Vol. 106, No.

949

The Americur Naturalist

May-June

1972

use anal mis'

Ne'rl Yok.
of the soft102:243-282.
JJlookltr'en

EDAPHIC

ARIDITY

AS A FACTOR IN ANGIOSPERM EVOIiUTION

D.Nrr, f. Axnr,non

l)cpaltrtrcuts

of Geology :rnd J-lotany, University

of Califolnia,

I)avis, California

95616

Stebbins (1952) has sholvn that there are several reasons why plant
eYolution would be relatively rapid irr arid. to semiaricl regions. First,
in areas where moisture is limited, diversity in local terrain, soil, and
other factors has a greater effect on the flora and vegetation than in
legions where moisture is adequate. Second, semiarid climates with their
regional diversity promote the division of medium- to large-sized populations into smalier units which are isolated from each other but can exchange
genes by occasional migration, and establish populations that may give
rise to new t&xa. And third, in dry regions, many different specialized
vegetative structures (e.g., reduced leaf size, specialized leaf covering
[scales, trichornes], deciiluous habit, deep root system, swollen trunks,
bulbs, etc.) can evolve r'.'hich may enable plants to withstand periods of
severe drought.
Stebbins invokecl considerabietopographic relief (e.g., Coast Ranges of
California), as well as aridity, to provide a diverse environmental frarnervorli for rapid. evolution. Fulthermore, hc also notecl that in clry regions
divergent evolution is not necessarily limited to adaptations to xerophytism
becausesome mesophytes, as in the genera Scorzonura, 'I'rapogtogon,Cassio,
In,ga, Aacsal,pina, Miruosa, and llatth,inia, appear to have been clerived
f r o m x e l o p h y t i c a n c c s t o r s I. - I e c o n c h r d 0 s :" T f t h e c l i m a t e . . . i s b e c o m i n g
:rogressively rnore moist, lve can cxpect that some of the xerophytes with
favorable structure rvill becorne adapteil to rnoist couditiorts.
In this
\\'ay, tlte flola of thc samc mesophytic legioll could conlc to contain membcrs of the srmefarnily or tribe, ali aclapted to esscntially similar conditions, but adaptcil in different ways and hence very differcnt frorn each
other because of their different evolutionary histories."
While I do not deny the efflcacy of topographic relief and climatic change
in promoting rapid ancl divergent evolutiorr, there ale nonetheless alternatives to Stebbins's analysis which have not been discussed previously. These
are suggested by the evolutionary potcntial lvhich may be inferrerl for
Iandscapes of crystaliine rock that provicled aid and semiarid edaphic
sites throughout l\{esozoic and Cenozoic tirncs, nclt only in seasonally dry
ereas but in wet tropical regions as l'ell.
EDAPIIIC

DESERTS AND SDMIDESERTS

Tirc possibility that barren to semibarrerr areas ancl their immediate

environs may have had an important evolutionary lole in angiosperm

311

312

TIII'

