Scientic! Horticulturae, 53 (1993) 261-267
Elsevier Science Publishers B.V., Amsterdam
Non-destructive estimation of area and variation in shape of leaf lamina in the fluted pumpkin
( Telfairia occidentalis) M.O. Akoroda Agronomy Department, University of lbadan, Ibadan, Nigeria (Accepted 24 August 1992 )
ABSTRACT Akoroda, M.O., 1993. Non-destructive estimation of area and variation in shape of leaf lamina in the fluted pumpkin (Telfairia occidentalis). Scientia Hortic., 53:261-267. Expanded and mature leaves of the fluted pumpkin ( Telfairia occidentalis Hook. ill. ), grown in fields, were studied to derive a predictive regression equation for use in estimating the leaf lamina area (L~,) during crop growth studies. A suitable equation based on the number of leaflets in a leaf (Nt) combined with the length (Lc) and maximum width, (We) of only the central leaflet in 156 leaves representing diverse seasonal, cultural, and genetic backgrounds with 2-5 leaflets was LA=O.9467+O.2475LcWc+O.9724LcWcNt (r2=0.92"**). Variation in the different attributes of lamina reduced the percentage of the variation in LA accounted for by use of the equation. Leaf shapes varied widely depending on the number of leaflets, extent of lamina separation, type of tip, degree c,f lobing, as well as the flatness or waviness of lamina surface. Overall, the central leaflet was largest with other leaflets being 80, 70, 64 and 54% the size of the central leaflet. Keywords: Fluted pumpkin; lamina area; leaf shape; regression; Telfairia occidentalis. Abbreviations: LA=lamina area of leaf (cm2); Lc=length of the central leaflet (cm); LL=leaflet leaf; Nt := number of leaflets per leaf; Wc= maximum width of the central leaflet (cm).
Leaf lamina area (LA) has been measured in many crop plants based on lamina length (Lc) and maximum width (We) together with various multiplicands based on these two linear dimensions (Wiersma and Bailey, 1975; Asif, 1977; Obiefuna and Ndubizu, 1979; Robbins and Pharr, 1987; Potdar and Pawar, 1991 ). Corresl~ondence to: M.O. Akoroda, Agronomy Department, University of Ibadan, Ibadan, Nigeria.
© 1993 Elsevier Science Publishers B.V. All rights reserved 0304-4238/93/$06.00
Non-destructive estimation of LA depends on two components: (1) the number of leaves (to be counted ); ( 2 ) the area of each individual leaf (which is estimated from linear dimensions of the leaf and the use of characteristic (s) of its shape). The linear measurements are then mathematically related to the LA by some pre-determined regression equation. Telfairia occidentalis is mainly grown for its edible leaves and young tender vines, which are relished as vegetables. The need to know LA in crop research is well understood; not only for crop growth and physiological studies, but also for monitoring the changes in LA that is consumed in a leafy crop such as Telfairia. In this way, the effect of environmental and imposed field conditions on LA can be measured and subsequent methods sought to control these factors to favour greater LA production. Varieties of Telfairia have not been classified, and it is difficult to ascertain the genotype of the sources of leaves or shoots. Under such circumstances, a general solution to LA estimation is necessary for practical purposes in growth studies. MATERIALS A N D M E T H O D S
In May-June 1987, LA was estimated using 140 leaves randomly sampled from 70 plants. For good representation of consumed leaves, one leaf from the older more mature lower leaves and one from the younger upper parts of each of the 70 plants was measured. These were placed fiat on graph paper, traced and the LA estimated. The relationship of the length of the central leaflet to the total LA of all leaflets was then examined by simple linear regression analysis. In September 1990, a random sample of 91 mature leaves was obtained from assorted farms in four areas of the University campus at Ibadan. These were from the following sources: Shaba (24); Ike ( 17 ); Ogbonna (24); Forestry (26). The LA of each leaf was determined electronically with a Li-Cor 3000 leaf area meter to two decimal digits. The length and m a x i m u m width of only the central leaflet was measured manually. In January 1992, another survey of mature leaves was made on various farms and 65 leaves were sampled. The length and m a x i m u m width of each leaflet was measured with a graduated ruler and their area electronically determined on a Li-Cor 3000 leaf area meter to two decimal digits. Various regression equations for estimating LA were considered based on the number of leaflets, Nt; the length, Lc, and m a x i m u m width, We, of the central leaflet, using 156 leaves from the above two sources. There were three leaves with two leaflets ( 1.9%), 106 with three leaflets (68.0%), six with four leaflets (3.9%), and 41 with five leaflets (26.3%). They represent a broad range of leaves of the species from different field and genetic backgrounds. The basic measurements and their squares were each correlated to the total
LA. F',quations with higher coefficients of determination (r 2) values were compared for single and multiple items before selecting one. Numerous pressed leaf samples were inspected and the most common shapes noted; and the role played by lamina shape irregularity, towards reducing the predictive value of the regression equations, was appraised. RESULTS
The length of the central leaflet alone (in the 1987 data) accounted for 77.4% of the variation in the LA of the central leaflet in younger leaves in the terminal portions of the vine; and 68.9% for the older leaves at the base of the vine. Using a similar type of analysis with 156 leaves ( 1990 and 1992 ), variation in length of the central leaflet explained 75.1% of the variation in the LA of the whole leaves of varied ages and positions on the vine of a wide diversity of plants. Up to 93.6% of the variation in the LA of the central leaflet alone was explained by an equation combining all six items: length (Lc), maximum width (Wc), LcWc, Lc2, Wc2, and Lc/Wc. However, only one out of the six items, LcWc alone explained as much as 93.1% of the variation in LA of the central leaflel:, being 99.5% as efficient as all six items together. Of ~he many simple, polynomial and multiple regression models tried, Table I presents the basic equations of the single variables, all ten variables comTABLE 1
Coefficients of determination (r 2) of basic regression models for estimating lamina area (LA) in mature leaves of Telfairia occidentalis ( n = 156) from length (L¢) and maximum width (We) of the central leaflet and the number of leaflets per leaf (Nt). All the r 2 values are significant at P=0.01 Regression model of leaf LA with single items
2. 3. 4. 5.
