ECOLOGICAL FACTORS IN WOOD EVOLUTION: A FLORISTIC APPROACH

SHERWIN CARLQUIST

Reprinted from AMERICAN JOURNAL OF BOTANY Vol. 64, No. 7, August 1977 Made in the United States of America Amer. J. Bot. 64(7): 887-896. 1977.

ECOLOGICAL FACTORS IN WOOD EVOLUTION: A FLORISTIC APPROACH1

SHERWIN CARLQUIST Claremont Graduate School, Pomona College, and Rancho Santa Ana Botanic Garden. Claremont, California 91711

ABSTRACT Wood florulas from southwestern Australia were analyzed to determine whether wood anatomy is sufficiently correlated with ecology so that vessel element features can be said to have a predictive value. Indices for vulnerability (vessel diam: vessels per sq. mm) and meso- morphy (vulnerability x vessel element length) were calculated for each species in the follow­ ing florulas: karri forest understory, coastal granitic slopes, bogs, sand heaths, and desert. Wood indices for the species studied and for each florula show that these florulas form a se­ quence in increasing xeromorphy in the order listed. Genera represented in more than one florula validate the trends. Data for Gyrostemonaceae. Loranthaceae, and Cupressaceae are calculated separately because these are succulents, epiparasites, and conifers, respectively. Comparison with categories from floras elsewhere in the world shows the flora of Western Australia as a whole to be relatively xeromorphic. The indices devised show promise of great reliability because correlations with rainfall, temperature, and other factors are very close. Functional nature of the vessel element is thereby believed to be clarified.

