Comparative ontogeny of a wild cucurbit and its derived cultivar.
Because most plants show metameric construction and open growth, development can be studied at two morphological levels: 1) whole plant ontogeny (germination through death) and 2) organogenesis (inception through senescence of individual organs, e.g., leaves) (Weston, 1988; Groff, 1989; but see also Tomlinson, 1984; Guerrant, 1989). At the level of whole plant ontogeny, evolutionary changes in form under domestication commonly involve changes in architecture (Coyne, 1980; Gottlieb, 1986; Pickersgill, 1986). For example, wild sunflower taxa are highly branched whereas most cultivated sunflowers are not (Hockett and Knowles, 1970; Harlan et al., 1973). At the level of organogenesis, evolution under domestication often involves meristic and structural changes related to dehiscence and dispersal (Schwanitz, 1966; Purseglove, 1968; Baker, 1972). However, the most common morphological organogenetic change in cultivars is an increase in size, a phenomenon loosely described in the crop literature as "gigantism" (Schwanitz, 1966; Baker, 1972; Evans, 1975, 1976; Hawkes, 1983).
The gourd family, Cucurbitaceae, is well suited for a study of development and morphological evolution. The family is highly variable morphologically and it includes several pairs of wild and cultivated taxa in which the wild taxa are considered progenitors of the cultivars. There are at least three such pairs within the genus Cucurbita (the pumpkins and squashes) (Decker-Walters et al., 1990), one pair being the two subspecies of Cucurbita argyrosperma (Merrick and Bates, 1989; Merrick, 1990). Subspecies argyrosperma var. argyrosperma, formerly C. mixta Pang. (Mabberley, 1985), is cultivated in eastern and southern Mexico for its large, nutritious seeds (Whitaker, 1968). This variety, hereafter referred to as argyrosperma, often grows in the same locales as the second subspecies of C. argyrosperma, subsp. sororia (Merrick, pers. comm.). This second subspecies, hereafter referred to as sororia, is wild and is the probable progenitor of the cultivated and weedy forms of the subspecies argyrosperma (Merrick, 1990; but see Wilson, 1989 for a different interpretation).
Despite the close phylogenetic relationship between argyrosperma and sororia, differences in their morphology are readily apparent. The most striking is in leaf size: leaves of argyrosperma are larger. The second is in the pattern of leaf blade shapes produced during ontogeny. During early phases of growth, both subspecies produce leaves that are superficially similar in shape; except for the first two or three leaves in sororia, these early, "juvenile" leaves are slightly lobed. However, later in the ontogeny of sororia, by the time the first carpellate flower is produced, the "adult" leaves are highly lobed. Argyrosperma, on the other hand, produces only slightly lobed leaves for its entire life; "adult" leaves are never highly lobed. The difference in leaf blade shapes produced at different phases of ontogeny indicates that sororia is heteroblastic, meaning that these differences in blade shape are genetically programmed changes in shoot components that occur as a normal expression of the plant's ontogeny (Goebel, 1900; Allsopp, 1965, 1967; Kaplan, 1973). A third morphological difference between subspecies is in flowering time: the cultivar flowers earlier in the growing season.
Because the cultivar flowers earlier, and because the "adult" leaf of argyrosperma resembles a smaller, slightly lobed, "juvenile" leaf of sororia, I hypothesized that the cultivar evolved through paedomorphosis and proportioned gigantism (sensu Alberch et al., 1979). In this scenario, the highly lobed leaf that characterizes the "adult" phase of development in sororia would have been lost during the evolution of argyrosperma. Supporting evidence for this hypothesis is that: (1) both subspecies should share similar patterns of whole plant ontogeny, i.e., positional relationships should be comparable between plants; (2) larger leaves of argyrosperma should be proportionally or allometrically larger versions of early leaves of sororia; and (3) all less lobed leaves should share a common developmental pathway (but see Jones, 1990; unpubl. data).
In this paper I present results relevant to the first two points. To address the first cirterion of shared patterns of whole plant ontogeny, I compare ontogenies of primary shoots emphasizing the "age" of production of the first carpellate flower, both in terms of time and morphology (i.e., nodal position). For logistical reasons, I restrict my comparison of whole plant ontogeny to the primary shoots of each subspecies. While this comparison encompasses only a portion of whole plant ontogeny, I describe the comparison as that between "shoot system ontogenies" in order to maintain a conceptual distinction between ontogeny and organogenesis.
Comparisons of ontogenies at either the level of the whole plant or individual vegetative shoots are frequently hampered by open growth and phenotypic plasticity (Guerrant, 1989). To compare the same phases of shoot ontogeny in both subspecies, I analyze: growth phases that can be distinguished by a discrete beginning and end point sensu Alberch et al. (1979). The beginning of this phase is defined as emergence from the soil, i.e., growth of the epicotyl beyond one cm, and the end of this phase as the position along the shoot of the first fertile carpellate flower.
Apparent heterochronic shifts in ontogeny, i.e., the appearance of characters at different nodal positions along the shoot in the derivative relative to the progenitor, could conceivably have two explanations in metameric organisms. One is that a character may become disassociated from other characters associated with a given metameric position. As a result, in the derivative, this character arises at an earlier or later position relative to the expression of other characters and would be shifted in position along the shoot. This condition is analogous to heterochronic shifts of traits in nonmetameric animals. Alternatively, in metameric organisms, differences in positions of characters could arise through an insertion or deletion of metamers within the developmental phase in question (cf. Kluge, 1988), so that while the absolute position of a particular character changes, it changes in conjunction with other characters associated with that position in the progenitor. To distinguish between these alternatives in plants, i.e., organisms that often exhibit plasticity in metamer number and open growth, one must be able to establish whether, within the growth phases being compared, metamers at the same position are developmentally equivalent. I use the expression of architecture and the pattern of meristem deployment (e.g., tendril formation and branching) along the primary shoot as a basis for comparing plants in terms of flowering and leaf shape variation.
The second criterion concerns the extent to which leaves of the cultivar are proportionately or allometrically similar in shape to earlier leaves of the wild subspecies. This analysis requires both archiving shape, and analysis of archived shapes. Several techniques for image acquisition and analysis have been used to describe two-dimensional shapes of leaves (e.g., Dickinson, 1986; Dickinson et al., 1987; Blue and Jensen, 1988; White et al., 1988; Ray, 1990; Lacroix and Posluszny, 1991). I use truss networks because they offer the following combination of features: (1) shapes are sampled redundantly so that measurement error is minimized (Strauss and Bookstein, 1982; Bookstein et al., 1985); (2) average shapes as represented by trusses can be easily generated and visualized; (3) shapes can be compared using both geometric and multivariate approaches; and (4) trusses permit the assignment of differences in shape to localized regions of the form (Dickinson, 1986).
