WOOD ANATOMY OF MYOPORACEAE : ECOLOGICAL AND SYSTEMATIC CONSIDERATIONS

Quantitative and qualitative features are presented for 28 collections of three genera (Bonhia, Er emophila, Myoporum); data on Oflia are also included since it is sometimes referred to Myoporaceae. Wood of all Myoporaceae represents variation on a basic plan: woods diffuse porous or semi-ringporous; vessels with simple perforation plates; lateral wall pits of vessels alternate and circular, with grooves interconnecting some pit apertures; vessels grouped to various degrees into radial multiples; imperforate tracheary elements all fiber-tracheids with pit cavities 1—5 m in diameter (wider on contacts with ray cells), nonseptate; axial parenchyma vasicentric scanty plus, in some species, tan gential paratracheal bands (sometimes terminal); rays multiseriate plus uniseriate (unisenate almost exclusively in two species of Eremophila); ray cells procumbent exclusively, upright and square in uniseriate rays and in tips of multiseriate rays; rays storied in some species of Eremophila, axial parenchyma variously storied; crystals present in ray cells of some species of Eremophila and in fibers of E. polyclada; amorphous deposits and starch often present in parenchyma. Oflia differs from Myoporaceae by having: septate libriform fibers rather than fiber-tracheids; large pits on vessel walls; axial parenchyma nearly absent; and erect cells predominant in rays. Ofha may be placed in Scroph ulariaceae. Leucophyllum differs from Myoporaceae by having: helical thickenings in vessels; vasi centric tracheids transitional from vascular tracheids; scarce axial parenchyma; and erect cells pre dominant in rays. Leucophyllum may be excluded from Myoporaceae. Wood anatomy ofMyoporaceae shows relationship between Myoporaceae, Scrophulariaceae, and Gesneriaceae, and is also indicative of derivation from woody ancestors. Wood of Myoporaceae reflects xeromorphy, especially in Ere mophila, but the foliar apparatus may partially preempt the role of wood in promoting safety, which is connoted by growth rings, narrow and numerous vessels, and grouped vessels.


INTRODUCTION
Three genera have traditionally been assigned to Myoporaceae: Bontia, Eremophila, and Myoporum.Bontia is a monotypic shrub or small tree from the West Indies (and possibly tropical Mesoamerica: Willis 1980).Eremophila (including Pho/idia, Pho/idiopsis) occurs in Australia.Including recently described species (Dill 1975;Henderson 1978;Chinnock 1979Chinnock , 1980Chinnock , 1986)), about 65 species of Eremophila are now known.Although Eremophila is not exclusively a desert genus as its name might suggest, it is indeed characteristic of dry habitats.Myoporum consists of about 30 species that range from eastern Asia to Australia, New Guinea, New Zealand, various Pacific islands (including the Hawaiian chain), and Mauritius (Willis 1980), a distribution that suggests infrequent events of seawater dispersal of the corky drupes.Myoporum can be found in dry lowland situations, but inland populations can enter moderately wet forest, as in the Kipuka Pualu (Hawaii) population referred to M. sandwicense var.fauriei by Webster (1951).The ecological range of Myoporaceae is thus basically a narrow one, with some incursions into moderately moist conditions.The basically maritime nature of the family is evident if we imagine that the flora of dryland Australia (i.e., habitats of Eremophila) bears a relationship to seacoasts in its salinity and in its geological history: central and southern portions of the continent represent relatively recent embayments.One Myoporum is even a mangrove of sorts (Toyama and Itow 1975).The ecological range of the family is broad enough to deserve comparison with variations in wood anatomy, with due attention to the participation of other factors related to regulation of the water economy of the plant.
