Wood Anatomy of Macaronesian and other Brassicaceae

The tendency for herbaceous families of dicotyledons to be represented on insular areas by species and genera more markedly woody than their continental relatives is exemplified by the family Brassicaceae. This tendency, although early noted by such authors as Hemslev (1885), has not been explored to any appreciable extent genetically, experimentally, or even ecologically for any of the families in which this phenomenon occurs. The present study is one of a series of papers in which I am attempting to discern what modes of anatomical structure characterize woody insular representatives of herbaceous groups, and with the aid of ecological correlations draw conclusions concerning the occurrence and nature of what may be termed "insular woodiness."


INTRODUCTION
The tendency for herbaceous families of dicortyledons to be represented on insular areas by species and genera more markedly woody than their continental relatives is exemplified by the family Bmssioaceae. This tendency, although early . noted by such , authors as Hemsley ( 1885) , has not been explored to any appreciable extent genetically, experimental, ly, or even ecologically for any of the families in which this phenomenon occurs. The present study is one of a series of papers in which I am attempting to discern what modes of anatomical structure characterize woody insular representatives of herbaceous groups, and with the aid of ecologica,l correlations draw conclusions concerning the occurrence and nature of what may be termed "insular woodiness." Crucifers offer an excellent subject for such a study, for the fami, ly as a whole seems clearly an herbaceous one, and primitive tribes do not contain the woodiest members of the family. One of the tribes somtimes regarded as primitive is Stanleyeae. The genus Stanleya is mosrtly herbaceous, and the shrubby species studied here, S. pinnata, can be said to be exceptional in this regard ( Rollins, 1939). The genus Romanschulzia seems to contain a greater constellation of primitive features ( Rollins, 1956), although it, too, can be called essermally herbaceous.
Within the related family Oapparaceae, woodiness may be primitive, although some genera appear to have increased woodiness during their evolution. The likelihood appears very great that none of the woody orucifers studied here represent vestiges of capparaceous woodiness in any oase, for they belong to tribes with numerous speciailized features, well removed from the capparid-like cruoifers, and they grow in geographical areas in which woodiness is · common in families previalently herbaceous elsewhere. The woody habit of Stanleya pinnata may well be a response to longer growing seasons in the southwestern United States. In this same region, there are many woody species of Asteraceae, and this parallel suggests thrut climate has been influential in the production of a shrubby habit.
innovations, so that a broom-like shrub with a short, woody caudex-like stem is formed. An exception to this is Parolinia, in which an upright elongate stem like that of a "typical" shrub is branched at various levels. Parolinia also has wood of a texture much more dense than that of other crucifers.
The Brassicaceae of the present s,tudy are , interesting in that unlike other phylads woody on islands, these insular crucifers are not mesic in their adaptations. Woody Lobe1ioideae (Carlquist, 1969a) are nearly all mesic in their ecological preferences, and show this <in t erms of wood anatomy. Likewise, an appreciable number of Asteraceae ( Carlquist, 1966) have entered such niches. The Hawaiian and Macamnesian euphorbias ( Carlquist, 1970b) have occupied a broad span of situations with respect to rainfall, although only a few could be caHed ohamcteristically rain-forest species. The woody species of Echium ( Carlquist, 1970a) are characteristic of drier situations in the Cal!1iary Islands and Madeira, and only E. pininana and E. candicans are laurel-forest species. Although this pair of species represents the most mesic adaptations of the genus, and they show this with respect to wood anatomy, the Macaronesian laurel forest cannot be ca:lled a rain forest. Brassicaceae on islands are distinctive in that the woody species studied here occur exclusively or nearly exclusively in xeric localities. Only one collection was made from a , laurel-forest location: Cheiranthus mutabilis, Cadquist 2645. This plant, which grows along the waterfaHs in the Ribeiro Frio district of Madeira, probably represents a recent invader of open rocky situations within the laurel forest, and the primary adaptation of the species as a whole, as shown by other collections, is xeric.