IN ATTEMPTING to analyze factors underlying the Correlations based upon wood anatomy and hab­ selective forces which have guided evolution of itat of given species seem a valid approach to conductive tissue in vascular plants, I have used production of these hypotheses. One may pro­ various types of correlation (1975a). This meth­ ceed to develop correlations either with a system­ od could be criticized as constituting indirect evi­ atic group or with a floristic unit. The former dence. The most direct type of evidence would method was used, for example, where compari­ appear to be experimental. Experimental work, sons between wood anatomy and species, species such as that initiated by Scholander et al. (1975), groups, and their ecology and habit were of­ represents valid lines leading to a potential syn­ fered (Carlquist, 1975b, in press). Onagraceae thesis between wood anatomy and physiological cover a wide range of habit from annual to ecology. Ideally, one would like to isolate physio­ arboreal, and range from tropical cloud forest logical factors and compare them to individual to dry areas and aquatic habitats, although Ona­ wood characteristics; however, wood cannot be graceae include no true desert shrubs or lianas. isolated from the many adaptations of the plant The almost perfect habital and ecological cor­ in which it exists or from the complex ecological relations with wood anatomy in that family regimes to which any given species is adapted in prompted extension in the use of quantitative ves­ its tolerance ranges. Foliar apparatus, for ex­ sel element features to such families as Penaea- ample, can be of overriding importance. High ceae (Carlquist and DeBuhr, in press). Penaea- diffusive resistance of leaves and consequent low­ ceae occupy a relatively wide range of habitats ered transpiration rate, crassulacean acid metab­ for a small family, but not as great as Onagra­ olism, or C4 photosynthesis are among the factors ceae. This fact is reflected in quantitative terms that can override xylem conformtion as an opti­ with regard to wood anatomy. mal design system for coordinating foliage with conductive efficiency and resistance to tension My concepts concerning the latter two families of water columns in tracheary elements. are an extension of my (1975a) hypothesis, and are as follows. Short, narrow vessel elements are Experimental work on woods, if it is to be done theorized to resist high tensions in water col­ meaningfully, must be structured on hypotheses. umns. Narrowness of vessels is, to some extent, inversely correlated to number of vessels per sq. mm. This inverse correlation is, however, by no 1 Received for publication 13 December 1976; revision accepted 6 April 1977. means a perfect one. By dividing mean vessel This study was aided by grants from the National Sci­ diameter by number of vessels per sq. mm of ence Foundation, GB-38901 and BMS 73-08055 A-l. transection, one finds a range of values, not a The author is indebted to Dr. Larry DeBuhr for as­ constant, within a family (Carlquist, in press; sistance in collection of data. Mr. A. S. George and Mr. Kevin Richards kindly invited me on an expedition far Carlquist and DeBuhr, in press). A low value for into the interior of Western Australia. this ratio could be interpreted as great "redun- 887 8KS AMERICAN JOURNAL OF BOTANY [Vol. 64 dancy" of vessels. The more numerous the ves­ and if species with different xylem formulations— sels per sq. mm. the less the chance that disabling even excluding vascular plants other than di­ of a given number of vessels by air embolisms cotyledons—can coexist in literally the same lo­ formed under water stress would seriously im­ cality. If wood characteristics of particular flor- pair conduction in a plant. This idea was sug­ ulas tend to conform to particular plans and to gested to me by Dr. Martin H. Zimmermann differ modally from those of other florulas, how­ (personal communication). A low value for this ever, we have a tool, just as in the systematic ap­ ratio would therefore indicate capability of with­ proach, for developing criteria for xylem adapta­ standing water stress or freezing. I have termed tion to varying ecological regimes. Even if these this ratio "vulnerability" (Carlquist, in press). tools fall short of utmost precision, we can use That ratio, multiplied by mean vessel element clear correlations in developing a holistic ap­ length, I termed "mesomorphy" in accordance proach to expressing water relation adaptations with the demonstration that higher values occur of individual species, incorporating foliar and I in species that seem, on various grounds, more other features in addition to wood anatomy. highly mesomorphic. Ideally, one would wish for both systematic The rationale underlying use of vessel element and floristic comparisons within a single study. ^ length is that short vessel elements are more re­ In the present study, as many genera and fam­ sistant to collapse or deformation