morphological characterization and relationships of … and... · morphological characterization...

MORPHOLOGICAL CHARACTERIZATIONAND RELATIONSHIPS OF WILD TOMATOES

(SOLANUM L. SECT. LYCOPERSICON)

Iris E. Peralta and David M. Spooner

ABSTRACT. Wild tomatoes (Solanum L. sect. Lycopersicon (Mill.) Wettst.) are native towestern South America. Different classifications have been based on morphologicalor biological species concepts. Molecular data from mitochondrial DNA restrictionsites, nuclear and chloroplast DNA restriction length fragment polymorphisms, andmost recently gene sequences of the single-copy nuclear GBSSI or waxy gene, alsohave been used to examine species relationships. This study is a companion to theGBSSI gene sequence study of the same accessions. It provides the first explicit useof morphological data to examine distinctness and relationships of all 10 wildtomato species (including the newly described S. galapagense), with a concentrationon accessions of the most widespread and variable species, S. peruvianum. Pheneticand cladistic analyses largely support the nine species outlined by the latest treatment by C. Rick, but demonstrate the distinct nature of the northern andsouthern Peruvian populations of S. peruvianum, and suggest that they may represent distinct species.

Key words: phylogeny, Solanum sect. Lycopersicon, tomato.

10

228

A FESTSCHRIFT FOR WILLIAM G. D’ARCY

Wild tomatoes are mainly charac-terized by anthers with sterileappendages, laterally con-nivent forming a flask-shaped

cone. All species are native to western SouthAmerica and distributed from central Ecuador,through Peru to northern Chile, and in theGalápagos Islands, where two endemic species,S. cheesmaniae and S. galapagense, grow(Darwin et al., 2003). Wild tomato species growin a variety of habitats, from near sea levelalong the arid Pacific coast to over 3300 m inthe numerous valleys of the western side of theAndes (Rick, 1973; Taylor, 1986). The wild ances-tor of cultivated tomatoes, S. lycopersicum, ismore widespread and perhaps more recentlydistributed into Mexico, Colombia, Bolivia, andother South American countries (Rick & Holle,1990). Wild species provide a wealth of usefulgenetic traits to improve cultivated tomatoes.The traditional breeding for pure lines in thecrop has contributed to its narrow genetic base(Stevens & Rick, 1986). All wild tomato speciesare diploids (2n = 2x = 24) and can be crossed(but sometimes with difficulty) to the cultivatedtomato. They are of great use in breeding pro-grams as sources of disease resistance and agro-nomic traits (Esquinas Alcazar, 1981; Laterrot,1989; Rick, 1982a, 1986a, 1987; Rick et al., 1987;Stevens & Rick, 1986). The tomato also serves asa model organism for genetic studies (Tanksley& McCouch, 1997; Hay et al., 2004).

Wild tomatoes exhibit great differences in mor-phological characters, mating systems, andhabitat preferences. In their natural habitats,wild tomatoes probably behave as annualsbecause frost or drought kills the plants afterthe first growing season, but they can be bien-nial or perennial herbs (Müller, 1940). All toma-to species are pubescent, and trichome typesand density are valuable characters to distin-guish species. Stems are slender and herbaceousthroughout, but also can undergo secondarygrowth at the base. Habits vary from erect toprocumbent to decumbent. Leaves are alter-nate, imparipinnate to bipinnate, with 2 to 6opposite or sub-opposite pairs of petiolate or

subsessile leaflets, and sometimes with addi-tional small sessile or subsessile interjectedleaflets. Leaf dissection distinguishes subspeciesor varieties of S. peruvianum. The basic inflores-cence includes monochasial, dichotomous, andpolychotomous cymes (Luckwill, 1943). Somewild tomato species develop inflorescencebracts and pseudostipules at the base of theleaf petiole. Flowers are typically yellow, thecalyx is divided in five lobes that are accrescentabout the fruits, and the corolla is stellate orstellate-rotate. The five anthers have an elon-gated sterile appendage at the apex and arelaterally connivent forming a flask-shapedcone, except in S. pennellii. Fruits are fleshyberries with two or rarely more locules, variablein size, shape, color, and pubescence.

Müller (1940) and Luckwill (1943) produced thetwo most recent and complete taxonomic treat-ments of wild tomatoes based on morphologi-cal concepts, and treated them underLycopersicon (Fig. 1). Müller (1940) divided thegenus into two subgenera: (1) Eulycopersicon C.H. Müll., with two species possessing glabrous,and red- to orange- to yellow-colored fruits; flatand silky pubescent seeds; bractless inflores-cences; and leaves without pseudostipules; and(2) Eriopersicon C. H. Müll., with four speciesbearing pubescent green or greenish white toyellowish and purple tinged fruits, frequentlywith a dark green, lavender, or purple stripe;thick glabrous seeds or pilose only at the apex;bracteate inflorescences; and leaves usuallywith pseudostipules. Müller (1940) alsodescribed L. glandulosum C. H. Müll. and classi-fied the highly polymorphic L. peruvianum intotwo varieties: var. dentatum Dunal and var.humifusum C. H. Müll.

Luckwill (1943; Fig. 1) adopted the two subgen-era recognized by Müller (1940) and proposed aphylogeny of Lycopersicon. He hypothesizedthat the two subgenera might have evolvedfrom an ancestral simple form characterized byimparipinnate leaves with 5 to 7 entire leaflets,few interjected leaflets, and probably no sec-ondary leaflets, unbranched inflorescences, and

MORPHOLOGICAL CHARACTERIZATION AND RELATIONSHIPS OF WILD TOMATOES

229

Figure 1. Comparison of classifications of Solanum sect. Lycopersicon. The numbers in parentheses represent numbers ofinfraspecific ranks (subspecies, varieties, and forms). The lines connect synonymous taxa.

Figure 2. Abstracted cladistic results of the 65 accessions of the 9 tomato species and 14 outgroup taxa examined in thephylogenetic analysis of the GBSSI gene sequences by Peralta and Spooner (2001). Nomenclature of Solanum subg. Potatoeaccording to Child (1990). We here informally label wild tomatoes in the clade containing all members of “subsect. Lycopersicon”(S. lycopersicum to S. pennellii) to provide a coordinate name for their formally described outgroups in section Juglandifolium(Rydb.) Child and subsection Lycopersicoides Child. Our use of these ranks may change in our monograph of tomatoes and theirrelatives (Peralta et al., in prep.). Numbers indicate bootstrap values, and decay values are indicated between parentheses.

230


undeveloped pseudostipules. Two different lin-eages diverged from the ancestral forms, onecharacterized by fruits with carotenoid pig-ments and the other by green fruits with antho-cyanin pigments. Luckwill (1943) agreed withMüller (1940) in the circumscription of sub-genus Eulycopersicon, but within Eriopersiconhe considered L. pissisi a distinct species, with L.peruvianum var. humifusum C. H. Müll. as itsbasionym (Fig. 1). He also proposed differentinfraspecific categories in L. esculentum, L.cheesmaniae, and L. peruvianum. Lycopersiconcheesmaniae, with yellow to orange fruits,bractless inflorescences, and leaves withoutpseudostipules, apparently was misclassified inthe subgenus Eriopersicon by Müller (1940) andLuckwill (1943).

D’Arcy (1972) treated tomatoes in the genusLycopersicon but did not address subgenericrelationships. More recently, Child (1990, Fig. 1)placed tomatoes in Solanum subg. Potatoe (G.Don) D’Arcy, sect. Lycopersicon (Mill.) Wettst.,subsect. Lycopersicon, mainly characterized byanthers with sterile appendages and laterallyconnivent forming a tube, and classified themin three series: Lycopersicon, Eriopersicon (C. H.Müll.) Child, and Neolycopersicon (Correll) Child(Fig. 1). The first two series correspond to thesubgenera Eulycopersicon and Eriopersiconused by Müller (1940) and Luckwill (1943).Series Lycopersicon includes herbs with few orno pseudostipules, ebracteate inflorescences,and fruits with carotenoid pigments, andEriopersicon includes herbs to subshrubs withpseudostipules, bracteate inflorescences, and fruits with no carotenoid pigments (fruitspure green or greenish white, dark green topurple striped). Series Neolycopersicon onlyincludes S. pennellii (Correll, 1958), a specieswith curved, markedly unequal length anthers loosely coherent and without sterileappendages. Child (1990) also proposedSolanum sect. Lycopersicon (Mill.) Wettst., subsection Lycopersicoides Child, and sectionJuglandifolium (Rydb.) Child as the closest rela-tives of subsection Lycopersicon.

Most recently, a new orange-fruited species oftomato was recognized, S. galapagense S.Darwin & Peralta (Darwin et al., 2003). Withintomatoes it shares its orange fruit color onlywith S. cheesmaniae, but is clearly distinguishedfrom it by a number of leaf and flower traits.This species, formerly recognized as S. cheesma-niae f. minor, is clearly related to the otherspecies with carotenoid pigments and is includ-ed in the present study.

Mating systems have played an important rolein the evolution of wild tomato species, varyingfrom allogamous self-incompatible, to faculta-tive allogamous and self-compatible, to autog-amous and self-compatible (Rick, 1963, 1979,1982b, 1986b; Table 1). The self-incompatibilitysystem in tomatoes is gametophytic and con-trolled by a single, multiallelic S locus (Tanksley& Loaiza-Figueroa, 1985).

The self-incompatibility system has shownstrong relationships between the extent of out-crossing and allelic polymorphisms, floral dis-play, and degree of stigma exsertion in wildtomatoes. Rick (1982b) investigated the geneticbases of self-compatibility, self-incompatibility,and flower characters, using interspecifichybrids between the self-compatible (SC) S.pimpinellifolium, used as recurrent parent, andthe two self-incompatible (SI) species, S.habrochaites and S. pennellii. He postulatedthat three independent genetic phases, mostprobably regulated by different unlinked genesor gene complexes, are essential for successfulfunctioning of the self-incompatibility system.These genes are operating on (1) prevention ofself-fertilization, (2) changes in the flowerorgans to ensure cross-pollination, and (3)development of secondary flower characters toattract pollinators. He concluded that the evo-lution of a mating system in wild tomatoes hasoccurred from self-incompatibility, as ancestralcondition, to self-compatibility, and probablynever reversed to self-incompatibility. Changesfrom self-incompatibility to self-compatibilityare expected to arise frequently and independ-


ently (Rick, 1982b). This trend has been found inS. habrochaites and S. pennellii, both with self-incompatible and self-compatible populations.The self-incompatible populations occupy thecenter of their species geographic distributions,and have higher genetic variation, larger flowerparts, and exserted stigmas. Self-compatiblepopulations occur toward the northern andsouthern edges of S. habrochaites and S. pen-nellii distribution, have less genetic variation,smaller flower parts, and little or no stigmaexsertion (Rick et al., 1979; Rick & Tanksley,1981). The change from self-incompatibility toself-compatibility has been reported in only onepopulation of S. peruvianum (Rick, 1986b).

In the self-compatible species, the extent of out-crossing and genetic variation is also related tofloral display and degree of stigma exsertion.Marginal populations of Solanum pimpinellifoli-um are highly autogamous with little or no genet-ic variation, small flower parts, and little or nostigma exsertion, while the central facultativeallogamous populations have high genetic varia-tion, larger corollas, and marked stigma exsertion(Rick et al., 1977). A comparison of different S.pimpinellifolium genotypes in experimental plotsin Peru showed that different outcrossing ratescould be largely attributed to differences in floralcharacters, especially the level of stigma exsertion,rather than to differences in numbers and typesof pollinators (Rick et al., 1978).

