Volume 46 Number 1 Article 5


Floristic analysis of the southwestern

Steven P. McLaughlin University of , Tucson

Follow this and additional works at: https://scholarsarchive.byu.edu/gbn

Recommended Citation McLaughlin, Steven P. (1986) "Floristic analysis of the southwestern United States," Great Basin Naturalist: Vol. 46 : No. 1 , Article 5. Available at: https://scholarsarchive.byu.edu/gbn/vol46/iss1/5

This Article is brought to you for free and open access by the North American Naturalist Publications at BYU ScholarsArchive. It has been accepted for inclusion in Great Basin Naturalist by an authorized editor of BYU ScholarsArchive. For more information, please contact [email protected], ellen_amatange[email protected]. FLORISTIC ANALYSIS OF THE SOUTHWESTERN UNITED STATES

Steven P. McLaughlin'

Abstract. —A study was made of the distributions of native, terrestrial, \aseular plants occurring in 50 local floras from throughout the Basin and Range and physiographic of the southwestern United States. The objectives of the study were to objectively define and describe the floristic elements —assemblages of species with roughly coincident geographic distribution —occurring in the southwestern United States and to deter- mine what such assemblages reveal about the floristic history of the . The total flora (native, terrestrial species only) of the Southwest is estimated at 5,458 species, 77% of which were recorded in 1 or more of the local floras. Nearly 22% of these species are endemic to the study region. A majority of the species were found to be relatively rare. The average range of a species included only 4 floras, and 90% of the species were recorded from 11 or fewer floras; only 81 species (1.5%) were recorded from 50% or more of the floras. Trees constitute 2% of the regional flora and have the widest average distribution; perennial herbs constitute 59% of the flora and have the most restricted distributions. Factor analysis was used to identify seven floristic elements for the region; a Great Basin element, a Mojavean element, a element, a Chihuahuan element, an Apachian element, and a Mogollon element. This factor analysis solution was shown to satisfy criteria of interprctabilit\ and consistency. The Mojavean, Colorado Plateau, and Apachian elements are believed to be autochthonous. The otlier four elements show high overlap in species composition with one or more adjacent . Each floristic element is mapped to show its geographic form and distribution. Analysis of these shows how the existence of objectively defined floristic elements is not contradictory to either the individualistic view of the distribution of a species or local continuity of vegetation and flora. The rarity of the majority of species and the clear association of floristic elements with rather narrowly circumscribed environments suggests that many Southwestern species have migrated little and are of rather recent, probably postglacial origin. Geographic "principles " derived from the distribution patterns of relativeh few, wide- spread, dominant, usually woody species may not be applicable to entire, regional floras.

Plant geographers have long recognized dence between floristic groups and floristic that plant species can be grouped on the basis provinces (e.g. , Gleason and Cronquist 1964), of similarities among their geographic distri- in which case the floristic group is the charac- butions. Such floristic groups have generally teristic flora of the floristic . More been termed "geographical elements ' by Eu- often floristic groups are conceived as assem- ropean phytogeographers (Stott 1981, Cain blages of variable and often overlapping areal

1947). The same concept is embodied in such extent, including a few wide-ranging types terms as "natural floristic areas" (Raup 1947), and many more narrowly defined types. A

"floristic assemblage" (Cain 1944), "floristic good example is the set of "areal types" used group" (Gleason and Croquist 1964), "areal by Whittaker and Niering (1964) in their anal- types" (Whittaker and Niering 1964), "cate- ysis of the flora of the Santa Catalina Moun- gories of geographic origin" (Stebbins 1982), tains of southern Arizona. Included are wide- or "directional classes" (Meyer 1978) of Amer- spread and overlapping "temperate," "wes-

ican authors. The term element is used uncrit- tern, "," and "southwestern" types ically in the American literatine as a synonym along with more restricted and largely dis- for taxon or taxa. crete "Sonoran," "Rocky ," "," A concept related to that of floristic assem- "Madrean, " and "Chihuahuan" types.

blages is that of floristic or biotic provinces. This chaos in terminology regarding floris-

Provinces in the sense of Dice (1943) are rea- tic assemblages is a consecjuence of the inade- sonably discrete areas with characteristic quate empirical and theoretical basis for such i:)hysi()graphy, , vegetation, flora, and concepts in the American tradition. Gleason fauna. There may be a one to one correspon- (1926) challenged the validity of the organis-

'Officc of .\ricl Studies. Ui

46 January 1986 McLaughlin: Floristic Anai 47 mic concept of associations as applied to vege- sis of vascular plant distributions recorded in tation by F. E. Clements and argued further local floras from the southwestern United that each species has its own individualistic States. environmental requirements and migrational history. It should follow, therefore, that each community of species is unique and any at- Methods tempt to identify repeatable associations must be highly subjective. The "individualistic con- Study Region cept" and the related "continuum concept" have been accepted by most American plant The study region includes the Colorado ecologists (Mcintosh 1967). Plateau and the Basin The ideas of Gleason were very broadly and Range Physiographic Province north of applied to plant by Mason (1947) the U.S. - International Boundary, and Cain (1944, 1947). Mason (1947) invoked both as mapped by Hunt (1967) (Fig. 1). This the individualistic of the species and region is bounded by the Sierra and the principles of past migrations of species (as southern on the revealed in the fossil record) to argue against west; by the on the north; the existence of persistent and recognizable and by the Rocky and southern floras. Cain (1947) believed that floristic ele- on the east. Only the southern ments could be recognized only relative to the boundary is artificial. This region will be re- particular area under investigation and not as ferred to simply as the Southwest. a universal system of floristic regions. He The majority of the region lies between the hnked the ideas of floristic groups and plant two great Western mountain systems — Sierra association types, and went on to argue Nevada on the west and to against the objective recognition of associa- the east. The entire region is arid to semiarid. tion types, implicitly casting doubt on the The Great Basin of the Basin and objective recognition of floristic assemblages Range Province experiences hot summers, of any form. cold winters, and mostly winter . The current attitude toward floristic assem- The southeastern portion of the Basin and blages is well illustrated in a statement by Range Province is at the opposite end of the Johnston (1977:356): climatic spectrum for the region, with hot summers, mild winters, and mostly summer "Recurrent distribution patterns" may simply be the precipitation. The most arid portion of the result of intuitive (but in this case unfortunate?) ten- dencies to lump and generalize, in the way the eye region centers around the lower Colorado tends to connect totally unrelated star-clusters into . meaningless but "recognizable" patterns in the night The study region has generally been recog- sky. nized as a natural one by biogeographers. One A recent text on ( and of the earliest formal treatments is that of Dice Gibson 1983) discusses the association/contin- (1943) and a recent one is that of Cronquist uum conflict but makes no reference to the (1982). Both recognize a semiarid natural re- problem of floristic elements. gion lying between the and the