than long ones ilies as possible which occur in more than one because of the strengthening furnished by the end florula were selected: the genera Acacia, Boro- walls, which are unpitted constrictions along the nia, Casuarina, Hibbertia, Hybanthus, and Pime- length of a vessel. Vessel element length, fur­ lea; the families Epacridaceae, Fabaceae, Myrta- thermore, is controlled by morphogenetic factors ceae, Proteaceae, Rutaceae, and Sterculiaceae. independent of those affecting diameter of ves­ for example. In a sense, I used both approaches sels and number of vessels per sq. mm of tran­ in my 1966 summary of wood anatomy of Astera- section. Other subsidiary characters which may ceae, but the ecological and habital categories be related to xeromorphy could be added: thick­ were relatively simplistic, and florulas were not ness of vessel walls, number of bars per perfora­ used. tion plate, and the nature of lateral wall pitting of vessels. However, the majority of dicotyle­ MATERIALS AND METHODS—During my 1974 dons have simple perforation plates. The thick­ field work in Western Australia. I obtained wood ness of vessel walls and the tendency for vessels samples from several distinctive habitats. West­ to be grouped in large aggregations or to be ern Australia contains well-marked ecological formed in a solitary fashion are characters sta­ zones and a flora that is characterized by con­ bilized in particular taxonomic groups rather than siderable adaptive radiation into these zones susceptible to a wide range of modification in (Carlquist, 1974). These considerations are dicotyledonous families at large. This is also true basic to my choice of southwestern Australia, but of lateral wall pitting of vessels. Although these no single region is ideal. Ultimately a variety of subsidiary features are worthy of consideration, different regions should be compared with each the three vessel element measurements cited as other. The comparisons offered by the "World utilized in the indices devised were chosen be­ flora" categories in Table 6, however, provide an cause they are applicable to virtually all dicotyle­ equally valid kind of perspective. donous woods. Wood samples of desert shrubs in Western The alternative to study of a taxon in the eco­ Australia were obtained from a transect beginning logical interpretation of wood structure is, as at Cocklebiddy on the Nullarbor Plain coast, mentioned, a floristic one. One can trace this ap­ ranging northwards through the Victoria Desert proach to such authors as Webber (1936), who and the Gibson Desert to the Rawlinson Range studied wood of chaparral and desert shrubs. near the southwestern corner of Northern Terri- « More recent representatives of the floristic ap­ tory. The remaining ecological zones sampled r proach include Novruzova ( 1968) and Versteegh were relatively small in extent by comparison. ( 1968). Novruzova's samples did not include a The karri (Eucalyptus diversicolor) forest under- broad ecological range, and only some of the de­ story shrubs were collected in the vicinity of Pem- sired quantitative data were collected. Versteegh bsrton and Manjimup. Coastal shrubs were all did not deal quantitatively with wood features and collected at a single station, Canal Rocks, where used altitudinal levels (with ecology only im­ they grew in deep pockets of soil on granitic plied) as a criterion. domes. Bog shrubs were collected in wet depres­ Both the floristic and systematic approaches sions between Albany and Nornalup. The sand are complementary, as implied by the work of heath shrubs sampled were obtained from the Baas (1973). One might question the validity of large sandplain inland from the Jurien Bay coast a floristic approach if, as 1 allege (1975a, p. near Badgingarra. The Western Australian sand 30-31). each species has its own series of ana­ heath is composed of aeolian acidic sands. The tomical solutions in coping with water relations. bogs present a problem with regard to their eco- Au«USt, 1977] CARLQU 1ST—FACTORS IN WOOD EVOLUTION 889

TABLE 1. Karri forest understory shrubs (Annual rainfall, ca. 150 cm)

Vessel Vessel Vessels element diam persq. length, Species Family Coll. no. „.:n mm nm V M

Acacia urophvlla Benth. Fabaceae 5563 57 31 262 1.84 482 Acacia sp. Fabaceae 5576 51 39 280 1.31 367 Bossiaea laidlawiaiui Fabaceae 5571 61 48 217 1.27 550 Tovcy & Morris Casuarina decussata Benth. Casuarinaceae 5569 42 71 433 0.59 255 Chorilaena quercifolia Endl. Rutaceae 557(1 53 100 389 0.53 206 Hibbertia furfuracea Dilleniaceae 6065 37 95 642 0.39 250 (R.Br.) Benth. H. tetrandra (Lindl.) Gilg Dilleniaceae 5581 34 86 743 0.36 267 Hoxea elliptica (Sm.) DC. Fabaceae 5582 42 74 249 0.57 142 Lasiopetalum floribundum Benth. Sterculiaceae 6074 30 145 279 0.21 59 Leucopogon verlicillalus R. Br. Epacridaceae 5564 32 98 460 0.33 152 Pimelea clavata Labill. Thymelaeaceae 5565 65 45 190 1.44 274 Trvmalium spathulatum Rhamnaceae 5577 48 57 474 0.82 389 (Labill.) Ostf.