The xeromorphic preferences of Macaronesian Bras, sicaceae are clearly indicated by their presence in lowland dry situations ( the site of the vast majority studied here), but also in dry alpine situations as well: Cheiranthus scoparius, C. tenuifolius, and Descurainia bourgeauana could be called Macaronesiian alpines. Sea bluffs, the characteristic habitat of Lepidium serra, Sinapidendron angustifolium, and S. frutescens, are physiologically dry localities, sometimes even miidly hailophyt!ic in their vegetation. For further descriptions of ecology and habit photographs and drawings, the reader can consult Burchard ( 1929), Carlquist ( 1970c), Hililebrand ( 1888), Lems ( 1960) , Lowe ( 1868), andSchenck ( 1907). Space does not, unfortunaitely, permit inclusion of habit photographs here.
The crucifers studied here seem good material for answering the question: what are · the characteristic modes of wood structure in a basicalJy herbaceous group, woody derivatives of which have, with few exceptions, retained a xeric or near-xeric ( oa. 25 inches of rain per year or less) adaptation? One must concede that information from woody species of Draba and Romanschulzia is necessary before a full assessment of wood anatomy of Brassicaceae can be made. One might also b enefit from knowledge of wood anatomy of cabbages on subtropical islands, like the Canary Islands or the West Indies. On such isiands, cabbages often become unbranched or little-branched rosette herbs with stems up to peThaps three meters tall. This phenomenon is a phenotypic response to the moderate maritime cli-ALISO [VoL. 7,No. 3 mates of these islands, and although it does not represent, as far as known, a genetic change to the rosette-tree habit, it forms an interesting parallel to those genera of other families ( e.g., Dendroseris of Asteraceae; Cyanea of Campanulaceae-Lobelioideae) that have deve~oped the rosette-tree habit during evolution on islands.
MATERIALS, METHODS, AND ACKNOWLEDGMENTS Wood specimens were collected in the field, with the exception of Stanleya pinnata, samples of which were obtained from plants cultivated in the Rancho Santa Ana Botanic Garden, Claremont, California. Wood samples were dried without artificial heat after they were collected. Basal portions of stems have been utilized unless other portions of the plant are indricated in the table. Authors of ta,conomic names for the taxa studied here are given in the table and omitted elsewhere. AU wood samples s· tudied are documented by voucher specimens in the Rancho Santa Ana Botanic Garden (RSA). Replicates of most of these specimens were prepared and distributed to other herbaria, pa11ticularly the Gray Herbarium.
Wood sections and macerations were prepared according to the usual techniques. The chamcteristics studied are typical of those of studies on secondary xylem, but have been influenced by features that seemed to show interesting variations in ·my previous studies on wood ana, tomy. For each quanuitative feature , analyzed, 50 measurementis were made and a mean obtained. Significant differences in wood ,anatomy within a plant oan occur. The data of the • table show ma· rked differences between base and upper branch of Descurainia briquetii, but differences among root, stem, and upper branch are relatively minor in Cheiranthus arbuscula and Sinapidendron frutescens.
I have followed the taxonomic , treatment of Lems ( 1960) for Canarian Brassicaceae and that of Lowe ( 1868 ) for Madeiran species. I suspect, however, that Lowe was overimpressed with minor differences, and that some of his varieties or even species might correspond with what we would consider eootypes , today. Taxonomy of insular species is often difficult, apparently because some genera speciate or diversify without the formation of genetic barriers among popula,tions that would typically accompany speciation in continental areas. Consequently, numerous intergrades can occur, desp:iite distinctiveness of extremes, and a confusing taxonomic situatiion often results.
For assistance in ooHecting activities, I wish to thank Mr. Gunther Kunkel; the late Dr. Komelius Lems provided much informa. tion helpful to my field work. Microtechnical assistance was provided by Mr. Gemld Benny, Mr. Arthur Gibson, and Mr. Timothy Magee. Professor Reed C. Rollins read a preliminary manuscript of this paper and provided useful suggestions and improvements.

