A Survey of Mucilaginous Testa in Chamaesyce

JORDAN, M. S. & W. J. HAYDEN (1992). A Survey of Mucilaginous Testa in Chamaesyce. Collect. Bot. (Barcelona) 21 :79-89. Seeds of Chamaesyce were studied for presence of mucilaginous testa. Samples were selected to represent all major systematic sections within the genus. Observations were made with a dissecting microscope following brief hydration of seeds; additional SEM observations of both intact and fractured dry seeds were made for selected species. With few exceptions, most of Boissier's subsections test positively for seed mucilage; however, mucilage is generally absent in subsections Gymnadeniae and "Sclerophyllae", for wich absence of mucilage is regarded as secondary loss from mucilagionus ancestors. Although mucilage production is associated with a well-defined subepidermal layer of macroesclereids, it is the surface layer of cells that actually secrete mucilage. Mucilage secreting cells and macroesclereids of mature testa are interpretted to develop from the epidermal layers of outer and inner integuments, respectively. Seed mucilages may play a role in seed hydration and/or seed dispersal.

Seeds of Chamaesyce were studied for presence of mucilaginous testa.Samples were selected to represent all major systematic sections within the genus.Observations were made with a dissecting microscope following brief hydration of seeds; additional SEM observations of both intact and fractured dry seeds were made for selected species.With few exceptions, most of Boissier's subsections test positively for seed mucilage; however, mucilage is generally absent in subsections Gymnadeniae and "Sclerophyllae", for wich absence of mucilage is regarded as secondary loss from mucilagionus ancestors.Although mucilage production is associated with a well-defined subepidermal layer of macroesclereids, it is the surface layer of cells that actually secrete mucilage.Mucilage secreting cells and macroesclereids of mature testa are interpretted to develop from the epidermal layers of outer and inner integuments, respectively.Seed mucilages may play a role in seed hydration and/or seed dispersal.

INTRODUCI10N
The value of seed and testa morphology in the systematics of Euphorbia L., sensu lato, is well-known (KROCHMAL, 1952;EHLER, 1976).Size, shape, surface texture, and presence or absence of a caruncle are all routinely employed to help distinguish species and other subgeneric taxa in this exceptionally diverse and speciose group.Seed features are especially useful in the segregate genus Chamaesyce S. F. Gray (see e.g., BURCH 1966;McVAUGH, 1961;WHEELER, 1941).
That seeds of some species of Euphorbia and Chamaesyce exude a mucilaginous surface layer following hydration has received sporadic comment in previous literature.MANDL (1926) described such a layer in species of Euphorbia subgenus Tithymalus, as did ROSSLER (1943), who also made note of seed coat mucilages in several species of Chamaesyce.Correll and Johnston (1970) mention the mucilaginous layer exuded by hydrated seeds of Chamaesyce revoluta (Engelm.)Small.CARLQUIST (1966CARLQUIST ( , 1980) ) has discussed the presence and absence of mucilaginous seeds in relation to the dispersal of Chamaesyce to the Hawaiian Islands.In only one species, C. macu/ata (L.) Small (JORDAN et al., 1985; reported as E. supina Raf.), has the structure of the mucilaginous layer been studied in detail.
The present study was stimulated by the observation that, while the ability to produce mucilaginous exudations from testa seems to be widespread in Chamaesyce, some species, in particular, C. mesembrianthemifolia (Jacq.)Dugand and certain Hawaiian species (CARLQUIST 1966(CARLQUIST , 1980;;KOUTNIK, 1987), lack this capacity.In order to ascertain the taxonomic distribution of mucilaginous testa in Chamaesyce, we describe here the results of a preliminary survey of all major species groups of Chamaesyce as classified by BOISSIER (1862).We also compare mucilaginous and non-mucilaginous seeds in the dry condition as viewed with SEM.

MATERIALS & MEIBODS
For the most part, seeds were obtained from herbarium specimens.However, seeds of a few species were obtained from mass collections of whole live plants dried in covered cardboard boxes in order to retain the ballistically dispersed seeds.