I analyze differences in leaf shape from two perspectives that are not directly influenced by differences in size (Lessa and Patton, 1989). The first, geometric shape, compares proportions of forms, measured as ratios of variables or angles. The second, multivariate shape, considers all variables jointly. Because I am interested in whether the apparent similarity in shape of slightly lobed leaves is due to similar allometric effects of size on shape in each subspecies, rather than whether the shapes of leaves are similar when size has been removed (e.g., Burnaby's technique), I analyze shape using principal components analysis (PCA) (Rohlf and Bookstein, 1987). When multivariate analyses of shape are based on PCA using variance-covariance matrices, the first component is typically interpreted as a "latent variable serving as a proxy for size" (Bookstein et al., 1985), or more precisely, an allometric size axis (Bookstein, 1989) if the coefficients of that component are of the same sign and similar in value (Jolicoeur, 1963; Pimentel, 1979). The second and subsequent components reflect variation in shape that is uncorrelated with the variation in shape "explained" by the first component (Bookstein, 1989). Because I am analyzing groups assigned a priori (i.e., sets of leaves from each subspecies), I use a multiple group principal component analysis (MGPCA) in which vectors are derived from a pooled within-subspecies dispersion matrix (Pimentel, 1979; Thorpe and Leamy, 1983) rather than from a total dispersion matrix from both subspecies, thus within group variation is not obscured by that between groups (James and McCulloch, 1990).
MATERIALS AND METHODS
Accessions. -- Accessions of argyrosperma and sororia were obtained from the collection of Laura Merrick (now part of the National Plant Germplasm System and housed at the USDA's Regional Plant Introduction Facility at Griffin, GA). Accession #340 of argyrosperma was collected by Merrick in 1982 from the Plan del Rio, Veracruz, Mexico. Accession #26595 of sororia was collected by Nee and Taylor in 1983 near Banos de Carrizal, Veracruz. Seed of each accession was increased by self or sib crosses in 1984 and again in 1985.
1985 and 1986 Field Conditions. -- Land was provided by the Department of Agronomy (1985) and the Department of Vegetable Crops (1986 and 1988) at the University of California, Davis. Plants were cultivated according to standard procedures for vine crops. (See Jones, 1990 for details). Pregerminated seeds were hand planted 1.37 m apart directly over a fertilizer band (ammonium phosphate) during the first week in June. Each subspecies was planted in blocks of 10 plants, and blocks were randomly arranged in the field. Plants were furrow irrigated on the day of planting and then every 10 to 14 days as needed. Twenty days after planting, beds were reformed so that plant rows occupied the center of each bed. At this time, the planted beds received a second banded application of ammonium sulfate. Plants were sprayed with the insecticide SEVIN approximately three and eight weeks after planting.
1988 Field Conditions. -- Cultivation practices followed those of 1986 except that the different subspecies were planted in separate rows. The field was planted in the first week of July. Seeds were planted in the center of rows; beds were not reformed so the second fertilizer application was sprinkled by hand in the furrows.
Leaf Collection. -- Throughout all three growing seasons, plants were collected from the field such that near equal spacing between plants within rows was maintained. Mature leaf blades on each plant were numbered sequentially starting with the first foliage leaf produced after the cotyledons as leaf #1. Each leaf was pressed separately in a plant press until dry. Internode lengths were measured for all internodes subtending all fully expanded leaves on the main stem of plants collected in 1986.
Leaf Production Rates. -- Lengths of the first 36 leaves were measured (in cm) nondestructively every day or every other day with flexible plastic rulers. Total length of blade plus petiole was recorded for leaves from the time they could be measured without damaging the apical bud (about 1 cm) until expansion was complete (1985 and 1986) or until they expanded beyond a reference length of 5 cm (1988). After the production of the first few leaves, leaf growth curves were approximately parallel, and leaves were initiated at relatively uniform intervals (Jones, 1990). However, because leaf growth was less uniform in argyrosperma, and therefore verged on violating the assumptions of the plastochron index (Erickson and Michelini, 1957), rates of leaf production for individual plants were calculated in two ways. One, the plastochron index was calculated for the whole plant and regressed on days from emergence (defined as when epicotyl length grew to greater than 1 cm) (Silk, 1980). Second, the number of leaves beyond the reference length was regressed on days from emergence. In both cases, leaf production through time was constant for individual plants, i.e., slopes of these relationships were linear (Jones, 1990), and could be used as rates of leaf production for individuals (Silk, 1980). Using either PI or leaf number through time gave the same results; those based on the second technique are presented.
Leaf production rates were compared between species with two-way ANOVA (Systat 1.0) that treated year and subspecies as random and fixed effects, respectively.
Leaf Shape Analysis. -- Sizes and shapes of each pressed leaf blade were recorded electronically with MORPHOSYS (C. Meacham and T. Duncan, University Herbarium, UC Berkeley). For each leaf image, measures made were area, perimeter, and a series of 26 point-to-point distance measures that recorded leaf shapes as a truss network (Strauss and Bookstein, 1982; Bookstein et al., 1985; Dickinson et al., 1987). Regardless of their shape, all leaves of sororia have five prominent veins that radiate palmately from their point of attachment to the mid-rib (Fig. 1A, 1C). Because these veins show strong correspondence in position and pattern of development (Jones, pers. obs.) across leaf positions, I considered them to represent serially homologous structures among leaves (i.e., points 1, 2, 3, 5, 7, 9, 11, 12 in Fig. 1 are homologous landmarks). The remaining points were placed along the outline in relation to the curvature of the outline and to the geometrically homologous points (points 4, 6, 8, 10 in Fig. 1). These points are considered pseudolandmarks (Bookstein et al., 1985; Dickinson et al., 1987; Rohlf, 1990). Point 1 was located at the juncture of the blade and petiole, the point from which major veins diverge. In all forms of leaves of sororia (Fig. 1A, 1C), points 2 and 12 were located along the outline at the tip of the most dominant vein in the lowermost lobe. Points 3 and 11 were placed at the tip of the next dominant vein, and points 4 and 10 at the point of greatest indentation (i. e., the base of the sinus) between the lower lobes and the distal mid-lobes. Points 5 and 9 were placed at the tips of the major veins in the mid-lobes, and 6 and 8 in the sinuses between these lobes and the terminal lobe. Point 7 was placed at the tip of the terminal lobe, and the distance between points 1 and 7 is used as blade length (LMLN). Figures 1B and 1D show the resulting trusses with associated labels indicating truss element distances.
In argyrosperma, all leaves have a central prominent vein ending at the tip of the terminal lobe, as well as prominent veins extending to the tips of the adjacent lateral lobes (Fig. 1E). These veins correspond in position to those in sororia. However, the correspondence between homologous veins in lower lateral lobes in leaves of argyrosperma with those in sororia was not immediately obvious, as there appeared to be an "additional" dominant vein in argyrosperma leaves at maturity (compare Fig. 1A with Fig. 1E).