The generic composition of Myoporaceae has been questioned.Ojtia, traditionally included in Myoporaceae, has been excluded on the basis of recent evidence from pollen (Niezgoda and Tomb 1975) and other features (Hallier 1905; Dahlgren and Rao 1971).Dahlgren and Rao believe Ojtia should be transferred to Scrophulariaceae.Metcalfe and Chalk (1950) have reported, with uncertainty, the presence ofintraxylary phloem in Ojtia; this has needed reinvestigation, since intraxylary phloem (which is lacking in both Myoporaceae and Scrophulariaceae) would indicate other affinities for the genus.A Madagascan species first assigned to Ojtia (Capuron 1972) has been segregated as a genus, Ranopisoa (Leroy 1977).Evidence from pollen morphology (Lobreau-Callen 1978) shows that Ranopisoa is like Myoporaceae in having three binorate ("endocolpar") colpae (furrows), whereas Ojtia has four colpae and no appreciable transverse furrow formation (Niezgoda and Tomb 1975).The position of Ojtia is considered in the present paper because of the availability of a relatively large wood sample collected by the senior author.
However, if Ojtia (which consists of two species from South Africa if Ranopisoa is segregated) is transferred to Scrophulariaceae, other questions remain.The tribe Leucophylleae of Scrophulariaceae has pollen that closely matches that of Myoporaceae (Niezgoda and Tomb 1975), although other features of Leucophyl/um, such as fruit morphology, show divergence from Myoporaceae.One notes, in this regard, Takhtajan's (1980) statement that Myoporaceae is a family with transitions to Scrophulariaceae, implying that some genera show intermediacy.The role of wood anatomy in such intermediacy is worthy of exploration.Wood anatomy of Leucophyllum is under study by Dr. David C. Michener, but observations based on wood of two species of Leucophyllum are included in the terminal section of this paper in an effort to help define Myoporaceae.
Conceding that Myoporaceae is closely allied to Scrophulariaceae, wood of Myoporaceae is of interest in defining the order and suborder to which these families belong.A survey of wood anatomy of tubiflorous dicotyledon families has been undertaken by the senior author.Although monographs of wood anatomy of several more families are contemplated, a summarization, in which Myoporaceae will be considered, should follow soon.Leading phylogenists place Myoporaceae close to Scrophulariaceae in the order Scrophulariales (or Bignoniales), as evidenced by the systems of Cronquist (1981), Dahlgren (1980), Takhtajan (1980), andThome (1976).

MATERIALS AND METHODS
Wood samples were available in dried condition only.The largest number of collections was contributed by the U.S. Forest Products Laboratory from their  --0 Samuel J. Record collection (SJRw) holdings, and by the Forestry Commission of New South Wales (SFCw).Field work by the senior author in Australia and South Africa yielded a few samples.Although Myoporum laetum is common in cultivation in southern California, both a sample from cultivation (Cariquist 1552) and one from a natural habitat near Wellington, New Zealand (SJRw 25470) were studied.The present study incorporates 20 species in three genera (Table 1).Thus, the surveys of Eremophila and Myoporum are not exhaustive, but they do cover wide ecological and taxonomic ranges within those genera.Polynesian Myoporaceae are identified in accordance with Webster (1951).
Wood samples were boiled in water, stored in 50% ethyl alcohol, and then sectioned on a sliding microtome.A few of the hardest samples benefited from an ethylene diamine treatment similar to that proposed by Kukachka (1977).Sections were stained with safranin; sections of many taxa were counterstained with fast green.Macerations of wood were prepared with Jeffrey's Fluid and stained with safranin.Means (Table 1) are based upon 20 measurements per feature except for vessel wall thickness and wall thickness and diameter of imperforate tracheary elements, in which means are based on an average of several measurements.Measurement of vessel diameter uses the widest diameter of each vessel and excludes the wall.Sections of most species were prepared by the junior author, who is also responsible for the data of Table 1 and for some qualitative observations.The senior author observed some qualitative features, prepared the photomicrographs, and constructed the text.This study was aided by a grant from the National Science Foundation, BSR 8419-169.

