Aliso: a Journal of Systematic and Evolutionary Botany Wood Anatomy of Cneoraceae: Ecology, Relationships, and Generic Definition

Wood anatomy of the three species of Cneorurn is described qualitatively and quantitatively. The species differ in features related to ecology and form a clear series in this regard. The wood features of the family can all be matched by some Rutaceae and Simarubaceae, and the characteristics of Cneoraceae are listed in this connection. Nearly as many features are shared by Cneoraceae with Anacardiaceae and Sapindaceae; certain distinctive features may be found in somewhat more distant families, such as Oxalidaceae. Resemblances between Cneoraceae and Euphorbiaceae are attributed at least in part to the fact that Euphorbiaceae comprise a highly heterogeneous family with respect to wood anatomy. Wood anatomy of the three species ofCneorurn diverges markedl y. These differences when tabulated show that the Cuban species C. trimerurn is the most distinctive. Cneorurn pulverulen/urn (Canary Islands) and C. tricoccon (northwestern Mediterranean coasts), although distinct in wood anatomy, resemble each other more closely than they resemble C. trirnerurn. Despite the distinct ive tetramerous flowers and hexacolpate pollen of C. pulverulenturn. a single genus seems advisable; C. trirnerurn cannot be readily segregated on the basis of gross morphology.


INTRODUCflON
Cneoraceae have been treated as a monogeneric family containing three species : Cneorum pulverulentum Vent .(lowlands of five of the Canary Islands: Bramwell and Bramwell 1974); C. tricoccon L. (Mediterranean coasts of Spain, France, Balearic Islands, Sardinia, and northern Italy: Tutin, Heywood, Burges, Moore, Valentine, Walters, and Webb 1968); and C. trimerum (Urb.)Chodat (Sierra Maestra, Oriente, Cuba: Sauget and Liogier 1951).These habitats range from rather moist tropical to dry Mediterranean; thus, comparisons between wood anatomy and ecology are appropriate.Heimsch (1942) studied wood anatomy of Cneoraceae in his investigation of the woods of the orders Gruinales and Terebinthales.Heimsch's work is accurate, but it is supplemented here with quantitative and certain qualitative details.Only twig material was available to Heimsch for two of the species, and mature stems of these are used in the present study.Heimsch's conclusion that wood ofCneoraceae resembles that of Rutaceae more closely than that of Zygophyllaceae is valid.A more detailed series of comparisons is attempted here in an effort to place Cneoraceae as precisely as possible.
The three species of Cneorum prove to have notably divergent patterns ofwood anatomy.This has led me to reassess the generic constitution of the family.The distinctive tetramerous flowers of C. pulverulentum led Tieghem (1899) to segregate that species as Chamaelea.Erdtman (1953) discovered that the species differs from C. tricoccon by having hexacolpate rather than tricolpate pollen, and affirmed this genus, which he renamed Neochamaelea for nomenclatural reasons.The evidence from wood anatomy complicates this picture, and the generic treatment of the family is therefore reassessed here.

MATERIAlS AND METHODS
Two collections of C. pulverulentum were made in the wild in 1968 on the island of Tenerife: Carlquist 2482 (Barranca del Infierno) and Carlquist 2530 (Teno).These were supplemented by a collection kindly provided by Dr. David Bramwell from the Jardin Botanico Canario "Viera y Clavijo," Tafira Alta, Gran Canaria.This latter collection has the merit of representing a specimen from cultivation, and therefore showing if features alter as a result of cultivation.The results obtained below show no differences.This validates the use of a specimen of C. tricoccon from cultivation: the specimen Carlquist 20-VI-1971 was cultivated in Orpet Park, Santa Barbara, California.However, the area of the Park where these plants grew received little or no artificial watering, perhaps accounting for the eventual death of most of the plants, after which the mature wood sample was taken.Although one tends to assume that localities under cultivation have greater moisture availability than those in the wild, this may not in fact be true.The wood sample studied of C. trimerum (Oxford 10768) was taken from the wild, and is probably a portion of the same collection studied by Heimsch (1942), who did not, however, cite the specimens he used.I am grateful to Dr. Jeff Burley of the Oxford Forestry Institute for loan of a wood-section slide of C. trimerum.
Wood samples were available in dried condition.Portions were boiled, stored in aqueous 50% ethyl alcohol, and sectioned on a sliding microtome.Wood of C. pulverulentum contains exceptionally thick-walled fibers, and one collection (Bramwell s.n.) was soaked in ethylene diamine prior to sectioning; this resulted in improvement of sectioning.Sections were stained with safranin and lightly counterstained with fast green.Macerations were prepared with Jeffrey's fluid and stained with safranin.
Means are based on the average of 25 measurements, except for vessel wall thickness, libriform fiber wall thickness, and diameter oflibriform fiber at widest point; for these three features, a few typical conditions were selected for measurement.Vessel diameter includes the thickness ofthe wall.The figure for vessels per group was obtained on the basis that a solitary vessel = 1.0, a pair of vessels in contact = 2.0, etc.The figure for vessels per mm-counts all pores rather than counting a group of pores as one.