AMI.I]iICN

NAI]URA.I,IST

evolution came to my attention on a recent visit to llrazil. There, highgrade gneissesancl granitic roclis of Archean age form barren to semibarren
tracts that are erltjrely sull'gunded by rvet tropicai rain forest' In adclition,
for the caa,tinga (tholn forest) of northsimila tra,cts for:m the srrbstl'rte
easteln llrazil, and they occur alsg in iterrnediate alcas whcre savann
vegetation is lvell clevelopecl.No one has previously raisecl for rliscussion
the evolutionary potential of such a lanclscape. It is of subcontinental
extent (Brazilian shield) ; is typified by persistent barren to semibarrcrr
is
bornhardts (inselbelgs), lorv hills, clomesrridges, ancl rock plains; and
simithat
found in diverse climates. li'urthermore, geologic evidence shows
lar landscapes of wide extent were preserrt on all the continents throughout
'lo set the background for the view that these
Mesozoieaud cenozoic times.
arid. to semiarid edaphic sites probably had an impoltant role in evtllution,
reference is macle first to ecological relations in Brazil, and then other
areas where similar terranes occur are noted briefly'
In the coastal parts of Brazil, where precipitation is high (over 2,000
mm, or 100f inches), regolith may be from 50 ft to over 200 ft deep
(Branner 1896, p.262). The transition to crystalline basement is abrupt,
usually taking place witliin r ferv rllillirneters. ]lecauso of the i[rpcrvious
nature of the basement, at times of torrential rrin the regolith on steeper
slopes may become saturated and unstable. As a I'esult' Iandslides occur
regularly and expose iarge patches of naherl clystalline br.sement.These
persist for long periocls because the roeks are excecdingly hard and resisla,nt, and the prod.ucts of rock weathering ilre washed-away as quickly as
they form on the steep (or' vertical) slopes. Sugarloaf at Rio de Janerio
is a farniliar example, and ntany otirers in tire region are sirnilal, extending
north i]lto the drier parts of Brazil as lve]l as to t]re south. A visit to sugartlrat it is bnsicaily a "clcscrt islanrl" isolated in a
Ioaf will clernonstr'rttt
tropical jungle. sugarloaf is barren mouolith (bornhardt) of granite-gneiss
that supports a few specics of brorneliads, orchids, cacti, vellozias, and simis
ilar petrophytes that attach in smail cr.eviccs.This exceeding)y xeric flora
situated. only a ferv tens of feet frorn a tlense tropical rain forest composed
Bactris, Ilatusteria, Caesalpinia, Cassia'
of. Anacard,iunt,, Asteroca,t'Aun1,,
Clusia, Covnbretium, Diospyros'
C'ssampelos,
Cebia, Cecropia, Clt'orisia,
M'imusoyts,Nactand'ra, Ocoteu,
Mi'mosa,
Ficus, IIelcinia, Inga, Jacaranda,
Psgclr,ortia, Rol'lti'nia, Srtltindus, Simaruba, B'iparuna, and many others
(Mcl-.,ean1919, p. 164-165). Above the level of the lowland rain forest
of
another type of forest commorrly rlevelops: it is more open, is composed
lower
The
(edaphically)'area.
smaller trees, and. Iives in a much drier
rain forest may gracle insensibly into it, though they are often sharply
separated because of the rapid change in soil depth as the slopes steepen'
Pedra
In other parts of tlie nearby area' as on the flat-topped mountain
.bhere are a more xeric type of forest and. associated tussock
d.a vavea,
grassland. They are iscIatedfrom the lower forest by a wail of inaccessible
in a cooler, more equable
|ranite-gneiss (Mcllean 1-919,p.6), and live
climate.

&"

The si
through
normal
interior
the rain
tion, and
decrease
developc
where ca
broacl pl
eattinga
sumac, c
deciduou
forrned.
is exceed
a hamme
Iligh, sti
Janerio i
plants e:i
is trappe
occasiona
This ar
seasonal
cipally tr
such rap
no moist'
factor in,
wise Jlre'
Slvador
blankctci
soil rvhic
on thc bf
rain forcl
hi)is of r!
caa.tinga.
the exact
the water
the rvate:
marily fr
forest to
Apalt
terI'anes
wpll

lffo

rrrystaJlirr
pr..diplain
times, mr

DDAPIIIC

re, highinibarren
a.dclition,
rli northsavanna
iscussion
,rtittental
nibarrett
1; and is
hat simiroughout
hat tltese
volution,
en other
er 2,000
ft deeP
j abriupt'
rpervious
r steePer
ies occur
rt. These
nd resis:ickly as
r Jarrerio
xtending
io Sugar,ted in a
ite-gneiss
and simic flora is
somposed
,,, Cassia,
liospyros,
t, Ocotea,
iy others
Lin forest
nposedof
lhe lower
r sharplY
J Steepen.
rin Pedra
d tussock
accessible
e equable