6. 7. 8. 9. 10. 11.
LA= - 116.57+ 105.720 N t - 5.4676 Nt2 LA= 5.845+ 19.145 N 2 - 0 . 3 3 5 2 (N 2 )2 LA= 154.14-25.983 Lc+2.1024 L~ LA= 35.696+0.579 L~ +0.00134 (L~)2 L A = 9 0 . 7 4 - 2 4 . 6 1 t¥c+4.9119 W 2 LA=24.647+ 2.4728 Wc2 +0.0060403 ( W~ )2 LA=7.8689+ 1.5307 LcWc+0.0024445 (LcW~) 2 LA=69.938+0.01143 (LcWc)2+0.000000059234 [ (L~W~)2] 2 LA=26.581 +0.45536 L~W~Nt-O.O00032272 (LcW~Nt) 2 LA= 103.63+0.0005363 (L¢W¢Nt)2--O.O000000001507 [ (LcW~Nt)2] 2 LA= --87.306-9.592 Lc--9.4859 Wc-2.2358 LcWc+ 99.269 Nt+ 1.2484 L 2 +3.1225 Wc+0.3841 LcWcNt-O.O0042757 [ (LeWd)2]2-0.000055535 [ (L~W~Nt)2] 2 - 13.693 N 2 LA = 0.9467 + 0.9724 Lc W~+ 0.2475 L~ WdVt
r2 (as%) 29.1 29.1 80.6 81.0 81.4 81.5 87.3 86.9 89.0 85.1
bined and the selected model based on only two items. The ten-item regression model explained 92.5% of the variation in total LA of the leaf. However, the two-item equation was selected as a practicable equation for rapid use in crop growth studies. This two-item equation was 99.3% as efficient as the tenitem equation, accounting for 91.8% of the variation of LA in all 156 leaves based on LA = 0.9467 + 0.2475Lc W~ + 0.9724Lc
WeNt (r 2 = 0.92*** )
where Lc, Wc and Nt are the length, the m a x i m u m width of the central leaflet and the n u m b e r of leaflets per leaf, respectively. Based on all 156 leaves, the relative sizes of leaflets decreased away from the central leaflet as follows: 100, 79.5, 69.5, 64.2 and 53.9% for a five-leaflet case. It is evident that the second and third leaflets of three-leaflet leaf are slightly unequal, so are the fourth and fifth leaflets in a five-leaflet leaf. Corresponding coefficients of variation in size for these leaflets were: 46.3, 47.2, 50.3, 51.5 and 55.5%. Telfairia leaf basically has a three- or five-leaflet pattern: a central leaflet plus one or two others on both sides of it (Figs. 1 (a), 1 ( b ) ) . With increased lobing in the basic three-leaflet pattern, a four- or five-leaflet stage is progressively attained often without increase in the n u m b e r of the leaflet stalks. The pattern of the basic ovate-lanceolate-oval leaf shape varied with: ( 1 ) relative size of the central leaflet to that of the side leaflet (compare Figs. 1 (h) and 1 (p) ); (2) degree of separation of side leaflets from one another (Figs. 1 (q), 1 (r), 1 ( s ) ) ; (3) size of leaflet(s) on one side relative to those on the other side or lamina symmetry (Figs. 1 (b), l ( s ) ) ; (4) extent of side leaflet(s) lobing (Fig. 1 (p); ( 5 ) type of tip leaflets (Figs. 1 (i), 1 (j), 1 (k), 1 (1), 1 ( m ) , 1 (n), 1 (o) ); (6) flatness or waviness of the leaflet lamina surface; ( 7 ) length to width ratio of leaflets; (8) the type of lamina margins (Figs. 1 (c), 1 (d), 1 (e), 1 ( f ) , 1 (g)). The shape of the side leaflets resembled their central leaflet to different extents. In both three- and five-leaflet cases, any two corresponding side leaflets rarely have the same size or exact form (Fig. 1 (t) ), but a generalized similarity becomes apparent when m a n y samples are considered. With the shape of the central leaflet as reference, side leaflets show greater lobing towards the basal half portion of their area, being slight, normal or deep (Fig. 1 ( p ) ) . Overall, increased lobing preceeds the separation of leaflets. The degree of lobing and lamina separation differed among side leaflets of three-leaflet cases, and only between fourth and fifth leaflets in five-leaflet cases. L a m i n a margins a m o n g leaves often occur in several forms on the same leaflet, or leaf, or on leaves of the same plant. However, one type is d o m i n a n t a m o n g leaves of a single plant or a family of plants with the same genetic relationship.