If each plant species is uniquely and inde- Rocky Mountains that extends into northern pendently distributed, and therefore recogni- Mexico. Cronquist's Great Basin Province es- tion of floristic assemblages must be inher- sentially includes Dice's Artemisian and ently arbitrary and subjective, why have so Navahonian provinces; Cronquist's Sonoran many phytogeographers nevertheless recog- Subprovince is composed of Dice's Mohavean nized such groups? Can phytogeographical el- and Sonoran provinces; and their Chihuahuan ements be described more objectively and, if regions are similar. The principal difference is so, how can they be reconciled with individu- that there is nothing in Cronquist's treatment alistic concepts and what do they mean in that corresponds closely to Dice's Apachian terms of the evolutionary histories of their Province of the uplands of southeastern Ari- species? The objective of this paper is to an- zona, southwestern , and north- swer these questions through a floristic analy- western Sonora. 48 Great Basin Naturalist Vol. 46, No. 1

Fig. 1. of study region. Southwestern United States. Numbers refer to local floras listed in Table 1; letters in boxes refer to peripheral regions listed in Table 6.

Data Base Complete bibliographic citations for 33 of the 50 local floras can be found in Bowers' Fifty local floras from the Southwest were (1982) annotated bibliography. Flora 42 is

examined (Table 1). These floras represent a based both on the publication referenced in systematic sample of the regional southwest- Bowers (1982) and on an unpublished check- ern flora and were selected to provide as thor- list distributed by Natural Bridges National ough and uniform coverage of the study re- Monument. Flora 13 consists of those plants

gion as possible (see Fig. 1). listed by Twisselman (1967) as occurring in January 1986 McLaughlin: Floristic Analysis 49

Table 1. Southwestern local floras used in floristic analysis. Map numbers refer to map locations in Figure L

Map 50 Great Basin Naturalist Vol. 46, No. 1 are in every case lower than the numbers gions were examined using factor analysis. given in the original reports, since several Factor analysis is a family of ordination tech- classes of species were not included in the niques used to examine the pattern and struc- present analysis: ture of large data matrices. Factor analysis has long been used in sociological and geographic a. . b. Strict aquatics, i.e., floating, submerged, and studies, and its applications for plant emergent plants of permanent lotic habitats have been discussed by Goodall (1954) and for (ponds, lakes, reservoirs, , etc.) phytogeography by Jardine (1972). Since it is c. Synonyms at the species level. essentially a nonhierarchical clustering proce- d. Species listed by authors as "likely" to occur in their areas. dure, factor analysis is a particularly appropri- Aquatic species were excluded because of ate tool for examining floristic afiinities (Jar- dine 1972). their limited value in comparative geographic Factor analyses are of two basic types. studies. Such species tend to be very wide- Con- sider a matrix that lists the s spread but are relatively uncommon in dry presences of species (rows) in n floras (columns). regions, depending on the occurrence of - A Q-mode analysis examine the relationships in able (often man-made) habitats. Kartesz and would an s X s matrix of correlations or similarities Kartesz (1980) and Welsh et. al (1981) were the s species; an analysis used to achieve consistency and uniformity in among R-mode the relationships and to identify synonymous would examine in an n x n of correlations or similarities plant . In the remainder of this paper, matrix among the floras. Ideally prefer flora of the study region will refer to the na- n one would to per- a analysis to investigate floristic tive, vascular, terrestrial flora as defined form Q-mode above. affinities, but two problems arise: there is no suitable coefficient similarity avail- In addition to recording the presence of or index able to species in each of the 50 local floras, manuals correlate species ranges when most of the and checklists for the region (Holmgren and species are very rare; and available factor analysis programs are limited to -100 variables Reveal 1966, Cronquist et al. 1972, 1977, 1984, Kearney and Peebles 1969, Martin and (species in this case), a number smaller than Hutchins 1980, Munz 1968, Correll and John- the smallest local flora used in this study. ston 1970) were consulted to determine what I therefore used an R-mode analysis based inter additional species occur in the Southwest. Fi- on similarity indices among the floras and pe- ripheral regions. The nally, again using regional manuals, I rec- R-mode and Q-mode solutions should orded occurrences of all southwestern species agree whenever clear-cut in each of seven adjacent or peripheral regions patterns actually exist in the data (Bryant et al. 1974). The similarity index used was that of (see Fig. 1): (A) Pacific region (roughly equiva- lent to the California floristic province), (B) Otsuka (as given in Simpson 1980):

Columbia Plateau, (C) Rocky Mountain re- ISotsuka = C,/(A,B/'^ gion, (D) Great Plains, (E) Sonoran where: region, (F) Sierra Madre region, taken here as A, = number of species in flora i; the area in between the Bj = number of species in flora j; Sonoran and Chihuahuan , and (G) C,j = number of species common to flo-

Chihuahuan Desert region. The resulting lists ras i and j. of southwestern species occurring in these Since ISytsuka can take only positive values be- peripheral regions were used as floras in the tween and 1, the analysis used here is, in the data analysis (see below) to examine the rela- terminology of Pielou (1984), one using stan- tionships between the flora Southwest and dardized but none-entered data. With data in that of the peripheral regions, and to distin- this form, the first several factors are usually guish allochthonous from autochthonous ele- unipolar, i.e., all or nearly all the factor load- ments. ings (see below) are of the same sign, all posi- tive or all negative. Furthermore, there are as Data Analysis many unipolar factors as there are qualita- Similarities in the species composition of tively different clusters of data points (Pielou the 50 local floras and seven peripheral re- 1984). —

January 1986 McLaughlin: Floristic Analysis 51

Table 2. Statistical summary of the native, terrestrial flora of the Colorado Plateau and Basin and Range provinces of the southwestern United States.