Specimens tested and their documentation are included in Table 1.Subsectional taxonomy follows BOISSIER (1862) with some modifications.We follow SUBlLS (1977) who transferred C. selloi (Kl.& Gke.) Croizat from subsection "Pleiadeniae" to subsection "Chamaesyce" and KOUTNIK (1987) who moved C. degeneri (Sherff) Croizat & Degener from "Sclerophyllae" to Gymnadeniae (Boiss.)Koutnik.Further, we tentatively include C. lata (Engelln.)Small in subsection "Chamaesyce", since MAYFIELD (1991) has argued for its exclusion from subsection "Acu~ae".We also depart from Boissier's (1862) treatment of these plants by according them generic status separate from Euphorbia and have used combinations in the genus Chamaesyce wherever possible.Although many species have nomenclatural combinations available in both genera, this is not true for all, and not all of Boissier's subsections have been formally adopted as subgeneric taxa of Chamaesyce; these circumstances pose nomenclaturally awkward consequences in communication of our morphological information.Some species, for example the recently described "Euphorbia" johnstonii Mayfield, appear so close to the boundary between Euphorbia and Chamaesyce that their ultimate inclusion in the latter may be doubted.Further, Boissier's subsections, although a useful framework for a preliminary survey, must be considered provisional at best in light of the years that have passed since their first proposal.For these reasons we forego perfunctory coining of nomenclatural novelties for taxa without published names in Chamaesyce and, further, as a temporary means of communication, we have resorted to placing such names in   Hayden 2050, ventral view, potentially mucilaginous. 8. C. glyptosperma, Hayden 934, dorsal view; note contrast between potentially muciJaginous cells at ridges and non-mucilaginous cells between ridges.9. C. mesembrianthemifolia, Hayden 709,ventral view,Hayden 3140,ventral view,, bars= 500 µm ; 8, bar = 1 mm.quotation marks.No new names, new combinations, nor new statuses are overtly intended in this publication.
Ability to produce a mucilaginous layer was assessed by placing several seeds from each collection in a drop of water and observing their reacti.onwith a Nikon SMZlO stereoscopic dissecting microscope at lOx to 40x.Seeds were kept moist and observations continued for a period of 5 minutes; if seeds showed no mucilaginous exudations within this time, mucilage was judged to be absent.Pbotomicrographs (Figures 1-6) were prepared from Kodak Tmax 100 film processed with Tmax developer.
Seeds of selected specimens were prepared for SEM as follows: Ory seeds were mounted directly on aluminum conductive tape affixed to specimen stubs.Intact seeds were mounted to expose both dorsal and ventral surfaces.Seeds of some species were sectioned transversely with a razor blade and mounted with the cut surface exposed.After mounting, seeds were placed on a 40 C warming tray in order to drive out residual moisture.Seeds were then sputter coated with 15 nm of gold-palladium mixture and observed with a Hitachi S-2300 SEM.Electronmicrographs  were prepared from 4" x 5" Kodak Tri-X film developed in Kodak HCl 10 developer at dilution "B.,,

RESULTS & DISCUSSION
Description of dry and hydrated seed sulfaces.Many, but not all, species of Chamaesyce tested produce mucilage upon hydration of the seed coat.Reactions of the species tested are listed in Table 1, which also includes data for several species based on previous literature.Mucilaginous seeds generally react within 10 seconds following exposure to water.Cells of the surface layer of testa form opaque cylindrical projections extending at roughly right angles from the surface.Thickness of the mucilage layer varies from species to species, and can be roughly correlated with seed size.Individual mucilage strands from large seeds, such as C. polygon{folia (L.) Small (Figu res 1, 2) or C. acuta (Engelm.)Millsp., approached 0.5 mm in length, whereas those of much smaller seeds, e.g., C. hirta (L.) Millsp., were less than 0.1 mm long.To the extent detectable at 40x mucilage layers observed were consistent with descriptions for C. macu/ata by JORDAN et al. (1985).We also confirm the report that the mucilaginous layer of this species can reform repeatedly under alternating cycles of wet and dry conditions (JORDAN et al., 1985); indeed, this capacity seems characteristic of all mucilaginous species examined.Seed size, shape, and surface