protrusion; Proximal Fruit End Shape comprises angle, shoulder height, and indentation area; Eccentricity comprises all eccentricity attributes in addition to fruit shape heart; and Blockiness comprises distal and proximal fruit end shape blockiness and fruit shape triangle. By selecting the corresponding group tab in the lower right corner of Tomato Analyzer, results for the selected group are displayed (Fig. 3).

Some attributes allow the user to select settings that maximize phenotypic diversity. User-defined settings are offered for upper and lower blockiness positions. The upper position is used for calculating proximal end blockiness. The lower position is used to calculate distal end blockiness. In addition, these new values will affect fruit shape triangle. The values for the blockiness positions equal the percentage of the height from the top of the fruit. These values can be changed to any number as long as they are both between 0 and 1 and lower position is greater than upper position. There are two settings to calculate distal end angles, referred to as macro and micro. The setting for macro level determines the percentage of the perimeter from the bottom where the angle will be measured, ranging from 5% to 50%. The micro level setting determines where the proximal angle is measured, ranging from 2% to 10% from the tip of the fruit.

The save function allows the user to save the manual adjustments and analyzed fruit shape attributes. Subsequently, when a user opens the original image file, the saved file will be opened and will display all of the adjustments.

There are two methods to export data obtained by Tomato Analyzer. In the first, one image is analyzed and the data for the shape attributes of individual fruit is exported to a .csv file suitable for loading into a spreadsheet or statistical analysis package. The second method is called batch mode and allows more than one image to be loaded and analyzed. In this scenario, the data for each attribute is exported to a .csv file as an average of all fruit in a single image. This method of export is most useful when conducting QTL analyses.

Phenotypic Analyses Conducted by Tomato Analyzer

To validate the accuracy and demonstrate the utility of the Tomato Analyzer software, phenotypic and genetic analyses were conducted on two F₂ populations derived from crosses between one of the extreme-shaped S. lycopersicum cultivars, Howard German or Banana Legs, and a small, round, wild relative, Solanum pimpinellifolium LA1589 (Fig. 4, A–C ). Phenotypic data were collected for fruit from both populations using Tomato Analyzer. Principal components analysis (PCA) was conducted to determine the major sources of variation among the morphological traits within and between these populations (Fig. 4D). The variation within each of the two F₂ populations could be explained by the first three principal components (PC), which combined represented 77.4% of the total variability (Fig. 4D). In addition, PC 3 demonstrated significant differences between the Howard German and Banana Legs F₂ populations, whereas PC 1 and PC 2 were not significantly different (Fig. 4D; Supplemental Table I). PC 1, representing 35.4% of the variation, was predominantly affected by attributes that evaluate the tapered shape of fruit, including the traits fruit shape triangle, proximal end blockiness, and horizontal asymmetry ovoid (loading values for contributing traits are listed in Supplemental Table II). PC 2, representing 24.9% of the variation, was primarily controlled by traits that contribute to fruit elongation such as fruit shape index, proximal end shape features, and distal end angle, as well as the fruit shape uniformity and homogeneity features circular, rectangular, and heart shape (Supplemental Table II). PC 3, representing 17.1% of the variation, was influenced by fruit size, homogeneity, and eccentricity characteristics, as well as distal end blockiness and indentation area (Supplemental Table II).

View larger version (40K): [in this window] [in a new window]   Figure 4. Phenotypic variation within the Howard German and Banana Legs populations. The fruit images in A, B, and C are depicted to scale. A, Image of S. lycopersicum cv Howard German. B, Image of S. lycopersicum cv Banana Legs. C, Image of S. pimpinellifolium accession LA1589. D, Graphical display of PC 1, 2, and 3 from analysis of fruit morphology traits in the Howard German (HG) x LA1589, and Banana Legs (BL) x LA1589 F2 populations. Clouds represent the means and SE of the first three PCs. The variation described by each of the PCs is listed in parentheses.