Florula average 46 74 385 0.62 239

logical status, a problem discussed in a later sec­ florula from Greenland published by Miller tion. (1975). The remaining "World flora" data in The rainfall figures in Tables 1-5 are from the Table 6 are based upon Carlquist (1975a). Atlas of Australian Resources (1952-1966). In Woods were sectioned and macerated accord­ Table 6, the figures for Gyrostemonaceae are ing to the usual techniques. The means of the based on 15 collections representing three genera lengths and diameters of vessels elements (tra- and seven species. An account of wood anatomy cheids in the case of conifers) were obtained of this family will be published elsewhere. The from 50 measurements per feature in each spe­ Loranthaceae of Table 6 represent four collec­ cies. The mean number of vessels per sq. mm of tions of Loranthus (sensu lato), parasitic on transection was obtained from 10 measurements acacias, from the desert transect. The conifers for each collection. The reader may question the included in Table 6 are Actinostrobus acumina- omission of such statistical details as standard tus, A. pyramidalis, Callitris canescens, and C. deviation and standard error—items which could robusta. Collection numbers in the tables are my easily have been calculated from the raw data in own, and voucher specimens are located in the hand. However, I find that such statistical analy­ herbarium of the Rancho Santa Ana Botanic Garden. The figures on Arctic shrubs in Table 6 sis is unwarranted in the present study because are compiled from data on eight species in a wood there are overriding sources of potential and ac­ cidentia! bias. Standard deviation and standard

TABLE 2. Coastal shrubs from Canal Rocks (Annual rainfall ca. 100 cm)

Vessel Vessel Vessels element diam persq. length, Species Family Coll. no. lira mm urn V \i

Acacia sp. Fabaceae 6083 49 61 224 0.80 179 Boronia alata Sm. Rutaceae 5541 42 163 358 0.26 93 Diplolaena dampieri Desf. Rutaceae 5544 32 146 525 0.22 116 Exocarpus odorafus (Miq.) DC. Santalaceae 5545 48 102 182 0.47 86 Exocarpus sparteus R.Br. Santalaceae s.n. 28 299 501 0.09 45 Guichenotia ledifolia J. Gay. Sterculiaceae 5546 28 196 254 0.14 36 Hibbertia cuneifortnis Dilleniaceae 6085 41 71 755 0.58 438 (Labill.) Gilg Hibiscus huegelii Endl. Malvaceae 6088 63 35 316 1.80 568 Pimelea sp. Thymelaeaceae 5454 4(1 107 196 0.37 73 Ricinocarpus sp. Euphorbiaceae 6087 43 76 318 0.57 181 Scaevola crassifolia Labill. Goodeniaceae 6084 37 119 368 0.31 114 Templetonia retusa (Vent.) R.Br. Fabaceae 6086 4ii 43 187 0.93 174 Florula average 41 118 342 0.35 120 890 AMERICAN JOURNAL OF BOTANY [Vol. 64

TABLE 3. Bog shrubs (Annual rainfall ca. 150 cm)

Vessel Vessel Vessels element cliam persq. Species Family Coll. no. um mm fem V M

Acacia mooreana W. V. Fitz. Fabaceae 6047 61 35 249 1.76 438 Boronia denticulata Sm. Rutaceae 5770 31 I 11 226 0.28 63 Boronia oxvantha Turcz. Rutaceae 5705 23 182 253 0.12 30 Cosmelia rubra R.Br. Epacridaceae 5675 25 765 587 0.03 IS lsopogon buxifolius R.Br. Proteaceae 5720 25 1 6? 304 0.15 46 Leptospermum crassipes Lehm. Myrtaceae 5724 64 140 333 0.49 163 Leptospermum firmum Myrtaceae 5708 35 207 559 0.17 95 (Schau.) Benth. Leucopogon assimilis R.Br. Epacridaceae 5707 25 324 482 0.08 39 Melaleuca micropltxlla Sm. Myrtaceae 5712 4:; 175 498 0.24 119 Melaleuca sp. Myrtaceae 6007 37 37 331 1.00 331 Persoonia longifolia R.Br. Proteaceae 6058 36 120 530 0.30 159 Phebalium argenteum Sm. Rutaceae 5710 3! 209 314 0.15 47 Pultenaea dasvphvlla Turcz. Fabaceae 5706 41 121 210 0.34 71 Stirlingia tenuifolia Proteaceae 5711 33 144 183 0.23 42 (R.Br.) Steud.

Florula average 37 195 361 0.19 69 error indicate statistical validity for cell popula­ from the results of Stern and Greene (1958). tions for which figures were actually obtained. The selection of species was dictated by routes of They cannot tell us anything about materials not travel and places where stops were made, hardly studied, however. For example, only one stem the ideal sampling criteria. The ecological limits per species was studied. Other portions of a given of a florula are a matter of arbitrary decisions, plant and other individuals of a given species and thus a source of human bias. Since each spe­ would have given different figures that would cies has a wood formulation that works conjunc­ render meaningless standard deviation and stan­ tively with a particular type of foliar apparatus, dard error on the single stem studied, judging one could say that an attempt should have been

TABLE 4. Saiul heath shrubs (Annual rainfall ca. 65 cm)

Vessel Vessel Vessels element diam per sq. length, Species Family Coll. no. um um M