Endemic species

Taxonomic group Families Genera Species no.

Ferns and allies 24 120 3 ( 2.5) Gymnosperms 3 7 36 2 ( 5.6)

Monocotyledons 10 136 723 63 ( 8.7) Dicotyledons 105 827 4579 1119 (24.4) Totals 127 994 5458 1187 (21.7)

The factoring program used was that of Nie Commonness as used in this paper refers to

et al. (1975). I used principal components fac- ubiquity rather than abundance (Preston tor extraction with orthogonal (varimax) rota- 1948), although ubiquitous taxa are also usu- tion, probably the most commonly used fac- ally abundant (Hengeveld and Haeck 1982, toring technique in ecological studies. Several Brown 1984). The Asteraceae, which is the aspects of the analysis were altered later to largest family, accounting for 17. 1% of the check on the consistency of the solution (see total species in the flora, is also the most com- below). mon family, with 78.6 species/flora. is the second most common family even

Results though it is represented by more than 100 fewer species than the Fabaceae, the third Description of the Flora most common family. The families Fabaceae, The flora of the study region is summarized , Scrophulariaceae, Boragi- in Tables 2-6. A total of 4,185 species were naceae, Polygonaceae, and Cactaceae are recorded from the 50 local floras; 1,273 addi- characteristic families of the region in that tional species native to the study region were they have both a large number of species and a identified from the literature. The estimated high proportion (>25%) of their species en- total regional flora (5,458 species) is probably demic within the region. The five most com- somewhat conservative, since the flora of the mon genera Eriogonum, Astragalus, Cryp-

intermountain region in and Nevada is tantha, Penstemon , and Phacelia — also have

still being compiled and monographed. The a high proportion of endemic species. Several sample of species in the 50 local floras, how- large genera common within but not charac- ever, probably represents at least 70%-75% teristic of the region, including Carex, Jun- of the regional flora. cus, and Salix, are mostly found in relatively Table 2 lists total species and endemic spe- mesic habitats. cies by taxonomic group. Within the The distribution for the 5,458 study region, 21.7% of the species are en- species occurring in the Southwest is shown in demic. Most of the endemic species are di- Figure 2. The average number of floras per cotyledons, and, in fact, nearly a quarter of species is 4.07. Only 81 species (1.5%) were the dicotyledons are endemic. The proportion recorded from one-half or more of the local of endemics in the study region is much lower floras (Table 4). Most of these widespread spe- than that in the California floristic province to cies have several subspecific taxa within the the west (47.7%) but higher than that of other region, indicating that they are genetically continental areas including (5.9%), variable species. The widespread species (9.0%), or the northeastern United listed in Table 4 are found throughout most of States (13.5%) (Raven and Axelrod 1978). western . The vast majority of There is a rather high ratio of species per species in the region, however, are rare. genus (5.5) in the Southwest, similar again to Nearly two-thirds of the species are recorded that of the California floristic province (5.6) from three or fewer floras. and higher than for most other continental The distribution patterns of southwestern areas (3.0-4.0) (Raven and Axelrod 1978). species, as revealed by their occurrence in the The most common families and genera 50 local floras examined, are probably reliable within the study region are listed in Table 3. evidence for the rarity of most species in this 52 Great Basin Naturalist Vol. 46, No. 1

Table 3. Commonest families and genera of native, terrestrial plants in the Colorado Plateau and Basin and Range provinces. January 1986 McLauchlin: Floristic Analysis 53

TaBI.K 4. Most eommonh' eiicoiiiitt'red species in 50 local floras fVoni the southwestern United State

Subspecific %of Subspecific %of Species" taxa'^ floras Species taxa floras

Artemisia ludoviciana 7 Sitanion hijstrix Descurainia pinnata Rhus trilohata Erigcron divcrg,ens canescens Aristida purpurea" Giitierrezia saroth rae Mentzclia alhicaulis Poafendlcriana Miinulus pittatus Vulpia octoflora Contjza canadensis Linum leivisii caespitosa Boutehua curtipendula Senecio douiilasii Onjzoi)sis lujincnoides Holodiscus (htinosus Chnjsothamnus nauseosus Lappula redotvskii Sporohohis cnjptandrus phacacantha Sporohohis airoides pauciflora Echinoceretis triglochidiatus Hehanthus ajinuus Leucelene ericoides ptdcheUum Rosa ivoodsii Cercocarpus nwntanus Popidus fremontii Erigeron pumihis Draha cuneifolia AlHonia iucarnata Bouteh)ua gracihs CastiUeja ch romosa Artemisia dracuncuhis BrickeUia caUfornica Symphoricarpos oreophiliiis 54 Great Basin Naturalist Vol. 46, No. 1

Table 5. Summarv of flora by life form.

Life form No. species Floras/Species

Trees —

January 1986 McLaughlin: Floristic Analysis 55

Figs. 3-6: 3, Plots of loadings for factor 1 — Great Basin floristic element. 4, Plot of loadings for factor 2— Colorado Plateau floristic element. 5, Plot of loadings for factor 3— Mojavean floristic element. 6, Plot of loadings for factor 4 — Chihuahuan floristic element.