ANATOMICAL DESCRIPTIONS
In , the table, qualrit, ative and quantitative da~a for woods of Brassicaoeae have been summarized. Most of the items are self-explanatm:y. However, abbreviations have been used in some instances. In the column headed "RAY CELL TYPES," the relative abundance of upright (erect), square, and procumbent cells as seen in radial sections are indicated: an uppercase letter indicates relative abundance, a lowercase letter relative paudty. Thus USp d enotes that upright and square cells are relatively abundant, procumbent ceHs rather less frequent. Under "sTORIED woon STRUCTURE," only lribriform fib ers are the concern ( rn· possibly also shorter, nucleated fibers that may resemble axial parenchyma ce1ls). Rays are not storied in the Brassicaceae studied. Thus "+" indicates storying in a least parts of the a~iaJ xylem of a section, "O" no ,appreciable storying. In the column headed "vESSEL WALL FEATURES," presence of grooves interconnecting adjacent pit aperhues is indicated by "g"; "e" denotes that most later, al-wall pits -are ellriptrical or elongate in face view, and "c" that most pits are circular in outline. In the column headed "GROWTH RING FEATURES," the elements in early wood differing from those in late wood are indicated: "mv" = more numerous vessels in early wood; "wv" = wider vesseils; "tf" = thinnerwalled libriform fib ers ( or poss, ibly nucleated fibers ) .

VESSEL ELEMENTS
Vessel-Element Length.-MetcaHe and Cha, lk ( 1950) indicated that orucifers have notably short vessel elements. As shown in rthe tabJe, this seems a very accurate ,appraisal. The longest average vessel-element length obtained is only 200.6 µ, ( Cheiranthus mutabilis) , the shortest 62.l µ, ( Descurainia briquetii). If one takes an avemge of the average lengths given in this column, one obtains the figure 117.4 µ, for the woody Brassicaceae studied h ere as a whole. The comparable figme for 328 species of Asteraceae ( Carlquist, 1966) is 235 µ,, which is appreciably longer, ailthough s, till w ell b elow the average figure for dicotyledons as a whole ( cf. Metcalfe and Chalk, 1950). Shortness of vessel elements is olearly indicated in the tangential seotrions presented here as Figs. 2,4,6,8,11,_ 13,17,20,22, and 25. One may ask why Brassicaceae have such notably short vessel elements. An explanation that might be offered concerns the fact that Brass, icaceae is an essentially herbaceous family, and that if the herbaceous ha bit in dicotyledons is an indicator of specia:li,m;tion compared with woody dicotyledons, shorter vessel elements would, in general, be expected in herbaceous groups. This, while p erhaps true, may be only a partial exp~anation. To be sure, the tribes of Asteraceae richest in h erbaceous species do appear to have shorter average vessel-element lengths: for example, Ambros,ieae ( 187 µ,), Anthemideae ( 145 µ,), Arctotideae ( 129 µ,), Cynareae ( 185 µ,) , and H elenieae ( 167 µ,) ( Carlquist, 1966) . However, the correlation b etween vessel-element length and ecoilogy cited in that paper and others ( Carlquist 1969b( Carlquist , 1970aWebber, 1936) seems clear. As mentioned above, the Bra:ssicaceae studied come from reilatively dry regions. The data from Cheiranthus aire interesting in this regard. Cheiranthus mutabilis, Carlquist 2645, comes from the moist laurel-forest areas of Madeira, whereas Cheiranthus scoparius, Carlquist 2496, represents a population from alpine regions of extreme drought on Tenerife. With regard to ecologica:l correlations as op-   :i: "'" :: ,. I-<;,. u ~ l:l ~ i>l i>l ~ el ::g;,.