Cyanostegia lanceolata Turcz. Dicrastylidaceae 5905 36 22 1 274 0.16 44 Cyanostegia sp. Dicrastvlidaceae s.n. = 6 252 257 0.14 36 Eremaeopsis pauciflora (Endl.) Druce Myrtaceae 5944 36 180 323 0.20 65 Eriostemon spicaius A. Rich. Rutaceae 5945 24 462 385 0.05 19 Geleznowia verrucosa Turcz. Rutaceae 5895 23 300 192 0.08 15 Halgania lavandulacea Endl. Boraginaceae 5744 21 614 263 0.03 S Hibbertia lineala Steud. Dilleniaceae 5940 34 217 605 0.16 97 Hybanthus bilobus C. A. Gardn. Violaceae 5748 23 482 302 0.05 15 H. floribundus (Walp.) F. Muell. Violaceae 5298 2') 246 512 0.12 61 Lachnostachys eriobotrya Dicrastylidaceae 5896 34 155 266 0.22 5') (F. Muell.) Druce Leucopogon australis R. Br. Epacridaceae 5549 32 209 427 0.15 '-4 Lysinema ciliaium R. Br. Epacridaceae 5942 is 522 340 0.04 14 Physopis lachnostachya C. A. Gardn. Dicrastylidaceae 5975 36 109 201 0.33 66 Pileanthus peduncularis Endl. Myrtaceae 5890 27 139 313 0.20 61 Pimelea sp. Thymelaeaceae 5939 34 183 276 0.19 52 Pityrodia bartlingii Dicrastylidaceae 5483 44 171 251 0.26 55 (Lehm.) Benth. P. old field ii (F. Muell.) Benth. Dicrastvlidaceae 5899 36 93 210 0.39 S2 Rhagodia preisii Moq. Chenopodiaceae 5440 37 133 95 0.28 27 Santalum spicatum (R.Br.) DC. Santalaceae 5911 48 70 271 0.68 185 Xylomelum angustifolium Kippist. Proteaceae 5889 57 41' 367 1.41 518 Florula average 33 240 307 0.14 43 August, 1977] CARLQUIST—FACTORS WOOD EVOLUTION 89 J

TABLE 5. Desert shrubs (Annual rainfall ca. 22 cm)

Vevel Vessel Vessels element diam per sq. length, Specie Family Coll. no. .vm mm /im V M Acacia burkitlii (F. Muell.) Benth. Fabaceae 5136 43 109 215 0.39 84 A. sowdenii Maiden Fabaceae 5140 (.1 57 167 1.06 177 A. tetragonophylla F. Muell. Fabaceae 5139 43 70 214 0.61 131 A. sp. Fabaceae 5137 54 57 163 0.95 154 A. sp. Fabaceae 5145 55 92 231 0.60 139 A triplex nitmmularia Lindl. Chenopodiaceae 5141 28 99 91 0.29 26 Casuarina pinaster C. A. Gardn. Casuarinaceae 5980 28 226 406 0.13 53 Eremophila latrobei F. Muell. Myoporaceae 5184 29 321 163 0.09 L5 Halgania cyattea Lindl. Boraginaceae 5167 37 136 192 0.27 52 H. sp. Boraginaceae 5156 27 298 175 0.09 16 Hvbamhus auianiiacus Violaceae 5170 33 328 295 0.10 29 (F. Muell.) Mclch. Kochia sedifolia F. Muell. Chenopodiaceae 5142 28 144 83 0.19 16 Myoporum sp. Myoporaceae 5134 36 125 234 0.29 67 Myoporum sp. Myoporaceae 5143 34 165 221 0.21 46 Prostanthera sp. Lamiaceae 5160 27 426 196 0.06 12 Sollya sp. Pittosporaceae 5276 28 413 430 0.07 30 Florula average 37 192 217 0.19 41

made, for example, to select species representing when compared with the woods of primitive di­ a range of foliar types within each florula. Epa- cotyledons (Table 6). The criteria for selection of cridaceae have exceptionally narrow vessels, quite those primitive woods were: more than 10 bars numerous per sq. mm of transection. Had a (mean) per perforation plate of vessel elements; larger number of Epacridaccae been selected for presence of true tracheids as the imperforate tra- any of the florulas, a greater degree of xero- chcary element; and presence of diffuse axial morphy for that florula would have appeared. parenchyma. Of the karri understory shrubs, only The difficulties of sampling when a floristic ap­ three have values of less than 200, suggested proach is used thus render sophisticated statistical above as a lower threshold for mesomorphy: analysis superfluous. The means and indices in Hovea elliptica, Lasiopetalum floribundum, and the tables, however, can be said to be valid when Leucopogon verticillatus. Of these, Hovea ellip­ interpreted cautiously. tica and Lasiopetalum floribundum could be From Table 6, one sees that a "V" value of described as not true karri understory shrubs, for approximately 1.0 to 2.5 indicates mesomorphy, they appear most commonly along road margins whereas a figure below 1.0 indicates redundancy in relatively exposed areas. Leucopogon verti­ of vessels and greater safety under conditions of cillatus is more typically an understory plant, but water stress. Values for "M" would indicate what may appear along road cuts as well. Leucopogon one would call mesophytes above levels of ap­ appears to share with all Western Australian Ep­ proximately 200, if one also concedes that suc­ acridaccae a very xeromorphic wood formulation, culents may be called mesophytes in terms of xy- however, and one can hypothesize that phylads lem structure. Xcrophytes in the classicial sense of Epacridaceae evolving into more mesic situa­ would have values of 75 or below. tions do not alter wood formulations nearly so much as foliar apparatus (Leucopogon verticilla­ tus has broad, thin leaves unlike the needle-like RESULTS—The florulas and other groups can leaves of most Leucopogon species). be reported in terms of the categories for the Western Australian flora listed in Table 6. Coastal shrubs—As Table 2 shows, V and M Karri forest understory shrubs—The values values for woods in this ecological category aver­ shown in Table 1 indicate that understory shrubs age about half the values for the karri understory shrubs. This can be seen in terms of individual of the karri forest are easily the most mesophytic genera (Acacia, Pirnelea) as well as in the trend dicotyledons in the Western Australian flora. of the figures. This habitat can be regarded as This accords not only with the greater annual intermediate between karri understory and sand rainfall of the karri forest areas, but also with heath in terms of rainfall, available water in the the finding by Gindel (1973) that despite tran­ form of runoff from granitic domes at Canal spiration in a wet forest, soil moisture remains Rocks, and the mitigating influence of maritime higher than in open areas. However, the karri humidity in modifiying temperature and tran­ understory shrubs are not notably mesophytic spiration extremes. 892 AMERICAN JOURNAL OF BOTANY [Vol. 64