ANATOMICAL DESCRIPTIONS
CNEoRUM PULVERULENTUM (Bramwell s.n.) (Fig. 1-5).-Growthrings weakly present; latewood vessels slightly narrower (Fig. I).In some years a thin band of terminal axial parenchyma (often only a single cell layer thick) is present.Vessels often in radial multiples (Fig. I).Mean number of vessels per group, 1.89.Mean number of vessels per mm-, 234.Mean vessel diameter, 24 /-tm.Mean vessel element length, 288 /-tm.Mean vessel wall thickness, 3.0 /-tm.Perforation plates simple.Lateral wall pitting of alternate circular bordered pits averaging 3.2 /-tm in diameter on vessel-vessel contacts; vessel-axial parenchyma pits sparser, about 4.0 /-tm in diameter.Grooves interconnecting apertures of pits adjacent in a helix present in walls of at least some latewood vessels .Imperforate tracheary elements are mostly libriform fibers with minute simple pits; some vasicentric tracheids also present (Fig. 3).Mean libriform fiber diameter at widest point, 14 ~m.Mean libriform fiber length, 618 ~m.Mean libriform fiber wall thickness, 4.0 ~m.A few vasicentric parenchyma strands are adjacent to vessels or vessel groups (strand = 2 cells in length).Terminal parenchyma bands one (rarely more) cell in thickness present at ends of some growth rings .Parenchyma cells also present in diffuse arrangement, but all diffuse parenchyma consists ofcrystalliferous strands in which each cell is thin walled and contains a single large crystal (Fig. 4, 5), sometimes with additional small crystals.Uniseriate rays much more common than biseriate rays, no triseriate rays observed (Fig. 2, 3).Ray cells all procumbent except for cells at tips of rays, which vary from procumbent to square (Fig. 4, 5).Ray cell walls about 1.5 ~m in thickness (Fig. 4 CNEORUM TRICOCCON (Carlquist 20-VI-1971) (Fig. 6-1O).-Growthrings present, vessels markedly wider in earlywood (Fig. 6).Vessels often grouped into diagonal aggregations (Fig. 6).Mean number of vessels per group, 6.0 .Mean number of vessels per mrn-, 591.Mean vessel diameter, 19 ~m .Mean vessel element length, 288 ~m.Mean vessel wall thickness, 2.4 ~m.Perforation plates simple.Vesselvessel pits about 3 urn in diameter.Helical thickenings present in all vessels as well as in vasicentric tracheids (Fig. 8).Imperforate tracheary elements are libriform fibers with minute simple pits; numerous vasicentric tracheids are also present within the diagonal vessel groups, where the vasicentric tracheids resemble the narrowest vessel elements in diameter.Mean libriform fiber diameter, 14 ~m.Mean libriform fiber length, 660 ~m.Mean libriform fiber wall thickness, 2.8 urn.Axial parenchyma cells 1-2 near some vessels, thus vasicentric scanty; these parenchyma cells in strands oftwo.Terminal parenchyma present (usually a single layer of cells) at the ends of some of the growth rings.A very small number of crystalliferous axial parenchyma strands with a diffuse distribution also present (Fig. 10).Multiseriate (biseriate plus triseriate) rays about as frequent as uniseriate rays (Fig. 7).Triseriate rays less frequent than biseriate rays.Most ray ceUs procumbent (Fig. 9), a few erect or square cells scattered throughout rays.Rhomboidal crystals (one per cell) present in some ray cells (Fig. 9).Bordered pits present on some ray cell walls, pits otherwise simple.Ray cells 1-1.5 #Lm thick (Fig. 9).Mean multiseriate ray height , 232 #Lm.Mean multiseriate ray width, 2.36 cells.Mean height uniseriate rays, 82 #Lm.Wood nonstoried.Tannin deposits in some cells.