:  the shortest vessel elements for Cichorieae. Because vessel-element length is controlled by length of fusiform cambial irnitials and thereby may be assumed to be a deep-seated charncteristic genetically, not readily subject to phenotypic modinca;tion, the dual explanation of shortness due to speciali:mtion of an herbaceous phylad and also due to adaptation to ecologicailly dry conditions may account for the £.gures in Brnssicaceae. However, one may still ask why herbaceous groups have shorter vessel elements. I am inclined to relate this correlation, also, to ecology, for many annuals and woody herbs accomplish growth, especially the later or last-formed secondary xylem, during warm, dry parts of the year when water near the surface of the soH~where their roots are located, as opposed to those of trees and sluubs-reaches minimal levels most rapidly. In foot, water-stress conditions appear decisive, together with other factors , in termination of the life span of many annual or biennial species. Vessel Diameter.-Although undoubtedly under genetic control, width of vessels seems subject to a greater degree of phenotypic modinca:tion than does vessel-element length. Vessels vary considerably in dliiameter within a single growth ring, for example. Thus, vessel diameter may be an even more sensiitive indicator of xeromorphy than vessel-element length ( Cadquist, 1969b). Vessel diameter in Brassioaceae does parnllel vessel-element length m· ther closely, suggesting that ecofogy is impo,rtiant to both diameter and , length. Notable are the figures for average vessel diameter in the alpine collection from T enerife of Cheiranthus scoparius ( 27 µ,), relatively wide diameter in the more nearly mes,ic populations of C. mutabilis ( 54 µ,, 50 µ, ). The average of averages for vessel diameters in Brnssioaceae in the present study is 40.3 µ,, which is rather narrow for dicotyledons as a whole ( cf. MetoaHe and Chalk, 1950), and appreciably nan,ower than th, at for Asteraceae as a whole ( 51 µ,: Carlquis,t, 1966). This suggests that the woody Brassicaceae studied have adapted to fewer mesic situa-tions than the woody Asteraceae summarized by the wl'iter ( 1966). In other families in which I have studied wood anatomy, wider vessels seem correlated with mesomorphy, narrower ones with xeromorphy, with the exception of virning or scandent species, which throughout dicotyledons have wider vessels than do their non-vining relatives. V essel Grouping.-As can b e seen from , the table, average number of vessels per group does not va1y over a wide gamut in crucifers. However, the species of more xeric localities ( Cheiranthus scoparius, Fig. 19; Parolinia ornata, Fig. 12; and Stanleya pinnata, Fig. 21) do have an appreciably higher £.gure. The most mesomorphic species studied here, Cheiranthus mutabilis ( Fig. 16) , has the lowest £.gure for vessels per group. This correlation is clearly shown in Astemceae, where woody species divided into mesic, dry, .and desert categories showed marked differences in this feature ( Carlquist, 1966) . Tm.s correiation is not as marked in some other families, for woody Goodeniaceae ( Carlquist, 1969b) and Echium of the Boragi-  ( Carlquist, 1970a) seem to show a much lower degree of vessel grouping, although the correkttion could still be said to hold within these families. Vessel grouping in Brassioaceae tends to take the form of short mdia· l chains in most species ( Fig. 1, 3, 7, 10, 12 ), but those with a higher degree of vessel grouping ( Fig. 19, 21) are oharaoterized by inegular pore clusters.