The Great Basin element is strongly corre- primarily the Sierra Nevada portion —both of lated with the flora of the Columbia Plateau pe- which have probably contributed species to the ripheral region to the north. Both are regions of Great Basin. The factor loading with the Rocky sagebrush desert and -forested mountain Mountain region is the higher of the two, sug- ranges. The Great Basin element is also strongly gesting that the contribution from there may correlated with the Rocky Mountain region to have been greater than that from the Sierra Ne- the east and the Pacific region to the west vada, as noted previously by Harper et al. (1978). —

56 Great Basin Naturalist Vol. 46, No. 1

2. Colorado Plateau Element. —The eral region and weakly correlated with all second factor (Fig. 4) is clearly associated with other peripheral regions. the Colorado Plateau. Its greatest representa- The four desert elements recognized tion is in the northern portion of the Colorado Great Basin, Mojavean, Chihuahuan, and Plateau about the confluence of the Green and Sonoran—correspond very closely to the four Colorado . The element extends beyond desert regions of the same names recognized the hmits of the Colorado Plateau into the by Shreve (1942). In fact, the formal analysis southeastern Great Basin. Its most abrupt at- presented here nicely confirms Shreve's im- tenuation occurs along the southern boundary pressions based on vegetation and flora. More of the Colorado Plateau in the recent interpretations that would include the area. as a of the Sonoran Unhke the Great Basin element, the Colo- Desert (Cronquist 1982; Turner 1982) are not rado Plateau element is not strongly corre- supported. Floristic evidence favors retention lated with any of the peripheral regions. The of the Mojave Desert as a biogeographic en- Colorado Plateau is also an area of relatively tity distinct from the to the high endemism (Fig. 10). The high incidence south. of endemics and the low similarities with pe- 6. Apachian Element. Factor 6 (Fig. 8) ripheral regions identify the Colorado Plateau occurs primarily in southeastern Arizona and element as an autochthonous element. . It can be termed 3. Mojavean Element. —The third factor an Apachian element in the sense of Dice (Fig. 5) is centered about the Mojave Desert (1943). The Apachian element extends all region of southeastern California, northwest- along the southern boundary of the Colorado ern Arizona, and . The Mo- Plateau along the Mogollon Rim into north- javean element, like the Colorado Plateau ele- western Arizona. An extension can easily be ment, is weakly correlated with all peripheral recognized in Trans-Pecos Texas and south- regions, including the Sonoran Desert region. ern New Mexico east of the .

The rapid attenuation of the element to the It is not clear from an analysis of floras from west, where it meets the California floristic north of the International whether this re- province, is notable. The Mojave Desert element is largely autochthonous or al- gion has the highest frequency of endemic lochthonous. The element could be centered species in the Southwest (Fig. 10). This ele- within the mountainous region of southeast- ment is clearly autochthonous. ern Arizona, southwestern new Mexico, and 4. Chihuahuan Element. —The fourth northeastern Sonora, the Apachian Biotic factor (Fig. 6) is identifiable as a Chihuahuan Province of Dice (1943), or it could be a north- element, located in the ern extension of a broader floristic element area of Trans-Pecos Texas in the extreme centered in the of southeastern portion of the study region. It southeastern Sonora, southwestern Chi- extends up the Rio Grande Valley and into huahua, and northwestern Diuango, the southeastern Arizona. In the latter region spe- Sierra Madre Occidental Biotic Province of cies with Chihuahuan affinities are frequently Goldman and Moore (1945). Two local floras associated with limestone substrates (Whit- from northwestern Mexico provide additional taker and Niering 1968, Wentworth 1982). data. White (1948) prepared a large flora (927 The Chihuahuan element is strongly corre- native, terrestrial species) for the Rio de lated with the flora of the Chihuahuan periph- Bavispe region of extreme northeastern eral region and is an extension of the flora of Sonora, including the Sierra del Tigre, often that region. The Chihuahuan element is also considered an outlying section of the Sierra moderately correlated with the floras of the Madre. Maysilles (1959) examined the flora Plains and Madrean peripheral regions. (506 native terrestrial species) of the pine 5. Sonoran Element. —The fifth factor of western in the central (Fig. 7) is concentrated in southwestern Ari- Sierra Madre. I have calculated the similari- zona and the Colorado Desert area of extreme ties between these two Mexican floras and five

southeastern California. It is highly correlated floras with high loadings on factor 6 (Table 7). with the flora of the Sonoran Desert periph- The Bavispe flora shows high similarity with January 1986 McLaughlin: Floristic Analysis 57

6 Figs. 7-10: 7, Plot of loadings for factor 5— Sonoran floristic element. 8, Plot of loadings for factor — Apachian floristic element. 9, Plot of loadings for factor 7 — Mogollon floristic element. 10, Distribution ofspecies endemic to the southwestern United States. Isolines are labeled with the proportion ofspecies in local floras occurring only in the study region.

all five floras in Table 7; the Western Durango pretation of factor 6 as an autochthonous ele- flora shows low similarity with the same five ment from the southwestern USA and north- floras. The Bavispe flora, in fact, shows signifi- eastern Sonora. The Sierra Madre Occiden- cantly greater similarity with the Skull Valley tal, like the Sierra Nevada and Rocky flora at the western end of the Mogollon Rim Mountains, might best be considered a floris-

region than it does with the Western Durango tic region separate from the Southwest re- flora. These comparisons support an inter- gion of this study. More floristic research in 58 Great Basin Naturalist Vol. 46, No. 1

Table 7. Comparison of selected southwestern floras with two floras from the Sierra Madre region of northwester Mexico.

Otsuka Similarity Index

Flora Bavispe Region January 1986 McLaughlin: Floristic Analysis 59

Fig. U. Floristic areas of the southwestern United States identified from a factor analysis of similarity coefficients among 50 local floras.

the Southwest that have been recognized for cedures or in the data are likely to result in at least 40 years. The factors found in this substantial changes in the final solution. To analysis are simply and sensibly inter- check the consistency of the factor analysis pretable. solution, I made several changes in the data

The second criterion is that of consistency. analysis and compared the results. Changes If relationships among the variables are weak included using oblique rotation instead of or-

or if there are in fact no correlated clusters of thogonal rotation, substituting Simpson's in- variables, small changes in the factoring pro- dex of similarity (Simpson 1960) for Otsuka's, '