Two closely related self-compatible species, S.chmielewskii and S. neorickii, illustrate anotherexample of changes in flower characters associat-ed with outcrossing and genetic variation.Solanum neorickii is exclusively autogamous withlow intra-population genetic variation and smallflowers with stigmas included in the anther tube.In contrast, the facultative allogamous S.chmielewskii exhibited higher levels of heterozy-gosity, larger flower parts, and exserted stigmas.Rick et al. (1976) postulated that S. neorickiievolved from S. chmielewskii. All populations of S.chilense are self-incompatible. The related speciesS. lycopersicoides, S. sitiens, S. ochranthum, and S.juglandifolium are exclusively self-incompatible.

In contrast to the morphological species con-cepts of Müller (1940) and Luckwill (1943), Rick(1960, 1963, 1979, 1986b) recognized speciesand proposed a supraspecific classificationbased primarily on the biological criteria ofmating systems, cytology, genetic variation, andcrossing relationships. He recognized ninespecies of Lycopersicon (Fig. 1; Rick, 1979),including two new species, L. parviflorum and L.chmielewskii (Rick et al., 1976), and L. pennellii(Rick, 1979). Furthermore, L. chilense, a speciesdescribed by Dunal (1852), was revalidated byRick and Lamm (1955), and L. glandulosum wasincluded as a synonym of L. peruvianum.

In order to determine the crossing relationshipsnecessary for his tomato breeding program,Rick (1963, 1979, 1986b) hybridized numerousaccessions of all tomato species in all possiblecombinations and found two major crossingcomplexes. The Esculentum complex includedseven species, mainly self-compatible and inter-crossable (Fig. 1; Table 1). Lycopersicon hirsutumand L. pennellii typically are self-incompatible,but with some self-compatible populations.Three species of the Esculentum complex havemostly glabrous, red- to orange- to yellow-col-ored fruits (L. cheesmaniae, L. esculentum, L.pimpinellifolium), while the others have pubes-cent, green fruits (L. chmielewskii, L. hirsutum,L. parviflorum, L. pennellii). Hybrids wereobtained in all combinations, in spite of thepresence of some unilateral incompatibility. ThePeruvianum complex, on the other hand,included the self-incompatible species L.chilense and L. peruvianum (Fig. 1; Table 1),both with pubescent green fruits. The barrierbetween these two complexes can be brokenonly by embryo rescue (Rick, 1979). The chro-mosomes of the F1 hybrids of successful inter-complex crosses generally showed completepairing at pachytene, and normal chiasmataformation and chromosome segregation atanaphase during meiosis. Rick (1979) concludedthat speciation in wild tomatoes took placemainly via gene substitution and, to a minorextent, by chromosomal differentiation. He alsopointed out that natural hybridization is very

231


TA

BL

E1.

232

Acc

essi

on

s o

f w

ild t

om

ato

es a

nd

ou

tgro

up

s, e

xam

ined

fo

r m

orp

ho

log

ical

(th

is p

aper

) an

d G

BB

SI g

ene

seq

uen

ce v

aria

tio

n (

Pera

lta

& S

po

on

er, 2

001)

.V

ou

cher

s ar

e d

epo

site

d a

t B

M, M

ERL,

PTI

S, a

nd

WIS

. We

use

d t

he

Sola

nu

meq

uiv

alen

ts o

f th

e Ly

cop

ersi

con

spec

ies

as r

eco

gn

ized

by

Ric

k et

al.

(199

0),

wit

h t

he

add

itio

n o

f th

e va

riet

ies

and

fo

rms

of

S. h

abro

chai

tes,

S. p

enn

ellii

, an

d S

. per

uvi

anu

mas

list

ed in

th

e d

atab

ase

of

ger

mp

lasm

at

the

C. M

.R

ick

Tom

ato

Gen

etic

s R

eso

urc

e C

ente

r (<

htt

p://

tgrc

.ucd

avis

.ed

u>

).

Map

Acc

essi

onB

reed

ing

Taxo

naL

ocal

ityb

Num

berc

Col

lect

ord

Syst

eme

Loc

alit

yf

Ing

rou

p, S

ola

nu

mse

ct. L

yco

per

sico

n

chs

G

LA16

6 U

nkn

ow

n

Au

t-SC

Ec

uad

or.

Gal

ápag

os

Isla

nd

s: S

anta

Cru

z, B

arra

nco

, N o

f Pu

nta

Ayo

ra,

1 km

N o

f Pu

nta

Ayo

ra, 5

0 m

chs

G

LA14

50

Un

kno

wn

A

ut-

SC

Ecu

ado

r. G

aláp

ago

s Is

lan

ds:

Isab

ela,

Bah

ia S

an P

edro

, 1/2

km

fro

m c

oas

t, 1

5 m

gal

G

LA

317

Un

kno

wn

A

ut-

SC

Ecu

ado

r. G

aláp

ago

s Is

lan

ds:

Bar

tolo

me,

15

m

chl

29

LA19

30

Ric

k et

al.

3172

A

llo-S

I Pe

ru. A

req

uip

a: Q

ueb

rad

a C

alap

amp

a, R

ío A

carí

, 500

m, 1

5°39

’S, 7

4°39

’W

chl

36

LA19

63

Ric

k et

al.

3204

A

llo-S

I Pe

ru. T

acn

a: R

ío C

aplin

a, 1

3 km

W o

f Ta

cna,

3 k

m f

rom

Pan

amer

ican

Hw

y.,

tow

ard

Bo

ca d

el R

ío, 2

00 m

, 18°

14’S

, 70°

33’W

chl

38

LA28

84

Ric

k et

al.

7609

A

llo-S

I C

hile

. An

tofa

gas

ta: A

yavi

ri, 2

2°02

’S, 6

8°07

’W

chm

25

LA

1306

(PI

379

029)

Ric

k et

al.

390

Fac-

SC

Peru

. Aya

cuch

o: T

amb

o, 1

2°56

’S, 7

4°01

’W, 1

km

E o

f Ta

mb

o, 3

100

m

chm

27

LA

1327

R

ick

et a

l. 41

1 Fa

c-SC

Pe

ru. A

pu

rím

ac: S

ora

cata

, 198

0 m

chm

26LA

2663

R

ick

et a

l. 61

01

Fac-

SC

Peru

. Cu

sco

: Tu

jto

hai

ya (

Up

per

Ap

urí

mac

), 25

00 m

lyp

c 5

LA12

26 (

PI 3

7904

5)

Ric

k et

al.

309

Au

t-SC

Ec

uad

or.

Mo

ron

a-Sa

nti

ago

: 3 k

m S

of

Sucu

a, 1

000

m

lyp

c 21

LA

1673

R

ick

et a

l. 19

85

Au

t-SC

Pe

ru. L

ima:

Ñañ

a, 1

1°59

’S, 7

6°50

’W, a

lon

g W

wal

l of

the

Esta

nci

a

hab

11

LA

1353

(PI

365

934)

Ric

k et

al.

437

Allo

-SI

Peru

. Caj

amar

ca: C

on

tum

azá,

265

0 m

, 7°2

2’S,

78°

49’W

hab

24

LA

1928

R

ick

et a

l. 31

70

Fac-

SC

Peru

. Ica

: Oca

ña, R

ío In

geni

o, 7

6 km

fro

m P

anam

eric

an H

wy.

, 266

0 m

, 14°

39’S

, 75°

15’W

hab

g

3 LA

1223

(PI

365

903)

R

ick

et a

l. 30

1 Fa

c-SC

Ec

uad

or.

Ch

imb

ora

zo: A

lau

sí, 2

°12’

S, 7

8°50

’W, 2

200

m

neo

7

LA22

00

Ric

k et

al.

5045

Au

t-SC

Pe

ru. A

maz

onas

: Cho

ipia

co, R

ío S

uche

, 6°1

3’S,

77°

51’W

, 16

km E

of

Chac

hapo

yas,

1900

m

neo

17

LA

247

Och

oa

1017

A

ut-

SC

Peru

. Hu

ánu

co: C

hav

inill

o, 9

°47’

S, 7

6°35

’W


neo

20

LA

1326

(PI

379

033)

R

ick

et a

l. 41

0 A

ut-

SC

Peru

. Ap

urí

mac

: Río

Pac

hac

hac

a, 1

980

m, 1

1°39

’S, 7

6°01

’W

pen

19

LA

1376

(PI

379

017)

R

ick

et a

l. 46

0 A

llo-S

I Pe

ru. L

ima:

Say

án, 1

000

m, A

gu

a Pe

rdid

a (R

ío In

gen

io),

11°0

6’S,

77°

39’W

pen

31

LA

716

(PI 2

4650

2)

Un

kno

wn

Fa

c-SC

Pe

ru. A

requ

ipa:

Atic

o, n

ear

Atic

o fr

om C

hala

alo

ng P

anam

eric

an H

wy.

, 16°

14’S

, 73°

39’W

pen

p

28

LA19

26

Ric

k et

al.

3168

A

llo-S

I Pe

ru. I

ca: 2

5 km

fro

m P

anam

eric

an H

wy.

, 120

0 m

, 14°

38’S

, 75°

04’W

per

6

LA21

85

Ric

k et

al.

5030

A

llo-S

I Pe

ru. A

maz

on

as: P

on

go

de

Ren

tem

a, in

th

e g

ap o

f th

e M

arañ

ón

, E o

f C

orr

al

Qu

emad

o, 4

00 m

, 5°2

6’S,

79°

39’W

per

9

LA21

72

Ric

k et

al.

5017

A

llo-S

I;po

ssib

lyPe

ru. C

ajam

arca

: Cu

yca,

161

km

fro

m O

lmo

s ju

nct

ion

wit

h t

he

Pan

amer

ican

Hw

y.,

SC o

r se

gre

ga-

11°3

1’S,

75°

07’W

ti

ng

SI/S

C

per

9

LA21

63

Ric

k et

al.

5008

A

llo-S

I Pe

ru. C

ajam

arca

: bet

wee

n C

och

abam

ba

and

Yam

alu

c, 2

–5 k

m N

o

f C

och

abam

ba,

180

0 –19

00 m

per

9

LA21

57

Ric

k et

al.

5002

Fa

c-SC

Pe

ru. C

ajam

arca

: Tu

nel

Ch

ota

no

, 23

km N

of

Ch

ota

, 160

0 m

per

11

LA

392

Ric

k et

al.

141

Allo

-SI

Peru

. Caj

amar

ca: L

lalla

n, 7

8 km

fro

m C

ar. P

anam

eric

an H

wy.

, 12

km W

of

Chile

, 900

m

per

h

11

LA21

52

Ric

k et

al.

4997

A

llo-S

I Pe

ru. C

ajam

arca

: Río

Jeq

uet

epeq

ue,

140

km

fro

m P

anam

eric

an H

wy.