To determine which veins in the lower lobes of argyrosperma were "additional" (i.e., lacking counterparts in sororia), I examined patterns of vein development in early stages of leaves at node 8 from both subspecies. Very young leaves were cleared in Histoclear and stained for two min in 1% basic fuchsin in 95% Etoh. Initially five palmate veins (a central vein and two pairs of lateral veins) were prominent in leaves of both subspecies (Fig. 2A, 2B). During growth, the lowermost prominent vein of leaf 8 in sororia became displaced basally due to growth of the intraveinal blade region between the central midvein and the next proximal major vein as well as between this vein and the lowermost prominent vein (#3 in Fig. 2C). Dissection of the blade in the lobed leaves occurred between these veins. In leaves of argyrosperma, the lowermost vein did not exhibit a basal displacement (compare #3 in Fig. 2C and 2D). Instead this vein retained orientation approximately 90 [degrees] from the midvein. A second vein gained prominence later in development as this lower basal region developed (arrow in Fig. 2D). Due to (1) similarity in position, (2) similarity in sequence and pattern of development of veins in early leaf development, and (3) the presumed progenitor/descendent relationship between sororia and argyrosperma, I considered points 1, 3, 5, 7, 9, and 11 (Fig. 1) to be biologically homologous (sensu Roth, 1988; Wagner, 1989) between blades of each subspecies. Using these criteria, the most basal prominent veins in the lower lobes of argyrosperma are "additional," but they are serially homologous with other prominent veins in leaves of both subspecies. Their presence is correlated with greater growth in the lower basal region in argyrosperma (compare silhouettes of leaf 8 in Figs. 4 and 5). Consequently, trusses for argyrosperma were constructed as in Figure 1F.
Mean Blade Shapes. -- Representative blade shapes shown in Figures 4 and 5 were chosen from the population of actual leaves at each position along the main shoot for each subspecies based on nearness of that leaf to mean blade area, perimeter, area to perimeter ratio, and subsinus dimension between the terminal and mid lobe.
Trusses representing mean leaf shapes were constructed from populations of leaves at each position (see Appendix for sample sizes) from untransformed mean truss element lengths. Mean trusses were constructed by triangulation mapping (Strauss and Bookstein, 1982). Using mean element lengths of untransformed variables to construct mean trusses produced trusses in which elements not directly used in the triangular reconstruction fell within 1 to 2 mm of the mapped points, thus within the margin of error of the mapping technique.
Size-scaled standard deviations on these mean truss shapes were calculated by scaling all blade lengths for each individual leaf to the mean blade length of leaves at that position along the shoot. This scaling factor was then used to calculate the scaled truss element lengths, after which standard deviations for those elements that measured the subsinus regions, i.e., SN2L, SN1L, SN1R, SN2R (Fig. 1D, 1F) were calculated. When plotted on mean trusses of leaves at specific positions, size-scaled deviations show the variation in lobing that would occur if the blade lengths of all leaves at that position along the shoot equalled the mean blade length (Fig. 7).
Geometric Shape A nalysis. -- I measured only fully expanded mature leaves in this study. Mean ratios were calculated from the ratio of untransformed truss element lengths per individual averaged over the population of leaves at that position for each type. Ratios were uncorrelated with size (i.e., blade length)(data not shown) and were normally distributed in all but a few cases (PROC UNIVARIATE, SAS, 1985a). They were analyzed both nonparametrically (Wilcoxon Rank Sum Test, SAS, 1985b) and parametrically (t-tests based on separate variances, SYSTAT 1.0). There were no differences in significance probabilities between results of either test; results of t-test are presented. Ratios among early, transition and late leaves within subspecies were compared using ANOVA (SAS, 1985b).
Multivariate Shape A nalysis. -- Truss networks, as constructed, reflected lateral symmetry of leaves, and as a result, truss lengths measuring the same dimensions on each half of a leaf were highly correlated and resulted in a high degree of collinearity within the data matrix (examined using PROC REG/ collin, SAS, 1985b). Consequently, all multivariate analyses are based on truss element distances archiving the right half of individual leaves.
To illustrate total variation among shapes of all leaves in multivariate space, a simple principal components analysis (PCA) based on a variance-covariance matrix of log-transformed data was performed on all even-numbered leaves from both subspecies (positions 4-36) (SAS, 1985b).
To examine the multivariate allometric relationships among variables in each subspecies, and to compare directions of the principal axes between subspecies, PCAs were generated for early leaves (4-16) of each group separately, again based on a variance-covariance matrix of log-transformed data. Static multivariate allometric coefficients were calculated from the coefficients of PRIN1 from these PCA for each subspecies. These coefficients were standardized to a value of isometry equal to unity by dividing each coefficient by [square root of (1/p)] where p = number of variables (Jolicoeur, 1963; Cheverud, 1982; Shea, 1985).
A multiple group principal components analysis (MGPCA) based on log-transformed truss data from early leaves was performed, using subspecies as groups, as described in Smith and Patton (1988): a variance-covariance matrix of residuals derived from an ANOVA (using GLM in SAS, 1985b) of log-transformed variables for each subspecies were analyzed with a PCA. The resulting eigenvectors of this analysis were multiplied by the original log-transformed variables using the SCORES procedure to restore the original within subspecies variation for plotting.
An assumption of MGPCA is that the principal axes of the components, i.e., the eigenvectors, are parallel among groups. The number of components to be tested for parallel eigenvectors was determined by examination of a standardized scree plot of the eigenvalues of the MGPCA. A dramatic decrease occurred between the first and second eigenvalues. Because the first eigenvector is generally interpreted as a size component, and the second as shape, both were examined. The cosine between the first eigenvectors of each group was determined by subjecting the coefficients of the eigenvectors calculated from separate PCAs of each group to cosine analysis using SIMINT of NTSYS (Rohlf, 1987). This calculated cosine was then jackknifed using the algorithm presented in Sokal and Rohlf (1981) to generate an estimated cosine and its standard error. A cosine value of one would indicate that the angle between the eigenvectors of the first component in each group was zero, and therefore that the directions of the principal axes for each eigenvector were parallel. This analysis was repeated for the second component of each group.
Shoot System Ontogeny: The Establishment of Architecture
In sororia, initial growth resulted in orthotropic rosette shoots with short internodes (Fig. 3). Internode length increased gradually along the shoot; the greatest relative increase occurred between nodes 4 and 5. As primary shoot length neared 10 to 15 cm, a gradual change in the orientation of growth from orthotropic (upright) to plagiotropic (horizontal) occurred. Lateral buds formed at all nodes, but typically only those between nodes 2 and 8 grew into prominent lateral shoots. Within three weeks from emergence, the plant had a sprawling, vining habit. The first tendril on the main axis usually appeared at node 6, although it was smaller and not as branched as tendrils at later nodes. Under typical field conditions, the first flowers of sororia were staminate; they appeared about node 16 (x [bar] = 15.8 [+ or -] 4.0 SD, N = 21) but usually aborted (Fig. 3). Fertile staminate flowers were produced at node 19 (x [bar] = 19.4 [+ or -] 2.9 SD, N = 21) and at each subsequent node until plants began producing carpellate flowers. The first carpellate flowers, usually produced at the 39th node (x [bar] = 38.8 [+ or -] 2.8 SD, N = 7) were usually fertile. They were produced, on average, at every third node from this point (flower production beyond node 45 was not recorded), although the actual number of nodes between carpellate flowers ranged from 0 to 4 in sororia.