AITIDITY

AND

ANC;IOSPIjIiM

llVOLUTION

313

The significance of dry edaphic sites becomesapparent as one progresses

through the region and realizes that they are not rare but make up a
normul part of the landscape. 1,'urthermorc, as one rnovcs to the drier
exa.mrlrl,in coasta,lBhia
interior these sitcs beeomemore flcqttcnt. .11'tlr:
(60-70
inches) yearly precipitathe rain forest receives about 1,000 mm
tion, and the soii is very deep. About 20-30 miles inland precipitation has
decreaseclto 600-700 mm (30-35 inchcs), atrrl in that arca tlrere is a welldevelopcclcaatinga. The soil is scarcely 7-10 cm (3-4 inchcs) deep in swales
rvhere caatinga is dcnsest, bt is thinlter to abscnt on the slopcn of the
broad plains or on the crests of lor,v hills and ridges. In such areas the
cattinga is semi-open and inhabits a rock platform rvith bromeliads, palms,
sunrac, cacti, aca(lia, zizyphus, atl similar plants-bgtlt evcrgr<lcrLanrl
ileciduous-growing out of roch crevices whele a little soil has lodged or
formed. The exposed rock is entirely fresh, has poor platy joiriting, antl
is exceedingly harcl-emitting a sharp bcll-like ring wlten rounclerl rvith
a irammer. The transition to soil is abrupt, taking place in scarcel.v2 mm.
Ifigh, steep-sided borhardts that are counterparts of Sugarloaf at Rio de
Janerio arc scattererl in the area. They are bare and support onlv a ferv
plants except at their cooler (solttr-facing) bases r,vhere sufficient runof{
is trappecl in the gr:us and shallow soil to support a dense caatilga. or,
shoTt-trce forest.
occasionrll.y,
moistttre irrlancl or to its
This arid flora is not due entirely to clecreased"
seasonalconcentration. As Smith (1945, p.298) observed,it "is due principaily to tlie impervious nature of thc frock] rvhich sheds t'rter rvith
such rapidity as to cause floods in pcriods of heavy rainfall and retains
no nroisture for fplants during] peliotls of r'lrought." Thrt this edaphic
factor increases the degree of aridity far bcyond that rvhich lvould otherrvise prevaii is consistent with relations observed along tlie roatl frorn
Slvador to Jacobina. llhe coastal regign which suppolts rain forest is
blanlieted by Cretaceous and Tertiary sedirneritary rocks that have a deep
'l'hc setlimentary cover laps out
soil u'lrich is under intellsc cultivatitill.
on tlrc basement about 25 rniles inlancl. It is thcre that lelict patches of
rain forcst quicirly disappear, intense cultivation ceases,and the rolling
hills of Archcan basementrock with only a fcw millimeters of soil support
caatinga. The change is abrupt ancl takes place lvithin scarcely a mile;
the exact transitiou is obscured now by centuries of land use. In any event,
the rvater-retaining properties of the sedimentary series, as compared. with
the Nater-shedding properiies of the crystalline basement, account primarily for the marked. and rapid. change fi'om a mesophytic evergreen
forest to the caatinga, which is adapted to severe aridity.
Apart from this landsctrpe,which is rvell dr:r'ulopeclin Brazil, similar
te.-anes occur elservilere in South Arucrica rrtl on otltcr eontirtcuts as
rvell (fig. 1) . These ancient landscapcs rvhich hrvebeen barved into old
crystalline loclis are lepl'esented rlow by dcfol'llctl and uplifted broad
pecliplains of earlier erosiott cycles. lTntil colnparatively lecent geologic
tirnes, much of tllis terl'ttrewas lelatively lorv a,ltrl wa.s interluptr:cl bV

el ^
ua=

THE

AII]IIiICAN

NATUITALIST

croslon res
rlomes,rid
and had r
Locally, h
to those ex
Sahara, tl
Australia.
Althoug
metamorpl
recailecl tl
Mountain)
vcr:y arid
Ilcucheria
Yucca, w)
sen1ibarre
r'yhercthc
ous harcl$
those fron
also probt
have been
qr

arninr

increasing
cvolved (l
rocks.
From h
cstablishc
floor spre
continentn
tuaterl by
late Meso
(1) Jura
IrIurasia
Gonclwan
ino

in

rp

wJiich mc
to threo
most extr
(5) intcn
irro fnr

ili rll art

ltc, 1.-21. I)istrilutior of shieltl i:llcas ilr tho SotLtltcrrt I"Iemisphclc (fronr
KinS 1967). Ilxposures of younger bascment rocks rc lot includcii' Edaphic
deserts ancl semideserts have been preset in eaeh area duiug angiosperm
(from
history. n, Distr.ibution of shield ].cs irl the Northclll I{crlisp}torc
Iting 1967). Exposures of younger bascrnent rocks not shorvn. Eclaphic deserts
and semirlescl.ts llave been present i}l cach area clul'ing rngiospet'm history.