ANATOMICAL RESULTS
Table 1 summarizes quantitative data for the taxa studied.These are discussed below, but observations on qualitative features are also incorporated in the paragraphs below.

Growth Rings
Myoporaceae are characteristically evergreen; thus, despite the seasonal dryness of some of their localities, less prominent growth rings are to be expected in Myoporaceae than in families in which plants are deciduous because of drought or cold.A diffuse porous condition is illustrated here for Bontia daphnoides (Fig. 1) and Myoporum laetum.Other diffuse porous taxa include M. acuminatum, M. montanum, M. sandwicense (all varieties), M. serratum, and M. stokesii.Fluctuation of vessel diameter to a very limited extent, but with some change in thickness of imperforate tracheary elements, was observed in Eremophila duttonii, Myoporum longifolia, and M. rapense (Fig. 19).Other Myoporaceae could be regarded as semi-ring-porous because of various degrees of fluctuation of vessel diameter.Semi-ring-porous conditions are illustrated here for Eremophila alternifolia (Fig. 5), E. bignoniflora (Fig. 7), E. latrobei (Fig. 20), Myoporum platycarpum (Fig. 17), and Ofiia africana (Fig. 21).Apparent demarcation of growth rings by conspicuous parenchyma bands was observed in Eremophila alternifolia (Fig. 5), E. bignoniflora (Fig. 7), E. freelingii, E. scoparia, and Myoporum platycarpum (Fig. 17).One should be cautioned, however, that parenchyma bands can occur at various points within growth rings as well as at the ends of growth rings.For example, in Myoporum laetum (Fig. 13) inconspicuous parenchyma bands are frequent within the growth rings.
Vessel density (number of vessels per mm 2 transection) is close to being inversely proportional to vessel diameter in Myoporaceae, although not without deviation.The figures in Table 1 show this, and require no comment, except to note that the range in figures is wide (454 to 23).
Vessel elements are not notably short in Myoporaceae, although they fall well below the median for dicotyledons at large (approximately 600 ~m: Metcalfe and Chalk 1950:xxiv).One must concede that the Metcalfe and Chalk sampling tends to favor arboreal woody species over shrubs and woody herbs.The range in mean vessel element length (165 to 421 pm) in Myoporaceae is not great, and many of the species cluster near the family mean, 256 ~m.
Simple perforation plates exclusively were found in all of the Myoporaceae studied.A few double perforation plates (two perforations per end wall, not indicating a vessel bifurcation) were observed occasionally in Ojiia africana (Fig. 23).Lateral wall pitting of vessels appears to consist of nearly circular to somewhat elliptical pits, arranged in alternate fashion.These pits were observed to be 2-3 ~m in diameter in Eremophila duttonii, E .polyclada, E .scoparia, and Myoporum rapense; 3.5-4.5 in Bontia daphnoides, Eremophila alternifolia, E. freelingii, E. latrobei, E. longifolia, E. oppositifolia, E. serratum, Myoporum acuminatum, M. laetum, M. platycarpum, and M. sandwicense var.wilderi; 4-5 ~m in Eremophila bignonijlora, Myoporum sandwicense var.fauriei, M. sandwicense var.sandwicense, and M. stokesii; only in Ojiia africana did pits range between 5 and 6.5 ~m in diameter (Fig. 24).Grooves interconnecting a few to numerous pit apertures in a helix were observed in Bontia daphnoides, Eremophila bignonijlora, E. latrobei, E. longifolia, E. serratum, Myoporum laetum, M. platycarpum, M. sandwicense var.sandwicense, and Ojiia africana (Fig. 24).Such grooves are figured for Myoporum laetum by Meylan and Butterfield (1978, Fig. 790 and especially Fig. 787).These grooves cannot be called "helical thickenings" because they represent indentations into, not thickenings of the vessel wall.Although stress has been laid on helical thickenings in the literature, helical grooves often occur without being accompanied by helical thickenings, as in some Asteraceae (Carlquist 1961(Carlquist , 1965)).Vessel wall thickness in Myoporaceae is shown in Table l.The range in thickness is not exceptional, but a loose correlation can be observed between vessel diameter and vessel wall thickness.
