Table 1.
Seed characters surveyed and their representative codes and states.
Continuous characters values are averages.
Fig 1.
Seed features revealed with scanning electron microscopy.
A. Epidermis surface, C. lupuliformis (subg. Monogynella). B–D. Overall seed morphology and surface of dry epidermis. B. Cuscuta europaea (subg. Cuscuta). C–D. Cuscuta cephalanthi (subg. Grammica). C. Entire seed. D. Detail of pitted epidermis of dry seeds. E–F. Different stages of epidermis rehydration. E. Cuscuta gronovii var. gronovii. F. Cuscuta cephalanthi. G–H. Hilum area (black arrows indicate hilar fissure). G. Cuscuta approximata. H. Cuscuta mitriformis. I–K. Embryo morphology. I. Filiform and coiled, C. pacifica. J–K. Globose toward the radicular end. J. Cuscuta nevadensis. K. Cuscuta microstyla. L–M. Longitudinal sections through the hilum area showing all the seed components. L. Cuscuta epithymum. M. Cuscuta globulosa. N–P. Seed coat anatomy. N. Cuscuta lupuliformis. O. Cuscuta alata. P. Cuscuta gronovii var. gronovii. Ep = epidermis; En = endosperm; E = Embryo; P1 = Inner or single palisade layer; P2 = Outer palisade layer. Scale bars. A, E, F = 200 μm; D = 40 μm; G, H = 100 μm; B, C, I–M = 0.5 mm; N–P = 50 μm.
Fig 2.
Summary of character evolution hypotheses.
A. Evolution of epidermis morphology. An invariant epidermis with rectangular, interlocked epidermal cells (Type I) is likely ancestral and characterizes subg. Monogynella. An epidermis with isodiametric cells that can alternate their morphology between dome-shaped and pitted (Type II) evolved in subgenera Cuscuta, Pachystigma and Grammica. B. Evolution of embryo and architecture of palisade layers outside hilum area. The seed coat with only one palisade layer (P1) outside hilum area in subg. Monogynella is likely ancestral, while a seed coat with two palisade layers (P1 and P2) in the remaining subgenera, is likely derived; one palisade layer reverted two times in subg. Grammica in C. microstyla (clade O) and sect. Denticulatae (clade E). The latter taxa also evolved an embryo with an enlarged radicular end, which likely functions as a storage organ. Ep = epidermis; En = endosperm.
Fig 3.
Seed features viewed with light microscopy.
A–D. Embryos. A. Cuscuta monogyna (embryo removed from the endosperm); note that in this species the embryo does not form a full coil. B. Embryo of C. tinctoria var. floribunda embedded in the endosperm. C. Developing embryo of C. nevadensis surrounded by the endosperm epidermis (the rest of endosperm was almost entirely consumed). D. Fully developed embryo of C. nevadensis (endosperm epidermis removed). E–H. Cuscuta lupuliformis (subg. Monogynella). E–G. Longitudinal sections through the hilum area of C. lupuliformis. E. Overview; black asterisk indicates position of water gap. F. Detail of hilum area; black arrow indicates position of water gap; note that two palisade layers (P1 and P2) are present. G. Tracheid-like structures embedded in a parenchyma tissue. H. Testa architecture outside hilum area with only one palisade layer; black arrow indicates linea lucida. I–K. Seed coat architecture with two palisade layers outside the hilum area. I. Incipient stage in the development of the two palisade layers in C. argentinana; at this stage, epidermis contains starch grains. J. Cuscuta europaea. K. Cuscuta cristata; note the presence of linea lucida in the inner palisade layer (P1). L. Parenchyma cells with lipids and starch in the enlarged portion of C. nevadensis embryo. M. Longitudinal section of rehydrated C. sandwichiana seed after 15 min in Aniline Blue; dye penetration is limited to the water gap (indicated with arrows). E = Embryo; En = endosperm; H = hilum; Ep = epidermis; Par = parenchyma; P1 = Inner or single palisade layer; P2 = Outer palisade layer. Scale bars. A–E = 0.5 mm; G, I–K = 50 μm; H, L = 25 μm; F, M = 100 μm.
Fig 4.
Analysis of variance for the breeding system categories and the average number of seeds per capsule (S/C).
A. Facultatively autogamous taxa have the highest S/C averages, but also the highest variation. B. Facultatively xenogamous group include species that possess an intermediate S/C number between the other two categories. C. Fully xenogamous taxa have the lowest S/C average and the least amount of variation; species in this category are self-incompatible.
Fig 5.
Regression tree analysis of number of seeds per capsule (NRSeedCapsule) and pollen/ovule ratios (P/O) used as an indicator of breeding systems.
The first split separated directly the leaf of 14% facultatively autogamous taxa with an average P/O of 226 (first leaf to the left). At the next node, the remainder species were divided depending on whether they had more or less than 2.5 S/C. 45% of taxa had more than 2.5 S/C and were split again depending whether they had more or less than 3.3 S/C. 14% of taxa had more than 3.3 S/C and were placed in the second terminal leaf, with a P/O of 746. 31% had less than 3.3 S/C and were separated in the third terminal leaf, with a P/O of 1010. Taxa with more than 2.3 S/C were found in the sixth terminal leaf, comprising 12% of the total, P/O of 1681. Taxa with less than 2.3 S/C were divided once more if they have more of less than 1.3 S/C. 7% of the total had less than 1.3 S/C, P/O 1012, while 22% had more than 1.3 S/C and P/O of 1369.
Fig 6.
Kaplan-Meier survival curves showing the proportion of Cuscuta seedlings surviving (log scale).
Species are listed in order of their increasing seed size: blue line = C. epithymum, the smallest seeds (0.89 mm in average); red line = C. costaricensis, intermediate sized-seeds (1.08 mm in average); green line = C. campestris, the largest seeds (1.28 mm in average). Standard error not shown for clarity.