View larger version (12K): [in this window] [in a new window]   Figure 5. LOD curves for distal fruit end angle measurements on chromosome 7. Graphical depiction of the LOD curves for distal fruit end angle measurements at the 2%, 5%, 10%, and 20% above the distal tip of the fruit. The results were derived from the Banana Legs F2 population. The x axis displays the molecular markers along the chromosome; the y axis displays the LOD score.

View larger version (40K): [in this window] [in a new window]   Figure 6. Variety of fruit that were subjected to phenotypic evaluations by Tomato Analyzer. A, Butternut squash. B, Yellow squash. C, Large jalapeno. D, Banana pepper. E, Chili pepper. F, Jalapeno. G, Grape. H, Strawberry (Fragaria spp.). I, Bartlett pear.

	DISCUSSION

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Software programs and computational methods have been developed to categorize and classify plant organ shapes such as fruit (Morimoto et al., 2000; Beyer et al., 2002) and seeds (Sako et al., 2001a, 2001b). Many of these applications were developed to describe shape uniformity and overall quality of fruit and seed lots. Despite the usefulness of categorizing objects for quality purposes, these methods could neither measure specific shape features, such as the degree of indentation of the proximal fruit end, nor be applied to quantitative genetic analyses. Recently, a method was employed to describe leaf shape by calculating the distance between coordinates along the leaf surface and subjecting the measurements to PCA to determine the greatest sources of variation (Langlade et al., 2005). In addition, the PCs were considered as traits in QTL analyses, and loci were identified contributing to the variation in leaf morphology (Langlade et al., 2005). The described method identifies overall changes in leaf shape but cannot identify individual features of shape, for instance the angle along the boundary or the position of the widest width of the leaf. Consequently, it is difficult to discern which locus controls a specific shape feature and vice versa which attribute is controlled by a specific locus. Tomato Analyzer, on the other hand, analyzes specific shape traits, which were combined and subjected to PCA and subsequent QTL analyses. Individual shape attributes that most significantly contribute to each PC and underlying QTL could be discerned (Supplemental Table II). One prime example of coincidence of individual trait QTL and PC QTL is offered by PC 2 and the attributes that most significantly contribute to this component. The major PC 2 QTL is located on chromosome 7 near marker cos103. This result suggests that the attributes comprising PC 2 are controlled, at least in part, by the same locus. Indeed, when we map the individual traits such as fruit shape index, distal fruit end shape angles, and the proximal fruit end shape features, which comprise PC 2, we identify QTL for these traits at a similar location on chromosome 7 (Fig. 5; data not shown). In addition, these QTL coincide with the sun locus that is known from our previous studies to control fruit shape index (Van der Knaap and Tanksley, 2001; Van der Knaap et al., 2004). Another interesting finding is that PC 3 is controlled in part by traits affecting fruit size (Supplemental Table II). The major size QTL are typically located on chromosomes 1, 2, 3, 7, and 11 (Lippman and Tanksley, 2001; Van der Knaap and Tanksley, 2003), whereas the QTL controlling PC 3 in both populations is located on chromosome 10 and not known to play a major role in controlling fruit size in tomato. This result suggests that employing PCA on several traits may lead to identification of novel QTL and/or improve significance of minor QTL.

In all, Tomato Analyzer is a flexible and comprehensive application that provides intuitive descriptors and output that facilitate the analysis of fruit morphology. Furthermore, our efforts to combine controlled vocabulary with mathematical descriptors into one software application make this a very useful tool for several applications. Tomato Analyzer allows for accurate and objective measurements of fruit shape attributes in a high-throughput manner and of traits that are nearly impossible to quantify manually. The application is specifically developed to analyze fruit shape QTL in tomato but could readily be applied to fruit of other species and other plant organs such as seed and leaves.