Pitting.-As shown by the colmm1 in the table labeled "VESSEL WALL FEATURES," elliptical or markedly elong,ate pits characterize the vessels of most orucifer woods. Suoh pirtting is shown for Sinapidendron angustifolium ( Fig. 9). Pits circular in outline charaoter<ize some species, as shown for Cheiranthus mutabilis in Fig. 18. In Stanleya pinnata ( Fig. 23), pits are closely crowded , and notably polygonal rather than circular in outline. Both circular and elliptical pits may be found in a single species. Predominance of circular pits seems related to mechanical strength. The most markedly sluubby species, such as Parolinia ornata and Stanleya pinnata, are characterized by circular pits. The suggestion that elongate pits tend to occur in predominantly herbaceous families in those species with growth forms not seemingly related to pronounced mechanical strnngth has been made earlier ( Carlquist, 1969a). Because most woody crucifers are • ail: most small sluubs, often branched from near the base, abundance of elliptical pits on this basis might be expected. Broader contact , areas among tr, acheary elements as provided by eHiptical pits would seem conducive to maximal water conductivity but provide a weaker wall, viewed in meoharnical t erms. Superficial examination of vessel walls in some cruoifor woods gives an appearance of elliptical pitting in some oases where the pits are, in fact, circular, but where pit apertures are interconnected in lateral series by grooves. This is shown for Cheiranthus mutabilis in Fig. 18. As the table indicates, such grooves are almost universally present in woody cruoifers, but they are not conspicuous, nor do marked forms of sculpturing sometimes called "tertiiary helical th, ickenings" occur in the crucifers under study. Presence of gmoves and other types of helical sculpture seems a clear indacator of xeromorphy in vessels of Asteraceae ( Carlquist, 1966 ) and other families (Webber, 1936).

LIBRIFORM FIBERS
No tracheids or fiber-tracheids are present in Bmssicaceae; pits of all imperforate elements are simple. This is expected for a rather speciaJized herbaceous family, and is true in such f.amHies as Boraginaceae, Asternceae, Campanulaceae, Lami,aceae, So1anaceae, etc. The length of libriform fibers in species of Brassicaceae ( see table ) forms a pamHel to that of vessel elements, as would be expected. Wall thickness and diameter of libriform fibers vary widely among the crucifers studied, however. Narrow wall tru.ckness is not always conelated with narrowness of libriform fibers. For example, the narrow fibers of Parolinia ornata ( Fig. 12, 13 ) have walls so thick of elliptical and circular pits.-Scale for Fig. 5-8 shown above Fig. 1. Scale for Fig. 9 shown above Fig. 9: three divisions represent 10 µ, each.

ALISO
[VoL. 7, No. 3 that virtually no lumina are present in these cellrs. This feature makes the wood of Parolinia ornata extremely hard and dense, the hardest wood among the crucifers studied here. Although bands of vessel elements in growth rings of wood of Stanleya pinnata are accompanied by few fibers, there are narrow, thick-walled fibers mixed with the wider vessels in alternating bands, accompanied by very thick-walled fibers; -these make the wood of Stanleya pinnata notably strong. An opposrite condition is shown for the basal region of Descurainia briquetii ( Fig. 24, 25). In such caudexliike bases, mechanical strength would be expected to have less significance than in upright, elongate sterns of shrubs such as Stanleya and Parolinia. Variations in wall thickness and diameter of fibers with respect to axial parenchyrna and growth rings are discussed under those respective headings below.

AXIAL PARENCHYMA
In alil the woods of Brass, ioaceae studied, scanty vasicent1ic parenchyrna is present. This takes the form of complete or incomplete sheaths, usually only one cell wide, around vessels and vessel groups. This condition can be seen most clearly in the photogmph of Crambe fruticosa ( Fig. 3). As viewed in longitudinal sections, axiail parenchyma is often present as strands of two cells, but perhaps more commonly is not subdivided. "Single-cell strands" might well be expected beoause the vessel elements they acoompany are so short.
Some of the crucifer woods seem clearly to show fiber dimorphism in relation to growth rings. Thin-waHed Hbriform fibers in ea-rly wood are to be expected in dicotyledonous woods, but where such fibers exhibit many or aH of the ohamcteristics of parnnchyma cells, marginal parenchyma may be said to be present. Such a condition can be sarid to exist in the two speoies of Crambe ( Fig. 1, 3), and M atthiola maderensis ( Fig. 10), but is most clearly shown in Sinapidendron ( e.g., Sinapidendron .angustifolium, Fig. 7; Fig. 8, lower left). Liquid-preserved wood s-amples would be required if one is to ascertain whether the short, parenchyma-like, thin-walled fibers are nucleated a,t maturity and should, therefore, be termed parenchyma ( or at least nucleated fibers). At any mte, the presence of a type of fiber dimorphism seems clearly present in the cruoifers cited, in which the phenomenon resembles the conditions descri.bed in Dubautia or H emizonia ( Carlquist, 1958), where the concept of fiber dimorphism as a source for formation of marginal parenohyma was originally described.