TABLE 6. Wood anatomy of Western Australian flortilas compared to categories from other areas

V» ssel diam, Vessels Vessel element World flora /tin per sq. mm length, um V M Mesic primitive woods 109 47 1385 2.29 3172 Rosette trees 79 31 412 2.25 1051 Vines and lianas 157 19 334 8.22 2745 Annuals 61 162 1X6 0.38 71 Desert shrubs 29 353 218 0.08 17 Stem succulents 72 64 259 1.33 344 Arctic shrubs V 559 245 0.10 25 Western Australian Flora Karri understory shrubs 46 74 385 0.62 239 Coastal shrubs 41 118 349 0.34 119 Bog shrubs 37 195 361 0.19 69 Sand heath shrubs J 2 240 307 0.14 43 Desert shrubs 37 192 217 0.19 41 Gyrostemonaceae 71 62 180 1.15 206 Loranthaceae 36 155 78 0.23 18 Conifers' (27) i 1596. (0.02) ' Figures for conifers are for tracheids rather than vessel elements.

Bog shrubs—These woods (Table 3) were col­ ber through March (Atlas of Australian Re­ lected in order to determine whether shrubs re­ sources, 1952-1966). During the extremes of spond in evolutionary patterns to what would these months, these bogs are actually equivalent seem, at first glance, a highly mesic habitat. One to sand heaths in the dryness of their porous, must note that although designated as "bogs'" in sandy bottoms. The bogs of southwestern Aus­ Australian floristic literature, these habitats do tralia are mesic only in having a longer period of not correspond to bogs in the sense of floristic water availability than do sand heath areas. Ob­ literature of the North Temperate Zone. The viously, extremes rather than averages in rainfall Australian bogs, like the "vleis" of Cape Prov­ and temperature influence wood structure. The ince. South Africa, are not comparable to bog bog species can be said to differ from those of habitats or other ecological categories described sand heath areas in their tolerance to inunda­ by northern hemisphere botanists. The Austra­ tion (the wood structure of Leptospermum eras- lian bogs are depressions with underlying hard- si pes suggests this) during the relatively brief pan which hold standing water during the winter rainy season. The xeromorphic structure of wood and early spring months, unlike depressions in in bog shrubs has a positive selective value for sand heath country. Bogs in Western Australia vary in depth and duration of standing water. As drought, but this structure is not of selective dis­ Table 3 shows, only one species. Acacia moor- advantage during periods of water availability. eana. appears to show mesomorphic response to this habitat. The remainder of the species can be Sand heath shrubs—The V and M values for said to show remarkably low V and M values. bog shrubs (Table 3) are, as might be expected, The explanation for this appears twofold. First, slightly higher than for sand heath shrubs (Table the bog areas of southwestern Australia are very 4). Although in a belt of rainfall with more than few and small in extent compared to dryland twice the rainfall of the desert shrubs (Table 5), areas, and are therefore unlikely to have been the sand heaths are, in essence, a sandy desert centers of preservation of a relictual mesic flora during the summer months. The water-holding (one notable exception might be Cephalotus fol- capacity of sand heaths is certainly minimal. licularis). The phytogeographic source for bog Summer extremes provide a water relations re­ species has very likely been the sand heath flora, quirement regime equivalent to a desert, a fact which is composed of species with similar appar­ indicated by the near identity of V and M values ent preference for acidity. The bogs, like the for sand heath shrubs and for desert shrubs. sand heaths, have a substrate of white granitic That the sand heath shrubs represent more xero­ sand, so this ecological shift would be a logical morphic wood structure than do the coastal one. One must remember, moreover, that even shrubs can be seen by comparisons within in­ bogs in this moderately high rainfall area are sub­ dividual genera such as Hibbertia, Pimelea, and jected to considerable heat and drought from No­ Eriostemon (a close relative of Boronia). The vember through April. For example, in the study vulnerability values of sand heath shrubs are area where bog species were collected, average remarkably low compared with the "World flora" rainfall falls below 7 cm per month from Novem­ figures of Table 6. Lower values can be found in dicotyledon assemblages only in the desert August, 1977] CARLQU 1ST—FACTORS IN WOOD EVOLUTION 893 shrubs (from North America) and Arctic shrubs their respective habitats (chiefly sand heath, with (from Greenland). some species in desert and coastal sands), but also because examination of bark and leaves of Desert shrubs—To be sure, the source of the Gyrostemonaceae shows them to be succulents. desert shrubs (Table 5) used in this study is a Succulents are rare in the Australian flora at long transect rather than a single area. The tran­ large. One might regard the underground tubers sect consists of (1 ) shallow clay-like lateritic so prevalent in genera such as Drosera, Chamae- soils overlying limestone on the Nullarbor Plain; scilla, etc. as forms of succulence, however. The (2) red sands of the Victoria Desert; (3) stony presumption that Gyrostemonaceae qualify as lateritic clays and sands of the Gibson Desert. succulents seems justified. If one compares fig­ The annual rainfall of these areas is so low (see ures for Gyrostemonaceae (Table 6) to those of Table 5) that water availability of clay versus "stem succulents" from the World flora at large sand would seem to be of little importance. (Table 6), one notes a virtual identity in quan­ Clay soils might permit the growth of small trees, titative features. such as some species of Acacia, however. All of the acacias represented in Table 5 came from the Loranthaceae—The Western Australian Lo- Nullarbor Plain. The Nullarbor Plain has slightly ranthaceae show a very low M value, as one higher rainfall and possibly greater water avail­ would expect, since epiparasites would be ex­ ability (clay soils above a limestone hardpan) pected to have wood more xeromorphic than that than do desert areas to the interior. In fact, if of their host plants (Carlquist, 1975a). This one substracts the acacias from Table 5. one ob­ xeromorphy takes a form different from that of tains much lower values for the desert assem­ blage (vessel diameter = 30 ^m; vessels per sq. desert shrubs, as the data on Loranthaceae in mm = 244; vessel element length = 226 /im; V Table 6 show. The Loranthaceae studied do not = 0.12; M = 27). These V and M values are have an exceptionally low figure for "vulnerabil­ sufficiently lower than those of the sand heath ity." This would be related to the probability shrubs so that the lower rainfall and extreme that mistletoes do not experience sharp seasonal heat of the Victoria and Gibson Deserts appear fluctuations in moisture availability. Water in reflected in wood data. xylem of the epiparasite would be expected to be under very high tension, however. If so, the Noteworthy with respect to the desert shrubs shortness of vessel elements in Loranthaceae is that the two with successive cambia (Atriplex would agree with my (1975a) hypothesis on re­ nummularia and Kochia sedifolia) conform in sistance of short vessel elements to high tensions quantitative features to woods of dicotyledons in water columns of the xylem. with normal cambia. In the sampling of woods with successive cambia in my earlier (1975a) Perforation plates—Table 7 shows the two study, the greater width of parenchyma bands dicotyledonous families in the flora of southwest­ lowered the number of vessels per sq. mm, and ern Australia in which scalariform perforation thus these woods appear more mesomorphic plates in vessels are known. To this list may be quantitatively than do the Australian desert woods added Byblis gigantea (Byblidaceae), in which a with successive cambia. In the desert Atriplex bar or two traversing a perforation plate may be and Kochia, parenchyma bands between products seen occasionally (Carlquist. 1976). The species of successive cambia are so reduced that the within each family in Table 7 are arranged in or­ woods agree in all quantitative details with woods der of decreasing number of bars per perforation of desert shrubs with normal cambia. The same plate. Undoubtedly other species of Hibbertia can be said for sand heath shrubs of the family and of Epacridaceae from southwestern Australia Dicrastylidaceae. All species of Dicrastylidaceae have scalariform perforation plates. Epacrida­ have successive cambia, but quantitative data for ceae, Byblis, and Hibbertia may be regarded as these woods fall within the range of woods of mesic elements that have become established in sand heath dicotyledons with normal cambia. mesic pockets of southwestern Australia and radi­ Epacridaceae are virtually absent from desert re­ ated into drier habitats. The remainder of the flora gions, probably because interior sands are alka­ of southwestern Australia is a "simple perforation line, whereas the sands of the sand heaths are plate flora." indicative of establishment by groups acidic, a condition preferred by Epacridaceae. with simple perforation plates and otherwise more xeromorphic wood features (Carlquist, 1975a). Gyrostemonaceae—As noted in the previous Epacridaceae, despite having predominantly section, removal of the Nullarbor Plain acacias scalariform perforation plates, are not incon­ gives a more accurate picture of desert wood gruous in a flora that can, as a whole, be called characteristics. The Gyrostemonaceae were de­ xerophytic. As the figures for Epacridaceae in Tables 1-6 show, they have very narrow vessels, liberately removed from the remainder of the numerous vessels per sq. mm, moderately short flora not only because their quantitative features vessel elements, and therefore, as a family, have differ markedly from those of other shrubs in 894 AMERICAN JOURNAL OF BOTANY [Vol. 64

TABLE 7. Species with SCalariform perforation plates in the flora of southwestern Australia