topography vary widely within Chamaesyce.No obvious surface features were found that correlate consistently with ability to produce mucilage.For example; seeds over 2 mm long include both mucilaginous, e.g., C. acuta, and non-mucilaginous examples, such as C. remyi (Boiss.)Croizat & Degener.Similarly, many species with seeds less than 1.5 mm long are mucilaginous, whereas, C. mesembrianthemifolia (Figure 9), with seeds ca 1.3 mm long, is non-mucilaginous.Mucilage production was found on smooth seeds, e.g., C. acuta (Figure 10), C. polygonifolia (Figures 1, 2), as well as on wrinkled seeds, such as C. glyptosperma (Engelm.)Small (Figures 5, 6, 8) and C. hypericifolia (L.) Small (Figure 7).As viewed with SEM, some species with mucilaginous seeds, such as C. nutans (Lag.)Small (Figures 3, 4), have convexly protruding surface cells forming a densely pusticulate surface (Figure 13), but so do some non-mucilaginous seeds, e.g., C. halemanui {Sherff) Croizat & Degener (Figure 14).Similarly, seeds with a more or less smooth surface of closely tessellated cells proved to be either mucilaginous, e.g., C. acuta (Figure 10), or non-mucilaginous, e.g., C. mesembrianthemifolia (Figure 9).Seeds with sharply ridged surfaces, e.g., C. glyptosperma (Figure 5), often produced mucilage more copiously from the ridge crests than from the intervening crevices (Figure 6); SEM of dry seeds of this species (Figure 8) reveals surface cells of the mucilaginous ridges to be larger and somewhat more protrusive than cells of the non-mucilaginous regions.
Although surface features alone proved uninformative, sections reveal a striking difference between mucilaginous and non-mucilaginous seeds.Mucilage-producing seeds such as C.  (MANDL, 1926;SINGH, 1959;SING H, 1969).In C. nurans, macrosclereids are 27-40 µm long and 8-15 µm wide.These macrosclereids are closely overlain with shorter ceUs (ca 12 µm tall) that give rise to the mucilaginous layer.From below, endosperm appears to extend to the base of the macrosclereid layer (Figure 11).While macrosclereids are correlated with mucilage production, there is no obvious causal connection between these two observations.Mucilage secretion is clearly a function of the surface layer.
In contrast, sectioned seeds of C. halemanui (Figure 14), C. mesembrianthemifolia (Figure 12), and C. remyi, all of which tested negatively for mucilage, also lack the prominent macrosclereid layer.Testa layers for these three non-mucilaginous species are not, however, identical.In C. mesembrianthemifolia a layer of brachysclereids ca 25 µm tall occupies a position comparable to that of the macrosclereids.Further, this species has a zone of thin-walled parenchymatous cells located between the brachysclereids and endosperm (Figure 12).Chamaesyce halemanui and C. remyi have an extremely thin testa, which appears as two poorly differentiated layers of relatively thin-walled cells directly enclosing the endosperm.
Developmental considerations.For the most part, all seeds within a collection and multiple collections of a species reacted similarly.Occasionally a single seed of an otherwise mucilaginous species failed to develop mucilage, further, some samples included both light and dark colored seeds with noteworthy differences in their capacity to produce mucilage.These examples of infrequent negative responses (no mucilage) in otherwise mucilaginous species can be attributed to differences in seed maturity.For example, the abundant light-colored mature seeds in a mass sample of C. maculata consistently develop mucilage; other, less abundant, darker seeds of the same collection frequently failed to produce mucilage.Dissection of young indebiscent capsules of C. maculata reveals dark seeds to be a sign ofimmaturity in this species.Although based on a much smaller sample size, similar results were observed in C. glyptosperma; see, for example, Figure 6, in which mucilage production varies considerably from seed to seed.Interestingly, opposite results were noted for C. hypericifolia, in which mature seeds are usually described as "brownish or reddish with a very thin whitish bloom" (CORRELL & JOHNSTON, 1970); darker seeds produced greater amounts of mucilage than did lighter seeds.It may be hypothesized that mucilage-producing ability develops late in testa ontogeny; thus weak mucilage production may indicate immaturity of the seed and copious mucilage (in mucilaginous species) is correlated with mature seed coloration.