Tyloses were observed in vessel elements of a few Myoporaceae: Bontia daphnoides, Eremophila bignonijlora, E. freelingii (Fig. 4), and E. mitchelli.In all of these, the tyloses are thin walled, spherical, and do not fill the lumina of vessels.Brown (1922) reports tyloses for Myoporum sandwicense.

Imperforate Tracheary Elements
According to Bailey (1936) or the definitions of the lAW A Committee on Nomenclature (1933), the imperforate tracheary elements of Myoporaceae must be called fiber-tracheids because pits on them bear vestigial borders.Pits are also relatively sparse.Pit cavity diameters range between 1 and 2 IJ.m in Eremophila bignonijlora, E. latrobei, E. oppositifolia, and E. scoparia; 2 and 3 IJ.m in E. alternifolia, E. duttonii, E. mitchelli, E. polyclada, E .serrulata, Myoporum acuminatum, M .montanum, M .platycarpum, and E. sandwicense var.sandwicense; 3 and 4 IJ.m in Bontia daphnoides, Eremophila longifolia, Myoporum laetum, M. sandwicense var.fauriei, and M .serratum; 4 and 5 IJ.m in Eremophila freelingii (Fig. 3) and Myoporum stokesii.However, note should be taken that these represent fiber-tracheid to fiber-tracheid pits within any given sample; fiber-tracheid to ray cell pits have pit cavities at least 1 IJ.m and often 2 IJ.m wider than those of fiber-tracheid to fiber-tracheid pits.This circumstance shows one should exercise caution in designating pit size in imperforate tracheary elements on the basis of a few observations.Length of apertures on the pits of the fiber-tracheids can be about the same as the pit cavity diameter, as in Myoporum laetum (Fig. 15; see also Meylan and Butterfield, 1978, Fig. 791).However, in most of the taxa studied, long slitlike pit apertures are common (Fig. 3).These apparent pit apertures probably are, to various degrees, the result of splitting due to dehydration (during microtechnical processing) of somewhat gelatinous walls.Notable gelatinous walls were observed in fiber-tracheids of Eremophila oppositifolia and Myoporum sandwicense var.sandwicense.Metcalfe and Chalk (1950) report imperforate tracheary elements in the family to have simple pits; this report is contradicted by the present study as well as by Meylan and Butterfield (1978).However, the pits of imperforate tracheary elements of Ojiia africana are apparently simple (Fig. 25).Counterstaining produces a fringe of deeply stained area adjacent to the pit apertures, but no border can be detected by means of light microscopy.
Fiber-tracheids in Myoporaceae are about 2.5 times as long as vessel elements (figures based on family means, Table 1), although there are some minor deviations from that ratio in particular species.
Diameter of imperforate tracheary elements at the widest point (Table 1) parallels vessel element diameter rather closely, as can be seen by examining transections (Fig. 1 , 5, 7, 13, 17, 19, 20, 21).Wall thickness of imperforate tracheary elements (Table 1) varies independently of cell diameter in Myoporaceae, although with narrow fiber-tracheid diameter, thicker appearance of walls is achieved with less wall thickness.Keeping that in mind, one can note relatively thin-walled fibertracheids in Bontia daphnoides (Fig. 2), Myoporum laetum (Fig. 13), and all varieties of M. sandwicense.The libriform fibers of Ojiia africana (Fig. 21) are notably thin walled.Relatively thick-walled fiber-tracheids are characteristic of all species of Eremophila (Fig. 5, 7, 20); this accounts for the hardness of those woods.Wall thickness of fiber-tracheids can vary according to location within a growth ring, as in Eremophila alternifolia (Fig. 5) or Myoporum rapense (Fig. 19).No septa were observed in fiber-tracheids of Myoporaceae.However, septa do occur in a portion of the libriform fibers of Ojtia africana (Fig. 25,left).No vasicentric tracheids or vascular tracheids were observed in either Myoporaceae or Ojtia.