Seedlings of argyrosperma also initially formed orthotropic rosette shoots with short internodes (Fig. 3). Internode length increased gradually along the shoot, with the greatest relative increases occurring between nodes 4 and 5, and 5 and 6. Internode elongation was correlated with a shift to plagiotropic growth in this subspecies as well. Prominent lateral shoots grew out from nodal position 2 or 3 to approximately node 8. The first tendril on the main axis was apparent around node 3. The first staminate flowers, which were fertile, appeared at node 12 (x [bar] = 12.0 [+ or -] 2.1 SD, N = 13) (Fig. 3). First carpellate flowers appeared at node 30 (x [bar] = 30.8 [+ or -] 1.9 SD, N = 7), several nodes earlier than in sororia, and in four out of five plants, the first carpellate flowers did not abort. There was a mean of 2.4 nodes between carpellate flowers, although the range was from zero to seven. Flower and fruit production were not recorded beyond node 45.
In sororia, the shapes of leaves along the main shoot changed as a normal expression of the plant's ontogeny (Fig. 4). The first two to three seedling leaves were small and lobed. The number of lobed seedling leaves and their degree of lobing was highly variable among plants and among accessions. Leaves 4 or 5 through 18 were slightly lobed, while leaves 18 through 26 marked the transition from less-lobed to highly-lobed. All subsequent leaves, including leaf 28, were highly-lobed.
In argyrosperma, all leaves along the main shoot were similar in shape to each other. These leaves were also visually similar to the early, less-lobed leaves of sororia (Fig. 5).
Area and Perimeter. -- In both subspecies, blade lengths (Fig. 6A) and areas initially increased with increasing leaf number. Beyond leaf 4 in sororia, mean blade areas ranged from 83 to 120 [cm.sup.2]. This relative constancy of mean areas occurred despite a general increase in blade perimeter with position. Increasing blade perimeter coupled with only slight increases in leaf area resulted in a gradual decrease in the ratio of [square root of area] to perimeter from early to later leaves (Fig. 6B). Thus later leaves were longer and more lobed. This difference in shape resulted in a larger ratio of [square root of area] to length in early leaves (Fig. 6C) which was significant [F(2,327) = 40.60, P < 0.0001].
Beyond leaf 4 in argyrosperma, mean blade areas ranged from 422 to 562 [cm.sup.2]. Both blade length (Fig. 6A) and area increased with position, and these were accompanied by only slight increases in perimeter. Thus in contrast to sororia, the ratio of [square root of area] to perimeter remained more constant across position (Fig. 6B), although this ratio was significantly larger in early leaves [F(2,193) = 17.12, P < 0.0001]. The relation of blade area to length (Fig. 6C) was also significantly larger in early leaves of argyrosperma [F(2,193) = 3.28, P < 0.04].
Mean Blade Shapes. -- Among early leaves of sororia (represented by leaf 8), the subsinus region between the terminal and the lateral lobes (SN1L and SN1R) was more variable than the subsinus region between the two lateral lobes (SN2L and SN2R) (Fig. 7A). This was also true for the leaves in the transition region (represented by leaf 20) (Fig. 7 B). Variation in SN1L and SN1R of transition leaves shows that shapes of these leaves ranged from being nearly as unlobed as average leaves at earlier positions to those with sinuses nearly as deep as leaves at later positions. Sinuses between the two lateral lobes (SN2L and SN2R) in these transitional leaves were less pronounced and less variable and resembled those of the earlier leaves. Later leaves, represented by leaf 28, showed uniform sinus indentation in all sinus dimensions, although SN2L and SN2R were more variable (Fig. 7C.)
Mean trusses of argyrosperma at representative early and late positions were very similar, as expected (Fig. 7D, 7E). Comparing these two trusses reveals subtle differences in both size and shape of early and late leaves: the most basal lateral veins indicated by BOTL and BOTR were displaced downward in leaf 8 relative to leaf 28, suggesting that, in later leaves, there was greater growth of the region basal to the basalmost veins rather than simple proportional growth to a larger size.
Geometric Shapes. -- Comparisons of ratios of sinus variables with blade length show that differences in proportion occur both among leaves at different positions within plants and between leaves at the same position between subspecies (Fig. 8). Both subspecies showed a decrease in the ratio of SN1R to LMLN with increasing leaf position along the shoot, indicating that in both subspecies, later leaves are more lobed (i.e., more indented in SN1). However this ratio fell markedly in sororia only. Ratios of SN1/ LMLN between leaves at the same position were significantly lower in sororia (t-tests; for all positions P < 0. 001), indicating that even early leaves of sororia (i.e., leaves 2-16) were significantly more lobed than early leaves of argyrosperma. This difference in the degree of lobing can be seen by scaling an early leaf of sororia to the same blade length as an early leaf of argyrosperma (Fig. 9). Thus leaves of argyrosperma are not simply "scaled up" versions of early leaves of sororia. Ratios of SN2R to LMLN paralleled those of SN1R/LMLN in both subspecies (data not shown).
Multivariate Shape Analysis. -- A plot of PRIN1 versus PRIN2 from a simple PCA based on all leaves illustrates the distribution in shapes and sizes of leaves in PCA space (Fig. 10). Within sororia, shapes of leaf blades exhibited a continuum from early to later leaves, apparent from their distribution along PRIN2. Leaves of argyrosperma were much less variable in shape than those of sororia. In fact, the subtle differences in shape between early and late leaves of argyrosperma as revealed by truss networks and ratios did not emerge in the PCA. Neither was a difference apparent between early and late leaves of argyrosperma when this subspecies was analyzed alone (data not shown). Consequently, for argyrosperma, leaves at early positions were considered representative of shapes of leaves at later positions in subsequent analyses.
Differences that emerged in patterns of multivariate (static) allometry indicated that relative to sororia, larger leaves of argyrosperma had wider terminal lobes (WDN) and larger subsinus regions (SN1R) (Fig. 11). Argyrosperma also differed from sororia in the location of the second subsinus region (SN2R) along the perimeter, as shown by the difference in the coefficients of SG8 and SG9 (Fig. 11). However despite these differences, the overall direction of the first principal axes of each subspecies were the same [observed cosine = 0.98 (11.5 [degrees]), estimated cosine = 1.05 [+ or - ] 0.008 SE ([or nearly equal to] 0 [degrees])], indicating that the way that size "explains" variation is the same in each set of leaves: they have similar patterns of static multivariate allometry. This conclusion is supported by the multivalriate comparison of shapes of early leaves in each subspecies based on MGPCA (Fig. 12). That MGPC1 represents a multivariate allometric size axis is indicated by coefficients of MGPC1 that are all positive and similar in magnitude (Table 1). The first component accounts for 59% of the total variance. Parallel principal axes, but separation along MGPC1, suggests that despite the difference in the leaf size, the way in which variation among all variables changes with size is the same in each group.