rrrgulittc
cyclcs iI'
.orn nri n o

of brselr
rYailablc

EDAPI{IC ARIDITY

AND ANGIOSPERM I]VOI'UTION

3l l

erosion residuals of harcl basement rocl(s that formed scatteled bornhalctts,

d.omes,ridges, and. stripped. low surfaces that wele barren to semibarren
and. hrtLonly thin soil (sec il]ustrations in l(ing 1967; willis 1936).
I=rocally,higher mountains cornprised the cores of ancient ranges similar
to those exposetl lrow as thc Tibctsi and l{oggar Mourltains of thc sriuthcrrt
central
Sahara, the Guiana llighlaniis, and the Macdonnell Range of
Australia.
youngcr
Although tire regions meiltioned rc of allcicnt age, arels of
rnetamorphic and granitic rocks also have dry edaphic sites' Here may be
(Stone
recallecl the numerous granite-gneiss domes scattered fronr Georgia
a
l\{ountain) to Virginia at altitudes near 1,500-1'800 ft' They support
l.er.) aricl flora inclucling spccies of Alliunt,, Arenaria, Aster, Btt,Lltostylis,
Ileu,clteria, Hypericnw,, Panutnt, L'oa, Opuntia, Sed'ttrtt, Yiguert, ancl
Yrtcco, rvliich have lveste1'n1a,thcr than castcrn rffirlitics'Thcy livc orl
semibarren to barren granite-gneissSlopes,chiefly of southerly cxposule)
\rhere they rrewithin a ferv te[s of rncters of the rich mcsopllvtic deciduattd
ous hardlvood forest. The dorrelandsof the soutlieu Sierra Nevacla,
havc
Califollia,
Baja
into
those from central San Diego County southward
tltc\r
.lso pr.obably aii r critical role in cvolutiort irr that I'egiott bc<raltsc
to
trerlcl
the
area,
have been exposed since the iate cetaceous. rlL tltis
to
rvithstancj
p)ants
able
greater aridity during the Tertiary has selectecl
trx rvir
I'c"r.'nri,rg clro'gt. Jtt seems hig1y probabJc tat t. lrew
of
basetrent
evolvecl ciid so in local clry sites provicle<l by exposeclslopes
roclis.
more firrnly
From his revierv of the landscapes of the earth, ancl as is now
suppol'ts oceanestablishedon thc basis of ncw additional evidencewhich
floorspreaililrg,I(irlg(1950,196?)lrass]iowntlrattheplainlantlsoftlre
cycles purlccontinents reveal a history of rjenuclational ancl clepositional
Frorn
equilibra'
ilew
established
tuatecl by brief tectonic episoclesthit have
6)
p.
5l6-41
(King
1967,
late Mesozoicto r.ccenttimc this history includcs
(Gondrvarran
alld
(1) Jurassic-Early Creticcousrviclespreadplanation
(postLaurasian) ; (2) partial planation during the Late Cretaceous
cycle resultGondt'ana and post-Laurasian); (3) a long early Cenozoic
frorn
surfaee
ancestral
the
represent's
ing in ver)' smooth planation tha't
(4)
one
can'ed;
subsequently
rvhich most of the world's scenery has been
(Miocene)
is
earliest
to three late cenozoic partial cycles, of which the
a'n11
bclrlngs;
tnost cxt0nsive antl to rvhich rtrost of lJrc lvolld' sccncrv
glaciation' Allorv(5) intense Quaternary orogeny rvith valley cutting and

: (fron
0daphie
iospcrtu
(from
aleserts
,'y.

ingforlocalvariation,themajorpatternofcycliclanclscapcsseemssimilar
$'hich opcrates to
ilr rll aleas an<l suggcsts a global ctltltrol liy tcctonism
evolutirlr, these
of
r:egulatelanclscapey goo"r^irg base levels. Tn teuns
provide the basis for
c),clesare significaut becausecleforrutionallcl uplift
exposing new tracts
Iernoving the overlying seilimentarV forlriations, thus
as
lionetlr areas l']rich beconc
()f baserne[t ter,r.ailrtlrat rrray bc }t,gariled
t ntl cxploitrttiotr'
a v l i l a l l l c f o r p l a r r t o c c u p a l . i o ta