ECOLOGICAL CONCLUSIONS
The woods of the specimens from cultivation seem to be altered little or not at all compared to the quantitative features observed in the wild specimens, and the comparisons undertaken here assume the woods studied are typical for the respective species.The Mediterranean coasts where C. tricoccon is native have warm, dry summers and cool, moderately wet winters with occasional frosts.At least with respect to temperatures, the Canary Island localities where C. pulverulentum occurs are more moderate, because these insular lowlands are frost-free.However, the Canarian localities probably receive no more rainfall than the C. tricoccon localities (comparisons cannot be made because recording stations on the Canary Islands are not close to the C. pulverulentum localities).Cneorum trirnerum occurs in exposed sites in montane eastern Cuba where the rainfall, humidity, and therefore water availability are greater than for either of the two other species.Frost is probably absent in the C. trimerum localities.Thus, in terms of increasing water availability, the three species form a series: C. tricoccon, C. pulverulentum, C. trimerum.The Mesomorphy ratio (vessel diameter times vessel element length divided by number of vessels per mm-: Carlquist 1977) continues to prove a reliable indicator because the three features are independent adaptations to safety in wood.For the species in this study, the Mesomorphy values are as follows: C. pulverulentum (collections averaged), 29; C. tricoccon, 9; C. trimerum, 207.The sequence of values parallels the climatic severity of the localities as described above.The large gap between C. trimerum and the other two species probably shows not so much the effect of frost, but of prolonged summer drought.The number of vessels per group is most elevated in the species from the most xeric localities, C. tricoccon.
Another wood feature that parallels the above series is presence of vasicentric tracheids.Cneorum trimerum lacks vasicentric tracheids; they are relatively few in C. pulverulentum (mostly, but not exclusively, in latewood); they are abundant in C. tricoccon.Presence of vasicentric tracheids was reported earlier in C. tricoccon (Carlquist 1985).Vasicentric tracheid presence is characteristic of shrubs in Mediterranean-type climates; presence and relative abundance of vasicentric tracheids are indicators of degree ofxeromorphy (Carlquist 1985).The prominent diagonal groupings of vessels in C. tricoccon is related to abundance ofvasicentric tracheids as well as to xeromorphy: diagonal vessel aggregations occur only in species that have vasicentric tracheids (or, in a few cases, very narrow vessels) and are also in relatively xeric areas (Carlquist 1987).
RELATIONSHIPS OF CNEORACEAE Heimsch (1942), on the basis of wood studies, claimed that Cneoraceae are nearer to Rutaceae than to Zygophyllaceae.The current systems most widely cited (Thorne 1976;Dahlgren 1980;Takhtajan 1980;Cronquist 1981) place Cneoraceae in an order usually termed Rutales (sometimes Sapindales) near Rutaceae and Simarubaceae.At least some members ofboth ofthese families share the following features with at least some species ofCneoraceae: vessels with simple perforation plates and alternate lateral wall pitting; libriform fibers with simple pits present; vasicentric tracheids present; axial parenchyma aliform-confluent or terminal plus diffuse crystalliferous strand parenchyma; rays predominantly (or exclusively) uniseriate, with cells procumbent or mostly so; crystals in some ray cells; axial parenchyma and vessels storied (data from Metcalfe and Chalk 1950; Carlquist 1985).The diffusely distributed crystalliferous strand parenchyma is particularly characteristic of many genera of Rutaceae (Heimsch 1942;Metcalfe and Chalk 1950).Indeed, one could match almost exactly the wood of Cneoridium dumosum (Nutt.)Hook.f. (Carlquist 1985, p. 48) with that of Cneorum tricoccon, although that should be construed more as a parallelism due to evolution of a rutalean plant in similar ecological conditions than to close genetic relationship between the two species.
Cneoridium is the only genus of Rutaceae reported to have "resin cells" in leaves (Metcalfe and Chalk 1950), although in both Cneoraceae and Simarubaceae, secretory cells presumably corresponding to that term are common in leaves (Metcalfe and Chalk 1950).The secretory cavities so common in Rutaceae do not occur in Cneoraceae, and the use of the term "oil gland" (Doggett in Heywood 1978) in Cneoraceae is a misnomer.
Pollen morphology does not clearly ally particular families with Cneoraceae, for nomenclatural reasons).The Cuban species C. trimerum was originally named as Cubincola, but this genus was described in Euphorbiaceae, and segregation from Cneorum was therefore not intended.When one compares the wood anatomy of the three species in Table 1, a notable picture emerges.Several conclusions may be reached from this comparison.The differences among the species are more numerous than is typical within most genera.The geographical disjunction of the three species is unusual, however, and the differences in wood anatomy may be related to those disjunctions.A few of the differences in Table I are related to ecology, as noted in the Ecological Conclusions (e.g., abundance of vasicentric tracheids), but most features in Table 1 do not bear a direct relation to ecology (e.g., crystal distribution, presence of storying).
The species most distinctive on the basis of its tetramerous corolla and hexacolpate pollen is C. pulverulentum.Wood of C. pulverulentum is more similar to wood of C. tricoccon than to that of C. trimerum.Thus if wood anatomy were used as a criterion, C. trimerum would be the species most worthy of generic segregation.That treatment would likely not be accepted because features ofgross morphology do not differentiate C. trimerum markedly from C. tricoccon.
Considering evidence currently available from wood anatomy as well as from other sources, the most advisable treatment would seem to be recognition of a single genus.Construction ofthree subgenera is a possible treatment, but definition of these offers much the same problems as does recognition of three genera.