60 Great Basin Naturalist Vol. 46, No. 1 and omitting the checklists for the seven pe- formations or plant association types. Each of ripheral regions from the data base. Each of the 50 local floras used in the analysis contains the alternative analysis produced a set of representative species of several elements; seven unipolar factors nearly identical to likewise the species and elemental composi- those depicted in Figures 3-9. The factor tion of the floras change in a continuous man- analysis solution does appear to satisfy the ner along geographic gradients. The situation criterion of consistency. is analogous to fronts and the particu- Simpson's index did appear to be superior lar state of the atmosphere at any particular to Otsuka's index for use in this type of analy- time — the existence of continuous gradients sis. The communalities, defined as the sum of in air pressure across a region is not contradic- the squared factor loadings for each variable, tory to the recognition of high and low pres- measure the proportion of variance of each sure cells in different areas of the region. variable accounted for by the final factor anal- Somewhat surprisingly the analyses failed ysis solution. With Otsuka's index the com- to recognize widespread elements. That is, munalities were correlated with the size of there were no factors that could be associated the flora (r = 0.43, p < .01); with Simpson's with extensive groups such as "northern, index the communalities and flora sizes were "western, " "boreal, " "temperate, " or "south- not correlated (r = 0.27, > .05). This means p western " areal types (Whittaker and Niering that the unexplained or residual variances 1964). There certainly are widespread spe- were associated with the smaller floras using cies, i.e., those of Table 4. Such species do Otsuka's index. Intuitively it seems more rea- constitute a significant fraction of any particu- sonable either for the residual variance to be lar local flora, but they constitute a small frac- ' associated with the larger floras, or for it to be tion of the regional flora. The average extent uncorrelated with flora size. Also, the final of a floristic area in this analysis includes 7 solution accounts for a larger proportion of the floras, but the average range of a species in- total variance (68.7%) when the similarities cludes only 4 floras. In fact, only 17% of the are expressed as Simpson's indices. species of the region occur in 8 or more floras, beyond the average range of one floristic area, and only 6% occur in 15 or more floras, be-

Discussion ' yond the average range of two floristic areas.

The flora of the Southwest can be described In many instances one could probably associ- in terms of seven floristic elements. These ate many of the subspecific taxa of the rela- elements can be objectively recognized and tively few widespread species with specific explicitly mapped. The derived maps (Figs. floristic elements. No widespread elements 3—9) show, however, not discrete imits with were recognized because the factors reflect clearly delineated boundaries, but rather geo- general trends among the ranges of the major- graphic features varying in intensity from a ity of species, which are not widespread. i central area of greatest development and di- The recognition of floristic elements, i.e., | minishing gradually outward. Each element is coincident patterns in species ranges, is not most clearly developed in a limited area that necessarily inconsistent with the individualis- can be characterized by topographic and cli- tic concept of species distributions. The latter matic features that vary from adjacent areas view states that no two species distributions dominated by other elements. Five of these are idetitical since each species has individual areas — Great Basin, Colorado Plateau, Mo- physiological tolerances and evolutionary his- jave Desert, Sonoran Desert, and C'hihua- tory. Indeed, very few of the over 5,000 spe- huan Desert —are widely recognized natural cies occurring in the Southwest show an exact geographic areas. The other two — Apachian correlation of presence and absence in the floristic area and Mogollon floristic area— are local floras. The existence of floristic elements readily associated with upland regions sepa- indicates that species have nonrandom, over- rating adjacent lowland desert areas. lapping patterns of area, not that they are Floristic elements are objectively defined found consistently in similar communities, assemblages, but they are not discrete, which is what the individuahstic concept ad- bounded units in the sense of Clementsian dresses. Many southwestern species are so anuary 1986 McLauchlin: Floristic Analysis 61 incommon that they may consistently occnr plant species are capable of migrating to all suit- ogether in similar areas, i.e., are members of able habitats and that such migrations are suff^i- he same local floras yet seldom or never occnr ciently rapid to keep pace with environmental ogether in the same community. changes. How then can floristic elements be in the face of such extensive migra- The problem of scale or scope is thus seen to maintained tions? )e crucial in assessing the validity of associa- ional concepts such as elements or assem- The fact is that observed patterns of plant dis- )lages of species. Whittaker (1970) presents tributions in the Southwest do not provide much graphs of overlapping, normal curves repre- evidence of extensive migration as a general enting the abundances of dominant, woody phenomenon. The uncommonness of the major- pecies in local areas to argue against the exis- ity of species indicates either that most plants are ence of vegetation association types. The rather poor migrators or that insufficient time }uadrats or plots of the ecologist, however, has elapsed for these species to complete their contain only a minute fraction of a regional migrations. Both are probably true for south- . As one expands the sample area from western species. )lots to local floras to floristic areas, the Two observations