Inflorescences of Cuscuta (Convolvulaceae): Diversity, evolution and relationships with breeding systems and fruit dehiscence modes

Morgan Glofcheskie, et al.

Inflorescence development and major architectural types in Cuscuta

Quantitative data obtained/used in this study were summarized in S2 Appendix. SEM imaging data did not reveal any additional details compared to stereomicroscopy, and was not included in the results.

Cuscuta has two functional types of shoot systems: exploratory, which forage in the plant community, and haustorial, which twine in a dextrorse direction around the stems of the hosts and produce haustoria. Vegetative branching occurs especially in the exploratory shoots and is sympodial. One to four (six) shoots can emerge successively from axillary buds at the base of the cauline scale-like leaves, which have a 2/5 phyllotactic alternate arrangement. In subgenera Monogynella and Cuscuta, “floral unit meristems” sensu [29] develop extensively on the exploratory stems, while in subgenera Pachystigma and Grammica, most of the cymes are concentrated on the haustorial stems.

Inflorescences are axillary; the primary axes originate from buds located underneath the reduced, cauline scale-like leaves, while the subsequent, higher order axes, arise from buds found underneath bracts. The inflorescence fundamental unit in all Cuscuta species is a monochasial cincinus or scorpioid cyme, and different architectural patterns emerge depending on whether the cymes are simple or compound, and contingent on the number of orders and the length of different axes. Inflorescence axes and pedicels attain maximum length at anthesis and remain size-invariant at fructification. Thus, in Cuscuta, infructescences are identical to inflorescences.

Two main rules that allow decoding of the inflorescence patterns are: 1) scale-like leaves or bracts always indicate a branching point and the origin of a cyme. This is made evident by their removal, which reveals underneath the flower buds at different stages of development. 2) The oldest flowers are the farthermost from the leaf scales or bracts. In the case of the compound cymes, the oldest cymes are the farthermost from the leaf-scale. Three major types of developmental patterns are evident in Cuscuta, and characterize subgenera.

Monogynella” type (Figs 1A, 2, 5A, 5B). Inflorescences in species of subg. Monogynella are compound monochasial scorpioid cymes. Characteristic to this type is the prolonged growth of the longest primary axes, which superficially imparts them the appearance of thyrses (racemes or spikes with cymes).


Fig 1. Diagrams of major types of inflorescences in Cuscuta.

A. “Monogynella” type is a compound monochasial scorpioid cyme in which the main axes (e.g., I and II) superficially resemble thyrses. B. “Cuscuta” type is a simple scorpioid cyme, in this case with sessile flowers. B1. Simple cyme seen laterally (arrow indicates the direction of flower development). B2. Cyme seen from the top. C. “Grammica” type is also a compound monochasial scorpioid cyme, but with shorter axes. b = bract; ls = leaf scale; I, II, III indicate the order of axes.


Fig 2. Development of “Monogynella” type of inflorescence.

A, B. The first inflorescence axis. A. Apical bud that grows vegetatively and generates lateral monochasial cymose axes. B. General view of first inflorescence axis. C–E. Progressive stages in the development of secondary axes. F, G. Simple scorpiod cymes developing on the secondary axes. H. More advanced stage in the development of the overall compound scorpioid cyme. I, II, III indicate axes of different orders. b = bracts; ls = leaf scale. Scale bars = 1 mm.

Cuscuta” type (Figs 1B, 3, 5C–5E). The species of subg. Cuscuta examined in this study have sessile flowers developing on a small, common receptacle, under a single leaf scale. The dense, spherical inflorescences superficially resemble a capitulum (but which is a racemose type of inflorescence). Flower formation follows a zig-zag pattern, with the youngest flowers emerging closest to the leaf scale and the oldest located farthest away, the inflorescence being a simple scorpioid cyme.


Fig 3. Development of shoot system and “Cuscuta” type of inflorescence.

A–E. Sympodial development of exploratory shoot system. A–D. Axillary exploratory shoot. A. Apex of lateral exploratory shoot. B. Fragment of (primary) exploratory shoot showing axillary (secondary) developing exploratory shoot. C, D. Distal part of exploratory shoot from B. E. First two orders of exploratory shoots. F–I. Development of inflorescence. F. Incipient stage of flower primordia emerging at the base of series of axillary exploratory shoots (1, 2, 3, 4). G–I. Sessile flowers developing in zig-zag to form a simple scorpioid monochasial cyme (leaf scale removed from I). ls = leaf scale; lsp = leaf scale primordia; f = flower. Scale bars = 1 mm.

Grammica”-type (Figs 1C, 4, 5F–5P). The inflorescence of all the species of subg. Pachystigma and most species of subg. Grammica is a compound monochasial scorpioid cyme with up to five orders of axes. In contrast to the “Monogynella” type, the primary axes have a limited growth, and thus, the semblance to a thyrse is not apparent.