Variation explained by the first component is orthogonal to and thus independent of variation explained by the second component in PCA. This second component is interpreted as a shape component when signs of the coefficients are mixed, thus reflecting variation in different directions among variables. That shape is reflected in MGPC2 is indicated by the mixed signs and greater range of values of the coefficients (Table 1). However, interpreting the spread of scores along MGPC2 (Fig. 12) in terms of shape difference is problematic because the angle between the second eigenvectors of each group considered independently is much greater than 0 [observed cosine = 0.56 (55.9 [degrees]), estimated cosine = 0.48 [+ or -] 0.098 SE ([or nearly equal to]61.2 [degrees])], indicating that the eigenvectors are not parallel. This finding alone shows that the variation reflected in this component is not the same in each group, and to the extent that this component is a shape component, that the features of leaf shape reflected in this component are different between subspecies. In other words, less lobed leaves of both subspecies share a common effect of size on shape as reflected by parallel first principal axes, but once this variation has been removed, the variation in shape that remains and is explained by the second principal component is unique to each subspecies. The nature of these unique shape differences is revealed by examination of the second eigenvectors from the separate PCAs of each subspecies (Table 2). In sororia, the second eigenvector is characterized by large coefficients, positive and negative, for variables that are associated with the degree of lobing between the terminal and lateral lobe, i.e., SN1R and SG7, and to a lesser extent, WDN. In argyrosperma, on the other hand, variables with the highest absolute values of coefficients all characterize the basal lobe, i.e., BR4, SG9 and SG10 (Table 2).
TABLE 1. Coefficients of the first two components of a Multiple Group Principal Component Analysis based on a variance-covariance matrix of log-transformed truss element lengths from the right half of trusses for early leaves (nodal positions 4-16). Subspecies constitute groups. Proportion of total variance explained by each component is given below. Variables MGPC1 MGPC2 LMLN 0.236 0.021 SN1R 0.146 0.371 LFLT2R 0.234 0.064 SN2R 0.239 0.149 LFLT3R 0.257 0.141 BOTR 0.255 0.152 SG6 0.089 0.169 SG7 0.234 0.147 SG8 0.307 0.180 SG9 0.288 -0.111 SG10 0.398 -0.721 WDN 0.241 0.128 BR3 0.366 -0.257 BR4 0.299 0.310 Proportion total variance 0.59 0.15 TABLE 2. Coefficients of the first two components of separate Principal Components Analyses for early leaves (4-16) of each subspecies. Variance-covariance matrices based on log-transformed truss element lengths for the right half of individual leaves. Proportion of total variance explained by each component presented below. sororia argyrosperma Variables PRIN1 PRIN2 PRIN1 PRIN2 LMLN 0.240 0.005 0.222 0.051 SN1R 0.123 0.436 0.203 0.111 LFLT2R 0.238 0.053 0.219 0.083 SN2R 0.240 0.148 0.229 0.111 LFLT3R 0.262 0.139 0.235 0.077 BOTR 0.260 0.148 0.232 0.041 WDN 0.066 0.217 0.151 0.004 BR3 0.242 0.158 0.204 0.083 BR4 0.305 0.160 0.302 0.333 SG6 0.276 -0.118 0.323 -0.093 SG7 0.405 -0.741 0.381 -0.208 SG8 0.267 0.082 0.159 0.131 SG9 0.340 -0.129 0.439 -0.664 SG10 0.305 0.243 0.274 0.574 Proportion of total variance 0.60 0.16 0.56 0.14
Leaf Production Rates
Despite differences in leaf shape and size between subspecies, rates of leaf production did not differ significantly between them (Table 3). In argyrosperma, successive leaves expand through the reference length at a mean rate of slightly more than one per day. In sororia, this interval is 0.99 ([+ or -] 0.06) days.
TABLE 3. Leaf production rates as estimated by the relationship between leaf number and days from emergence. Data shown are rates of individual plants. F statistic is from a two-way ANOVA using year and plant type as main effects. Year sororia [R.sup.2] argyrosperma [r.sup.2] 1985 0.896 0.996 1.096 0.991 1.037 0.944 0.972 0.977 0.986 0.991 0.974 0.962 1986 0.974 0.998 1.125 0.994 0.988 0.998 1.099 0.993 0.991 0.996 1.017 0.992 0.956 0.998 1988 1.115 0.996 0.921 0.983 1.045 0.997 1.052 0.984 1.091 0.999 1.042 0.986 0.922 0.999 1.217 0.983 1.059 0.997 1.068 0.998 Source df SS F P Type 1 0.014 2.497 0.132 Year 2 0.007 0.619 0.550 Type . year 2 0.009 0.818 0.458 Error 17 0.094
Heterochrony and Shoot System Ontogeny
Two criteria are necessary to establish that indeterminantly growing metameric organisms share a common ontogeny. The first is that the same phase of ontogeny should be compared in both organisms. In this study, I assumed that the interval of whole plant ontogeny between initial growth of the epicotyl and production of the first carpellate flower was developmentally comparable between subspecies. The second criterion is to determine whether, within the growth phase being compared, metamers at the same position are developmentally comparable. To assess positional correspondence, both plants must exhibit characters whose nodal position along the shoot reflects whole plant ontogeny and these characters must be independent of the characters of interest, e.g., flowering. I propose that in the case of these subspecies, metamers at equivalent nodal positions represent comparable stages of shoot system ontogeny. This assertion is based on architectural similarity between the primary shoots of each subspecies. Both subspecies began growth as orthotropic shoots with short internodes and later shifted to longer internodes. In both subspecies, the pattern of internode elongation was very similar and was accompanied by a change from orthotropic to plagiotropic growth. As well, both produced prominent lateral branches at early nodes. Tendrils also arose at similar positions. Furthermore, rates of leaf production were not significantly different between subspecies, despite differences in sizes of leaves. Consequently, real time and biological time (Gould, 1977) were equivalent in terms of shoot system ontogeny. Given the premise that similarity in architecture and in growth rates signifies that nodal position reflects comparable estimates of plant development both in time and position, then the appearance of flowers at earlier nodal positions in the subspecies reflects a heterochronic shift in the onset of the reproductive phase in terms of shoot system ontogeny. Hence, the cultivar is paedomorphic to the wild subspecies through a progenetic acceleration of the position of fertile carpellate flowers.
An alternative argument is that earlier flowering in the cultivar arose through a deletion of metamers. In plants with variable growth pattern where nodal position does not accurately reflect comparable phases of development, a hypothesis of deletion or insertion of metamers may be untestable. Nevertheless, it is conceivable that in plants with consistent growth patterns in the same environment in which an independent nodal marker of whole plant ontogeny exists, for example architecture, it may be possible to recognize when metamers have been inserted to or deleted from a developmental program relative to that of the progenitor. For example, it is possible that the earliest nodes of sororia, bearing lobed seedling leaves, were evolutionarily deleted from the ontogeny of the cultivar. This case would be analogous to predisplacement in the Alberch et al. (1979) formalisms. However, even if one assumes that as many as 4 nodes were lost from the early phases of ontogeny in argyrosperma, argyrosperma still produces its first carpellate flower at an earlier position on average than would be expected from a deletion alone. A second alternative, which cannot be dismissed entirely, is that a deletion in the ontogeny of cultivar occurred between node 10 (the upper architectural marker) and the node of the first carpellate flower (K. Landa, pers. comm.). That this deletion did not occur is suggested by similar patterns of early leaf development along the shoot in both subspecies (Jones, unpubl. data).
Heterochrony and Leaf Shape
In argyrosperma, subtle differences in leaf blade shape occurred along the shoot. Later leaves were larger, with more growth in the blade base, and slightly less growth in subsinus regions, as shown by the ratio SN1R/ LMLN. Nevertheless, when all shape variables were considered jointly in a PCA, positional differences in shape were not detectable.