	MATERIALS AND METHODS

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 

Software Implementation

The software application has been implemented in C++ using Visual Studio 6.0 and runs on the Windows operating system. The image processing library Computer Vision and Image Processing (CVIP) 3.7c is used for image I/O. Modifications to the software were done using Visual Studio 2003 with source code control provided by SourceSafe. The source code cross indexer LXR was used to create an online indexable and searchable version of the software code. The program is free for academic purposes and can be downloaded from our laboratory Web site: http://www.oardc.ohio-state.edu/vanderknaap/.

Plant Material

Two F₂ populations were constructed from crosses between one of two Solanum lycopersicum cultivars (Banana Legs or Howard German) and a wild species, Solanum pimpinellifolium accession LA1589. The Banana Legs population consisted of 99 plants, whereas the Howard German population contained 130 plants. Both populations were grown simultaneously in the same greenhouse in the summer of 2003. Up to eight fruits were harvested from each plant. Fruit were weighed, cut longitudinally, and scanned at 300 dpi resolution. The images were saved as jpeg image files for phenotypic analyses. Each image contained fruit from only one plant. If eight fruit per plant were harvested at once, these fruits were scanned and saved as one image. If fewer than eight fruit per plant were harvested at once and additional fruits were harvested later, both images were combined using Adobe Photoshop 7.0 (Adobe Systems) prior to analysis with Tomato Analyzer.

Phenotypic Analyses

Tomato Analyzer was used to conduct the phenotypic analyses. Manual adjustments, including modification of fruit boundaries and rotation of individual fruits, were made to the images, if necessary. If boundaries were distorted, for example by an attached seed, they would be modified using the software. If fruit were scanned at an angle, manual rotation was required to properly align them. Adjusted images and associated results were saved as a separate file with the extension .tmt. Batch analyses allowed analysis and exportation of selected shape attribute data for at least 100 images at once. The values for each attribute were averaged for all fruit per image and exported into .csv file.

Genotypic and Statistical Analysis

Total genomic DNA was isolated from young leaves as described by Bernatzky and Tanksley (1986) and Fulton et al. (1995). The genetic map was constructed with a combination of RFLP and PCR-based markers. Additional information on RFLP and PCR-based markers, including map location and primer information, can be found on the Solanaceae Genomics Network Web site (http://www.sgn.cornell.edu). In addition, marker cos103 is the same marker as Lp103E7-R in Van der Knaap et al. (2004). A molecular linkage map was constructed with MAPMAKER v2.0 and the Kosambi mapping function (Kosambi, 1944; Lander et al., 1987).

PCA and analysis of variance were conducted with SAS V8 (SAS Institute). QTL analysis was performed by composite interval mapping (Zeng, 1993, 1994) using model six with five marker cofactors selected by forward regression and a 10-cm window size, as implemented in Windows QTL Cartographer v2.0. Log of the odds (LOD) scores greater than 4.0 were considered significant. Additive, dominance (d) effects and percent of phenotypic variance explained by the QTL (R²) were estimated with Windows QTL Cartographer at highest probability peaks.

	ACKNOWLEDGMENTS

 
Drs. R. Pratt and D.M. Francis are acknowledged for their critical reading of the manuscript, and Dr. P. Jaiswal for insightful comments and suggestions on the trait ontology terms. Also, we acknowledge J. Moyseenko for excellent plant care and assistance in genotyping, and B. Stricker and R. Drushal for their contributions on Tomato Analyzer.

Received January 25, 2006; returned for revision March 14, 2006; accepted March 15, 2006.

	FOOTNOTES

 
¹ This work was supported by the National Science Foundation (grant no. DBI 0227541).

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Esther van der Knaap (vanderknaap.1{at}osu.edu).

^[W] The online version of this article contains Web-only data.

www.plantphysiol.org/cgi/doi/10.1104/pp.106.077867.

^* Corresponding author; e-mail vanderknaap.1{at}osu.edu; fax 330–263–3887.