, 2 k

m W

o

f Sa

n J

uan

, 215

0 m

, 7°1

7’S,

78°

29’W

per

h

11

LA25

48

PE-2

9 A

llo-S

I Pe

ru. C

ajam

arca

: La

Mu

yun

a, R

ío J

equ

etep

equ

e, 8

70 m

per

10

LA

1396

R

ick

et a

l. 48

0 A

llo-S

I Pe

ru. A

maz

on

as: B

alsa

s, 4

°25’

S, 8

0°17

’W

per

13

LA

2553

PE

-31

Allo

-SI

Peru

. Caj

amar

ca: B

alco

nci

llo-S

an M

arco

s, 2

650

m, 7

°20’

S, 7

8°11

’W

per

12

LA

441

Ric

k et

al.

128

Allo

-SI

Peru

. La

Lib

erta

d: C

erro

Cam

pan

a, W

of

Pan

amer

ican

Hw

y. N

of

Tru

jillo

, 350

m,

5°05

’S, 7

9°17

’W

per

12

LA

1984

R

ick

et a

l. 32

25

Allo

-SI

Peru

. La

Lib

erta

d: O

tuzc

o (

Río

Mo

che)

, 2 k

m o

n s

ide

road

E o

f O

tuzc

o, 2

800

m,

7°54

’S, 7

8°35

’W

per

13

LA

2328

R

ick

et a

l. 51

72

Allo

-SI

Peru

. La

Lib

erta

d: A

rica

pam

pa,

230

0–25

00 m

, 7°4

8’S,

77°

43’W

per

14

LA

1982

R

ick

et a

l. 32

23

Allo

-SI

Peru

. An

cash

: Hu

alla

nca

, Río

San

ta, n

ear

the

hyd

roel

ectr

ic p

lan

t, a

bo

ut

4 km

b

eyo

nd

LA

1981

, 140

0 m

, 8°4

9’S,

77°

52’W

per

15

LA

1626

U

nkn

ow

n

Allo

-SI

Peru

. An

cash

: mo

uth

of

Río

Ru

pac

, 170

0 m

per

16

LA

1360

(PI

365

952)

R

ick

et a

l. 44

4 A

llo-S

I Pe

ru. A

nca

sh: P

aria

coto

, 149

0 m

, 9°3

1’S,

77°

53’W

per

18

LA

111

Ric

k et

al.

41

Allo

-SI

Peru

. Lim

a: S

up

e, a

lon

g P

anam

eric

an H

wy.

at

km 1

58, 5

0 m

, 10°

48’S

, 77°

44’W

per

18

LA

110

Ric

k et

al.

39

Allo

-SI

Peru

. An

cash

: Caj

acay

, 5 k

m W

of

Caj

acay

, 250

0 m

, 10°

10’S

, 77°

26’W

233


per

18

LA

1365

(PI

365

953)

R

ick

et a

l. 44

9 A

llo-S

I Pe

ru. A

nca

sh: C

aran

qu

illo

c, 2

450

m

per

19

LA

1379

(PI

379

018)

R

ick

et a

l. 46

3 A

llo-S

I Pe

ru. L

ima:

Cau

jul,

1500

m, 1

0°48

’S, 7

7°02

’W

per

21

LA

1274

(PI

365

940)

R

ick

et a

l. 35

6 A

llo-S

I Pe

ru. L

ima:

Pac

aib

amb

a, 1

440

m

per

21

LA

370

Ric

k et

al.

115

Allo

-SI

Peru

. Lim

a: H

uam

pan

i Qu

ebra

da,

200

m

per

g

21

LA36

4 R

ick

et a

l. 10

9 A

llo-S

I Pe

ru. L

ima:

9 k

m W

of

Can

ta, 2

100

m, 1

2°35

’S, 6

9°09

’W

per

g

21

LA12

83 (

PI 3

6594

2)

Ric

k et

al.

365

Allo

-SI

Peru

. Lim

a: S

anta

Cru

z d

e La

ya, 2

000

m

per

21

LA

103

Ric

k et

al.

28

Allo

-SI

Peru

. Lim

a: C

ajam

arq

uill

a, o

n b

oth

sid

es o

f d

ry w

ash

, 100

m, 6

°25’

S, 7

9°14

’W

per

22

LA

107

Ric

k et

al.

33

Allo

-SI

Peru

. Lim

a: H

acie

nd

a Sa

n Is

idro

, 5°3

3’S,

80°

49’W

per

23

LA

1339

(PI

365

949)

R

ick

et a

l. 42

3 A

llo-S

I Pe

ru. L

ima:

Cap

illu

cas,

170

0 m

, 13°

39’S

, 12°

45’W

per

23

LA

444

Ric

k et

al.

198

Allo

-SI

Peru

. Ica

: Ch

inch

a, s

ou

ther

n o

uts

kirt

s, 1

00 m

, 17°

20’S

, 71°

00’W

per

24

LA

1305

(PI

379

015)

R

ick

et a

l. 38

9 A

llo-S

I Pe

ru. H

uan

cave

lica:

20

km E

of

Ticr

apo

, 296

0 m

per

24

LA

1910

R

ick

et a

l. 31

52

Allo

-SI

Peru

. Hu

anca

velic

a: T

icra

po

, Tam

bill

o (

Río

Ica)

, 74

km f

rom

Ica,

230

0 m

per

28

LA

1913

R

ick

et a

l. 31

55

Allo

-SI

Peru

. Ica

: Tin

gu

iayo

g (

Río

San

ta C

ruz)

, 15

km f

rom

Pan

amer

ican

Hw

y., 9

00 m

per

28

LA13

31 (

PI 3

6594

6)

Ric

k et

al.

415

Allo

-SI

Peru

. Ica

: Naz

ca, 1

500

m, 1

4°50

’S, 7

4°57

’W

per

29

LA

1937

R

ick

et a

l. 31

79

Allo

-SI

Peru

. Are

qu

ipa:

Qu

ebra

da

Torr

ecill

as, (

Río

Ch

apar

ra),

21 k

m f

rom

Ch

apar

ra, 6

4 km

fr

om

Pan

amer

ican

Hw

y., 2

500

m, 1

5°55

’S, 7

4°07

’W

per

30

LA

448

Ric

k et

al.

202

Allo

-SI

Peru

. Are

qu

ipa:

Ch

ala,

15°

52’S

, 74°

16’W

per

30

LA

1945

R

ick

et a

l. 31

86

Allo

-SI

Peru

. Are

qu

ipa:

Car

avel

í, 71

km

fro

m P

anam

eric

an H

wy.

, 8 k

m b

efo

re

reac

hin

g C

arav

elí,

1940

m, 1

5°46

’S, 7

3°22

’W

per

32

LA

1474

U

nkn

ow

n

Allo

-SI

Peru

. Are

qu

ipa:

Lo

mas

de

Cam

aná,

130

0 m

, 16°

22’S

, 73°

01’W

per

33

LA

1973

U

nkn

ow

n

Allo

-SI

Peru

. Are

qu

ipa:

Yu

ra, 1

6°12

’S, 7

1°42

’W

per

35

LA31

56

Un

kno

wn

A

llo-S

I Pe

ru. M

oq

ueg

ua:

Om

ate

Ag

ricu

ltu

ral V

alle

y

per

34

LA

1954

R

ick

et a

l. 31

95

Allo

-SI

Peru

. Are

qu

ipa:

2–

4 km

W o

f M

olle

nd

o, 5

0 m

, 17°

02’S

, 72°

01’W

TA

BL

E1

CO

NT

INU

ED.

234

Map

Acc

essi

onB

reed

ing

Tax

ona

Loc

alit

ybN

umbe

rcC

olle

ctor

dSy

stem

eL

ocal

ityf


per

34

LA

454

Ric

k et

al.

208

Allo

-SI

Peru

. Are

qu

ipa:

8 k

m S

of

Río

Tam

bo

cro

ssin

g, 1

200

m

per

36

LA

2964

U

nkn

ow

n

Allo

-SI

Peru

. Tac

na:

Qu

ebra

da

de

Los

Bu

rro

s

per

37

LA

2744

R

ick

et a

l. 75

02

Allo

-SI

Ch

ile. T

arap

aca:

So

bra

ya, (

Aza

pa)

, 31

km f

rom

Ari

ca, 4

00 m

pim

2

LA12

37 (

PI 3

6591

0)

Ric

k et

al.

318

Au

t-SC

Ec

uad

or.

Esm

eral

das

: Ata

cam

es, 5

m

pim

8

LA15

81

Ric

k et

al.

1861

Fa

c-SC

Pe

ru. L

amb

ayeq

ue:

Pu

nto

Cu

atro

, 1 k

m W

of

Pan

amer

ican

Hw

y.

pim

23

LA

1606

R

ick

et a

l. 18

86

Au

t-SC

Pe

ru. I

ca: T

amb

o d

e M

ora

, Hu

auca

s, 1

3°28

’S, 7

6°12

’W

Ou

tgro

up

, So

lan

um

sect

. Lyc

op

ersi

cum

sub

sect

. Lyc

op

ersi

coid

esC

hild

lyc

36

LA23

86

Och

oa

1424

8 A

llo-S

I Pe

ru. T

acn

a: C

hu

pap

alca

, 318

0 m

sit

39

LA28

76

Ric

k et

al.

7601

A

llo-S

I C

hile

. An

tofa

gas

ta: C

hu

qu

icam

ata

235

a Sp

ecie

s ab

bre

viat

ion

s [a

nd

Lyc

op

ersi

con

syn

on

yms]

are

:

chl =

S. c

hile

nse

(Du

nal

) R

eich

e [L

yco

per

sico

n c

hile

nse

Du

nal

];

chm

= S

. ch

mie

lew

skii

(C. M

. Ric

k, K

esic

ki, F

ob

es &

M.H

olle

) D

. M. S

po

on

er,

G. J

. An

der

son

& R

. K. J

anse

n [

L. c

hm

iele

wsk

iiC

. M. R

ick,

Kes

icki

, Fo

bes

&M

. Ho

lle];

chs

= S.

ch

eesm

ania

e(L

. Rile

y) F

osb

erg

[L.

ch

eesm

ania

eL.

Rile

y];

gal

= S

. gal

apag

ense

S. D

arw

in &

Per

alta

[L.

ch

eesm

ania

ef.

min

or

(Ho

ok.

f.)

C. H

. Mü

ll.]

;

hab

= S

. hab

roch

aite

sS.

Kn

app

& D

. M. S

po

on

er [

L. h

irsu

tum

Du

nal

];

hab

g =

S. h

abro

chai

tes

f. g

lab

ratu

mC

. H. M

üll.

;

lyc

= S.

lyco

pers

icoi

des

Dun

al [

L.ly

cope

rsic

oide

s(D

unal

)A.C

hild

ex

J.M

.H.S

haw

];

lyp

= S

. lyc

op

ersi

cum

L.;

lyp

c =

S. ly

cop

ersi

cum

L. v

ar. c

eras

ifo

rme

(Du

nal

) D

. M. S

po

on

er, G

. J.

An

der

son

& R

. K. J

anse

n [

L. e

scu

len

tum

Mill

. var

. cer

asif

orm

eD

un

al];

neo

= S

. neo

rick

iiD

. M. S

po

on

er, G

. J. A

nd

erso

n &

R. K

. Jan

sen

[L.

par

vifl

o-

rum

C. M

. Ric

k, K

esic

ki, F

ob

es &

M. H

olle

];

pen

= S

. pen

nel

liiC

orr

ell [

L. p

enn

ellii

(Co

rrel

l) D

’Arc

y];

pen

p =

S. p

enn

ellii

var.

pu

ber

ulu

mC

orr

ell;

per

= S

. per

uvi

anu

mL.