NON-DESTRUCTIVE ESTIMATION OF LEAF LAMINA AREA
Fig. 1. Leaf and leaflet form of three- or five-leaflet (LL) of Telfairia occidentalis (bars are 1 cm lonl;): Whole leaf form (A) 3-LL (most common form); (B) 5-LL (note the three leaflet stalk); Lamina margins: (C) indented; (D) undulate; (E) wavy; (F) spiny; (G) entire. (H) lamina of central leaflets (blackened) with varied size, shape and symmetry of surface area. Leaflet tip: (I) arc; (J) acute; (K) emarginate; (L) pointed; (M) knob; (N) obtuse; (O) pointed-knob. (P) lamina of side leaflets with varied size and types of lobing. Incomplete separation of lamina and varying degrees of lobing: (Q) slight; (R) partial and complete; (S) partial; (T) relationship of form and area in two side leaflets ofa 3-LL. DISCU SSION
The: literature on non-destructive or linear estimation of LA indicates that the more serrated, lobed or irregular the shape or pattern of the leaf lamina, the more difficult it is to predict its area from one or two linear attributes.
This is because every minute aspect of lamina shape or pattern is subject to variation. The more the sources of variation, the less would any one source of variation sufficiently account for the overall variation in LA. Using attributes of the central leaflet to predict the LA of the whole leaf implies that we may only approach, but not exceed, the predictive level attained for the LA of the central leaflet itself. The level to which the LA of the central leaflet can be accounted for therefore sets a limit to the predictive power of the regression equation for estimating the LA of the whole leaf when using only attributes of the central leaflet. Plantains and banana have similar and regular leaf lamina shapes. Nondestructive estimation by linear measures of L and W resulted in regression equations that explained 53.3% of plantain LA, where L A = 0 . 8 LW (Obiefuna and Ndubizu, 1979); in contrast to 96.0% for banana for LA=0.0266+0.7629 LW (Potdar and Pawar, 1991 ). Both studies were based on 40-45 leaves. However, in less regularly shaped leaves, okra midrib length (X) explained 80.5% of variation in LA, where L A = l l 5 X - 1 0 5 0 (Asif, 1977 ); and in cucumber with a very similar lamina shape, 99.0% of variation in LA was explained based on L A = 14.61 - 5 . 0 L + 0 . 9 4 L2+0.047 W+0.63 W2 _ 0.62 LW (Robbins and Pharr, 1987 ). In soybeans with trifoliate leaves, it was possible to predict total leaf area from just the length and width of the terminal leaflet as LA=0.411 +2.008 LW with r 2 of 98.3% (Wiersma and Bailey, 1975 ). Their use of only the central or terminal leaflet is similar to the use of the central leaflets for LA estimation in leaves of Telfairia with mostly three leaflets. The failure of the regression line to account for more than 91.8% of the variation in LA must be owing to variation in the eight variables of lamina shape, which differ for leaves on the same plant or among plants of the same and different fruit families. Increasing variation in leaflet size away from the central leaflet as well as differing degrees of lamina separation and lobe development in side leaflets attests to greater difficulties in LA estimation as the number of leaflets increases. ACKNOWLEDGEMENTS
The technical assistance of Clement Shaba, Ike Ijeoma, Andrew A. Efisue, Phineas Ntawuruhunga and Rasak A. Gbadamosi in this study is acknowledged with gratitude. The author also thankfully remembers all the farmers that gave us free access to inspect their fields and also harvest leaf samples. REFERENCES Asif, M.I., 1977. Estimation of leaf area in okra (Abelmoschus esculentus (L.) Moench.) Trop. Agric. (Trinidad), 54: 192.
NON-DESTRUCTIVE ESTIMATION OF LEAF LAMINA AREA
Obiefuna, J.C. and Ndubizu, T.O.C., 1979. Estimating leaf area of plantain. Scientia Hortic., 11: 31-36. Potdar, M.V. and Pawar, K.R., 1991. Non-destructive leaf area estimation in banana. Scientia Hortic., 45: 251-254. Robbin,;, N.S. and Pharr, D.M., 1987. Leaf area prediction models for cucumber from linear measurements. HortScience, 22:1264-1266. Wiersma, J.V. and Bailey, T.B., 1975. Estimation of leaflet, trifoliate and total leaf areas of soybeans. Agron. J., 67: 26-30.