316

THI] AMEI],ICAN NATURALIST

EVOIJUTIONABYR,OI,EOF EDAPIIIC AR'IDITY

The geologic and geomorphic eyidence reviewed by lfing (1967) makes

it abundantiy clear that dry edaphic sites composed of hard. crystalline
rocks have been available for plant occupation throughout angiosperm
history in tropical, ternperate, ancl polar regions (fig. 1). Ir'ulthermore,
nelv sites have become available as lands were elevatecl ancl stripped of
sedimentary cover. These areas were exposed graclually and in a patchlike
manner, resulting in a seies of disconbinuolls baserrrentsites prior to the
final stripping of the entire surface (see illustration in Willis 1936). On
other occasions large areas of basement have been flooded by seaways or
buried under extensive floodplains, thus eliminating sites and. many of the
taxa restricted to them.
The presenee of dry edaphic sites of crystalline basement would favor
all the positive features that tend to accelerate evolution under aridity, as
elaborated by Stebbins (1952). Moderate changes in local relief would
accentuate microclimatic differences wherever barren to semibarren surfaces rvee present, antl rvould favor the juxtapositiorr of divergent populations. fn furms of adaptation to clrought, the gradients of aridity in these
edaphic dry areas would be comparable with rain shadows discussed earlier
from the standpoint of rapid evolution (Axelrod 1967).'We can visualize
a series of isohyets around each dry edaphic area, grading from the boders
(humid-subhumid) to more xeric (scmiarid-arid) conditions toward the
center. fn each major climatic zone, therefore, edaphic sites rere available
to taxa adapted to all grades and intensities of rlrought, as weII as to the
accompanying thermal differences (hot, warm, mild, cold) that oecurred
with changing altitude and latitude.
Differences irr terrain, soil depth, anil barren to semibarren sites would
tend to disrupt spatial relations of popuiations and lead to the emergence
of discrete units. In this connection, it is noted again that areas of crystalline basement providing arid sites would. regularly have a spotwise or
patchy distribution during stages of uplift and unroofing of overlying
sedimentary rocks. This rvould result in disrupted. populations and lead
to isolation in areas of slightly or mod.erately different climatic conditions
owing to differcnces in local reiief ancl altitude. Furthermore, differences
in populations with respect to their inherent adaptations to a moderate
d.egreeof drought would. be expected in some cases to have high selective
value for further adaptation to more arid conditions-to live in local
deserts anil semideserts provided by these edaphic sites.
The topographic, climatic, and edaphic (arid) diversity in areas of
crystalline temane, together with local discontinuities in outcrop, seems
adequate to explain the fragmentation of populations and the ernergence
of incipient new trxa resulting froru isolation. Ncr,v populations that were
adapted to live in diverse ways untler conditions of someu'hat greater
aridity would norv be able to cleploy into nearby edaphicaliy arid sites
rvherever there 1l'cre local differences in microclimate, iletcrmined by their

position
baruen ri
ancestral
may be r
with distr
marshes,
I{owever,
the normr
relatively
array of
relatively
of the flc
great div
The pr,
such as t]
tion of pc
semia,rid
logical er
late Tert:
there, to ;
India, Af
the role r
climates r
Transverr
elevateclt
has been
fnasmucb
ephemera
and sumr
the clcser
ranean cl
opportun.
The fo:
climatic r
climates.
mesophyt
tinually
evidence.
to the cli
moister a:
is that w
(and war
mclister a
adapt to
cessfully
o c c ur w i t

EDAP}IIC AIT,IDITY AND ANGIOSPERM EVOLTION

67) makes
crystalline
rrgiosperm
;:thermore,
,ripped of
ra,tchlike
ior to the
1 9 3 6 ) .O n
eaways or
rny of the
ruld. favor
rlidity, as
ief would
rrren surrit populay in these
led earlier
r visualize
're borders
ward the
: available
as to the
. occurred
tes would
3mergence
rf crystal,otwise or
overlying
and lead
conditions
lifferences
moilerate
r selective
: in local
areas of
'op, seems
ernergence
that were
it greater
arid sites
I by their