RAY PARENCHYMA
Uniseriate rays are rela:tively infrequent in cruoifer woods, and can be said to be virtually absent. A few uniseriate rays, wiJth a maximum height of about three cells, can be seen in the photograph of Descurainia briquetii position of fibers, also from a tangential seotion, showing a rhomboid crystal at left.-Scale for magnification of Fig. 10-13 shown above Fig. 1. Scale for magnification of Fig. 14-15 shown above Fig. 9.  Fig. 16-18. Cheiranthus mutabilis, Carlquist 2648. -Fig. 16. Transection. Note large vessels, a few of which (below) are occluded by deposits. -Fig. 17. Tangential section. Non-lignified thin-walled ray cells have collapsed. -Fig. 18. Portion of vessel wall; grooves interconnect pit ap ertures of adjacent circular pits. -Fig. 19-20. Cheiranthus scoparius, Carlquist 2496.-Fig. 19. Transection. The very small vessels and their aggregations into pore multiples are prob- ( Fig. 25). There are interesting variations in width of multiseriate rays. Notably wide rays occur in Cheiranthus arbuscula, C. scoparius ( Fig. 20), Lepidium serra ( Fig. 6 ), and Sinapidendron angustifolium ( Frig. 8). Exceptionally nanow rays characterize Parolinia ornata ( Fig. 13 ) and Descurainia bourgeauana. Notably short mys ( less than 0.3 mm in height ) also characteiize these two species. Rays taller than 1.0 mm, although relatively frequent in dicotyledons as a whole ( cf. Metcalfe and Chalk, 1950) , occur in only a few of the Brnssioaceae studied. Descurainia preauxiana is the only species studied in which average ray height exceeds this figure, although species of Cheiranthus ( Fig. 17, 20) have rays neai'ly as tall. Shortness of rays in Brassicaceae as a whole may be related to shortness of fusiform cambial initials. At any rate, such shortness does run counter to the often-expres· sed concept tihat herbs have wide, tall rays, littfo altered during secondary growth from tall, wide pith rays ("primary rays").
A surprisingly large proportion of Brassicaceae have a predominance of upright (erect) ray cells with respect to procumbent cells ( see table). Only in Parolinia ornata is this situation reversed. Preponderance of erect ray cells has been cited as a possible criterion of juvenilism ( Carlquist, 1962 ), and seems to characterize mys of dicotyledonous groups that have become woody secondarily, such as lobelioids ( Carlquist, 1969a ) .

STORIED STRUCTURE
As can be seen from the table, aH of the Brassicaceae studied oan be said to exhibit growth-ring phenomena. Most of these patterns represent relatively minor modifications and are probably prnduced by fluctuations in av,ailabili:ty of waiter dmiing a season ( e.g., Matthiola maderensis, Fig. 10) or over a series of seasons. Thinner-walled fibers are the only conspicuous indicators of growth rings in most species. The most strongly-mairked growth rings were observed in Stanleya pinnata (Fig. 21), in which bands of fewer, markedly wider vessels mixed with thick-walled libriform fibers alternate with hands of numerous, very narrow vessels associated with axial parenchyma cells. Because Stanleya pinnata is the only continental crucifer studied here, and it comes from a temperate region, its pronounced growthring phenomena are to be expected. Stanleya grows in areas with marked seasonal contrasts in both temperahire and waiter availability in comparison with the more modernte climates in which the insular oruoifers grow.