seem to support the view grain" in species composition becomes finer that plants can and do migrate to nearly any- md patterns emerge. At the scale of conti- where within the range of their physiological lents, I doubt that any biogeographer fails to tolerances. First, many (but not all) seeds and ecognize the existence of distinctive biotas. fruits do possess obvious adaptations for disper- The best explanation for groups existing at sal. Second, the fossil record provides concrete egional levels while continuity prevails at lo- evidence that some species have migrated it ;al levels involves barriers to migration. Such widely. Nevertheless, from these observations )arriers are largely lacking for terrestrial does not follow that all or most plant species )lants at the level of quadrats and plots, but actually have migrated extensively. hey become more common and more effiec- Migration is a matter of probabilities. The vast ive as one increases scale. At the level of majority of seeds dispersed by plants travel very provide very effective bar- short distances (Levin and Kerster 1974). The kers to migration; at the level of floristic areas occasional long-distance transport of a single opography and climate provide sufficiently seed or fruit is unlikely to result in successful effective barriers to promote regional differ- colonization within a stable, competitive com- mtiation of the flora into floristic elements, munity. Most importantly, the likelihood of suc- mportant barriers in the Southwest that cessful migration is dependent on the number of ;erve to segregate one or more floristic ele- seeds of the species dispersed per unit of time nents include the Mogollon Rim along the compared to other potential migrators. The ma- louthern boundary of the Colorado Plateau, jority of plant species in the Southwest occur in he Wasatch Plateau in central Utah, the high- three or fewer local floras and may simply not be ands of southeastern Arizona and southwest- abundant enough to generate a flux of seed suffi- ern New Mexico (including the rather low-ly- ciently great to make extensive migration likely, ng in this area), and the even over considerable spans of time. )road zone of increasing basal elevation in Most of the actual examples of extensive mi- ;outhern Nevada. gration in the fossil record involve dominant, Migration has been depicted in the litera- usually woody species. These plants constitute a ure as a process that should effectively pre- small fraction of the southwestern regional flora. clude the formation or recognition of floristic Derivation of general principles for all species of elements. Gleason (1926) argued that both plants from data on the current and former distri- ndividualistic tolerances and extensive mi- butions of species such as redwoods, grations of species were inconsistent with the , and other trees is a questionable recognition of recurring associations of spe- practice that may have led to some unwarranted cies. Good (1931) and Mason (1936) elevated conclusions. to a "principle" the assertion that extensive If the capacity of most species to migrate to all migrations oifloras had occurred in the past. areas suited to their physiological tolerances Gleason and Cronquist (1964) stated that most were as pervasive as typically assumed, I would 62 Great Basin Naturalist Vol. 46, No. 1 expect the average range of a species to be sponse to . My analysis of mod- much wider. One could argue that species in ern plant distributions in the entire regional the Southwest have very narrow physiological flora, however, casts some doubt on the gen- tolerances that effectively restrict them to rel- erality of the consensus model outlined atively few habitats within a limited area, but above. Most of the local floras included in this that seems rather unlikely given the similar analysis are from areas encompassing exten- nutritional needs of plants and the apparent sive elevational relief Southwestern moun- lack of coincidence between most species and tain ranges, particularly in the Basin and very specific habitats. It seems much more Range Province, possess considerable habitat likely that barriers to migration, competition, diversity in association with elevation, typi- and low rates of seed dispersal interact to cally going from desert at the base to pinyon- greatly curtail the ability of most species to woodland, pine forest, or even migrate. forest at the top (e.g., Lowe 1964). Old spe- Over the past several years a general model cies with extensive migrational histories of southwestern phytogeographic history has would have left relictual populations scattered emerged that includes several postulates: (1) within and between these ranges. Only a - most species present today had evolved by the gle, small population is required to establish a end of the Tertiary, (2) these species have species as a member of a local flora. The con- undergone extensive and intensive migra- sensus model would predict that the ranges of tions in response to climate most species would include many such scat- changes, and (3) modern plant distributions tered, relictual populations, resulting in a reflect both this migrational history and the modern flora with most species widely dis- tolerances of species for modern climatic con- tributed throughout the region and little dif- ditions. ferentiation of the floristic assemblages within the region. The concept of great species longevity is supported by Levin and Wilson (1976), who Many species, particularly those of high el- estimated the mean duration for herbaceous evations, do show distribution patterns con- species to be 10 million years, with a mean sistent with the consensus model, but they do speciation rate of 1.15 hneages per million not constitute a large fraction of the Southwest years. Woody species were estimated to be flora. For the consisting of small even more persistent and to have even slower populations in only a few floras, i.e., the ma- rates of speciation. Consistent with this view, jority of species in the Southwest today, ex-