Axial Parenchyma
In all Myoporaceae, vasicentric scanty parenchyma is present.The sheathing of vessels may range from partial, only a few cells adjacent to a vessel, as in Bontia daphnoides (Fig. 1) or Eremophila latrobei (Fig. 20), to a complete sheath, as in Myoporum platycarpum (Fig. 17).Strands of axial parenchyma are composed mostly of two cells, but sometimes three or four, as in Bontia daphnoides (Fig. 2).In addition to vasicentric scanty parenchyma, various kinds of tangentially aggregated cells may be seen.In its least conspicuous form , where parenchyma cells look similar to fiber-tracheids in transection as in Myoporum laetum (Fig. 13), the bands are several cells wide and can be termed confluent, as was done by Metcalfe and Chalk (1950) and Meylan and Butterfield (1978).Some of the tangential bands are narrow and often occur in the terminal part of growth rings, as in Eremophila alternifolia (Fig. 5) or E. bignoniflora (Fig. 7).These bands are not always terminal, however.In addition to the two species just noted, tangential bands of axial parenchyma were observed in Eremophila longifolia, E. polyclada, E .scoparia, and M. sandwicense var.sandwicense.Axial parenchyma is very scarce in Ojtia africana.

Rays
Myoporaceae characteristically have both multiseriate (including biseriate) and uniseriate rays, although multi seriate rays are nearly lacking in Eremophila alternifolia (Fig. 5) and E .oppositifolia.Multiseriate rays appear to be more common than uniseriate rays in Bontia daphnoides (Fig. 2), Eremophila bignoniflora (Fig. 8), E. longifolia, Myoporum acuminatum, M .montanum, M. sandwicense var.fauriei, M. sandwicense var.wilderi, M. serratum, and M. stokesii.In all species not named in this paragraph, multiseriate and uniseriate rays are about equally prevalent, as shown for Myoporum laetum (Fig. 14) and M .platycarpum (Fig. 18).In Bontia daphnoides, multiseriate ray portions interconnected by uniseriate portions can be seen (Fig. 2); this condition is not common in other Myoporaceae.Such rays were termed "zusammengesetzt" rays by Metcalfe and Chalk (1983), although that term is confusing, since "zusammengesetzt" can perhaps be most accurately translated as "compound" or "aggregate," yet the rays just described should not be confused with the compound rays of Fagaceae or Casuarinaceae; we suggest the term "interconnected multiseriate rays" for the condition seen in Bontia daphnoides.
If multi seriate rays are more common than uniseriate rays in a wood sample of Myoporaceae, the multi seriate rays also tend to be relatively wide (e.g., more triseriate than biseriate rays); thus, the data on ray width in Table 1 are indirectly related to abundance of multiseriate versus uniseriate rays.This is an ontogenetic matter, since rays tend to become wider with age, although not in such species as Eremophila alternifolia, where rays remain nearly exclusively uniseriate.Ray heights are also shown in Table 1.Ray heights appear closely related to vessel element length, as was shown statistically for Staphyleaceae (Carlquist and Hoekman 1985).The species of Myoporaceae with the tallest rays, Myoporum laetum (568 ~m) does have the longest vessel elements for the family also.This sample probably came from a relatively old stem.The second tallest rays reported here are those of Ojtia africana (535 ~m) (Fig. 22), which ranks below a number of Myoporaceae in vessel element length; this is worth noting with respect to the relationships of Ojtia.