Fig 4. Development of “Grammica” type of inflorescence.

A–D. Cuscuta gronovii. A1. Shoot. A2. Incipient stage of axillary inflorescence. B–D. Progressive stages in the development of compound monochasial scorpioid cyme resembling a panicle. E–J. Cuscuta campestris. Different stages of “glomeruliform-subglomeruliform” inflorescence development. K–M. “Rope” inflorescence of Cuscuta glomerata. K. General view of early stage. L. Flower primordia (indicated with arrows) emerging from haustorial stems. M. Developing flowers. ls = leaf scale; b = bract; Hs = haustorial stem; I, II, II, orders of axes; scale bars = 1 mm.


Fig 5. Variation of inflorescence architecture in Cuscuta.

A, B. Subgenus Monogynella. A. Cuscuta lupuliformis. B. C. lehmanniana (photo by Vladimir Kolbintsev). C–E. Subgenus Cuscuta. C. C. planiflora. D. C. epithymum. E. C. babylonica (photo by Miguel García). F, G. Subgenus Pachystigma, “umbelliform-corymbiform” inflorescences. F. C. africana (photo by Miguel García). G. C. angulata (photo by Miguel García). H–P. Subgenus Grammica. H, I. “Umbelliform-corymbiform” inflorescences. H. C. sidarum. I. C. erosa (photo by Jillian Cowles). J–K. “Racemiform- paniculiform” inflorescences. J. C. corymbosa var. grandiflora. K. C. tinctoria var. floribunda. L, M. “Glomeruliform -subglomeruliform” inflorescences. L. C. volcanica. M. C. obtusiflora var. glandulosa. N. Flowers solitary or in fascicles, C. grandiflora. O. “Rope”, C. glomerata (photo by Richard Lutz). P. Mimetism of C. howelliana inflorescence developed inside the inflorescence of Eryngium castrense (Photo by Carol Witham).

The variations encountered in subg. Grammica refer to the number of orders, the number of axes per order, and their length relative to the length of the pedicels. Grammica type is the most diverse within genus Cuscuta (Fig 5). Based on their superficial resemblance to racemose inflorescences, following the literature, the compound cymes of subgenera Pachystigma and Grammica were further informally categorized as “umbelliform-corymbiform” (Fig 5F–5I) and “racemiform- paniculiform” (Fig 5J and 5K) when flowers were long-pedicelled. Compound cymes with sessile or very short pedicellate flowers were scored as “glomeruliform -subglomeruliform” (Fig 5L and 5M).

Exceptions of solitary flowers or simple scorpioid cymes with just a few flowers (“fascicles”) evolved only in a few species from several clades of subg. Grammica (Fig 5N; see next section). Another exception within subg. Grammica is the “rope”, which is a unique flower-aggregation architecture that was observed in C. glomerata (sect. Oxycarpae). Flower buds emerge endogenously directly from the haustorial shoots, irrespectively to the reduced leaves (Fig 4K–4M). No axes or pedicels develop and, as a result, the parasite stem appears as a dense “rope” of flowers twining around the stem of the host (Fig 5O).

Inflorescence character evolution

In a first reconstruction of ancestral character states only the three major types of inflorescences were scored. Likelihood reconstruction suggested that “Monogynella” type is ancestral (proportional likelihood = 0.6962; Fig 6). At the second node, “Grammica” type had a higher likelihood (proportional likelihood = 0.5328; Fig 6) compared to the other two types of inflorescences. In the parsimony reconstruction, all three major types were indicated as most parsimonious (MPR) at the first and second node; with Grammica type being MPR only at the third node (results not shown).


Fig 6. Summary of subgeneric character evolution hypotheses for the three main types of inflorescences using likelihood.

Monogynella” type was suggested to be ancestral (proportional likelihood = 0.6962), followed by “Cuscuta” and “Grammica” types.

In a second character evolution analysis, we also added as character states “rope”, “solitary or fascicles” as well as the morphologies resembling racemose types for the subg. Pachystigma and Grammica (Fig 5). Some species of subg. Grammica possess inflorescences that correspond to multiple character states (polymorphism), and only the parsimony reconstruction could be used. The four species of subg. Pachystigma examined are all “umbelliform-corymbiform” (Fig 7). While at the first nodes of subg. Grammica, both “umbelliform-corymbiform” and “glomeruliform-subglomeruliform” are indicated as MPR, at the deeper nodes, only the latter dominate (Fig 7). Overall, each of 15 clades of subg. Grammica is characterized by one or two major “types” accompanied sometimes by the recurrent evolution of a third “type”. The most common type of inflorescence in subgenus Grammica is glomeruliform-subglomeruliform. “Solidary-fascicle” evolved in three major clades of subg. Grammica (sections Subulatae, Lobostigmae and Denticulatae; Fig 7). A convergent trend of shortening of the total axes was noticeable at the scale of the entire genus (S1 Fig), but this trend was not observed for the length of the pedicels (S2 Fig).