Tidwell et al. (1972) hypothesized that the tinction, rather than migration, is the more flora at the beginning of the Pleistocene in the likely response to severe climate change. Intermountain Region was basically the same These species are unlikely to possess suffi- as exists today. cient ecotypic, genetic diversity or suffi- The best evidence that many dominant spe- ciently large populations required to keep cies have indeed undergone extensive migra- pace with the changing environment. tions in the Southwest at the end of the Pleis- Speciation, however, may also be acceler- tocene comes from fossil packrat middens ated during periods of severe climate change (Van Devender and Spaulding 1979, Wells (Stebbins 1947, Axelrod 1981). Many south- 1983). Many workers have textraploted these western species probably originated in post- results back to previous glacials and inter- glacial time in response to the new, more arid glacials, concluding that the repeated climate environments created by the changes from changes occurring during the Pleistocene in Pleistocene to Holocene . Such an topographically heterogeneous intermoun- interpretation has been applied to Atriplex tain environments would have kept species (Stutz 1978) and seems likely for many of the populations in constant movement (Cronquist species of large southwestern genera such as 1978). Erio^onum, Astragalus, Penstemon, Gilia,

It is diflicult to conceive how recognizable Ca7nissonia, Cryptantha, Phacelia, Opuntia, floristic assemblages could survive through and others. the Pleistocene with all this movement, fluc- Do we need two models of evolutionary tuation, and shuffling of populations in re- response to climate change, one for common January 1986 McLauchlin: Floristic Analysis 63 species and one for rare species? I think not. species. Floristics and vegetation analysis are Common plant species consist of many popu- complementary approaches to understanding lations of subspecies, varieties, or . the distribution and abundance of plants. During periods of severe climate change, Phytogeography needs both. some of these ecotypes will migrate, many will go extinct, and some will differentiate to Acknowledgments form new ecotypes, varieties, subspecies, and species in new habitats. As a result, the range I thank R. M. Turner, J. H. Brown, J. E. of the species before the climate change (i.e., Bowers, and M. A. Kurzius for reading and in the fossil record) will be different from its commenting on the manuscript. M. range after the climate change. But migration Beauchamp, J. E. Bowers, D. Clemons, R. A. of populations or ecotypes existing before the Fletcher, T. R. Van Devender, and R. D. climate change and persisting through the cli- Worthington helped in the acquisition of flo- mate change need account for only part of the ras and unpublished checklists. The illustra- total change in the range of a species. In other tions were prepared by M. A. Kurzius. words, the most likely response of any partic- ular small population to rapid environmental Literature Cited change is ; the probability of a spe- AxELROD, D I. 198L Holocene climatic changes in rela- cies extinct depends in large part on becoming tion to vegetation disjunction and speciation. how many populations it has. Amer. Nat. 117:847-870. AxELROD. D I. AND P H Raven. 1985. Origins of the The seven floristic elements identified in Cordilleran flora. Biogeogr. 12:21-47. the present analysis are clearly associated J. Beatley, J C 1975. Climates and vegetation pattern with modern, Holocene environments. They across the Mohave/ transition of represent centers of postglacial differentiation southern Nevada. Amer. Midi. Nat. 93:53-70. Bowers, E. 1982. Local floras of the Southv^^est, 1920- and speciation within the depleted flora sur- J 1980: an annotated bibliography. Great Basin Nat. in viving at the end of the Pleistocene the 42:105-112. effectively isolated Southwest, partially but Brown, J H 1984. On the relationship between abun- from adjacent elements by modern climatic dance and distribution of species. Amer. Nat. 124:255-279. and topographic barriers. Brown, J H , and A C Gibson. 1983. Biogeography. C. My analysis of specific similarities among V. MosbyCo., St. Louis, Missouri. local floras of the Southwest is consistent with Bryant, E H , B Crandall Stotler, and R E Stotler 1974. A factor analysis of the distribution of some a hypothesis of Holocene differentiation of the Puerto Rican liverworts. Canadian J. Bot. flora. modern regional An analysis of generic 51:154.5-1554. similarities these floras might better among Buchanan. H , and R Graybosch 1981. Revised check- reflect the Tertiary and Pleistocene migra- list of the vascular plants of Bryce , Utah. Great Basin Nat. 41:109-120. tional history of the flora, since the majority of Cain, S A 1944. Foundations ofplant geography. Harper the are doubtless of Tertiary genera age. My and Row, New York. interpretation of the evolution of the south- 1947. Characteristics of natural areas and factors in western flora is also consistent with the con- their development. Ecol. Monogr. 17:185-200. Clemons, D. 1984. Anza-Borrego Desert Area, San cept of punctuated equilibria, as opposed to Diego : plant list. Unpublished report, slow, gradual phylogenetic change, which California Native Plant Assoc. , Chap- would be more consistent with the consensus ter. model discussed above. CoRRELL, D S., and M. C JOHNSTON. 1970. Manual of the vascular plants of Texas. Texas Res. Found., Ren- American plant very ecology has been ner. much a science of dominants. This bias has Cronquist, a 1978. The biota of the Intermountain Re- been carried over into plant geography, which gion in geohistorical context. Great Basin Nat. Mem. 2:3-15. has emphasized the distribution of vegetation 1982. Map of the floristic provinces of North almost to the exclusion of flora. I have tried to America. Brittonia 34:144-145. show in this paper how a study of the entire Cronquist, A , A H Holmgren, N H. Holmgren, and J flora of a region, weighing all species equally, L Reveal. 1972. Intermountain flora. Vol. 1. Ge- ological and botanical history of the region, its can reveal patterns that may not follow from or plant geography, and a glossary. The vascular be in agreement with principles and models cryptograms and the gymnosperms. Hafner Pub. derived solely from study of the dominant Co., Inc., New York. 64 Great Basin Naturalist Vol. 46, No. 1

Levin, D. A., H. W. Kerster. 1974. in Cronquist, a , a. H Holmgren, N H Holmgren, J L and Gene seed Reveal, and P. K. Holmgren. 1977. Intermoun- plants. Evol. Biol. 7:139-220. tain flora. Vol. 6. The monocotyledons. Columbia Levin, D. A., and A. C. Wilson. 1976. Rate of evolution in University Press, New York. seed plants: net increase in diversity and chromo- 1984. Intermountain flora. Vol. 4. Subclass Asteri- some numbers and species numbers through dae (except Asteraceae). New York Bot. Card., time. Proc. Nat. Acad. Sci. USA 73:2086-2090. Bronx, New York. Lewis, M E. 1971. Flora and major plant communities of Dice, L. R. 1943. The biotic provinces of North America. the -East Humboldt Mountains, with special University of Michigan Press, Ann Arbor. emphasis on Lamalle canyon. Unpublished re- Gleason, H. a. 1926. The individualistic concept of the port, Humboldt National Forest, Elko, Nevada.

plant association. . Torrey Bot. Club 53:7-26. Lloyd, R M , and R S Mitchell. 1973. A flora of the Gleason, H A., and A. Cronquist. 1964. The natural White Mountains, California and Nevada. Univer- geography of plants. Columbia University Press, sity of California Press, Berkeley. New York. Lowe, C H 1964. Arizona landscapes and habitats. Pages 1-132 in C. H. Lowe, ed.. The vertebrates of Goldman, E. A , and R T. Moore. 1945. The biotic Arizona. University of Arizona Press, Tucson. provinces of Mexico. J. Mammal. 26:347-360.

Good, R D.O 1931. A theory of plant geography. New Martin, W C , andC R Hutchins 1980. A flora of New

Phytol. 30:149-171. Mexico. J. Cramer, Vaduz, New Mexico. GOODALL, D. W. 1954. Objective methods for the classifi- Mason, H. L 1936. Principles of geographic distribution as cation of vegetation. III. An essay in the use of applied to floral analysis. Madroiio 3:181-190. 1947. Evolution of certain floristic associations in factor analysis. Australian J. Bot. 2:304-324. . Ecol. Monogr. 17:201-210. Harper, K T , D C Freeman. W K Ostler, and L G

Klikoff. 1978. The flora of Great Basin mountain Maysilles, J H. 1959. Floral relationships of the pine ranges: diversity, sources, and dispersal ecology. forests of western Durango, Mexico. LInpublished Great Basin Nat. Mem. 2:81-103. dissertation, LIniversity of Mich., Ann Arbor. McTntosh, R. p. 1967. The continuum concept of vegeta- Hengeveld, R., and J. Haeck. 1982. The distribution of tion. Bot. Rev. 33:130-187. abundance. I. Measurements. J. Biogeogr. 9:303- 316. Meyer, S E 1978. Some factors governing plant distribu- tions in the Mojave-intermountain transition , J. S. 1982. A floristic and vegetation analysis of the Newberry Mountains, Clark County, Nevada. zone. Great Basin Nat. Mem. 2:197-207. Unpublished thesis. University of Nevada, Las MuNZ, P A. 1968. A California flora. With supplement. Vegas. University of California Press, Berkeley.