The pattern of leaf shape change along the shoot in sororia reflects a marked and unusual expression of heteroblasty in which both the earliest seedling leaves (1-4) and the later "adult" leaves are lobed. The importance of lobing in these early leaves is unclear, given high variation in both number of lobed seedling leaves and in the extent of lobing both within and among accessions (Jones, pers. obs.). At least the first two or three leaves are initiated in the seed prior to germination (Jones, pers. obs.) and these leaves may be responding to maternal factors present in the seed at the time of leaf morphogenesis (A. Fone, pers. comm.).
Much less variable among plants was the subsequent pattern of leaf shape expression. The next set of early leaves (4-16) were smaller than later leaves with respect to blade length and less-lobed. At approximately leaf 18, fully expanded leaves began to show the transition from the less-lobed to the highly-lobed leaf form, and the transition was complete by leaf 26. The consistency over three years in the position along the shoot of different leaf shapes suggests that the pattern of heteroblastic expression is not highly sensitive to yearly environmental fluctuation under these growth conditions. However there is slight variation among individuals in both the actual node at which transition leaves first appear, the depth of the sinus in these leaves, and the number of leaves involved in the transition.
Based on a subjective, visual assessment of similarity in early less-lobed leaves of sororia to all leaves of argyrosperma, and on the acceptance that nodal position represents comparable stages of shoot system ontogeny, I hypothesized that later leaves of argyrosperma evolved through processes of paedomorphosis in which the juvenile leaf form was retained at all later positions in the cultivar. The later lobed leaves of sororia would have been lost during the evolution of the cultivar. Such a process of paedomorphosis can be analogized to neoteny in unitary organisms, although the concomitant loss of lobed seedling leaves precludes a simple case.
The shapes of less-lobed leaves of argyrosperma and early leaves of sororia are visually, albeit subjectively, similar. However, ratios revealed that early leaves of argyrosperma are not simply isometrically (i.e., geometrically) "scaled-up" versions of early leaves of sororia and hence do not reflect a simple case of proportioned gigantism (Alberch et al., 1979).
Although differences in leaf morphology between subspecies are not accounted for by a simple proportional increase in size, it is possible that these differences arose from an allometric effect on proportion that is common to each group. Allometric shape is a concept based on proportional differences in form that are correlated with changes in size, because during growth, the rate of growth of one variable is not equal to the rate of growth of a second (Gould, 1966). Results of the MGPCA and the angular comparisons of the first eigenvectors of separate PCAs show that although the leaves of each subspecies had different final sizes, they shared a common multivariate size axis (MGPC1) such that the allometric effect of size on shape was the same in both groups. However, the fact that the second components of separate PCAs for each group were not parallel indicates that once the size/shape variation accounted for by MGPC1 is removed from the analysis, the remaining variation reflects differences in shape that are unique to each group. In sororia, this difference arises from variation in lobing, while in argyrosperma, this difference arises from variation in the basal lobe. Therefore, larger, less-lobed leaves in the cultivar have not arisen solely through a simple enlargement either allometrically or geometrically, of the juvenile leaf of the progenitor. Such an enlargement would be necessary to interpret the morphological differences between the two as a simple case of neoteny. On the other hand, each set of less-lobed leaves does share a common allometric axis. This suggests that leaf shape in argyrosperma is due in large part to common determinants of shape shared with its progenitor, but also to determinants of shape novel to the cultivar.
Functional Correlates of Size and Shape of Leaves Along the Primary Shoot
Gigantism is reported in the crop literature as the most common expression of morphological evolution in cultivated plants, particularly in parts that are harvested (Schwanitz, 1966; Evans, 1976). This increase in size of harvested parts has occurred through a disproportionate allocation of resources to these parts (Evans and Dunstone, 1970; Evans, 1975) and through pleiotropic increases in plant size (Hawkes, 1983). For example, leaf size is positively correlated with the size of other parts in several species (e.g., Simmonds, 1968; Evans and Dunstone, 1970; Duarte and Adams, 1972; Burris et al., 1973; Evans, 1976; Midgley and Bond, 1989). Because in argyrosperma, seeds are most frequently consumed (Whitaker, 1968), it is likely that larger leaves have evolved as a correlate of selection for larger seeds.
Due to both their larger size and less lobed shape, leaf areas of argyrosperma are nearly four times greater than those of sororia. From a functional perspective, larger leaves provide greater surface area for photosynthate production necessary to sustain larger fruits and seeds. Gifford and Evans (1981) and references therein have suggested that increases in leaf area of cultivars contribute more significantly to increased yields over the wild species than inherent changes in C[O.sup.2] exchange rates per unit leaf area. However, it is not clear that the production of less-lobed leaves at later positions in the cultivar are a necessary correlate of greater leaf areas and hence larger fruit and seed sizes in this species. For example, one of the other cultivated varieties, var. callicarpa, grows in the more arid habitats of central and northern Mexico and the southwestern United States (Merrick, 1990). This variety produces large leaves with considerable variation in the extent of lobing (Merrick, pers. obs.), but plants of this variety also bear large fruits.
As originally conceived by Goebel (1900), ontogenetic differences in leaf form along the shoot, as occurs in sororia, are marked expressions of heteroblasty and are associated with "juvenile" and "adult" phases of plant growth. A heteroblastic series of shape changes such as that in sororia, from less-lobed to lobed leaves, may have several functional consequences for the plant (see also Ray, 1990). Harper (1989) proposed that if the value of a leaf to a plant depends on the fate of its exported photoassimilates, then "early exports from a leaf are therefore of potentially greater value to the future growth of the plant than the same material exported from the same leaf later in life." This reasoning could be extended to leaves produced early versus later in the life of the plant: early leaves contribute more carbon to future leaves that in turn generate more exported material than do later leaves. Thus, the greater the starting capital, i.e., area of early leaves, the greater the potential for reinvestment of the carbon fixed by those leaves. (See also Guerrant, 1989.) If position of the leaf along the shoot constrains blade length, as suggested by the pattern of increasing blade lengths in both subspecies along the shoot, then an earlier, less-lobed leaf would have a greater surface area than a lobed leaf at a given leaf blade length. That the area of early leaves relative to blade length is important to these subspecies is suggested by the fact that in both sororia and argyrosperma, the ratio of the square root of area to length is significantly larger in early leaves than in later leaves.
A remaining question for sororia is what is the consequence of producing lobed leaves at all. Lobed leaves may facilitate more favorable leaf energy budgets in "stressful" or unpredictable environments (Gates et al., 1968; Smith, 1978; Givnish, 1979, 1987; Nobel, 1983). As well, producing lobed leaves at later positions may reduce self-shading substantially (Niklas, 1989; and see also Wells and Meredith, 1986; Wells et al., 1986; Meredith and Wells, 1987).