Myoporaceae show a predominance of procumbent cells in multi seriate rays.By far the majority of species have multi seriate rays with erect cells only in uniseriate wings; uniseriate rays in these species are composed of upright and square cells, with relatively few procumbent cells.This type of ray histology is seen in Fig. 2, 4, 8, and 18. Exceptions to this pattern can be observed in Eremophila alternifolia (Fig. 5), E. oppositifolia, and Myoporum montanum; in all of these species, uniseriate rays are predominant and are composed mostly of procumbent cells, with square (rarely erect) cells also present.Bordered pits on tangential walls of ray cells in the central portions of multi seriate rays were observed in all Myoporaceae; they are shown here for Myoporum laetum (Fig. 16).

Storied Structure
In Eremophila bignonif/ora (Fig. 8), E. mitchelli, and Myoporum platycarpum (Fig. 18), storied structure is visible in rays, although not in all rays.Fiber-tracheids in these species do not show storying, presumably because fiber-tracheids elongate so much in Myoporaceae (approximately 2.5 times the length offusiform cambial initials, assuming that vessel elements are the same length as fusiform cambial initials).Storying of rays in other Myoporaceae is absent or only vaguely present (Fig. 2,6,14).Storying of some axial parenchyma may be seen in some of the parenchyma cells, although generally only a portion, in the Myoporaceae for which tangential parenchyma bands were observed.One must assume that storied fusiform cambial initials are not uncommon in Myoporaceae.

Crystals and Other Intracellular Substances .
Rhomboidal crystals can be observed in ray cells of several Myoporaceae.They were reported by Metcalfe (1935) for Eremophila mitchelli.The presence ofthese is confirmed, on the basis ofthe present study.Rhomboidal crystals can be found, one per cell, in occasional ray cells of E. alternifolia (Fig. 9).However, ray cells of E. duttonii (Fig. 10) commonly contain one crystal per cell as do ray cells of E. oppositifolia.These crystals have a characteristically hexagonal shape and show layering, with one or more spaces near the center of many crystals.
Crystals were observed in axial xylem cells of Eremophila polyclada (Fig. 11,12).The cells that contain these crystals are fibriform, but bear no evident pits; walls are thinner than those offiber-tracheids, but secondary wall materials appear to occlude the cells in the interstices between the crystals; the cells are not chambered or septate.The term "crystal-bearing fiber" is suggested for this unusual mode of occurrence.The crystal-bearing fibers may be considered idioblastic, but they can occur in tangential groupings like the axial parenchyma cells which also occur in E. polyclada.
Amorphous deposits, often appearing as resinlike deposits, may be seen in ray  cells and axial parenchyma of Eremophila longifolia, Myoporum acuminatum, and M. rapense.These deposits were reported by Metcalfe (1935) for Eremophila mitchelli and may occur in a number of Myoporaceae.
Starch grains were figured in axial parenchyma for Myoporum laetum by Meylan and Butterfield (1978).In the present study, starch grains were observed in axial parenchyma and ray cells of Eremophila alternifolia, Myoporum platycarpum, and M. stokesii.Starch grains may be assumed to occur much more widely in woods of Myoporaceae, and micro technical methods (e.g., use of ethylene diamine) may have destroyed them in some species studied here.

ECOLOGICAL CONCLUSIONS
The ratio termed "Mesomorphy" in Table 1 can be viewed as a possible indicator of ecology of Myoporaceae.If calculated for each genus, the figure would be: Bontia, 181 ; Eremophila, 53.1; Myoporum, 160; Ojiia, 222.Note should be taken of two species of Myoporum with unusually low figures for this ratio, M. platycarpum and M. rapense.Myoporum platycarpum is unusual within its genus in being an Australian desert species (most species can be called predominantly maritime); the low ratio for M. rapense is unexpected and may represent an unusual sample.If these two species had been excluded from Myoporum, the Mesomorphy figure for the genus would have been 387, more than twice that for Eremophila, and indicative of the maritime and interior dryland habitats, respectively, for the two genera.The highest figure within Myoporum occurs in M. sandwicense var.jauriei, from a pocket of rain forest (Kipuka Pualu) near Kilauea, Hawaii.The lowest figures within Eremophila occur in E. latrobei, from near Norseman, Western Australia, and E. oppositifolia, from near Mildura, Victoria.Both of those localities annually experience as great climatic extremes as any habitat where Eremophila is found.Although the Mesomorphy ratio figures clearly follow the ecology of the species of Myoporaceae, the figures are not as low as one might expect, especially for Eremophila, if one may judge by other groups for which this ratio has been calculated.One can speculate that wood of Eremophila and perhaps the other Myoporaceae do not exhibit as great safety in construction as one would expect in habitats. of the various species because some other factor, such as the evergreen and semisucculent leaves, may play an important role in water economy of the family.