[L.

per

uvi

anu

m(L

.) M

iller

];

per

g =

L. p

eru

vian

um

f. g

lan

du

losu

mC

. H. M

üll.

;

per

h =

L. p

eru

vian

um

var.

hu

mif

usu

mC

. H. M

üll.

;

pim

= S

. pim

pin

ellif

oliu

mL.

[L.

pim

pin

ellif

oliu

m(L

.) M

iller

];

sit

= S.

sit

ien

sI.

M. J

oh

nst

. [L.

sit

ien

s(I.

M.J

oh

nst

.) J.

M.H

.Sh

aw];

bM

ap n

um

ber

s co

rres

po

nd

to

Fig

ure

1; G

ref

ers

to u

nm

app

ed a

cces

sio

ns

fro

mth

e G

aláp

ago

s Is

lan

ds

loca

ted

ab

ou

t 10

00 k

m w

est

of

Ecu

ado

r.

c LA

nu

mb

ers

are

fro

m t

he

C. M

. Ric

k To

mat

o G

enet

ics

Res

ou

rce

Cen

ter;

PI n

um

-b

ers

are

the

sam

e U

nit

ed S

tate

s Pl

ant

Intr

od

uct

ion

Nu

mb

er d

up

licat

e ac

ces-

sio

ns

hel

d a

t th

e U

nit

ed S

tate

s D

epar

tmen

t o

f A

gri

cult

ure

gen

eban

k in

Gen

eva,

New

Yo

rk (

Gen

eva

Plan

t G

enet

ic R

eso

urc

es U

nit

, NE-

9).

dTh

e co

llect

or

abb

revi

atio

n in

th

e d

atab

ase

of

the

C. M

. Ric

k To

mat

o G

enet

ics

Res

ou

rce

Cen

ter

is S

AL,

an

ab

bre

viat

ion

use

d b

y R

ick

for

his

co

llect

ion

sal

on

e o

r w

ith

oth

ers

to in

dic

ate

“So

uth

Am

eric

an L

yco

per

sico

n.”

e Allo

= a

llog

amo

us,

Au

t =

auto

gam

ou

s, S

C =

sel

f-co

mp

atib

le, S

I = s

elf-

inco

mp

at-

ible

.

f Lo

ng

itu

de

and

lati

tud

e d

ata

add

ed b

y th

e au

tho

rs.

236


unlikely, and only a few cases have been report-ed of spontaneous introgression between thetwo closely related L. esculentum and L.pimpinellifolium (Rick, 1986b).

In addition, Rick (1963, 1986b) recognized 40informally designated races or ecotypes in thepolymorphic S. peruvianum. A few of these arewidespread coastal races, but the majority arelocally distributed montane races. He proposedthat strict gametophytic self-incompatibility andgeographic isolation drove differentiationamong the S. peruvianum races (Rick, 1986b).Hybridization showed an almost complete repro-ductive barrier between variety humifusum, arace growing at high altitude (2500 m) in theJequetepeque River in northeast Peru, and mostof the other S. peruvianum populations.Nevertheless, the reproductive isolation was notcomplete, due to the existence of some bridgeraces (races theoretically able to maintain geneflow between northern and southern races)located in central Peru (Rick, 1963, 1986b).

Studies of additional collections and crosses ofnorthern populations of Lycopersicon peru-vianum allowed Rick (1986b) to identify fourgroups of races that were isolated by reproduc-tive barriers. Three groups of races occur innorthern Peru: the Chamaya-Cuvita group ofraces, the Marañón group of races, andChotano-humifusum group of races. The fourthgroup of races comprised all the remainingraces of L. peruvianum from central and south-ern Peru. Rick (1986b) confirmed the crossingbarriers between the northern and southernraces, and found that some of the northernraces crossed to a limited degree with L.chilense, L. hirsutum, and L. esculentum. Basedon these findings, Rick (1986b) hypothesizedthat the Río Marañón races of L. peruvianumwere ancestral to all other wild tomatoes.

The Endosperm Balance Number (EBN) cross-ability phenomenon was analyzed for Rick’stwo wild tomato complexes by Ehlenfeldt andHanneman (1992). The EBN hypothesis(Johnston et al., 1980; Ortiz & Ehlenfeldt, 1992;

Hanneman, 1994) postulates that in theabsence of stylar barriers, the success or failureof a cross is determined primarily by a 2:1maternal to paternal balance in the endosperm,independent of ploidy. The EBN is determinedusing standard crossing species with knownEBNs. The EBN data supported the hypothesis oftwo intra-fertile groups as proposed by Rick(1979). The Esculentum complex showed unifor-mity of EBN values, which can be compared tothe 2x(1EBN) species in potato. On the otherhand, the Peruvianum complex showed variablevalues for EBN, but most comparable to2x(2EBN) potato species (Ehlenfeldt &Hanneman, 1992). These authors hypothesizedthat the Esculentum and Peruvianum complexesare separated by a system analogous to the2x(1EBN) S. commersonii and 2x(2EBN) S. cha-coense crossability groups. This isolating mech-anism may restrict or suppress gene flow amongsympatric populations, and may play a role inthe reproductive isolation in tomatoes like inthe L. peruvianum var. humifusum races fromnorthern Peru (Ehlenfeldt & Hanneman, 1992).

Molecular data from chloroplast DNA restric-tion site and sequence data (Spooner et al.,1993; Bohs & Olmstead, 1997, 1999; Olmstead &Palmer, 1997; Olmstead et al., 1999) firmly placetomatoes and potatoes as sister taxa. The sys-tematic placement of tomatoes under Solanum,as adopted here, is beginning to gain accept-ance in the taxonomic literature (Bohs &Olmstead, 1997, 1999; Olmstead & Palmer,1997), but it is still controversial and highlightscompeting goals and hypotheses of classifica-tion (Peralta & Spooner, 2000).

Peralta and Spooner (2001) examined interspe-cific relationships of all 10 wild tomato species,with a concentration on the highly polymorphicand widespread species Solanum peruvianum, with DNA sequence data of the single-copynuclear encoded GBSSI gene. Their outgroupresults were concordant with the cpDNA restric-tion fragment length polymorphism (RFLP) phy-logeny of Spooner et al. (1993), supporting S.juglandifolium, S. lycopersicoides, and S. sitiens


as outgroups. The ingroup of wild tomatoesshowed a basal polytomy and a terminal clade.The basal polytomy was composed of the self-incompatible green-fruited species S. chilense,the central to southern Peruvian populations ofS. peruvianum, S. habrochaites, and S. pennellii.The terminal clade contained the northernPeruvian populations of S. peruvianum(also self-incompatible, green fruits), S.chmielewskii, and S. neorickii (self-compatible,green fruits), and the self-compatible and red-to orange- to yellow-fruited species S. cheesma-niae, S. galapagense, S. lycopersicum, and S.pimpinellifolium (Fig. 2). The results supportedallogamy, self-incompatibility, and green fruitsas plesiomorphic in tomatoes.

We wished to examine morphology as an inde-pendent and taxonomically accessible data set,using the same accessions in the GBSSI study(Peralta & Spooner, 2001). All prior classifica-tions used morphology for intuitive taxonomicjudgments, and this is the first explicit use ofmorphology for phenetic and cladistic analysesin tomatoes. The goals of this study are: (1) toexamine interrelationships among all 10 wildtomato species with phenetic and cladisticanalyses of morphological data, and (2) to com-pare our results with the morphological (Müller,1940; Luckwill, 1943) and biological (Rick, 1979;Rick et al., 1990) classifications; mitochondrialDNA restriction site results (McClean & Hanson,1986); nuclear DNA RFLP results (Miller &Tanksley, 1990), cpDNA restriction site results ofPalmer and Zamir (1982) and Spooner et al.(1993), and GBSSI DNA sequence results (Peralta& Spooner, 2001). Our results show diagnosticcharacters useful for species differentiation fora monographic study of Solanum sect.Lycopersicon, currently in progress.

MATERIALS AND METHODSSPECIESFor phenetic analyses, we analyzed a total of 66accessions of all 10 species of tomatoes (Table 1),including two species of Solanum subsect.Lycopersicoides supported as a close outgroup

of tomato (Rick, 1979, 1988; Child, 1990;Spooner et al., 1993). All the accessions, repre-senting much of the ingroup variation, wereobtained from the C. M. Rick Tomato GeneticsResource Center, Department of VegetableCrops, University of California, Davis. The lateCharles Rick, former curator of this gene bank,kindly provided advice on the choice of acces-sions, based on geographic distribution, mor-phology, genetic diversity, and breeding behav-ior (Fig. 3; Table 1). He also advised us on criticalmorphological characters to differentiate taxa(Table 2). The accessions represent nearly thesame as those used in the GBSSI gene phyloge-netic analysis of Peralta and Spooner (2001).

Three accessions per species were analyzed forSolanum chmielewskii, S. chilense, S.habrochaites, S. neorickii, S. pennellii, S.pimpinellifolium, two accessions for S. lycoper-sicum var. cerasiforme and S. cheesmaniae, oneaccession of S. galapagense, and 41 accessionsfor S. peruvianum. One accession of S. lycoper-sicoides and one accession of S. sitiens wereincluded as closely related non-tomato compar-ison species. For cladistic analyses, we deter-mined taxon-specific characters (Tables 3, 4)based on results from the character state varia-tion and phenetic analyses.

The plants were grown in the fall, becausesome wild tomato species need short photope-riods to flower. Seeds were planted in green-houses of the University of Wisconsin, Madison,in mid August 1999. Two weeks later, seedlingswere transferred to 30 cm diameter plastic potswith sterilized soil. Three plants per accessionwere randomly distributed in two replicates,and a total of six individuals were evaluated peraccession. Stem characters and leaves weremeasured from the ninth node, when plantsdeveloped more than 12 nodes and were usual-ly between 60 and 80 cm high. Most of theplants began to flower by the end of Septemberor the first week of October. Flowers were col-lected from mature inflorescences when theplants were in full bloom, during the months ofOctober, November, and December. In order to

237

238


STEM CHARACTERS1. Stem diameter at base. 2. Internode length between the 9th and 10th node (cm). 3. Number

of branches in the first 20 cm at flowering time. 4. Number of nodes before branch bifurcation. 5. Plant height at branch bifurcation (cm). 6. Pseudostipules: absent (0); present (1). 7. Number of leavesper sympodium.

LEAF CHARACTERS8. Leaf length (cm). 9. Ratio: leaf length/leaf width. 10. Ratio: leaf length/length of leaf from widest

point to leaf apex. 11. Leaf petiole length (cm). 12. Terminal leaflet lamina length (cm)*. 13. Ratio: lengthof terminal leaflet lamina/width of lamina. 14. Ratio: length of terminal leaflet lamina/length of termi-nal leaflet lamina from its widest point to apex. 15. Terminal leaflet petiolule length (cm)*. 16. Widthof terminal leaflet lamina from a point 5 mm below the apex (cm). 17. Length of primary dorsal leafletlamina (cm). 18. Ratio: length of primary dorsal leaflet lamina/width of lamina. 19. Ratio: length of primary dorsal leaflet lamina/length of primary dorsal leaflet lamina from its widest point to apex. 20. Length of primary dorsal leaflet petiolule. 21. Width of primary dorsal lateral leaflet lamina from apoint 5 mm below the apex. 22. Number of lateral leaflets. 23. Number of interjected leaflets. 24. Number of secondary leaflets. 25. Number of tertiary leaflets.