	LITERATURE CITED

TOP ABSTRACT RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

 
Alba R, Payton P, Fei Z, McQuinn R, Debbie P, Martin GB, Tanksley SD, Giovannoni JJ (2005) Transcriptome and selected metabolite analyses reveal multiple points of ethylene control during tomato fruit development. Plant Cell 17: 2954–2965[Abstract/Free Full Text]

Bernatzky R, Tanksley SD (1986) Toward a saturated linkage map in tomato based on isozymes and random cDNA sequences. Genetics 112: 887–898[Abstract/Free Full Text]

Beyer M, Hahn R, Peschel S, Harz M, Knoche M (2002) Analyzing fruit shape in cherry (Prunus avium L.). Sci Hortic (Amsterdam) 96: 139–150[CrossRef]

Bruskiewich R, Coe EH, Jaiswal P, McCouch S, Polacco M, Stein L, Vincent L, Ware D (2002) The Plant Ontology Consortium and plant ontologies. Comp Funct Genomics 3: 137–142[CrossRef]

Doebley J (2004) The genetics of maize evolution. Annu Rev Genet 38: 37–59[CrossRef][Web of Science][Medline]

Frary A, Doganlar S (2003) Comparative genetics of crop plant domestication and evolution. Turk J Agric For 27: 59–69

Fulton TM, Chunwongse J, Tanksley SD (1995) Microprep protocol for extraction of DNA from tomato and other herbaceous plants. Plant Mol Biol Rep 13: 207–209[Web of Science]

Grandillo S, Ku HM, Tanksley SD (1999) Identifying the loci responsible for natural variation in fruit size and shape in tomato. Theor Appl Genet 99: 978–987[CrossRef][Web of Science]

International Plant Genetic Resources Institute (1996) Descriptors for Tomato (Lycopersicon spp.). International Plant Genetic Resources Institute, Rome

International Union for the Protection of New Varieties and Plants (2001) Guidelines for the Conduct of Tests for Distinctness, Homogeneity and Stability (Tomato). International Union for the Protection of New Varieties and Plants, Geneva

Kosambi DD (1944) The estimation of map distances from recombination values. Ann Eugen 12: 172–175

Lander ES, Green P, Abrahamson J, Barlow A, Daly MJ, Lincoln SE, Newburg L (1987) MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1: 174–181[CrossRef][Medline]

Langlade NB, Feng X, Dransfield T, Copsey L, Hanna A, Thebaud C, Bangham A, Hudson A, Coen E (2005) Evolution through genetically controlled allometry space. Proc Natl Acad Sci USA 102: 10221–10226[Abstract/Free Full Text]

Lemaire-Chamley M, Petit J, Garcia V, Just D, Baldet P, Germain V, Fagard M, Mouassite M, Cheniclet C, Rothan C (2005) Changes in transcriptional profiles are associated with early fruit tissue specialization in tomato. Plant Physiol 139: 750–769[Abstract/Free Full Text]

Lippman Z, Tanksley SD (2001) Dissecting the genetic pathway to extreme fruit size in tomato using a cross between the small-fruited wild species Lycopersicon pimpinellifolium and L. esculentum var. Giant Heirloom. Genetics 158: 413–422

Morimoto T, Takeuchi T, Miyata H, Hashimoto Y (2000) Pattern recognition of fruit shape based on the concept of chaos and neural networks. Comp Electron Agric 26: 171–186

Mueller LA, Solow TH, Taylor N, Skwarecki B, Buels R, Binns J, Lin C, Wright MH, Ahrens R, Wang Y, et al (2005) The SOL genomics network: a comparative resource for Solanaceae biology and beyond. Plant Physiol 138: 1310–1317[Abstract/Free Full Text]

Paris HS, Yonash N, Portnoy V, Mozes-Daube N, Tzuri G, Katzir N (2003) Assessment of genetic relationships in Cucurbita pepo (Cucurbitaceae) using DNA markers. Theor Appl Genet 106: 971–978[Web of Science][Medline]

Sako Y, McDonald MB, Fujimura K, Evans AF, Bennett MA (2001a) A system for automated seed vigour assessment. Seed Sci Technol 29: 625–636