q1n

position rvith respect to bornhardts, domes, koppies, and semibarren to

barren ridges or slopes as compared with those in adjacent areas where
ancestral or elatecl taxa lived. 'Ihese semixeric and xeric eclaphic sites
may be regariled as pioneer areas, comparable in an evolutionary sense
rvith disturbcd areas (roadsicles,old ficlds, talus slides, riverine floodplains,
matshes, etc.) where evolution of diverse taxa is proceeding rapidly today.
flolr,ever, nerv tax in the latter areas tend to disappcar quickly through
the normal lllocess of plant succession.J3y contrast, arid edaphic sites are
relatively persistent and hence would become well stocked with a great
array of taxa that ileployecl into them from adjacent more mesic (though
relatively tlry) areas. That this has happened is apparent from the richness
of the floras of dry tropical regions, including the desert, and from the
great diversity of adaptive types that live there'
'Ih.e precccling evidence suggests it is not necessary to call on high relicf,
such as the Coast Ranges of California, to play a guiding role in the evolution of populations during most of angiosperm history. Actually, the largest
semiarid. regions have always been in the subtropics (savannas), ancl geological evidence shorvs that relief there was not gleat from Mesozoic into
Iate Tertiary time (I(ing 196?). Furthermore, evolution did. not stagnate
there, to judge from the rich savanna flora of central America, the Guianas,
India, Africa, Brazil, ancl Australia. There is much eviclcnce to show that
the role of topographic diversity in promoting e'i'olution in the semiarid.
climates of California is relrtivelyreccnt: the Coast Ranges as well as the
Transverse Ranges, the Peninsular ll,anges, and the Sierra Nevada were
elevated chiefly during the Quaternary. Since then, evolution in this region
has been primarily among herbaceous groups and at a loltr taxonomic level.
Tnasmueh as the Mediterranean climate to which they are adapted is
ephemeral, most of these taxa will probably disappear as the oceans warm
and summer rains return (Axelrod 1971). Nonetheless, it is evid.ent that
the close colrelation between mountain building, the appearance of Mediterranean climate, and rapid evolution of herbaceous taxa demonstrates the
opportunistic nature of evolution, as Stebbins has stressed.
The foregoing evidence also inclicates that it is not necessary to invoke
climatic change as the stimulus for readaptation of xerophytes to moister
climates. In the first place, since the taxa that have readapted to more
mesophytic conditions are of diverse ages, we would have to call on eonwhich there is no
tinually fluetuating climate deep in the tropics-for
evidence. While Some minor readaptation may have occurred in response
to thc climatic fluctuations of the Qnatcrnary, eviclencc of readaption to
moister aleas at that time has not been documented. The point to be stressed
is that while reaclaptation could occur in response to a trencl to moister
(and rvarmer) climate, retrclaptation coulcl lso ottct.tt'witlrout it. Since
moister areas are always adjaceut to eclaphically dry sites' taxa cou}cl re'
adapt to them as variable populations appearecl which could compete successfully in moister sites in these landscapes, irt whicli arid edaphic deserts
Ftlrtherlnot'e' aS new el'osiort
6ccttr rvitlritl a Stone'Sthrorv of I'airr fol'r.ls1,.