INFLORESCENCE CHARACTERS26. Inflorescence bracts: absent (0); present (1). 27. Number of branches per mature inflorescence.

28. Number of flowers per branch*. 29. Length of the mature inflorescence axis (cm). 30. Ratio: numberof flowers/length of inflorescence axis. 31. Peduncle length (cm). 32. Ratio: total inflorescencelength/peduncle length.

FLOWER CHARACTERS33. Pedicel length (cm). 34. Ratio: pedicel length from base of the pedicel to articulation/pedicel

length. 35. Sepal length. 36. Ratio: sepal length/width. 37. Length from center of corolla to apex ofcorolla lobe (cm). 38. Ratio: length from center of corolla to apex of corolla lobe/length from center ofcorolla to base of corolla lobe. 39. Width of corolla lobe at base of corolla lobe junction (cm)*. 40. Ratio:length from a line drawn across the widest point of corolla lobe to lobe apex/width of corolla lobe atbase of corolla lobes junction. 41. Anther length. 42. Anther appendage length. 43. Anther filamentlength. 44. Anther appendage curvature: straight (1), slightly curved to 40˚ (2), curved more than 45˚ (3).45. Anther attachment: free (0), attached (1). 46. Anther dehiscence: pores (0), longitudinal split (1). 47. Anther color: white (0), yellow (1). 48. Style length exsertion from apex of anthers to stigma apex.49. Stigma width. 50. Ovary length. 51. Style length (cm).

FRUIT CHARACTERS52. Diameter at its widest dimension. 53. Ratio: diameter at widest dimension/diameter at

narrowest dimension*. 54. Ratio: diameter at widest dimension/fruit length. 55. Fruit color: green (1),pale green (2), yellow green (3), orange (4), red (5). 56. Fruit stripe: absent (0); present (1). 57. Fruitpubescence: glabrous to glabrescent (1), pubescent to pilose (2), hirsute (3).

PLANT TRICHOMESSee Luckwill (1943) for illustrations of trichome types, and materials and methods for the method

of scoring.

58. Type I. 59. Type II. 60. Type III. 61. Type IV. 62. Type V. 63. Type VI. 64. Type VII*. 65. Type VIII*. 66. Trichome density.

TABLE 2. Morphological characters used in the phenetic analysis of wild tomatoes and outgroups. All measurementsare in millimeters, except as noted below. Characters marked with an asterisk (*) were not significantlydifferent (P = 0.05) between at least two species under the Tukey-Kramer HSD test.

obtain fruits from the self-incompatible (SI)accessions, plants were pollinated manuallyusing pollen of different individuals belongingto the same accessions. Self-compatible (SC)accessions produced fruits naturally. Fruits wereharvested approximately two months after pol-lination. Vouchers are deposited at DAV, MERL,PTIS, and WIS.

DATA MEASUREMENTFor phenetic analyses, we assessed a total of 66characters (50 quantitative and 16 qualitative)for 6 individuals of 66 accessions (over 26,000data points). The quantitative characters includ-ed 14 ratios that assessed shapes of differentplant organs (Table 2). Trichomes were studiedfrom young stems by making thin slice cuts 1 or2 cm below the apex with a razor blade.Trichome types were identified according to theclassification of Luckwill (1943) using a trans-mission light microscope (100× magnification).An arbitrary scale based on percentages:1%–20% = 1, 21%–40% = 2, 41%–60% = 3,61%–80% = 4, 81%–100% = 5, was used toscore the frequency of different trichomes(characters 58 to 65). The diameter and heightof the thin stem slice were measured under amicroscope, and the number of trichomesaround the slice circumference was counted.Trichome density percentage (D = n/S 100; character 66) was estimated as the total numberof trichomes (n) divided by the cylinder square(S = π × diameter/2 × height), represented by thethin stem slide.

For cladistic analyses, we chose characters thatwere shown by our analysis of character statevariation (below) to have non-overlapping or nearly non-overlapping states (Table 3).

DATA ANALYSIS: CHARACTER STATE VARIATION AND PHENETICSThe mean value of the six plants per accessionwas used as the Operational Taxonomic Unit(OTU). The mean, range, and standard devia-tion were estimated for each character.Significant character state differences (P = 0.05)

between any two pairs of species were evaluat-ed using the Tukey-Kramer test with JMP statis-tical software (SAS, 1995).

Cluster analyses were produced by NTSYS-pcR

version 1.70 (Rohlf, 1992) only on the subset of66 characters found to be statistically significantamong groups by one-way ANOVA. Averagesfor each character were standardized (STAND),and similarity matrices, using average taxonom-ic distance (DIST), Manhattan distance (MANHAT), Euclidean distance (EUCLID), andproduct-moment correlation (CORR) were generated. Clustering was performed using theunweighted pair-group method (UPGMA) inSAHN. Cophenetic correlation coefficients(COPH and MXCOMP) were used to measuredistortion between the similarity matrices andthe resultant three phenograms (Rohlf & Sokal,1981; Sokal, 1986). Stepwise discriminant analysis (SDA) was performed on all 66 characters by SAS Version 7 (SAS, 1998), usingthe mean values.

Four entities in Solanum peruvianum (thenorthern and southern populations, f. glandu-losum, and var. humifusum), S. cheesmaniae,and S. galapagense (= S. cheesmaniae f. minor),were recognized as OTUs to estimate the mean,range, and standard deviation. In addition tothese taxa, two forms of S. habrochaites(f. typicum and f. glabratum) and two varietiesof S. pennellii (var. typicum and var. puberulum)were considered in the cluster analysis and inthe SDA. The forms and varieties of the lattertwo species are mainly differentiated by theirpubescence.

DATA ANALYSIS: CLADISTICSWe used two subsets of characters that repre-sented all (26 characters, data set 1), or onlythose characters lacking division into ranges ofstates (15 characters, data set 2) (Tables 3, 4).Parsimony analyses were conducted with PAUP*4.0d64 (Swofford, 1998). Solanum juglandifoli-um, S. lycopersicoides, S. ochranthum, and S.sitiens were used as outgroups based on Rick

239


240


Figure 3. Distribution of the wild tomato accessions used in this study. Map numbers correspond to generalized map localities inTable 1. The line indicates the geographic distribution of S. peruvianum north (N) and south (S).


241

1.* Number of nodes before branch bifurcation: 1, > 30; 2, 17–20; 3, 10–15; 4, < 10. CI = 1, RI = 1.2. Number of leaves per sympodium: 1, > 5; 2, 3; 3, 2. CI = 0.667, RI = 0.857.3. Pseudostipular leaves: 1, absent; 2, present. CI = 1, RI = 1.4.* Number of lateral leaflet pairs: 1, 4–5; 2, 2–4. CI = 0.5, RI = 0.5.5.* Number of secondary leaflets: 1, 0–5; 2, 5–10; 3, 10–20; 4, 40–50. CI = 1, RI = 1.6. Tertiary leaflets: 1, absent; 2, present. CI = 1, RI = n.a.7.* Leaflet width 5 mm below the apex: 1, > 15 mm; 2, < 10 mm. CI = 1, RI = n.a.8.* Inflorescence peduncle length: 1, > 7 cm; 2, 5–7 cm; 3, 3–5; 4, < 3 cm. CI = 0.750, RI = 0.833.9.* Branches per inflorescence: 1, > 4; 2, 2 often 3; 3, 1–2; 4, 1. CI = 1, RI = 1.

10. Bracts in inflorescence: 1, absent; 2, present. CI = 0.5, RI = 0.8.11. Anthoclades (see Spooner et al., 1993): 1, 3.7; 2, 3.6. CI = 1, RI = 1.12.* Pedicel articulation: 1, mid to upper position; 2, variable; 3, basal. CI = 1, RI = 1.13. Corolla shape: 1, symmetrical; 2, asymmetrical. CI = 1, RI = n.a.14.* Length from center of corolla to apex of corolla lobe:1, >1.5 cm; 2, 1–1.5 cm; 3, < 1 cm. CI = 0.667, RI = 0.667.15. Anther color: 1, white; 2, yellow. CI = 1, RI = 1.16. Anther lateral interlocking papillae: 1, absent; 2, small; 3, elongate. CI = 1, RI = 1.17. Anther connation: 1, separate; 2, free but holding together; 3, tightly connate. CI = 1, RI = 1.18. Anther appendages: 1, absent; 2, present. CI = 1, RI = 1.19. Anther tube: 1, straight; 2, anther and appendages curved; 3, anther curved. CI = 1, RI = n.a.20.* Style exsertion: 1, > 1.8; 2, 1–1.8; 3, 0.5–1; 4, 0.5–0. CI = 0.750, RI = 0.857.21. Fruit color: 1, green; 2, yellow to orange; 3, red. CI = 1, RI = 1.22.* Fruit size: 1, > 25 mm diam. at widest point; 2, 15–25 mm diam. at widest point; 3, 5–15 mm diam.

at widest point; 4, < 5 mm diam. at widest point. CI = 1, RI = 1.23. Fruit pericarp: 1, thick and hard pericarp; 2, thin and leathery; 3, thin and membranaceous. CI = 1, RI = 1.24. * Seeds: 1, more than 3 mm; 2, less than 3 mm. CI = 1, RI = 1.25. Compatibility: 1, self-incompatible; 2, self-compatible. CI = 1, RI = 1.26. Breeding: 1, allogamous; 2, autogamous/facultative allogamous; 3, autogamous. CI = 0.667, RI = 0.750.

TABLE 3. Morphological characters used in cladistic analyses of wild tomatoes and outgroups. All 26 characters are usedin data set 1; those marked with an asterisk (*) are eliminated for data set 2 containing 15 characters.Consistency Index (CI) and Retention Index (RI) were estimated for each character. The RI is not applicable (n.a.)for autapomorphic characters.

TABLE 4. Data matrix of cladistic character states corresponding to Table 3.

Characters and Character States

Species 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

Solanum lycopersicoides 1 1 2 1 1 1 2 1 1 2 1 1 1 2 1 1 1 1 1 1 1 3 2 1 1 1

S. sitiens ? 1 2 1 1 1 2 1 1 2 1 1 1 2 1 1 1 1 1 1 1 3 2 1 1 1 S. ochranthum 1 1 2 2 1 1 2 2 1 1 2 2 1 2 2 2 2 1 1 1 1 1 1 1 1 1 S. juglandifolium 1 1 2 2 1 1 2 2 1 1 2 2 1 2 2 2 2 1 1 1 1 1 1 1 1 1

S. pennellii 3 3 2 2 1 1 1 2 3 2 1 3 2 1 2 2 3 1 3 1 1 3 3 2 1 1 S. habrochaites 2 2 2 2 1 1 2 3 3 2 1 1 1 1 2 3 3 2 1 2 1 3 3 2 1 1 S. chilense 3 3 2 1 1 1 2 1 2 2 1 1 1 1 2 3 3 2 1 2 1 3 3 2 1 1 S. peruvianum

south 3 3 2 2 1 1 2 2 2 2 1 1 1 1 2 3 3 2 2 2 1 3 3 2 1 1 S. peruvianum

north 3 3 2 2 1 1 2 2 4 2 1 1 1 2 2 3 3 2 1 3 1 3 3 2 1 1 S. chmielewskii 3 3 2 2 1 1 2 4 4 2 1 1 1 2 2 3 3 2 1 3 1 3 3 2 2 2 S. neorickii 3 3 2 2 1 1 2 4 4 2 1 1 1 3 2 3 3 2 1 4 1 3 3 2 2 3 S. cheesmaniae 4 2 1 2 2 1 2 4 4 1 1 1 1 2 2 3 3 2 1 4 2 3 3 2 2 3 S. galapagense 3 2 1 2 4 2 2 4 4 1 1 1 1 2 2 3 3 2 1 4 2 4 3 2 2 3 S. pimpinellifolium 3 2 1 2 2 1 2 4 4 1 1 1 1 2 2 3 3 2 1 3 3 3 3 2 2 2 S. lycopersicum 3 2 1 2 3 1 2 4 4 1 1 1 1 2 2 3 3 2 1 4 3 2 3 2 2 2

OU

TG

RO

UP

SIN

GR

OU

PS

242


(1979, 1988), Child (1990), Spooner et al. (1993),and Peralta and Spooner (2001). Solanum jug-landifolium and S. ochranthum are supportedby GBSSI results as the sister taxa to tomatoes,but were not included in the phenetic analysisdue to their slow growth in the greenhouse. Weobtained flower, fruit, and seed cladistic charac-ters of these species from the literature or fromherbarium specimens.

