Proceedings of the 7th International Symposium on Machinery and Mechatronics for Agriculture and Biosystems Engineering (ISMAB) 21-23 May 2014, Yilan, Taiwan

QUANTIFICATION OF FLORAL SHAPE VARIATION IN THE F2 HYBIRDS OF TWO STREPTOCARPUS SPECIES WITH CONTRASTING POLLINATION SYNDROME Yun Tsou1, Hao-Chun Hsu2, Cheng-Chun Wang1, Yan-Fu Kuo1*, Chun-Neng Wang2, 3, Michael Mӧller4 1

Department of Bio-Industrial Mechatronics Engineering, 2 Institute of Ecology and Evolutionary Biology, 3 Department of Life Science, National Taiwan University, No.1, Sec. 4, Roosevelt Rd., Taipei City 106, Taiwan (R.O.C.) 4 Royal Botanic Garden Edinburgh, 20A Inverleith Row, Edinburgh EH3 5LR, United Kingdom *Corresponding Author-- Voice: +886-2-3366-5329, Email: [email protected] Abstract: Floral shape variation is of the central focus to understand flower development and evolution on angiosperm families. This study quantified the floral shape variation among segregating F2 generation individuals between a keyhole shape butterfly pollinating species Streptocarpus johannis, and an open tube fly pollinating species Streptocarpus rexii. In the process, the front-view and side-view images of the corolla were captured. Image processing algorithms were applied to acquire landmarks (i.e., characteristic points) of the flowers. Next, generalized Procrustes analysis was applied to the landmarks to quantify floral shape. The floral shape variation was then investigated using principal component analysis. It is shown that the shape variation of the F2 Streptocarpus flowers can be effectively captured and quantified by the proposed approach. Key Words: Geometric morphometrics, Floral morphology, Generalized Procrustes analysis

INTRODUCTION Quantification of variation in floral shape between closely related species yet shifts in pollination syndrome shall enable us to understand how pollinator-mediated selection acts on flower evolution. Genus Streptocarpus species exhibit great diversity in floral shape in response to pollinators such as fly, bee, birds, and butterfly (Harrison et al. 1999; Möller and Cronk, 2001). Streptocarpus rexii is fly pollinated species with open tube corolla, whereas a closely related species Streptocarpus johannis is moths or butterflies pollinated with keyhole The authors are solely responsible for the content of this technical presentation. The technical presentation does not necessarily reflect the official position of the Chinese Institute of Agricultural Machinery (CIAM), and its printing and distribution does not constitute an endorsement of views which may be expressed. Technical presentations are not subject to the formal peer review process by CIAM editorial committees; therefore, they are not to be presented as th refereed publications. Citation of this work should state that it is from the 7 ISMAB paper. EXAMPLE: Author's Last th Name, Initials. 2014. Title of Presentation. The 7 ISMAB May 21-23, 2014. Yilan, Taiwan. For information about securing permission to reprint or reproduce a technical presentation, please contact CIAM at [email protected] or the Chinese Institute of Agricultural Machinery, c/o Department of Bio-industrial Mechatronics, National Chung Hsing University, 250 KuoKuang Road, Taichung 40227, Taiwan.

191

shape corolla (Fig. 1). These two species with drastically different floral shape can be intercrossed. The segregating F2 individuals show continuous variation on floral shape between fly and butterfly pollination syndromes. This allows us to quantify how shape is altered during the transitions between wide-open to keyhole shape corolla tube.