318

TIIE AMERICAN NATURALIST

cycles commenced and splead wideiy acr.osscontinents, numelous areas of

ancient clystalhne terranes wele continually exposecl for occupa,tion by
plants. such areas occurrecl in wet tropical areas, savanna regions, clry
'l'he opportunititrs hcrc
ltrclcooler antl moistcr regions.
l,emperate)ntit,ttr1es,
have at least in palt been pl'ovicled by a "restless earth," exposing the
basement of the sialic crust.
Tire lreceding rclations also lrn'can important bcarirrg on solllc problcrns
raised. by early angiosperm evolution. As suggested earlier (Axelrod 1970,
p. 292), they evident)y evolved in the Arctrean terrane in the uplands of
the rvarmer parts of Gondvanaland. Climates were seasonally dry, as demonstrated by saline deposits ancl aeolian sandstones fr:om the surrounding
Iowlands, and by geologic evidence which shows that arid edaphic sites
occurred throughout the areas of Arehean crystalline terrane. Tn view of
the diversity of relief, climate, and edaphic conclitions, carly angiosperm
populations would naturally be broken up into diverse scattered gloups.
Different adaptive mocles,defined by the nature of the root system, water
requirements, leaf size, leaf thickness, a1cl numeros other adaptive featules, can result in selection for very clifferent conditions in localized areas'
As a result, many very divergent taxa probably could arise in a brief time'
some of thesemay haYebeen unique adaptive typcs-and it is infelred that
some have persisteclin scarcely modifiecl form, as exemplified by members
of the Aizoaeeae, caetaceae, crassulaceae, Didieraceae, Fouquieriaceae,
Euphorbiaceae (tlibe Euphorbicae), and many others'
As Gondrvanaland broke up during middle cretaceous and later times.
to
there was a general trend to moister, milcler climate as the ratio of sea
Somc
widely'
spreacl
seas
Cenomanian
the
and
as
land. surface increasecl
ciitaxa that hacl evolvecl in drier areas probably then adapted to moister
no
they
though
even
features
ancicrrt
mates and retained some of their
dry
in
p'43),
(1952,
noted
has
Jonger had "need." for them. As Stebbins
prirnitive
areas leaoes of ISatthinia with sepalate leaflets appear to be more
.rvith
large-leaved
forest-and
bifid lalge leaves in the rain
lhan Ba,uhinia
(derivedg) speciesof Bau,hinia are iu rocks of Cenomanian age' Mcljean
occu'"ence of C'h'orisia
ifSfO, p. 167) also com'retrted.o' the atiornalous
(thorny trunk and branches) in t.he rain forest near Rio de Janerio' and
'
chorisia may also haYe ailapted to moister areas during the cenomanian,
a
though the fossil record does not yet provicle eviclence for it. Although
from
shift
trend to increased rnoisture woulcl be effcctive in promoting a
whendry to moister envirouments, readaptation also may have occurred.
morphoever suitable combinations of new aclaptive features (physiologic'
side in
by
sicle
cxist
may
sites
dry
and
humid
for
arose,
logic, or biotic)
ancient crystalline telranes.
tirnes'
As moister clinratcs sprcacl rviclcly cluig Ceotnania lrfl latcr
extinct
became
cloubt
no
to
drought
adapted
nllmerous taxa that lrrdbeen
to have
as their rreastended to shrirLk ancl disappear' But some appear
were
which
sites
e<laphic
drv
tlle
irr
lersistetl,arrtl they rllI)haYe r-ltlricso
unique
these
wltel'c
Ll'eas
]atitr1es,
,idespread throughot lorver to mitftllc

ItD

forms still oc<

Cretaccousant
than they are
thcn suppoltct)
Baja Californi
the Kalahari.
areas, anil in 1
probably rvere
clomes, ancl ri'
climates began
taxa in these a
expanding dry

Ancient terr
sided mountai
dry edaphic si
enabiing edap
savanna ancl r
mates.
Similar dry
history. Their
moisture have
the origin of r
These aricl
readapted to r
as in Cenomal
local refugia i
persisted.to th
Inasmuch. ai
the preceding
the fossil plan
that are alrea
adaptive typei
edaphic sites r

'Ihe present
my study of N
National Scier
lived in volca
explain some
The manus<
Jtavcn, a.rrrlG.

EDAP}IIC AIIIDITY AND ANGIOSPERM]]VOLUTION

ous areas of
cupation bY
regions, dry
.unities here
xposing tlie
nc problems
relrod 1970,
uplands o{
lry, as demsurrounding
dapliic sites
ln vlew or
angiosperm
rred grouPs.
'stem, water
rlaptive feaalized areas.
l brief time.
nferred that
by members
rquieriaeeae,
lnter times,
lio of sea to
irlcly. Some
moister clirgh they no
43), in drY
re primitive
Iarge-leaved
qe. l\{cliean
of Clt'orisia
Tanerio, and
)enomanian,
Although a
r shift from
ulreil whengic, morphoe by side in
later times,
came extinct
rear to have
rvhich were
these uniqlre

319

forms still occur today. The fossil record shows that during the Late
Cretaceor:sancl early Tcrtiary, the trcsent rlry regions 'wele more moist
than they are today (Axelrocl 1950). Alcas of the present tropical cleselts
thcn supported forests, a,sinrjica,tcd liy tlrc fossil recortl in sotltllcrn Alizona,
Baja California, coast-ccntral l'eru, rvcsl,ct'nTndia, the Saharan region, and
the l(alahari. I)uring tle Late Cretaceous ancl early Tertiary in these
arcas, and irr borclering lcgions as wrrll, plants of rrrortl arid rcquirenrcnts
probably were confined chiefly to clrier eclaphic sites provided by hills,
d.omes,and ridges composed of haril, rcsistant crystalline rocks. As clry
climates began to expand following the carly Eoeene, the rclict xeromorphic
taxa in these arid sites, as well as ne\\, taxa, could thon spread out into the
expanding dry areas.