In summary, this comparative study of ontogeny at the shoot level supports the hypothesis that the cultivar is progenetic with respect to its wild progenitor in terms of precocious flowering. Although less-lobed leaves of argyrosperma are visually similar to early leaves of sororia, they are not solely proportionately or allometrically larger versions of these leaves. Thus a hypothesis of simple neoteny in leaf shape evolution is not supported by the quantitative analysis. However, the similarity in patterns of multivariate allometry in each subspecies suggests that leaf blade shape in argyrosperma arises from strong determinants of shape that are shared by both the cultivar and its progenitor.
I thank D. R. Kaplan for his thoughtful insights and criticism throughout this project. L. Merrick generously provided seed and expertise on Cucurbita. I also thank the Department of Agronomy and especially the Department of Vegetable Crops at UC Davis for providing field plots. I am most grateful to F. Zinc for logistical support and assistance. Early drafts of this manuscript were improved by the criticisms of J. Coleman, P. Diggle, L. Feldman, T. Kahn, S. Morse, E. Lessa, M. Koehl, and R. Robichaux. I also thank P. Diggle, T. Dickinson, E. Guerrant, and M. Pigliucci for thoughtful comments on later drafts, and E. Lessa, R. Strauss, and M. Pigliucci for statistical advice. This work was supported in part by grants from Sigma Xi and the Chancellor's Patent Fund of UC Berkeley.
Alberch, P., S. J: Gould, G. F. Oster, and D. B. Wake. 1979. Size and shape in ontogeny and phylogeny. Paleobiology 5:296-317. Allsopp, A. 1965. Heteroblastic development in cormophytes, pp. 1172-1221. In W. Ruhland (ed.), Encyclopedia of Plant Physiology, Vol. 15. Springer Verlag, Berlin, Germany. --. 1967. Heteroblastic development in vascular plants, pp. 127-171. In M. Abercrombie and J. Brachet (eds.), Advances in Morphogenesis. Academic Press, N.Y., USA. Baker, H. G. 1972. Human influences on plant evolution. Econ. Bot. 26:32-43. Blue, M. P., and R. J. Jensen. 1988. Positional and seasonal variation in oak (Quercus; Fagaceae) leaf morphology. Am. J. Bot. 75:939-947. Bookstein, F. L. 1989. "Size and shape": A comment on semantics. Syst. Zool. 38:173-180. Bookstein, F., B. Chernoff, R. Elder, J. Humphries, G. Smith, and R. Strauss. 1985. Morphometrics in Evolutionary Biology, Special Publication 15. The Academy of Natural Sciences of Philadelphia, Philadelphia, PA USA. Burris, J. S., O. T. Edje, and A. H. Wahab. 1973. Effects of seed size on seedling performance in soybeans. 11. Seedling growth and photosynthesis and field performance. Crop Sci. 13:207-210. Cheverud, J. M. 1982. Relationships among ontogenetic, static, and evolutionary allometry. Am. J. Phys. Anthropol. 59:139-149. Coyne, D. P. 1980. Modification of plant architecture and crop yield by breeding. HortScience 15: 244-247. Decker-Walters, D. S., T. W. Walters, U. Posluszny, and P. G. Kevan. 1990. Genealogy and gene flow among annual domesticated species of Cucurbita. Can. J. Bot. 68:782-789. Dickinson, T. A. 1986. Topodeme differentiation in Ontario taxa of Crataegus (Rosaceae: Maloideae): Leaf morphometric evidence. Can. J. Bot. 64:2738-2747. Dickinson, T. A., W. H. Parker, and R. E. Strauss. 1987. Another approach to leaf shape comparisons. Taxon 36:1-20. Duarte, R. A., and M. W. Adams. 1972. A path coefficient analysis of some yield component interactions in field beans (Phaseolus vulgaris L.). Crop Sci. 12:579-582. Erickson, R. O., and F. J. Michelini. 1957. The plastochron index. Am. J. Bot. 44:297-305. Evans, L. T. 1975. Crops and world food supply, crop evolution, and the origins of crop physiology, pp. 1-22. In L. T. Evans (ed.), Crop Physiology--Some Case Histories. Cambridge University Press, Cambridge, UK. --. 1976. Physiological adaptation to performance as crop plants. Phil. Trans. R. Soc. London B. 275:71-83. Evans, L. T., and R. L. Dunstone. 1970. Some physiological aspects of evolution in wheat. Aust. J. Biol. Sci. 23:725-741. Gates, D. M., R. Alderfer, and S. E. Taylor. 1968. Leaf temperature of desert plants. Science 159:994-995. Gifford, R. M., and L. T. Evans. 1981. Photosynthesis, carbon partitioning, and yield. Annu. Rev. Plant Physiol. 32:485-509. Givnish, T. 1979. On the adaptive significance of leaf form, pp. 375-407. In O. T. Solbrig, S. Jain, G. B. Johnson, and P. H. Raven (eds.), Topics in Plant Population Biology. Columbia University Press, N.Y., USA. --. 1987. Comparative studies of leaf form: Assessing the relative roles of selective pressures and phylogenetic constraints. New Phytol. (Suppl.). 106: 131-160. Goebel, K. 1900. Organography of Plants. I. General Organography. Reprinted by Hafner Publishing Co., Inc., N.Y., USA. Gottlieb, L. D. 1986. The genetic basis of plant form. Phil. Trans. R. Soc. London B. 313:197-208. Gould, S. J. 1966. Allometry and size in ontogeny and phylogeny. Biol. Rev. 41:587-640. --. 1977. Ontogeny and Phylogeny. Harvard University Press, Cambridge, MA USA. Groff, P. A. 1989. Studies in whole plant morphology. Ph.D. Diss. University of California, Berkeley, CA USA. Guerrant, E. O., JR. 1982. Neotenic evolution of Delphinium nudicaule (Ranunculaceae): A hummingbird-pollinated larkspur. Evolution 36:699-712. --. 1989. Heterochrony in plants: The intersection of evolution, ecology and ontogeny, pp. 111-133. In M. McKinney (ed.), Heterochrony and Evolution: A Multidisciplinary Approach. Plenum Press, N.Y., USA. Harlan, J. R., J. M. J. De Wet, and E. G. Price. 1973. Comparative evolution of cereals. Evolution 27:311-325. Harper, J. L. 1989. The value of a leaf. Oecologia 80:53-58. Hawkes, J. G. 1983. The Diversity of Crop Plants. Harvard University Press, Cambridge, MA USA. Hockett, E. A., and P. F. Knowles. 1970. Inheritance of branching in sunflowers, Helianthus annuus L. Crop Sci. 10:432-436. Hufford, L. D. 1988. Roles of early ontogenetic modifications in the evolution of floral form of Eucnide (Loasaceae). Bot. Jahrb. Syst. Pflanzengesch. Pflanzengeogr. 109:289-333. James, F. C., and C. E. McCulloch. 1990. Multivariate analysis in ecology and systematics: Panacea or Pandora's box? Annu. Rev. Ecol. Syst. 21:129-166. Jolicoeur, P. 1963. The multivariate generalization of the allometry equation. Biometrics 19:497-499. Jolicoeur, P., and J. E. Mosimann. 