Bar per Species Coll. no. Habitat plate, mean Dilleniaceae Hibbertia telrandra (Lindl.) Gilg 5581 karri understory 25.3 H. cuneiformis (Labill.) Gilg 6085 coastal scrub 23.7 H. furfuracea (R. Br.) Benth. 6056 karri openings 9.5 H. lineata Sleud. 5940 sand heath 7.0 Epacridaceae Cosmelia rubra R. Br. 5675 bogs 16.6 Sphenotoina dracophylloides Sond. 5691 montane scrub 14.8 Andersonia echinocephala (Stschegl.) Druce 5692 montane scrub 8.0 Leucopogon assimilis R. Br. 5707 bogs 2.3 Lysinema ciliatum R. Br. 5942 sand heath 1.1 Leucopogon verticillatus R. Br. 5564 karri understory 1.1 Leucopogon australis R. Br. 5549 sand heath 0.4 Leucopogon atherolepis Sischegl. 5690 montane scrub 0.0 remarkably low V and M indices. Consequendy, conduction than libriform fibers, which are pres­ Epacridaceae may be said to be adapted (or "pre- ent in woods of most dicotyledonous families. adapted") to xeric situations. The bars on the Tracheids are potentially much better for con­ perforation plates in Epacridaceae from south­ duction than are libriform fibers because of western Australia are not thin and wiry, as in greater pit membrane area. Thus, even if a very such tropical mesophytes as Illicium; they are high proportion of vessels of Epacridaceae were wide and bordered. The fewer the bars, the wider blocked by air embolisms formed under water they are; also, the wider the perforations between stress, the tracheids would represent a supple­ the bars. The occurrence of wide, relatively few mentary conductive system that could conduct an bars and large perforations in the generally xeric appreciable volume of water. (compared with wet forest) habitats of the spe­ Precisely the same considerations as offered cies in Table 7 could be cited in support of my for Epacridaceae would hold true for Hibbertia hypothesis that few, wide bars with large per­ as well. Unfortunately, my sample of sand heath forations represent a modification for xeromorphy species (which develop very small amounts of in a phylad with scalariform perforation plates wood and therefore tend to be overlooked during (Carlquist, 1975a, p. 160). The figures of Table wood collecting) consists of only a single spe­ 7 and the wood of Byblis gigantea (Carlquist. cies. Hibbertia lineata, however, has the fewest 1976) suggest that groups with relatively primi­ and widest bars per perforation plate and the tive wood show accelerated specialization during widest perforations of any of the species studied. adaptation to the generally dry climate of south­ Perhaps other sand heath species of Hibbertia western Australia. will prove to be similar, but additional study is Epacridaceae seem well suited in wood struc­ highly desirable. With respect to woods of Epa­ ture and foliage to xeric conditions in Western cridaceae and Hibbertia in Western Australia, one Australia. Only Leucopogon verticillatus, from must also take into account the nature of the the moist karri understory. has broad leaves, and foliar apparatus. Many Western Australian spe­ cies of Hibbertia have linear or ephemeral leaves. these probably represent a secondary expansion Those with moderately broad leaves may well of the lamina during evolutionary response to have high diffusive resistance, which would limit shady conditions. Although only a scattering of transpiration just as effectively as reduced leaf species of Leucopogon have been surveyed, this area. Unfortunately, diffusive resistance of leaves genus can be said to have predominantly simple has been measured for only a few angiosperm perforation plates. The fact that Leucopogon is species, but would surely be an extremely sig­ the largest genus of Epacridaceae in Western nificant measurement with regard to water rela­ Australia and has radiated so successfully in sand tions and wood structure. heath areas is very likely associated with the re­ duced number of bars per perforation plate, in accordance with my (1975a) hypothesis regard­ Other vessel features—The three quantitative ing ecological distribution of scalariform versus features related to vessel measurements may be simple perforation plates. An additional factor modified by other features. Greater vessel wall worthy of note is that the imperforate elements thickness can be an expression of xeromorphy, in the wood of Epacridaceae are all tracheids. presumptively functioning in resistance to higher Tracheids are, at least in theory, much better for water tensions, as in Larrea (Carlquist, 1975a). August, 1977] CARLQUIST—FACTORS IN WOOD EVOLUTION 895