Figure 1. The image of Streptocarpus rexii and Streptocarpus johannis. Corolla shape must be quantified with high precision to accurately determine its variation. Typically, floral images are captured. A set of characteristic coordinate points, referred to as landmarks, are selected and used to present the floral contour (Adams et al., 2004). The landmarks are subject to geometric morphometrics (GM; Adams et al., 2004) for quantitative shape analysis. Various studies have applied GM to assessing the diversity of lip shapes on orchid species Dactylorhiza (Shipunov et al., 2004), to correlating the variation in shape and size of organs within the Antirrhinum species group (Feng et al., 2009), and to identifying pollinator preference in floral shape patterns on Erysimum mediohispanicum (Gomez et al., 2006). Tube opening and angle between two lateral lobes are major floral traits which affect pollinators’ approaching on the flower (Mitchell, 1994; Cronk and Möller, 1997; Gong and Huang, 2009). Thus, in this study, we defined and examined opening score and lateral lobe angle to facilitate the analysis. The objective of this study was to quantify floral shape variation among the F2 individuals of photographed front or, “face,” views and side views of the corollas, thereby enabling landmarks to be identified using image-processing algorithms. GM was conducted to quantify and demonstrate floral shape variation. Thin plate spline analysis was then applied to demonstrate the floral shape variation. MATERIALS AND METHODS FLOWER SAMPLES All plant materials were supplied from the research collections held at the Royal Botanic Garden Edinburgh. Two sexual compatible Streptocarpus species with distinct floral shape, open-tube flowered S. rexii and keyhole type flowered S. johannis were crossed to produce fertile F1 plants. Then a single F1 plant was self-pollinated to generate F2 plants. Figure 2 show F2 flower samples.

192

Figure 2. Sample images of F2 Streptocarpus flowers. IMAGE ACQUISITION The specimens were first pinned on a black board. Their images were captured with a regular digital camera. The face-view images were captured with the camera confronting the plane of unfolded petal lobes. The side-view images were taken with the camera confronting the dorsiventral plane of the flowers. All images were acquired at the stage when the corolla was fully unfolded and mature. A total of 65 floral images were collected. LANDMARK IDENTIFICATION Floral landmarks are categorized as being primary and secondary (Zelditch et al., 2004). The primary landmarks are readily recognizable points, such as the intersections between petals or sepals, whereas the secondary landmarks are equally spaced points between 2 conjunctive primary landmarks. Five primary landmarks were defined for the face-view and side-view images, respectively. The primary landmarks were the intersection of 2 consecutive petal contours. Figure 3 shows the landmarks and their assigned numbers. In the side-view images, the primary landmarks were the intersections of the sepal and tube, and the 2 consecutive petals. The secondary landmarks were 5 and 10 equally-spaced points between 2 conjunctive primary landmarks in the face-view and side-view images. As a result, 30 face-view and 25 side-view landmarks were collected for each flower.

(a) (b) Figure 3. Numbers assigned to the primary and secondary landmarks in the (a) face-view image and (b) side-view image. The landmarks were identified using image processing algorithms. These algorithms were implemented with a program written in C++ with Qt Creator (Nokia) and OpenCV (Intel). The landmark identification involved 4 steps: foreground segmentation, contour line detection, primary landmark selection, and secondary landmark determination. The details of the program can be found in Kuo et al. (2013). SHAPE VARIATION ANALYSIS GM was performed to evaluate floral shape variation. Here the “shape” is defined as the form that is unaffected by changes in translation, scale, or rotation of the images. The

193

landmark-based GM involves two processes – general Procrustes analysis (GPA) and principal component analysis (PCA). GPA was performed to remove non-shape information from the landmark coordinates. The final coordinates obtained are called GPA landmarks. PCA was then applied to the GPA landmarks for dimension reduction. The first few principal components (PCs) account for most of the variation in the landmarks and can well summarize the variance in shape with little loss of information. The analyses of variation were then performed on only the first few PCs. OPENING SCORE AND LATERAL LOBE ANGLE Opening scores and lateral lobe angle were defined. For the opening score, middle points between the corresponding dorsal and ventral landmarks in the side-view images were identified (Fig 4(a)). Define  as the summation of the length between each pair of adjacent landmark middle points, and  as the length of the segment between Landmarks 12 and 14. The opening score d was defined as  divided by :

d

 . 

(1)

A larger opening score indicates a wider opening flower tube. For the lateral lobe angle, 2 lines were obtained from each image, including a line linking the center of the flower to Landmark 10; another line links the center of the flower to Landmark 22. The lateral lobe angle θ was defined as the sharp angle between the 2 lines.

(a) (b) Figure 4. Illustration of (a) opening score and (b) lateral lobe angle. RESULTS & DISCUSSION LANDMARK IDENTIFICATION Figure 5 shows the original, foreground, contour images, as well as the landmarks.