The number of vessels per group, a feature not incorporated into the Mesomorphy ratio, may be considered as corroborative of the trends that ratio demonstrates.Thus, Bontia averages 1.66 vessels per group, but Eremophila is appreciably higher, 2.61.However, this range is not so great as in other groups with a comparable ecological span, as can be seen in the figures for Asteraceae (Carlquist 1966).A modifying effect of the evergreen foliage of Myoporaceae may again be hypothesized.
Growth ring presence is also indicative of relative xeromorphy of woods.In Myoporaceae, the range of diffuse porous to semi-ring-porous is not great.Of the seven species of Myoporaceae cited as semi-ring-porous, only two have vessels wider than the mean for the family, showing that semi-ring-porous species do have greater wood xeromorphy, but not by a wide margin.
Helical sculpture has been considered to represent a form ofxeromorphy (Webber 1936), but the form of it seen in Myoporaceae (grooves interconnecting pit apertures) is the least pronounced manifestation of this phenomenon, and it occurs in a minority (eight taxa) of the representatives studied.
Thus, all of the features currently used as indicators of wood xeromorphy are consistent with each other and consistent with the interpretation that woods of Myoporaceae show an ecological range, but do not exhibit as pronounced safety as do woods of other families that occupy similar habitats.Study of foliar apparatus of Myoporaceae with respect to transpirational characteristics would be very desirable.

SYSTEMATIC CONCLUSIONS
The woods of Myoporaceae reveal a number of instances in which species characters are demonstrated.The most distinctive of these is represented by the crystal-bearing fibers of Eremophila polyciada.This mode of crystal occurrence was observed in no other species of the family.Rhomboidal crystals, one per cell, were seen in some or many of the ray cells of a few species of Eremophila: E. alternifolia, E. duttonii, E. mitchelli, and E. oppositifolia.The occurrence of tangential bands of axial parenchyma appears limited to some taxa: Eremophila alternifolia, E. bignoniflora, E .polyciada, E. scoparia, E. serrulata, Myoporum laetum, M. montanum, M. platycarpum, and M. sandwicense var.sandwicense.
The taxonomic history of Ojiia has been traced by Dahlgren and Rao (1971), who decided that Ojiia should be included in Scrophulariaceae rather than Myoporaceae.Their decision seems defensible on the basis of the present study.Ojiia possesses the following features not observed in Myoporaceae: occasional double perforation plates; relatively large pits on lateral vessel walls; septate libriform fibers (rather than nonseptate fiber-tracheids); very scarce axial parenchyma; and predominance of erect cells in rays.The last-named feature , as well as the moderately tall rays of Ojiia, suggests some juvenilism in ray structure (despite the relatively mature nature ofthe sample studied, a woody stem 2.6 cm in diameter).These ray features reflect a degree of herbaceousness in Ojiia, whereas woods of Myoporaceae have rays characteristic of typically woody dicotyledons.Metcalfe and Chalk (1950) report intraxylary phloem in Ojiia with some degree ofuncertainty; our material does not show any intraxylary phloem, and thus there is as yet no evidence for suggesting inclusion of Ojiia in any of the families in which intraxylary phloem characteristically occurs (e.g., Solanaceae).The sum of wood features is compatible with the idea of Dahlgren and Rao (1971) that Ojiia should be included in Scrophulariaceae.