Cladistic analyses were performed on both datasets 1 and 2, using Wagner parsimony,unordered character states, and all characterswere equally weighted. The heuristic searcheswere performed using branch and bound andfurthest addition sequence. The amount ofhomoplasy for both trees and individual charac-ters was evaluated with the consistency index(CI) of Kluge and Farris (1969) and the retentionindex (RI) of Farris (1989). Consensus trees weregenerated from all most parsimonious trees.Bootstrap (Felsenstein, 1985) and Jackknifeanalyses, using 1000 replicates (using tree bisec-tion reconnection and saving all multiple trees),were conducted to estimate the internal rela-tive support for each branch.

RESULTSCHARACTER STATE VARIATIONThe Tukey-Kramer HSD test determined that 61of the 66 characters were significantly differentbetween at least two taxa (Table 2). The means,ranges, and standard deviations of the 30 char-acters showing most variation among species arepresented in Figure 4. Separately, distribution ofcharacter states of “northern” and “southern”populations of S. peruvianum, chosen on thebasis of the GBSSI results (Peralta & Spooner,2001), were also compared using the Tukey-Kramer HSD test. Thirty-one characters were sig-nificantly different between these two groups ofS. peruvianum. The means, ranges, and standarddeviations of the 18 most important charactersseparating these geographic groups of S. peru-vianum are illustrated in Figure 5.

PHENETIC RESULTSThe same dendrogram was produced by DIST(Fig. 6) and EUCLID, both with a cophenetic cor-relation coefficient of 0.93, which was onlyslightly higher than those produced by MAN-HAT, 0.91; CORR was 0.75. Cophenetic correla-tions between 0.8 and 0.9 can be interpretedsubjectively as good fits to the cluster analysis,and those between 0.7 and 0.8 as poor fits(Rohlf, 1992).

The DIST phenogram has four main groups (Fig.6A–D). The outgroups, S. lycopersicoides and S.sitiens, were clustered as the external branch(group D), followed by S. galapagense, andthen a group of all three accessions of S. pen-nellii (group C). The self-compatible, red- toorange- to yellow-fruited species (S. lycoper-sicum, S. cheesmaniae, and S. pimpinellifolium)formed a third cluster (group A), but with theexclusion of the distinctive S. galapagense. Thefourth group (B) included the remainingspecies.

Within group B, Solanum neorickii and twoaccessions of S. chmielewskii clustered together,to the exclusion of one accession of S.chmielewskii (LA 1306) that grouped with S.peruvianum. All accessions of S. chilense formeda group that also contained one accession of S.peruvianum (LA 1982). The three accessions ofS. habrochaites formed a separate group.Concordant with the GBSSI sequence data, twomajor groups were recognized within S. peru-vianum: “the northern” and “the southern”(Fig. 6 N, S). Only three S. peruvianum acces-sions were placed outside the northern andsouthern clusters (LA 2172, LA 1379, LA 1982).

We also present the CORR phenogram (Fig. 7)despite a lower cophenetic correlation (0.75; vs.DIST, 0.93), because it provides better pheneticsupport for S. cheesmaniae and S. galapagense,and north and south populations of S. peru-vianum. Unlike the DIST phenogram, it placesthe two outgroups, S. lycopersicoides and S.sitiens, as an internal branch with one of two


main clusters (A). The self-compatible, red- toorange- to yellow-fruited species: S. cheesmani-ae, S. galapagense, S. lycopersicum, and S.pimpinellifolium grouped together within thecluster A. The three accessions of S.habrochaites formed a separate group, and alsothe three S. pennellii accessions clusteredtogether. The other main branch (B) includes S.peruvianum, S. chilense, S. neorickii, and S.chmielewskii. The CORR phenogram, unlike theDIST phenogram, showed a better clustering ofthe northern and southern S. peruvianumgroups (Fig. 7), and maintained all accessions ofS. cheesmaniae and S. galapagense together.Like the DIST phenogram, S. peruvianum (LA1982) clustered with S. chilense. The two acces-sions of S. peruvianum f. glandulosum clusteredwith the southern S. peruvianum. Solanumneorickii, S. chmielewskii, and the two acces-sions of S. peruvianum var. humifusum clusteredwith the northern S. peruvianum.

The SDA of all taxa discriminated 30 characters.The 10 best characters, in decreasing order ofdiscriminative utility, were: (1) anther attach-ment, (2) anther color, (3) pseudostipules, (4)number of nodes, (5) number of tertiaryleaflets, (6) number of leaves per symposia, (7)trichome type III, (8) anther length, (9) plantheight, and (10) number of secondary leaflets.

The SDA of the northern and southern populationsof S. peruvianum discriminated 11 characters, (1)number of inflorescence branches, (2) style exser-tion, (3) dorsal leaflet petiolule, (4) fruit color, (5)ratio: articulation length/pedicel length, (6) anthercurvature, (7) internode length, (8) trichome type V, (9) pedicel length, (10) trichome type IV, and (11)terminal leaflet length.

CLADISTIC RESULTSThe analysis using all 26 characters (Tables 3, 4)produced 28 equally parsimonious trees, length54 (Fig. 8). The strict consensus tree (Fig. 9) hasa CI = 0.870, Consistency Index excluding unin-formative characters (CIU) = 0.857, and RI =0.900. Bootstrap analysis using 10,000 replicatesgave 96% support for the ingroup. Jackknife

values (10,000 replicates) were similar or some-times lower, but also support these groups (Fig. 9). In the ingroup Solanum pennellii is supported as basal. The relationships amongthe self-incompatible species S. chilense, S. habrochaites, and S. peruvianum south werenot resolved, and the three taxa formed a poly-tomy in all 28 equally parsimonious trees (Fig.9). Solanum peruvianum north appeared asbasal to S. chmielewskii and S. neorickii (Fig. 9)in the 28 equally parsimonious trees with76(64)% bootstrap (jackknife) support. Solanumchmielewskii and S. neorickii appeared alwaystogether and supported in a basal position inthe same clade as the monophyletic groupformed by S. cheesmaniae, S. galapagense, S.pimpinellifolium, and S. lycopersicum (Fig. 9).

When the breeding characters (Table 3: 25, 26)were excluded, 24 trees were obtained (length= 50; CI = 0.880, CIU = 0867, RI = 0.902). A strictconsensus tree of these 24 trees is identical tothe strict consensus tree using all 26 characters(Fig. 9). The analysis using the subset of 15 char-acters (Table 3) produced 10 equally parsimo-nious trees, length 26. The strict consensus tree(CI = 0.846, CIU = 0.818, RI = 0.895) is topologi-cally the same that was obtained using all 26characters (Fig. 9), except S. peruvianum northis now part of the polytomy with Solanumchilense, S. habrochaites, and S. peruvianumsouth.

CONCORDANT MORPHOLOGICAL CLADISTIC RESULTSThe cladistic results showed the following pat-terns of relationships: (1) Within the four out-group taxa, Solanum ochranthum and S. juglan-difolium were sister taxa and S. lycopersicoidesand S. sitiens were sister taxa. (2) Within thetomato ingroup, S. pennellii was basal. (3)Solanum cheesmaniae, S. galapagense, S.pimpinellifolium, and S. lycopersicum were ter-minal taxa and monophyletic. (4) Solanumchmielewskii and S. neorickii appeared alwaystogether in a basal position in the same clade asS. cheesmaniae, S. galapagense, S. pimpinelli-

243

244


Figure 4. Means, ranges, and one standard deviation of the mean for the 30 of 66 characters examined in this study showing thegreatest differences among taxa of Solanum chilense = chl; S. chmielewskii = chm; S. cheesmaniae = chs; S. galapagense = gal [=S. cheesmaniae f. minor]; S. habrochaites = hab; S. lycopersicoides = lyc; S. lycopersicum var. cerasiforme = lypc; S. neorickii = neo;S. pennellii = pen; S. peruvianum = per; S. peruvianum f. glandulosum = perg; S. peruvianum var. humisufum = perh; S. pimpinel-lifolium = pim; S. sitiens = sit.


245

Figure 4 continued.

246


Figure 5. Means, ranges, and one standard deviation of the mean for the 18 of 61 characters examined in this study showing thegreatest differences among S. peruvianum accessions from northern Peru = N, and southern Peru = S.


247

Figure 6. UPGMA dendrogram (DIST similarity option) based on 61 morphological characters. Species codes, accession numbers,and map locations as in Table 1.

248


Figure 7. UPGMA dendrogram (CORR similarity option) based on 61 morphological characters. Species codes, accession numbers,and map locations as in Table 1.


249

Figure 9. Strict consensus tree of the 28 most parsimonious trees (length 54) based on all 26 unordered morphological characters.Percentage of 10,000 bootstrap replicates, followed by percentage of 10,000 jackknife replicates (in parentheses), is given abovebranches. The same strict consensus topology was obtained in the analysis excluding the two breeding characters, and nearly the same topology (except for S. peruvianum N that was part of the basal polytomy with S. peruvianum S, S. habrochaites, and S. chilense) in the analysis of 15 characters (25, 26, Tables 3, 4).

Figure 8. One representative phylogram of 28 equally parsimonious trees, length = 54, based on all 26 unordered morphologicalcharacters. CI = 0.870, CIU = 0.857, RI = 0.900.

250


folium, and S. lycopersicum. (5) The self-incom-patible S. chilense, S. habrochaites, and S. peru-vianum south form a basal polytomy in theingroup (the 15-character data set included S.peruvianum north in this basal polytomy).