Testa features observed in the present study are generally consistent with previous developmental literature, including studies of various species of Chamaesyce (e.g., KAIALE, 1954;M UKHERJEE, 1961;MANGALY et al., 1979), Euphorbia (e.g., BoR & KAPIL, 1975;Singh, 1959;SINGH, 1969), and other Euphorbiaceae (e.g., BoR & BOUMAN, 1974).Ontogenetic studies in Euphorbia and Chamaesyce show the ovules to be bitegmic, with portions of both integuments contributing to the mature testa.Each integument consists of three to four layers of cells initially, but during maturation some layers are lost.The ontogenetic literature suggests that outer epidermis of the inner integument always persists, fo rming the layer of macrosclereids that constitute the major protective layer of mature testa; we found these macrosclereids to be prominent in mucilaginous species (Figu res 11, 13).It is interesting to note that, in at least some species, characteristic surface features such as crests, ridges, and reticulations have been attributed to unequal radial (anticlinal) elongation of macrosclereids nature of the wide zone of thin-walled parenchymatous ceUs located below the brachysclereids of this species (Figure 12) cannot be determined at this time.Conceivably, these cells could be derived from subepidermal regions of the inner integument or from nucellus.
The embryological literature (papers cited above) further suggest that as many as three layers of thin-walled cells derived from the outer integument may persist as the superficial covering over the sclerified layer of mature testa.Of the several species we sectioned and viewed with SEM, distinct well-defined cells are detectable at the surface of most seeds, but not in subsurface layers; i.e., only the epidermis of the outer integument is retained as a distinct cell layer.It is these epidermal cells that produce mucilage in mucilaginous species.Chamaesyce acuta, however, is notably different, with two distinct cell layers, presumed subepidermal cells of the outer integument, located between the surface and the macrosclereid layers.We interpret seeds of subsection Gymnadeniae as extreme products of testa reduction, consisting of a weakly sclerified layer and a thin, non-mucilaginous epidermis.Of course, we acknowledge that inference of developmental events from examination of matu re structure is fraught with pitfalls.We thus urge careful studies oftesta development in species which depart from the norm established by the earlier embryological literature.
Taxonomic and phylogenetic considerations.Mucilaginous seeds are present in at least some species of all but one of Boissier's subsections; mucilaginous seeds were not found in any species classified in subsection "Sclerophyllae".Lack of mucilage is also widespread in subsection Gymnadeniae.Only two subsections, " Acutae" and Gymnadeniae, were found to contain both mucilage-producing and non-mucilaginous species, otherwise, Boissier's subsections are consistent in regard to this character.
Testa anatomy and mucilage production has the potential to offer insight into aspects of the phylogeny of Chamaesyce.From presently available evidence, mucilage secretion and well-defined macrosclereids appear to be plesiomorphic for Chamaesyce, since both of these features are also found in species of Euphorbia (MANDL, 1926; ROSSLER, 1943).Gymnadeniae and " Sclerophyllae" share certain obvious synapomorphies with most other subsections of Chamaesyce, notably C4 photosynthesis (WEBSTER et al., 1975;ROBICHAUX & PEARCY, 1984) and terminal differentiation of the seedljng epicotyl (DEGENER & CROIZAT 1938;KOUTN IK, 1987;HAYDEN, 1988); absence of mucilage and weak sclerification of testa in Gymnadeniae and " Sclerophy llae" is thus most parsimoniously viewed as synapomorphic loss in these subsections.The general absence of seed mucilage in Gymnadeniae and "Sclerophyl/ae" underscores their apparently close relationship (KOUTNlK, 1987).Absence of mucilage in C. angusta (Engelm.)Small, however, seems to be a different case.This C 3 species (WEBSTER et al., 1975) is reputedly near the transition between Euphorbia and Chamaesyce (MAYFIELD, 1991) and shows no obvious relationship with species of Gymnadeniae or "Sclerophyllae"; its lack of mucilage, if not an artifact of seed immaturity, is, most likely, the result of parallel loss (homoplasy).