Holmgren, A H., and J. L. Reveal 1966. Checklist of the . No date. Checklist of plant spe- vascular plants of the Intermountain Region. cies collected from Natural Bridges National Mon- USDA For. Ser. Res. Pap. INT-32, Interm. For. ument. Unpublished report. Natural Bridges Range Exp. Sta., Ogden, Utah. Natl. Mon. Hulten, E. 1968. Flora ofAlaska and neighboring territo- 1973. Checklist of the vascular plants of Joshua ries. Stanford Unversity Press, Stanford, Califor- Tree National Monument, San Bernadino and nia. Riverside , California. Unpublished re- Hunt, C B. 1967. Physiography of the United States. W. port, Joshua Tree Natl. Mon.

H. Freeman, , California. NiE, N H , C H Hull, J G Jenkins, K Steinbrenner, Jardine, N. 1972. Computational methods in the study of and D H Bent 1975. Statistical package for the plant distributions. Pages 381-393 in D. H. Valen- social sciences. McGraw-Hill, New York. tine, ed.. Taxonomy, phytogeography, and evolu- PlELOU, E C 1984. The interpretation of ecological data. tion. Academic Press, London. John Wiley & Sons, New York. Johnston, M. C. 1977. Brief resume of botanical, includ- PooRE, M E D 1955. The use of phytosociological meth- ing vegetational, features of the Chihuahuan ods in ecological investigations. IV. General dis-

desert region with special emphasis on their cussions of phytosociological problems. J. Ecol. uniqueness. Pages 335-359 in R. H. Wauerand D. 44:28-50. H. Riskind, eds.. Transactions of the symposium Preston, F W 1948. The commonness, and rarity, of on the biological resources of the Chihuahuan species. Ecology 29:254-283. desert region. USDI Nat. Park Ser. Trans, and Raup, H M 1947. Some natural floristic areas in boreal Proc. Ser. 3. America. Ecol. Monogr. 17:221-234. T., Kartesz, J. AND R Karte.sz 1980. A synonymizcd RyWEN, P H , AND D I AXELROD 1978. Origin and rela- checklist of the vascular flora of the United States, tionships of the California flora. Univ. of California , and . University of North Car- Pub. Bot. 72:1-134. olina Press, Chapel Hill. Rogers, B S., and A Tiehm. 1979. Vascular plants of the

Kearney, T H , and R H Peebles. 1969. Arizona flora. Sheldon National Wildlife Refuge. U.S. Fish and

2d ed. with supplement by J. T. Howell and E. Wildl. Serv., Portland, . McClintock. University ofCalifornia Press, Berke- Shreve, F 1942. The desert vegetation of North Amer- ley. ica. Bot. Rev. 8:195-246. KuRZius, MA. 1981. Vegetation and flora of the SiKES. S ,andJ . 1975. Davis Mountains vegetation Grapevine Mountains, National study. Pages 1-23 in Moiuit Livermore and Saw- Monument, California-Nevada. Unpublished the- tooth Mountain: a natural area survey. Div. Nat. sis, University of Nevada, . Res. and Environ., University of Texas, Austin. '

anuarv 1986 McLaugmlin: Floristic Analysis 65

IMPSON, G. G 1960. Notes on the measurenient offaimal Van Devender, T R . and W G Spaulding. 1979. De- velopment of vegetation and climate in the south- resemblance. Amer. J. Sci. 258A:300-311. 1980. Why and how: some problems and methods . Science 204:701-710. in historical biology. Pergamon Press, Oxford. Warnogk. B H 1967. Specimens in the herbarium of Big FEBBlNS. G. L., Jr 1947. Evidence on rates of evolution Bend National Park. Unpublished rept., Big Bend from the distribution of existing and fossil plant Natl. Park. species. Ecol. Monogr. 17:149-158. Wells. P V 1983. Paleobiogeography of montane 1982. Floristic affinities of the high Sierra Nevada. in the Great Basin since the last glaciopluvial. Madrono 29:189-199. Ecol. Monogr. 53:341-382. rOTT, P. 1981. Historical plant geography. George Allen Welsh, S. L.. N. D. Atwood, S. Goodrich, E. Neese, K. and Unwin, London. H Thorne, and B. Albee. 1981. Preliminary in- ruTZ. H. C 1978. Explosive evolution in perennial dex of Utah vascular plant names. Great Basin Atriplex in western America. Great Basin Nat. Nat. 41:1-108. Mem. 2:161-168. Wentworth, T R 1982. Vegetation and flora of the Mule

\YE. A C. 1983. Flora of the Stansbury Mountains, Mountains, Cochise County, Arizona. J. Arizona- Utah. Great Basin Nat. 43:619-646. Nevada Acad. Sci. 17:29-44. S. S 1948. vegetation flora of the region HORNE, R. F., B. A. Prigge, and J. Henrickson. 1981. A White, The and flora of the higher ranges and the Kelso Dunes of of the Rio de Bavispe in northeastern Sonora, the eastern Mojave Desert in California. Also Mexico. Lloydia 11:229-302. 10:71-186. Whittaker, R. H 1970. Communities and .

[DWELL, W. D., S. R. RUSHFORTH, AND D. SiMPER. 1972. Macmillan Co., New York.

Evolution of floras in the Intermountain Region. Whittaker. R H , AND W A Niering. 1964. Vegetation Pages 19-39 m A. Cronquist, A. H. Holmgren, N. of the Santa Catalina Mountains, Arizona. I. Eco- logical classification distribution species. H. Holmgren, and J. L. Reveal. Intermountain and of J. flora. Vol 1. Hafner Pub. Co., Inc., New York. Arizona Acad. Sci. 3:9-34.

LiRNER, R. M. 1982. Mohave desertscrub. Pages 157-168 Whittaker, R H . and W A Niering. 1968. Vegetation in D. E. Brown, ed., Biotic communities of the of the Santa Catalina Mountains, Arizona. IV.

American southwest—United States and Mexico. Limestone and acid soils. J. Ecol. 56:523-544. Desert Plants 4. Worthington, R D 1980. Flora of Hueco Tanks State WISSELMAN, E D 1967. A flora of Kern County, Califor- Historical Park. Unpublished rept., Texas Parks Wildl. nia. Wasmann J. Biol. 25:1-395. and Dep.