Fig 7. Parsimony reconstruction of all the types of inflorescences, including the racemose analogies in subg.

Pachystigma and Grammica, revealed extensive convergent evolution in the latter subgenus.

Principal component analysis of inflorescence traits and correlations with pollen-ovule ratios, and fruit width

Principal Component 1 accounted for 36.81% of the variance and component 2 accounted for 25.79% of the variance (Table 2). Cumulatively, both components 1 and 2 accounted for 62.6% of the variance. The biplot showed that all the variables had positive loadings on the first component, with corolla diameter and corolla tube length having larger positive loadings than the pedicel length and total length of axes (Fig 8). Therefore, component 1 can be regarded as a general flower size-related variable, with taxa that have large positive PC1 scores exhibiting on average larger corolla diameter, longer corolla tubes, total axes and pedicels, while taxa with large negative principal component scores showing on average shorter corolla tubes and smaller corolla diameters and shorter total axes/pedicels.

The second principal component had positive loadings for pedicel length and total axes length, and negative loadings for corolla diameter and corolla tube length (Table 2). Therefore, component 2 can be seen as an overall axes length-related variable; taxa that have large positive principal component 2 scores, on average, have longer pedicels and longer total axes, but smaller corolla diameter and shorter corolla tubes, while taxa with large negative principal component 2 scores tend to have shorter pedicels, and shorter total axes, but longer corolla tubes and larger corolla diameter.

Using principal components 1 and 2, correlations between the principal component scores and pollen-ovule ratios, and fruit width were examined using Spearman’s correlation. Statistically significant positive correlations were observed between both component 1 scores and pollen-ovule ratios, and with fruit width (Fig 9A and 9C). Pollen-ovule ratios had the largest positive correlation with a rho value of 0.533 (Table 3). This indicates a pattern according to which larger flowers with longer pedicels and total axes also have large pollen-ovule ratio values, and fruit widths.


Fig 9. Correlation plots between components and reproductive traits.

A. Between component 1 and pollen-ovule ratios. Equation of the line of best fit: PO = 7.655 × (Comp 1)2 + 305.662 × (Comp 1) + 974.417. B. Between component 2 and pollen-ovule ratios. Equation of the line of best fit: PO = 54.93 × (Comp 2)5–80.28 × (Comp 2)4–358.95 × (Comp 2)3 + 448.34 × (Comp 2)2 + 162.83 × (Comp 2) + 765.47. C. Between component 1 and fruit width. D. Equation of the line of best fit, Fruit width B = 0.3327 × (Comp 1) + 2.7887. Equation of the line of best fit, fruit width B = −0.07643 × (Comp 2) + 2.78871.

Negative correlations were observed between component 2 scores and both pollen-ovule ratios, and fruit width (Fig 9B and 9D). The correlations for pollen-ovule ratios and component 2 were significant, but the correlation between fruit width and component 2 was not statistically significant (Table 3). Therefore, taxa with small flowers, long pedicles and total axes lengths tend to have smaller pollen-ovule ratios, while the size of their fruits is largely independent.

Relationships between fruit variables and between inflorescences and fruit

A Kruskal-Wallis Rank Sum test was performed to compare the effect of the mode of dehiscence on median fruit width and length. Fruit dehiscence character states dehiscent (DE), dehiscent and irregular dehiscent type A (DE + IrA), indehiscent (IN), and indehiscent and irregularly dehiscent type B (IN + IrB) were included as categorical variables. Overall, we rejected the null hypothesis that the median fruit width was the same across all dehiscence groups (Kruskal-Wallis χ2 = 11.69, p-value = 0.009). Pairwise comparisons indicated that the only significant differences in median fruit widths were located between IN and DE groups, and between IN + IrB and IN groups (Fig 10). Overall, no statistically significant differences were found between modes of dehiscence and median fruit length scores (χ2 = 6.689, p-value = 0.082), nor between median pedicel length and the modes of dehiscence groups (χ2 = 5.603, p-value = 0.133).


Fig 10. Boxplot of fruit width for DE, DE+IrA, IN, and IN+IrB modes of dehiscence (3 groups were removed as they had 3 taxa or less).

Boxes show the middle 50% of component 1 values for the modes of dehiscence, horizontal lines in the boxes show the median, circles are outliers, and the whiskers show the minimum and maximum values. Significant differences were noted for at least two of the groups of dehiscence examined. Pairwise comparisons using Wilcoxon rank sum test revealed significant differences between IN and DE (p-value = 0.033), and IN+IrB and IN (p-value = 0.033).

Source link