CONCORDANCE OF MORPHOLOGICAL PHENETICS AND CLADISTICSThe phenetic results are not designed to showphylogenetic relationships. When similaritiesdue to shared ancestral characteristics or homo-plasy exceed similarities due to shared derivedcharacteristics, the phenogram will not repre-sent the true phylogeny, but if there were nohomoplasy, the phenogram constructed by thephenetic analysis would correspond to the truephylogenetic tree (Futuyma, 1998). The similari-ties of phenetic and cladistic results show: (1) Theoutgroup taxa were differentiated in bothanalyses. (2) Solanum cheesmaniae, S. galapa-gense, S. pimpinellifolium, and S. lycopersicumwere also consistent “groups.” (3) Solanum peru-vianum was differentiated in two groups fromnorthern and southern Peru, with 76% boot-strap support in the cladistic results (Fig. 9). (4)Solanum chmielewskii and S. neorickii groupedwith the northern S. peruvianum but not with S.cheesmaniae, S. galapagense, S. pimpinellifoli-um, and S. lycopersicum as in the cladistic results.(5) Solanum chilense grouped with the southernS. peruvianum in the phenetic results (Fig. 7), butin the cladistic results S. chilense appeared as abasal polytomy with southern S. peruvianum andS. habrochaites (Fig. 9).

CONCORDANCE OF MORPHOLOGICAL CLADISTICS AND GBSSI CLADISTICSThere are striking concordances between themorphological cladistic results shown here andthe GBSSI cladistic results of Peralta andSpooner (2001). (1) Within the four outgrouptaxa, Solanum ochranthum and S. juglandifoli-um were sister taxa and S. lycopersicoides and S.sitiens were sister taxa. (2) Within the tomatoingroup, S. pennellii was supported as a basaltaxon. (3) Solanum habrochaites also appearedin a basal position. (4) Solanum peruvianum

south showed a close relation with the last twotaxa and also with S. chilense. (5) Solanumcheesmaniae, S. galapagense, S. pimpinellifoli-um, and S. lycopersicum were terminal taxa andmonophyletic. (6) Solanum chmielewskii and S.neorickii were basal to S. cheesmaniae, S. gala-pagense, S. pimpinellifolium, and S. lycoper-sicum. GBSSI and morphological results showeda closer relationship of these three last taxawith S. chmielewskii and S. neorickii, which arealso related to the northern populations of S.peruvianum.

DISCUSSIONPHENETICSThe above analysis of character state variationand phenetic relationships provided useful datato (1) help support existing species, (2) supportthe northern and southern populations ofSolanum peruvianum as separate taxa, (3) showcharacter state variation useful for speciesdescriptions, and (4) show character states use-ful for cladistic analyses.

Although the range of values for many charac-ters did not show discrete breaks to differenti-ate species, some have breaks in variability use-ful for species differentiation and cladisticanalysis. The most important morphologicalcharacters, determined by the simple statisticsand SDA, agreed with those previously used intaxonomic diagnoses and construction of keys(Müller, 1940; Luckwill, 1943; Rick et al., 1977,1990; Taylor, 1986). Qualitative flower characterstates, such as white, free anthers with large fil-aments, and more than five leaves per sympodi-um were only shared by the species of subsec-tion Lycopersicoides. Development of pseu-dostipules and inflorescence bracts, number ofinflorescence branches, anther curvature, colorand pubescence of the fruits, and trichometypes have taxonomic value within wild toma-toes. Solanum pennellii traditionally has beenrecognized by its particular anther structure,lacking the sterile tip, poricidal dehiscence, andslightly asymmetric flowers.


Although most of the quantitative charactersevaluated had continuous distribution of statesacross taxa, some of them were useful for speciesdiagnosis. Short plants with short internodelengths and early formation of inflorescence (8thto 9th node) characterize Solanum cheesmaniae,while in the shrubby, low-branched, and tallerplants of S. lycopersicoides, inflorescences weregenerated after the 35th node. Solanum sitienscan be identified by its smaller and narrowleaflets. Solanum neorickii was recognized by itssmall flowers with short anther tips, short inflo-rescence axis, and a high number of flowers inrelation to the axis length. Diagnostic charactersfor S. chilense were large peduncles, leaves withfew secondary and tertiary leaflets, and largeflowers and fruits. Taller, very robust, and pubes-cent plants characterized S. habrochaites, whichproduces inflorescences after the 18th node, andhas leaves without secondary leaflets. In S. pen-nellii, the width of the leaflet lamina, as well asanther length and structure, are valuable diag-nostic characters. Solanum lycopersicum pro-duced the largest fruits and numerous secondaryleaflets, while S. pimpinellifolium presented thelargest ratio between corolla lobe length andcorolla lobe width, indicating a stellate shape.

The phenetic analysis distinguished the “north-ern” and “southern” groups in S. peruvianum.This result was postulated by morphology andcrossing relationships (Rick, 1963, 1979, 1986b),and was also supported by the GBSSI phylogenyon the same accessions (Peralta & Spooner,2001). The SDA selected similar sets of charac-ters to those used previously to characterize thenorthern races, such as unbranched inflores-cence, straight anther tubes, reduced styleexsertion, and 5 to 7 leaflets (Rick, 1986b).Although most of the quantitative charactershad a continuous distribution of states acrossthese northern and southern populations, someof them (plant height, internode length, inflo-rescence axis length, leaf length, size of leaflets,and sepal length) showed that taller and morerobust plants characterized the northern acces-sions. In contrast, the southern accessions canbe differentiated by the presence of few sec-

ondary leaflets, branched inflorescences, brightyellow larger corollas, bright yellow larger andcurved anthers, and exserted styles.

CLADISTICSConcordance among phenetic and cladisticresults might indicate that the amount of homo-plasy does not outnumber similarities due toshared derived characteristics in wild tomatoes.The following discussion focuses on cladisticresults of morphological, GBSSI, and othermolecular data to comment on prior phyloge-netic hypotheses in tomatoes (above). Cladisticresults of morphological features confirmingroup relationships shown by nuclear RFLPdata (Miller & Tanksley, 1990), cpDNA restrictionsite data (Palmer & Zamir, 1982), and GBSSI DNAsequence data (Peralta & Spooner, 2001).Phenetic and cladistic analyses of morphologicalcharacters showed the distinct nature of thenorthern and southern Peruvian populations ofSolanum peruvianum, a result consistent withGBSSI sequence data. The morphology shownhere, as well as the nuclear RFLP (Miller &Tanksley, 1990), cpDNA restriction site (Palmer &Zamir, 1982), and GBSSI data (Peralta & Spooner,2001), do not agree with the mitochondrial DNAphenetic results (McClean & Hanson, 1986).

Monophyly of the self-compatible species withpigmented fruits, Solanum cheesmaniae, S. gala-pagense, S. pimpinellifolium, and S. lycoper-sicum, has been shown by the cpDNA phylogeny,GBSSI data, and nuclear RFLP data, and here con-firmed by morphology. Furthermore, S. neorickiiand S. chmielewskii were placed with northernraces of S. peruvianum as recently shown bycpDNA, nuclear RFLP, and GBSSI data. Solanumchmielewskii and S. neorickii appeared support-ed together in a basal position in the same cladeas S. cheesmaniae, S. galapagense, S. pimpinelli-folium, and S. lycopersicum. Our morphologicaland GBSSI results showed the northern acces-sions of S. peruvianum as closely related to S.neorickii, S. chmielewskii, and the three red- toorange- to yellow-fruited species of theEsculentum complex, discordant with Rick’s(1979) classification (Fig. 1). Furthermore, the

251

252


morphological results, as well as cpDNA, nuclearRFLP, and GBSSI data, are discordant with theplacement of S. pennellii and S. habrochaites inthe Esculentum group (Rick, 1979). On the otherhand, Solanum chilense showed close relation-ships with the southern accessions of S. peru-vianum, as indicated by crossability (Rick, 1979),cpDNA, and GBSSI phylogeny.

Morphological and molecular data supportallogamy, self-incompatibility, and green fruitsas plesiomorphic in tomatoes. The basal taxaalso share characters such as branched andbracted inflorescences and leaves usually withpseudostipules. These results do not supportthe phylogenetic hypothesis proposed byLuckwill (1943), where all tomato speciesevolved from an ancestral stock withunbranched inflorescences, leaves with fewleaflets, and without pseudostipules.

Large and attractive flowers have also beenassociated with the presence of a self-incom-patibility system, and with the extent of out-crossing and genetic variation. Rick (1982b) pos-tulated that three independent genetic systemsare operating on the prevention of self-fertil-ization, changes in flower parts to ensure crosspollination, and development of secondaryflower display characters to attract pollinators.Interestingly, a recent linkage analysis andquantitative trait loci (QTL) mapping of repro-ductive behavior and floral traits in wild toma-toes showed that major QTLs for several charac-ters important to pollination biology harbor inthe same region of the self-incompatibilitylocus, S, on chromosome 1. These QTL resultssuggest that a gene complex is controlling bothgenetic and morphological mechanisms ofreproduction, and this gene complex may havebeen conserved since early periods of plant evo-lution or may reflect a convergent evolutionaryprocess (Bernacchi & Tanksley, 1997).

Several characters of the corolla, anthers, andstigma (Table 2), associated with floral display,pollinator attraction, and reproduction, wereincluded in this morphological study. The self-

incompatible accessions of S. habrochaites andS. pennellii (Table 1) showed larger corollas andanthers as well as more exserted stigmas thanthe self-compatible accessions. Similarly, in theself-compatible S. pimpinellifolium one faculta-tive allogamous accession has larger flowerparts and stigma exsertion than the two autog-amous accessions. Changes in flower size andstigma exsertion also occur in the two siblingself-compatible species, S. chmielewskii and S.neorickii. The accessions of the first speciesshowed larger flower parts, while the acces-sions of the exclusively autogamous S. neorickiihave smaller flower parts and stigmas includedin the anther tube. The southern self-incompat-ible accessions of S. peruvianum, like the out-crossing genotypes of S. habrochaites, S. pen-nellii, and S. pimpinellifolium, also displayedfloral traits for pollination attraction. Larger,bright yellow corollas, larger, curved and asym-metrical anthers, and exserted styles are fea-tures that probably correlate with the high out-crossing rates found in S. peruvianum (Rick etal., 1977). Actually, branched inflorescenceswith many showy open flowers (usually 8 to 10)act as the attractive unit for pollinators (Rick etal., 1978). In contrast, the northern S. peru-vianum accessions have unbranched inflores-cences, smaller corollas, short and straightanther tubes, and reduced stigma exsertion.The differences found in inflorescence andflower morphology, and crossing behaviorbetween northern and southern S. peruvianumraces, raise an interesting question about therole that environmental factors might play insuch differentiation. Northern races are wide-spread from the coast to interior uplands in theAndes (north of 8°S) in more mesic-subtropicalhabitats (Rick, 1986b). In contrast, the southernraces are distributed along the arid Pacific coast(south of 8°S) to northern Chile, and in the adja-cent Andean Mountains (Rick, 1986b). The pres-ent hyper-arid climates between 5°S and 28°Sresult primarily from the rain shadow effects ofthe Andes and the subtropical convergence ofwinds, and the drying effects of the coldHumboldt Current. Paleobotanical, paleonto-


logical, and geological evidences suggest thatsuch harsh climatic conditions developedrecently during the Holocene (Arroyo et al.,1988). The Neotropical plant communities thatoccupied the area before the climatic changewere displaced toward more southerly andboreal latitudes, and a new flora became adapt-ed to the new xeric conditions (Villagran et al.,1983). A similar scenario can be hypothesized toexplain the differentiation between S. peru-vianum groups. The southern populationsbecame adapted to the unstable, very arid con-ditions of the coastal Pacific range. Under suchharsh climatic conditions, lower pollinator avail-ability and an increase of anemophily in theplant communities have been reported (Arroyoet al., 1983). Selection pressures to ensure cross-ing in self-incompatible plants may have drivenchanges in flower traits that increase insectattraction. The shifts of secondary flower char-acters are correlated with increase in levels ofheterozygosity in populations (Rick et al., 1977).An opposite change toward reduction of flowerorgans and attractiveness has been reported inthe self-compatible S. cheesmaniae. Thisendemic species of the Galápagos Islands growsin isolated, unstable, and dry habitats, wherevery few pollinator species are found. The S.cheesmaniae populations evolved toward strictautogamy, and genetic variation was onlyfound between populations (Rick, 1986b).