1960. Size and shape variation in the painted turtle, a principal component analysis. Growth 24:339-354. Jones, C. S. 1990. The developmental basis of leaf shape variation in a wild and cultivated subspecies of Cucurbita. Ph.D. Diss. University of California, Berkeley, CA USA. Kaplan, D. R. 1973. Comparative developmental analysis of the heteroblastic leaf series of axillary shoots of Acorus calamus L. (Araceae). La Cellule 69:253-290. Kluge, A. G. 1988. The characterization of ontogeny, pp. 57-81. In C. J. Humphries (ed.), Ontogeny and Systematics. Columbia University Press, N.Y., USA. Lacroix, C,. R., and U. Posluszny. 1991. A morphometric analysis of leaf development in Vitis riparia, and grape cultivars Concord and Vivant. Can. J. Bot. 69:400-406. Lessa, E. P., and J. L. Patton. 1989. Structural constraints, recurrent shapes, and allometry in pocket gophers (genus Thomomys). Biol. J. Linn. Soc. 36: 349-363. Lord, E. M. 1982. Floral morphogenesis in Lamium amplexicaule L. (Labiatae) with a model for the evolution of the cleistogamous flower. Bot. Gaz. 143:63-72. Mabberley, D. J. 1985. 'Die Neuen Pflanzen von C. H. Huber Freres & Co. in Hyeres'. Taxon 34:448-456. McKinney, M. L., and K. J. McNamara. 1991. Heterochrony: The Evolution of Ontogeny. Plenum Press, N.Y., USA. Meredith, W. R., and R. Wells. 1987. Sub okra leaf influence on cotton yield. Crop Sci. 27:47-48. Merrick, L. C. 1990. Systematics and evolution of a domesticated squash, Cucurbita argyrosperma, and its wild and weedy relatives, pp. 77-95. In D. M. Bates, R. W. Robinson, and C. Jeffrey (eds.), Biology and Utilization of the Cucurbitaceae. Comstock Publishing Associates, Ithaca, NY USA. Merrick, L. C., and D. M. Bates. 1989. Classification and nomenclature of Cucurbita argyrosperma: Baileya 23:94-102. Midgley, J., and W. Bond. 1989. Leaf size and inflorescence size may be allometrically related traits. Oecologia 78:427-429. Niklas, K. J. 1989. The effect of leaf-lobing on the interception of direct solar radiation. Oecologia 80: 59-64. Nobel, P. S. 1983. Biophysical Plant Physiology and Ecology. W. H. Freeman, San Francisco, CA USA. Pickersgill, B. 1986. Evolution of hierarchical variation patterns under domestication and their taxonomic treatment, pp. 191-209. In B. T. Styles (ed.), Infraspecific Classification of Wild and Cultivated Plants. The Systematics Association Special Volume No. 29, Clarendon Press, Oxford, UK. Pickersgill, B., and C. B. Heiser, JR. 1976. Cytogenetics and evolutionary change under domestication. Phil. Trans. R. Soc. London B. 275:55-69. Pimentel, R. A. 1979. Morphometrics, The Multivariate Analysis of Biological Data. Kendall/Hunt Publ. Co., Dubuque, IA USA. Purseglove, J. W. 1968. Tropical Crops--Dicotyledons, Vol 1. Longmans, Green and Co. Ltd., London, UK. Raff, R. A., and G. A. Wray. 1989. Heterochrony: Developmental mechanisms and evolutionary results. J. Evol. Biol. 2:409-434. Ray, T. S. 1990. Metamorphosis in the Araceae. Am. J. Bot. 77:1599-1609. Rohlf, F. J. 1987. NTSYS-pc: Numerical Taxonomy and Multivariate Analysis System for the IBM PC Microcomputer (and Compatibles). Applied Biostatistics Inc., Setauket, NY USA. --. 1990. Morphometrics. Annu. Rev. Ecol. Syst. 21:299-316. Rolhf F. J., and F. L. Bookstein. 1987. A comment on shearing as a method for "size correction". Syst. Zool. 36:356-367. Roth, V. L. 1988. The biological basis of homology, pp. 1-26. In C. J. Humphries (ed.), Ontogeny and Systematics. Columbia University Press, N.Y., USA. SAS. 1985a. SAS User's Guide: Basics, Version 5. SAS institute, Cary, NC USA. --. 1985b. SAS User's Guide: Statistics, Version 5. SAS Institute, Cary, NC USA. Schwanitz, F. 1966. The Origin of Cultivated Plants. Harvard University Press, Cambridge, MA USA. Shea, B. T. 1985. Bivariate and multivariate growth allometry: Statistical and biological considerations. J. Zool., London (A). 206:367-390. Silk, W. K. 1980. Plastochron indices in cantaloupe grown on an irrigation line source. Bot. Gaz. 141: 73-78. Simmonds, N. W. 1968. Change of leaf size in the evolution of the tuberosum potatoes. Euphytica 17: 504-506. Smith, M. F., and J. L. Patton. 1988. Subspecies of pocket gophers: Causal bases for geographic differentiation in Thomomys bottae. Syst. Zool, 37: 163-178. Smith, W. K. 1978. Temperatures of desert plants: Another perspective on the adaptability of leaf size. Science 201:614-616. SAS, R. R., and F. J. Rohlf. 1981. Biometry, 2nd. W. H. Freeman and Company, N.Y., USA. Strauss, R. E., and F. L. Bookstein. 1982. The truss: Body form reconstructions in morphometrics. Syst. Zool. 31:113-135. Thorpe, R. S., and L. Leamy. 1983. Morphometric studies in inbred and hybrid House mice (Mus sp.): Multivariate analysis of size and shape. J. Zool., London 199:421-432. Tomlinson, P. B. 1984. Homology: An empirical view. Syst. Bot. 9:374-381. Wagner, G. P. 1989. The biological homology concept. Annu. Rev. Ecol. Syst. 20:51-69. Wake, D. B., and A. Larson. 1987, Multidimensional analysis of an evolving lineage. Science 238: 42-47. Wells, R., and W. R. Meredith. 1986. Normal vs. okra leaf yield interactions in cotton. 11. Analysis of vegetative and reproductive growth. Crop. Sci. 26:223-228. Wells, R., W. R. Meredith Jr., and J. R. Williford. 1986. Canopy photosynthesis and its relationship to plant productivity in near-isogenic cotton lines differing in leaf morphology. Plant Physiol. 82:635-640. Weston, P. H. 1988. Indirect and direct methods in systematics, pp. 27-56. In C. J. Humphries (ed.), Ontogeny and Systematics. Columbia University Press, N,Y., USA. Whitaker, T. W. 1968. Ecological aspects of cultivated Cucurbita. HortScience 3:9-11. White, R. J., H. C. Prentice, and T. Verwijst. 1988. Automated image acquisition and morphometric description. Can. J. Bot. 66:450-459. Wilson, H. D. 1989. Discordant patterns of allozyme and morphological variation in Mexican Cucurbita. Syst. Bot. 14:612-623.
|Printer friendly Cite/link Email Feedback|
|Author:||Jones, Cynthia S.|
|Date:||Dec 1, 1992|
|Previous Article:||Patterns in tree balance among cladistic, phenetic and randomly generated phylogenetic trees.|
|Next Article:||Optimum brood size: tests of alternative hypothesis.|