SUT[I[ARY

Ancient terranes of granitic and tnetamorphic rocks regularly form steepsidecl mountains (bornharclts), domes, ridges, and plains. They provicle
dr' sl-ph'c sites in elimatic regios artging froln wet tropical to rlesert,
enabling eclaphic deserts to occur in proximity to tropical rain forest,
savaltrta arrcl otlier types of vegetation, includirrg tlrr.sein temperate climates.
Similar dry edaphic sites have becn present throughout angiosperm
histoly. Their scittered clistribution arrd steep gradients of decreasing
lloisture have proviclcd environmcntal opportunitics that have cncouraged
the origin of taxa adapted to varying graclcs of drought.
Tlrcse at.icl sitcs uray lso havc scrvctl ls loci frorn wlrich sotno ttxa
reaclaptedto nearby moister sites. IJuring pcriods of climatic amelioration,
as in cenomanian ancl later times, dry eclaphic sites appear to have been
local refugia for unique taxa adapted to clrought whose ilcccnclants have
persisted to the present.
fuasmuch as evolutionarv rate woulil tend to accelerte in local clry sitcs,
the preceding lelations nlay account fol thc rarity of "missing links" in
the fossil plant record, for the "sudden appearance" in the record of taxa
that are alread"y flily eyolved, ad for the occurrence today of unique
adaptive types that hl,veno fossil record becausc they developed in rlry
edanhic sites remote from areas where most of the fossil recorcl accumulated.

ACKNOWLIIDGMITNTS

The present pa,per is an outgrolvth of problems of paleoecology raised by

my study of Miocene floras in Nevada, a resealch project supported by the
I.{ational Science }'orurdatiou (grants GB 7480' GI} 19844)' Those flor:as
lir,ed in volcanic terranes that had dry edaphic sites, which appear to
explain some of the unusual paleoecological relations there'
i.h. *ur,orcript has been revierved critically b.v l)otrld liyhos, Prrl,e
Ravc.n,antl G. Tredyat.clStebbins, to \vh<llnthanhs treextclt<]cd.

a-a

320

THE

AM]IIIICAN

NATUIi,ALIST

Vol. 106, No.

LITER,AIIURE CITED

Axelrocl, D. I. 1950. nvolution of clcscrtvcgctation in wcstcln NortlL Arnorica. Carncgic

Inst. Washington Pub. 590:215-306.
1967.Drought, tliastrop)risnr,anrl qua,ntunrevolution. .llvolution 21:201-209.
1970. Mesozoic paleogeographyand early angiospelm history. llot. llcv. 36:
277-3L9.
1971.Eistoy of the Mediterraneanecosystemin califoilia. 1 IL Mooney [etl..],
Convergencein stucturo of ccosystcmsin Meditcrl'nciln climatos. SpringerYerlag, Ber)in (in press).
Branner, J. C. 1896.Decompositionof r.ocksin Brazil. Geol. Soc. Amer. Bull. 7 :255-314.
King, L. C. 1950. The study of thc worltlts plainlancls: a new approach in geomorphology. Quart. J. Geol. Soc. T,onclon106:101-127.
1967. The rnorphology of the earth: a stualy antl synthesis of worltl scenel'y.
Oliver & Boycl, Lontlon. 699 P.
Mclean, R. C. 1.919.Studies in the ecology of tropical rainforest: with speeial refeecc
to ths forests of South Brazil. J. I'col. 7:5-54, I2l-172'
smith, L, B. 1945. The vegetation of Brazil, p, 297-902.1 F. Yertloorn [ccl.], Plants
and plant science i:r Latin Ameica' Chronica Botanica, Waltha^rn, Mass'
Stebbins, G. L. 1952. Aidity as a stimulus to evolution. Amcr' Natur' 86:33-44'
willis, B. 1"936.East African plateaus ancl rift valleys. carnegie Inst. Washington Putr'
470. 358 p.

SEX R,AT

The top
of attenti
cording to

selective
males anci
parents (e.
apparent,

1961;
Osorio,
burt 1969;
Wildish I
Accordi
natural
two sexes.
progeny o
be the

note that
create a d.i
as differen

periolof
lreigh (
the period
female bi
factor to

Restricted
migration
have all b
tlie sex
The
pr0xl

openrngs,
compre

1 9 6 8 ;O s o r
* Coltribu