Sako Y, Regnier EE, Daoust T, Fujimura K, Harrison SK, McDonald MB (2001b) Computer image analysis and classification of giant ragweed seeds. Weed Sci 49: 738–745

Smith BD (1997) The initial domestication of Cucurbita pepo in the Americas 10,000 years ago. Science 276: 932–934[CrossRef][Web of Science]

Van der Knaap E, Sanyal A, Jackson S, Tanksley SD (2004) High-resolution fine mapping and fluorescence in situ hybridization analysis of sun, a locus controlling tomato fruit shape, reveals a region of the tomato genome prone to DNA rearrangements. Genetics 168: 2127–2140[Abstract/Free Full Text]

Van der Knaap E, Tanksley SD (2001) Identification and characterization of a novel locus controlling early fruit development in tomato. Theor Appl Genet 103: 353–358[CrossRef][Web of Science]

Van der Knaap E, Tanksley SD (2003) The making of a bell pepper-shaped tomato fruit: identification of loci controlling fruit morphology in Yellow Stuffer tomato. Theor Appl Genet 107: 139–147[Web of Science][Medline]

Vincent PLD, Coe EH, Polacco ML (2003) Zea mays ontology: a database of international terms. Trends Plant Sci 8: 517–520[CrossRef][Web of Science][Medline]

Yamazaki Y, Jaiswal P (2005) Biological ontologies in rice databases: an introduction to the activities in Gramene and Oryzabase. Plant Cell Physiol 46: 63–68[Abstract/Free Full Text]

Zeng ZB (1993) Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc Natl Acad Sci USA 90: 10972–10976[Abstract/Free Full Text]

Zeng ZB (1994) Precision mapping of quantitative trait loci. Genetics 136: 1457–1468[Abstract]

This article has been cited by other articles:

				I. Fernandez-Silva, E. Moreno, I. Eduardo, P. Arus, J. M. Alvarez, and A. J. Monforte On the Genetic Control of Heterosis for Fruit Shape in Melon (Cucumis Melo L.) J. Hered., March 1, 2009; 100(2): 229 - 235. [Abstract] [Full Text] [PDF]

				M. J. Gonzalo, M. T. Brewer, C. Anderson, D. Sullivan, S. Gray, and E. van der Knaap Tomato Fruit Shape Analysis Using Morphometric and Morphology Attributes Implemented in Tomato Analyzer Software Program J. Amer. Soc. Hort. Sci., January 1, 2009; 134(1): 77 - 87. [Abstract] [Full Text] [PDF]

				A. Darrigues, J. Hall, E. van der Knaap, D. M. Francis, N. Dujmovic, and S. Gray Tomato Analyzer-color Test: A New Tool for Efficient Digital Phenotyping J. Amer. Soc. Hort. Sci., July 1, 2008; 133(4): 579 - 586. [Abstract] [Full Text] [PDF]

				L. Depypere, P. Chaerle, K. V. Mijnsbrugge, and P. Goetghebeur Stony Endocarp Dimension and Shape Variation in Prunus Section Prunus Ann. Bot., December 1, 2007; 100(7): 1585 - 1597. [Abstract] [Full Text] [PDF]

				I. Paran and E. van der Knaap Genetic and molecular regulation of fruit and plant domestication traits in tomato and pepper J. Exp. Bot., November 23, 2007; (2007) erm257v1. [Abstract] [Full Text] [PDF]

				M. T. Brewer, J. B. Moyseenko, A. J. Monforte, and E. van der Knaap Morphological variation in tomato: a comprehensive study of quantitative trait loci controlling fruit shape and development J. Exp. Bot., April 1, 2007; 58(6): 1339 - 1349. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Supplemental Data Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (14) Citing Articles via Google Scholar Google Scholar Articles by Brewer, M. T. Articles by van der Knaap, E. Search for Related Content PubMed PubMed Citation Articles by Brewer, M. T. Articles by van der Knaap, E. Agricola Articles by Brewer, M. T. Articles by van der Knaap, E.