In other groups, such as Asteraceae, larger group­ have greater "safety" under conditions of freez­ ings of vessels with increasing xeromorphy may ing as well as water stress. A high degree of re­ occur (Carlquist, 1966). This would be a func­ dundancy is not necessarily primitive It can be tional correlation explainable by the hypothesis found in such a highly specialized dicotyledon that grouped vessels resist water tensions more as Loricaria thuyoides, which has very narrow effectively by mutual support than do isolated vessels and a large number of vascular tracheids vessels (Carlquist, 1975a). Features such as —an almost conifer-like wood (Carlquist, 1975a). these may well contribute to xeromorphy of woods in particular groups. If so, the vessel CONCLUSIONS—Those skeptical of analysis of features of prime importance in Tables 1-6 may xylem function by means of anatomical correla­ be expected to vary with relation to these mod­ tions may point to the fact that within each of the ifying vessel features. florulas studied here, there is not strict conform­ ity to a narrow quantitative range. The mitigat­ Conifers—Although not really comparable to ing features of foliar structure (e.g., cuticle thick­ dicotyledonous woods, conifers represent a xylem ness; succulence) and physiology (crassulacean plan ideally adapted to water stress conditions. acid metabolism; C4 photosynthesis) are obvious If "redundancy" of conducting cells in the wood sources for departure from a particular xylem is a measure of viability of xylem when air em­ formulation. Other factors, such as the other ves­ bolisms form, conifers are unexcelled. In addi­ sel features mentioned above and the nature of tion, air embolisms, when they occur, can be the root system (spread, depth, cork covering, localized within individual cells in the case of tra- presence of mycorrhizae) are additional reasons cheids, but in dicotyledonous vessels elements, for departure from a uniform plan within a flor­ they can spread from one vessel element into an ula. Ranges within each florula are, however, of entire vessel. If one utilizes tracheids instead of a relatively small order of magnitude compared vessels as a basis for calculating a "V" ratio, one with World flora categories (Table 6). The aver­ can see that an extremely low vulnerability is ages derived from each florula suggest correla­ demonstrated by conifers of Western Australia tions between xylem anatomy and habitat. The (Table 6), which are probably not markedly dif­ values for these indices fall in precisely the same ferent in this respect from other conifers, al­ sequence as the probable mesomorphy or xero­ though tropical conifers (Agathis, Araucaria, Podocarpus) tend to have somewhat wider tra­ morphy of the habitats. In turn, woods of all of cheids and therefore fewer tracheids per sq. mm the florulas of Western Australia conform to a of transection. Conifer xylem represents a con­ relatively xeromorphic pattern, with V and M ductive system successful in dry areas of Aus­ values at best well below those of primitive woods tralia only if transpiration rates are low, as they chiefly from tropical mesic regions. The usefulness presumably are in the microphyllous Cupressa- of the V and M indices as ecological indicators ceae of Western Australia. Although much low­ seems well substantiated on the basis of system­ er in a "V" ratio than vessel-bearing dicotyledons, atically-oriented as well as floristically-oriented the lowest ratio in the latter group is in Arctic approaches (or a mixture of the approaches). shrubs (Table 6). This is not surprising, for Byblis, Hibberita, and Epacridaceae may rep­ water in xylem of these shrubs is undoubtedly resent establishments of mesomorphic groups completely frozen during the winter. Air bub­ that have been able to radiate subsequently into bles are therefore present in all xylem elements drier environments. Among non-woody dicoty­ when spring thawing occurs. Probably these ledons, one might add Cephalotus, Drosera, and Arctic shrubs have mechanisms for resorbing or Lentibulariaceae. The vast bulk of the dicotyle­ otherwise expelling air from vessels, but great re­ donous flora of Western Australia, however, has dundancy of conductive cells in secondary xylem specialized xeromorphic woods. This majority of is undoubtedly functionally valuable in this re­ the flora may have established itself in moderately gime. Species with scalariform perforation plates dry to dry areas of Western Australia, and radi­ and with simple perforation plates are represented ated mostly into drier habitats but also, in an in about equal bumbers in Miller's (1975) Green­ appreciable number of instances, into more mesic land wood florula, so the nature of perforation habitats such as the karri understory as well. plates is probably a minor consideration in con­ Radiation into more mesic habitats by phylads ductive characteristics. There is, however, a large appears to have occurred in other floras, such as proportion of conifers and of dicotyledon woods that of the Hawaiian Islands (Carlquist, 1974, with tracheids or fiber-tracheids as imperforate 1975a), where Asteraceae would be a prime ex­ tracheary elements (Empetraceae, Eriaceae, Rosaceae) in Miller's florula. As mentioned in ample. We tend to think in terms of increasing the preceding section, dicotyledonous woods with adaptation to xeromorphy as a generalization— tracheids rather than libriform fibers as imper­ and this may be true in the majority of instances. forate tracheary elements might be expected to However, this phylesis is probably reversible un­ der special circumstances.

• 896 AMERICAN JOURNAL OF BOTANY [Vol. 64

LITERATURE CITED . AND L. DF.BUHR. 1977. Wood anatomy of Penaeaceae (Myrtales): comparative, phylogenetic. An AS oi- AISIRMIAN RISOIRCTS. 1952-1966. De­ and ecological implications. Bot. J. Linn. Soc. (in partment of National Development. Canberra (parts press). published variously, many not credited to particular authors). GINDEL. I. 1973. A new ecophysiological approach to forest-water relationships in arid climates. Dr. W. BAAS. P. 1973. The wood anatomical range in Ilex Junk B.V.. Publishers. The Hague. (Aquifoliaceae) and its ecological and phylogenetic significance. Blumea21: 193-258. MILLER. H. J. 1975. Anatomical characteristics of some woody plants of the Angmagssalik District of CARLQUIST, S. 1966. Wood anatomy of Compositae: southeast Greenland. Medd. Gr0enl. 198(6): 1-30. a summary, with comments on factors controlling wood evolution. Aliso 6(2 >: 25-44. NOVRUZOVA. Z. A. 1968. The water conducting system . 1974. Island biology. Columbia University of trees and shrubs in relation to ecology. Iztatel'tsvo Akademii Nauk Azerbaijan S. S. R.. Baku. Press, New York and London. . 1975a. Ecological strategies of xylem evolu­ SCHOLANDER. P. F.. H. HAMMII, E. D. BRADSTREET, AND tion. University of California Press. Berkeley. E. A. HEMMINGSEN. 1965. Sap pressure in vascu­ . 1975b. Wood anatomy of Onagraceae, with lar plants. Science 148: 338-345. notes on alternative modes of photosynthate move­ STERN. W. L.. AND S. GREENE. 1958. Some aspects of ment in dicotyledon woods. Ann. Mo. Bot. Gard. variation in woods. Trop. Woods 108: 65-71. 62: 3X6-424. VFRSTEEGH, C. 1968. An anatomical study of some . 1976. Wood anatomv of Byblidaceae. Bot. woody plants of the mountain flora in the tropics Gaz. 137: 35-38. (Indonesia). Acta Bot. Neerl. 17: 151-159. . 1977. Wood anatomy of Onagraceae: addi­ WEBBER, I. E. 1936. The woods of sclerophyllous and tional species and concepts. Ann. Mo. Bot. Card, desert shrubs and desert plains oi" California. Amer. (in press i J. Bot. 23: 181-188.