Functional hypotheses.Two hypotheses, seed hyd ration and seed dispersal, have been proposed to explain the adaptive value of mucilaginous testa in Chamaesyce and Euphorbia.JORDAN et al. (1985) suggested that seed mucilage in C. maculata plays a role in water absorption and thus, presumably, in germination.Although some seeds of this species may germinate shortly after dispersal, greatest rates of germination are achieved following cold stratification (KREUGER & SHANER, 1982).This experimental result conforms with observations in nature (W.J. Hayden, unpublished); in eastern North America, seedlings from naturally dispersed seeds are generally encountered only in early summer.Yet, in another study (HAYDEN, 1988), high rates of germination were obtained after just one week of stratification at 5 C. Thus, even though sufficient water absortion for germination can occur within a period as short as a week, many seeds in nature remain dormant until they have overwintered.Hydration appears to be a necessary, but may not always be a sufficient, factor for germination in this species.Details of the role of mucilage in hydration of seeds of Chamaesyce have yet to be investigated.
The dispersal function of mucilaginous euphorb seeds has been discussed by CARLQUIST (1966CARLQUIST ( , 1980)).The dispersal hypothesis is based largely on comparative data of Hawaiian and other Pacific species, and is, in part, a negative argument involving the contexts in which loss of dispersibility occurred.Noting that sticky mucilaginous seeds are widespread in Chamaesyce, Carlquist postulated that ancestors of Hawaiian Chamaesyce (subsection Gymnadeniae) reached the islands as dispersed seeds attached to feathers of birds.Chamaesyce celastroides (Boiss.)Croizat, a species oflow altitude inland habitats and found on all major islands of the chain, bas mucilaginous seeds, hypothetically the retention of a primitive character.Interestingly, KOUTNIK (1987)   , 1966).Both the presence of Chamaesyce species in the remote Hawaiian Islands and loss of mucilaginous seeds in many of them is, therefore, testimony to the effectiveness of mucilage in dispersal.
Loss of seed muciJage in the closely related subsection "Sclerophyllae'' is proposed to be a different story.Many species of "Sclerophyllae'' occur as plants of coastal strands, and are presumably dispersed as drift seeds (CARLQUIST, 1966) in which mucilage would be superfluous.Seeds of C. atoto (Forst.)Croizat are reputed to float in sea water by virtue of air spaces in its testa (GUPPY, 1906).Seeds of C. mesembrianthemifolia examined in this study also float in seawater, presumably by virtue of the broad zone of thin-walled parenchymatous cells (Figure 12) located below the band ofbrachysclereids, as described above.Loss of mucilage in "Sclerophyllae'' may well be a consequence of an adaptive shift to dispersal by floatation.
Studies on seed viability following floatation in sea water may well provide insight into the evolution of subsection "Sclerophyllae'', especially in regard to its putatively close relationship with Gymnadeniae.
also interpreted this species as most primitive among the species of Gymnadeniae.Other species of Gymnadeniae, such as C. clusiifolia (Hook.& Arn.) Arthur, C. degeneri, C. halemanui, C. remyi, and C. rockii (C.N. Forbes) Croizat & Degener, have lost the capacity to form mucilaginous exudations.Despite the necessity of efficient dispersal for colonization, dispersibility is mal-adaptive once a species becomes established in an island setting (CARLQUIST