(a) (b) Figure 5. Landmark identification of the (a) face-view and (b) side-view images

194

PRINCIPAL COMPONENT ANALYSIS Figure 6 shows the percentage of total variance of the PCs for the (a) face-view and (b) side-view images. The first 2 PCs of the face-view images, referred to as F-PC1 and F-PC2, account for 17.33% and 15% of the total variance. The first 2 PCs of the side-view images, referred to as S-PC1 and S-PC2, account for 36.78% and 18.6% of the total variance.

(a) (b) Figure 6. Variance percentage of each PC in the (a) face-view and (b) side-view images. Figure 7 visualizes the floral shape variation resulting from changes in PC values. For each PC, the shapes corresponding to the mean and mean plus/minus 2 standard deviations (STDs) of the PC values are shown as red contours. The blue line and arrow represent the magnitude and direction of the change from the mean. The figure indicates that F-PC1 corresponded to the degree of the expansion between the two lateral petals. Smaller degree of expanding occurred at larger PC1 values. The variation of F-PC2 represented the degree of length-width ratio of tube dilation. Larger degree of the length-width ratio occurred at larger F-PC2 values. S-PC1 associated with the degree of distal end protruding of dorsal tube. Smaller degree of the protruding occurred at larger PC1 values. S-PC2 corresponded to the degree of curvature and compression of the tube. Larger PC2 values referred to a narrow and curved corolla tube. -2 STD

Mean

+2 STD

F-PC1

F-PC2

S-PC1

S-PC2 Figure 7. Floral shape variation corresponding to changes in PC values.

195

OPENING SCORE AND LATERAL LOBE ANGLE Figure 8 shows the histograms of (a) opening score and (b) lateral lobe angle. The mean lateral lobe angle is 143.4 , and the standard deviation is 5.7. The mean opening score is 0.44, and the standard deviation is 0.075.

(a) (b) Figure 8. (a) Opening score and (b) lateral lobe angle histogram. CONCLUSIONS Floral shape variation was studied in the F2 population of a Streptocarpus hybrid obtained by crossing two species of the plant. GM showed that PC1 of face-view and side-view images accounted for 17.33% and 36.78% of the total variance of floral shape variation. Lateral lobe angle and opening score of tube were also defined and calculated. The floral shape variation was effectively quantified in the current study. The proposed approach can be applied in the examination of floral genotype-phenotype association in future work. REFERENCES Adams, D.C., Rohlf, F.J., Slice, D.E., 2004. Geometric morphometrics: ten years of progress following the 'revolution'. Ital. J. Zoolog. 71, 5-16. Cronk, Q., & Möller, M., 1997. Genetics of floral symmetry revealed. Trends in ecology & evolution, 12(3), 85-86. Feng, X., Wilson, Y., Bowers, J., Kennaway, R., Bangham, A., Hannah, A., Coen, E., Hudson, A., 2009. Evolution of allometry in Antirrhinum. Plant Cell 21, 2999-3007. Gong, Y. B., & Huang, S. Q., 2009. Floral symmetry: pollinator-mediated stabilizing selection on flower size in bilateral species. Proceedings of the Royal Society B: Biological Sciences, 276(1675), 4013-4020. Harrison, C. J., Möller, M., & Cronk, Q. C., 1999. Evolution and development of floral diversity in Streptocarpus and Saintpaulia. Annals of Botany, 84(1), 49-60. Kuo, Y.F., Weng, L.K., Lee, T.K., Hsu, H.C., Lin, T.T., Wang, C.N., 2013. Quantitative evaluation of the floral shape variation in Sinningia speciosa domestication, pp. 1933-1940 Mitchell, R. J., 1994. Effects of floral traits, pollinator visitation, and plant size on Ipomopsis aggregata fruit production. American Naturalist, 870-889. Möller, M., & Cronk, Q. C. B., 2001. Evolution of morphological novelty: a phylogenetic analysis of growth patterns in Streptocarpus (Gesneriaceae). Evolution, 55(5), 918-929. Shipunov, A.B., Fay, M.F., Pillon, Y., Bateman, R.M., Chase, M.W., 2004. Geometric morphometrics as a tool for understanding Dactylorhiza (Orchidaceae) diversity in European Russia. American Journal of Botany 91, 1419-1426. Zelditch, M., Swiderski, D., Sheets, H., Fink, W., 2004. Geometric Morphometrics for Biologists: A Primer. Elsevier.

196