ALISO
[VoL. 7, No. 3 Stanleya pinnata shows marked accumulation of res, in-like materials in vessels ( Fig. 21 ) . Other Brassicaceae in which such deposirt:s in vessels were observed (although to a lesser extent than in Stanleya) included Cheiranthus arbuscula, C. scoparius, Descurainia briquetii, D. preauxiana, and Parolinia ornata. These species are all native to relatively dry, sunny areas, and these chemically unidentified deposits may be related to that factor. CRYSTALS Crystals are infrequent in the Brnssicaceae s,tudied. Prismatic and rhomboidal crystals birefringent in polarized light were observed in ray cells of Descurainia briquetii ( Fig. 26). Minute rhomboidal crystals that showed no birefringence with my polarizing equipment were seen in fibers of Parolinia ornata ( Fig. 15 ), but in no other ceHs of that species .

DISCUSSION AND SUMMARY
The wood anatomist, or for that matter the comparative plant anatomist, has been accustomed to presentation of his data in terms of differences and similarities among species, genera, etc. These data are sometimes expressed in terms of taxonomic criteria, or so accepted by taxonomisits reading such data. I have become increasingly distressed about this tendency, because although harmless in some cases, it is definitely misleading in others ( and apparently especially so when computerized) . At best, it represents a myopria in which the bases for the differences and simiJari.ties am not sought. In wood anatomy, particular features may be the result of different growth forms ( of.ten between two species within a genus ) , they may represent the resu1t of phylogenetic specialization with respect to dicotyledons as a who~e, they may represent different degrees of mesomorphy or xeromorphy, they may stem from different degrees of juveniliism or adulthood, or they may result from such faobors as different physiological prncesses, single-gene differences, or even be artifacts. To interpret such a wide variety of expressions as simple taxonomic criteria seems to me a dubious procedure.
A few ex, amples from the present sh1dy will suffice to illustrate this point. Although Matthiola maderensis ( Fig. 10, 11 ) and Parolinia ornata ( Fig.  12-15 ) belong to the same tribe, Matthioleae, their woods are as different quantitatively and qualitatively as those of almost any other species of crucifers selected at random from the present study. Does this mean that &s particular pair of species should not be placed in the same e:ibe? Decidedly not. in the family), M atthiola an annual or biem1ial; Parolinia grows in subdesert areas, Matthiola grows during the wet season in seasonally moist localities and is therefore a mesomorphic genus in essence.
This case can be put even more strongly by saying that of the Brassicaceae studied, any given species falls quantitatively and qualitatively within the range of xylem expressions found in Asteraceae. Are these two families closely related? Obviously not-and many other families could be shown to overlap completely with respect to secondary xylem features. Srimilarities between woods of Brassicaceae and Asteraceae probably derii.ve merely from the facts that both families have woods wiith numeiious specialized features, both fami:lies have at least a large proportion of ancestors derived from herbaceous ancestor, s, and both families have numerous species in relabively dry, pioneer habitats. Relevant dat, a from evolution, growth from, etc., are all-too-rarely invoked by the wood anatomist. Can we expect blind use of wood characteristics, with or witJ1out a computer, to oounteract these deficiencies? Clearly not. If a wood anatomist does not understand the significance of his own data, others unfamiliar with wood anatomy can hardly be expected to lessen confusion and may merely increase it.