As noted in the Introduction, another group of uncertain affinity that may impinge upon Myoporaceae is the tribe Leucophylleae (containing Eremogeton, Faxonanthus, and Leucophyllum) ofScrophulariaceae.Niezgoda and Tomb (1975) demonstrated pollen features that link Leucophylleae with Myoporaceae.We have studied material of two species of Leucophyllum, L.frutescens (Berl.)I. Johnston (Carlquist 15913, RSA) and L. minus Gray (Michener 4394, RSA).These materials reveal presence in Leucophyllum of the following wood features that do not occur in Myoporaceae as far as presently known: vessels with prominent helical thickenings; vasicentric tracheids present; imperforate tracheary elements other than the vasicentric tracheids with apparently simple pits (and therefore libriform fibers); axial parenchyma scarce at least near the smaller vessels; and erect cells common in rays.The vasicentric tracheids are like those of Lamiaceae and Verbenaceae, present in latewood but so abundantly that they exceed the typical occurrence of vascular tracheids (last few layers of latewood where no vessels occur), so that they are intermixed with vessels.This mode of occurrence has been regarded as transitional between vascular and vasicentric tracheids, and for purposes of definition has been considered an instance of vasicentric tracheid occurrence because of the key feature ofadjacence to vessels (Carlquist 1985).In Leucophyllum there are numerous libriform vessel elements intermediate between ordinary vessel elements and vasicentric tracheids.The features by which Leucophyllum differs from Myoporaceae on the basis of material thus far studied seem appreciable, and deter one from adding Leucophyllum to Myoporaceae at present.These features, however, do not decisively place Leucophyllum in a particular family.No occurrence ofvasicentric tracheids has as yet been reported in Scrophulariaceae (Carlquist 1985), although the difference between the vascular tracheids of some xeric Scrophulariaceae and the vasicentric tracheids of Leucophyllum is not great.
Wood features seem to place Myoporaceae in Scrophulariales (Bignoniales of some authors).Some Scrophulariaceae agree with Myoporaceae in having fibertracheids with small bordered pits rather than libriform fibers (Metcalfe and Chalk 1950).Fiber-tracheids also occur in one genus of Gesneriaceae, Coronanthera (Carlquist and Hoekman 1986), and wood anatomy of Gesneriaceae is compatible with relationship between Gesneriaceae and Scrophulariaceae.In turn, that pair of families shows no feature that would rule out relationship with Myoporaceae.Wood anatomy ofMyoporaceae is more specialized than that of those two families in its storied wood structure and lack of any scalariform perforation plates (present in Gesneriaceae and Scrophulariaceae only in metaxylem and early secondary xylem of certain species).There seems no reason to contradict the concept of leading phylogenists (Thorne 1976;Dahlgren 1980;Takhtajan 1980;Cronquist 1981) that Myoporaceae is a member ofScrophulariales.The rays ofMyoporaceae seem to indicate a woody ancestry for the family; at least some segments of Gesneriaceae and Scrophulariaceae may have had an herbaceous ancestry (Carl-Quist and Hoekman 1986).
Legend for columns: VD = mean inside diameter of vessels at widest point, I'm; V/ MM = mean number of vessels per mm 2 of transection; VL = mean vessel element length, I'm; VWT = mean vessel wall thickness, I'm; TD = mean diameter of fiber-tracheids at widest point, I'm; TL = mean length of fiber-tracheids, I'm; TWT = mean wall thickness of fiber-tracheids, I'm; MRH = mean height of multiseriate rays, I'm; MRW = mean width of multiseriate rays at widest point, ...., I'm; URH = mean height ofuniseriate rays, I'm; MESO = Mesomorphy ratio (vessel diameter times vessel element length divided by number of vessels per mm2).

Table I .
Wood characteristics of Myoporaceae.