Ecology obviously guides wood evolution, but often indirectly and with numerous alternatives, compromises, , and shifts in direction. Wood features may be related , to more than a single factor. To illustrate the complexity of this situation, I will attempt to cite which characters are, in my opinion, related to particular factors: Level of Specialization of the Family Brassicaceae.-Because herbaceous dicotyledonous families are, in general, more specialized than their most closely related truly woody counterparts, we may expect a relative abundance of specialized features in Brassicaceae. Woody species of this family are all or nearly all probably derived from more herbaceous ancestors. High level of specialization for the family as a whole is shown by the very short vessel elements with exclusively simple perforation plates and nearly transverse end walls; absence of tracheids and fiber-tracheids in favor of libriform fibers; scanty vasicenhic axial parenchyma; and absence of uniseriate rays. Presence of storying in two genera is a specialized feature (Bailey, 1923), but one that can occur sporadically in families containing many fewer specialized features than do those of Brassioaceae. Ecology.-The very short vessel elements of the crucifer woods of the present study seem related, in part, to the fact that these woody crucifers grow mostly in xeric loca1ities. \Vithin the famHy, more mesomorphic species have longer vessel elements ( Cheiranthus mutabilis), more xeric species shorter ones ( Cheiranthus scoparius ), but the range is not grnat, probably because the woody crucifers studied have not adapted to as wide a gamut of ecological nii.ches as have such genera as Euphorbia and Scaevola in the Hawaiian Isliands. Narrowness of vessels in crucifers is also exceptional compared to those of dicotyledons at large, and this seems most easily interprnted as an expression of xeromorphy. Presence of grooves interconnecting pit apertures is suggestive of another xeromorphic indicator: it clearly seems to be in such other famiHes as Asteraceae and Gooden-iaceae. Number of vessels per group increases in the morn markedly xeromorphic orucifers ( Cheiranthus scoparius, Parolinia ornata, Stanleya pinnata). Conspicuousness and complexity of growth rings is directly related to ecological ex;tremes, so it is not surprising that the only continental species included in this study ( Stanleya pinnata) has the most clearlydemarcated growth-ring phenomena. Deposits of an unidentified res, in-like material are most abundant in vessels of the more xeromorphic cruoifers.
In citing ecological correlations, however, I again call attention to the fact that woody speoies of Draba and Romanschulzia are not included in this study, and these might pl'ove informative in broadening conclusions. Growth Form.-Most cruoifers in this study are subshrubs, branched from near the base. The clearest exceptions, where renewal buds can be more than a meter , above the surface of the ground, ad'e Parolinia ornata and Stanleya pinnata. Mechanical strength by virtue of thick-walled fibers is pronounced in these two species. These two species also have lateral-wall vessel pitting in the form of civcular pits exclusively, whereas the remainder mostly have elliptical pits predominantly or exolusively, some with nearscalariform conditions. Presence of elliptical pits is regarded here as a feature, p erhaps preserved into older stems from metaxylem by a kind of juveniHsm, which is not disadvantageous because the growth forms of most woody crucifers do not seem to stress maximal mechanical strength. Parolinia ornata has exceptionaUy short, narrow rays that may possibly be related to increased mechanical strength of this species. Juvenilism.-In addition to the relative abundance of elliptical pits cited above, predominance of erect ray cells in most woody oruoifers might be explained as a form of juveniilism, or paedomorphosis. Wide, tall rays might be expected in a basically herbaceous group of dicotyledons. Such rays do not occur in Bmssicaceae: p erhaps short ray initials in the cambium bear a correlation with short fusiform cambial initials characte1istic of these woods. Taxonomic Criteria.-Under this heading one can cite a residue of features that are not obvious adaptations to particular factors ; to be sure, some of the characterisHcs cited above could be utilized taxonomically, given a proper perspective. For ex;ample, presence of storying in two genera of the same tribe, the Brassiceae ( Crambe, Sinapidendron) as opposed to absence in the other crucifer genera studied may be significant. However, presence or absence of crystals of particular sizes and shapes and in particular ceH types ( Parolinia ornata, Descurainia briquetii) is the sort of characteristic easily used , taxonomioaHy-provided comparable mate1ials am studied. However, there are no wood features one could cite properly as species, genus, or even family characteristics on the basis of the present study----or very likely ever. One cannot really compare xylem of a markedly woody shrub to that of an ephemeral annual any more than one could exb,act taxonomic value from comparison of a proembryo of one species with a mature embryo of another. The present study oan only be a survey of some Bmssicaceae woody enough to provide oonvenient objects for investigation of secondary xylem. True comparability and a monographic approach must, in my