Genetic mechanisms ensuring crossing mayhave played a role to maintain gene flowamong southern S. peruvianum populations, ashas been supported by hybridization studies(Rick, 1963, 1979, 1986b). In contrast, the north-ern populations growing in more stable habi-tats developed isolation mechanisms that pre-vent gene flow in sympatric populations. Inaddition to the prezygotic barriers, postzygoticisolating mechanisms have been found in thisgroup (Ehlenfeldt & Hanneman, 1992).

A possible scenario in the evolution of wild toma-toes can be hypothesized taking into account theadaptation to changing environments, as well as

the interaction with biotic factors such as avail-ability of pollinators. The shift toward compati-bility and reduction of floral structures alsoplayed an important role in the speciationprocess. Putative SI ancestral populations proba-bly occupied a wide area of distribution in centralPeru. After the drastic climatic change during theHolocene and the formation of thePeruvian–Chilean desert, a process of adaptationto the new habitats may have occurred leading todifferentiation and speciation. Some populations,adapted to more mesic and humid conditions,colonized the northern areas. Changes from SI toSC may have originated autogamous popula-tions, further accentuating isolation and proba-bly genetic drift, and contributed to differentia-tion and speciation. Putative ancestors from theRio Marañón area may have originated self-com-patible taxa, as suggested by Rick (1986b). On theother hand, populations growing in arid habitatsmay have differentiated and migrated followingthe Pacific coast and colonized the adjacent west-ern slopes, but were limited in their altitudinaldistribution by frost. Selection pressure to ensurecrossing in the SI populations may have drivenevolutionary changes, and genetic differentiationmost probably occurred by gene substitution(Rick, 1986b). Reproductive isolating barriersarose to prevent gene flow between sympatricpopulations. Analogously, putative ancestors of S.juglandifolium and S. ochranthum becameadapted to more wet and warm environments,while ancestors of S. sitiens and S. lycopersicoidesevolved in the desert.

Some taxonomic problems remain, and morestudies are needed to elucidate the relation-ships between Solanum chmielewskii and thenorthern S. peruvianum races, and between S.chilense and the southern S. peruvianum races.Palmer and Zamir (1982), based on a cpDNAphylogeny, suggested that S. chmielewskii andS. chilense could be considered at the subspecif-ic level within S. peruvianum. Müller (1940) con-sidered S. chilense as a variety of S. peruvianum,while Luckwill (1943) included it as the sub-species puberulum. Our morphological results,

253

254


based on few S. chmielewskii and S. chilenseaccessions, provided good diagnostic charac-ters. Nevertheless, analysis of more collectionsand herbarium specimens, especially those fromoriginal types, will help to clarify taxonomicproblems. Studies of S. peruvianum infraspecif-ic categories as well as the species recognized asL. glandulosum and L. pissisi are currently inprogress. Rick and Lamm (1955), based on thephotograph of the original collection of L. pis-sisi, concluded that this specimen was not relat-ed to S. peruvianum. Nevertheless, the identityof L. pissisi remains unsolved.

Our morphological results support existingspecies and support the northern and southernpopulations of S. peruvianum as separate taxathat may represent two distinct species.

Morphological data showed valuable diagnosticcharacters for species differentiation and phylogenetic relationships in wild tomatoes.These results are currently being used to prepare a monographic study of Solanumsections Lycopersicon, Lycopersicoides, andJuglandifolium. This monograph will include anew classification based on a synthesis of dif-ferent data sets, reflecting the phylogeneticrelationships of wild tomato species.

NOTE IN PROOFThis paper showed the need to reconsider the species limits of Solanum peruvianum;this was accomplished by Peralta et al. (2005)who recognized four species within Solanum peruvianum sensu lato, necessitating the publication of two new names.

ACKNOWLEDGMENTSThis paper represents a portion of a Ph.D. thesissubmitted to the Plant Breeding and PlantGenetics Program at the University of Wisconsin,Madison. The authors especially thank CharlesRick for choosing and providing the accessionsfrom the C. M. Rick Tomato Genetic ResourcesCenter; committee members Paul Berry, RobertHanneman, Michel J. Havey, Kenneth J. Sytsma,

and Phillip W. Simon for advice; Harvey Ballard,Peter Crump, Claudio Galmarini, Brian Karas, JoeKuhl, Sabina Lara Cabrera, Jason Lilly, CindyMuralles, and Sarah Stevenson for technical orstatistical advice, and Sandra Knapp for review.The Fulbright-LASPAU program, the NationalScience Council of Argentina (CONICET), theNational University of Cuyo (UNC), and theUSDA funded research. Names are necessary toreport data. However, the USDA neither guaran-tees nor warrants the standard of the product,and the use of the name by the USDA implies noapproval of the product to the exclusion of oth-ers that may also be suitable.

LITERATURE CITEDArroyo, M. T. K., J. J. Armesto & R. Primack.

1983. Tendencias altitudinales y latitu-dinales en mecanismos de polinizaciónen la zona andina de los andes templa-dos de Sudamérica. Revista ChilenaHist. Nat. 56: 159–180.

—, F. A. Squeo, J. J. Armesto & C. Villagran.1988. Effects of aridity on plant diversi-ty in the northern Chilean Andes:Results of a natural experiment. Ann.Missouri Bot. Gard. 75: 55–78.

Bernacchi, D. & S. D. Tanksley. 1997. An inter-specific backcross of Lycopersicon escu-lentum × L. hirsutum: Linkage analysisand a QTL study of sexual compatibilityfactors and floral traits. Genetics 147:861–877.

Bohs, L. & R. G. Olmstead. 1997. Phylogeneticrelationships in Solanum (Solanaceae)based on ndhF sequences. Syst. Bot. 22:5–17.

— & —. 1999. Solanum phylogeny inferred fromchloroplast DNA sequence data. Pp.97–110 in M. Nee, D. E. Symon, R. N.Lester & J. P. Jessop (editors), SolanaceaeIV, Advances in Biology and Utilization.Royal Botanic Gardens, Kew.


Child, A. 1990. A synopsis of Solanum subgenusPotatoe (G. Don) D’Arcy [Tuberarium(Dun.) Bitter (s.l.)]. Feddes Repert. 101:209–235.

Correll, D. S. 1958. A new species and somenomenclatural changes in Solanum sec-tion Tuberarium. Madroño 14: 232–236.

Darwin, S. C., S. Knapp & I. E. Peralta. 2003.Taxonomy of tomatoes in theGalápagos Islands: Native and intro-duced species of Solanum sectionLycopersicon (Solanaceae). Syst.Biodiversity 12: 29–53.

D’Arcy, W. G. 1972. Solanaceae studies II:Typifications of subdivisions ofSolanum section Tuberarium. Ann.Missouri Bot. Gard. 59: 262–278.

Dunal, M. F. 1852. Solanaceae. Pp. 1–690 in A. L.P. P. de Candolle (editor), Prodromussystematis naturalis regni vegetalis 13.Treuttel and Würtz, London.

Ehlenfeldt, M. K. & R. E. Hanneman Jr. 1992.Endosperm dosage relationshipsamong Lycopersicon species. Theor.Appl. Genet. 83: 367–372.

Esquinas Alcazar, J. T. 1981. Genetic resources oftomatoes and wild relatives. Rep. Int.Board Pl. Genet. Res. No. AGP.IBPGR/80/103: 1–65. Rome.

Farris, J. S. 1989. The retention index and homo-plasy excess. Syst. Zool. 38: 406–407.

Felsenstein, J. 1985. Confidence limits on phylo-genies: An approach using bootstrap.Evolution 39: 783–791.

Futuyma, D. J. 1998. Evolutionary Biology, 3rd ed. Sinauer, Sunderland,Massachusetts.

Hanneman, R. E., Jr. 1994. Assignment ofEndosperm Balance Numbers to thetuber-bearing Solanums and their closenon-tuber-bearing relatives. Euphytica74: 19–25.

Hay, A., J. Craft & M. Tsiantis. 2004. Plant hor-mones and homeoboxes: Bridging thegap? BioEssays 26.4: 395–404.

Johnston, S. A., T. P. M. Den Nijs, S. J. Peloquin &R. E. Hanneman Jr. 1980. The signifi-cance of genetic balance to endospermdevelopment in interspecific crosses.Theor. Appl. Genet. 53: 262–269.

Kluge, A. G. & J. S. Farris. 1969. Qualitativephyletics and the evolution of anurans.Syst. Zool. 18: 1–32.

Laterrot, H. 1989. The tomato. Advantages anduse of wild varieties for varietal cre-ation. Rev. Hort. 295: 13–17.

Luckwill, L. C. 1943. The genus Lycopersicon: Anhistorical, biological, and taxonomicalsurvey of the wild and cultivated toma-toes. Aberdeen Univ. Stud. 120: 1–44.

McClean, P. E. & M. R. Hanson. 1986. Mito-chondrial DNA sequence divergenceamong Lycopersicon and relatedSolanum species. Genetics 112:649–667.

Miller, J. C. & S. D. Tanksley. 1990. RFLP analysisof phylogenetic relationships andgenetic variation in the genusLycopersicon. Theor. Appl. Genet. 80:437–448.

Müller, C. H. 1940. A revision of the genusLycopersicon. U.S.D.A. Misc. Publ. 382:1–28 + 10 pl.

255

256


Olmstead, R. G. & J. D. Palmer. 1997. Implications forphylogeny, classification, and biogeogra-phy of Solanum from cpDNA restrictionsite variation. Syst. Bot. 22: 19–29.

—, J. A. Sweere, R. E. Spangler, L. Bohs & J. D.Palmer. 1999. Phylogeny and provision-al classification of the Solanaceaebased on chloroplast DNA. Pp. 111–137in M. Nee, D. E. Symon, R. N. Lester & J.P. Jessop (editors), Solanaceae IV,Advances in Biology and Utilization.Royal Botanic Gardens, Kew.

Ortiz, R. & M. K. Ehlenfeldt. 1992. The impor-tance of Endosperm Balance Numberin potato breeding and the evolutionof tuber-bearing Solanum species.Euphytica 60: 105–113.

Palmer, J. D. & D. Zamir. 1982. Chloroplast DNAevolution and phylogenetic relation-ships in Lycopersicon. Proc. Natl. Acad.Sci. U.S.A. 79: 5006–5010.

Peralta, I. E. & D. M. Spooner. 2000. Classifica-tion of wild tomatoes: A review.Kurtziana 28: 45–54.

— & —. 2001. GBSSI gene phylogeny of wild toma-toes (Solanum L. section Lycopersicon[Mill.] Wettst. subsection Lycopersicon).Amer. J. Bot. 88: 1888–1902.

—, S. K. Knapp & D. M. Spooner. 2005. Newspecies of wild tomatoes (Solanum sec-tion Lycopersicon: Solanaceae) fromnorthern Peru. Syst. Bot.: in press.

Rick, C. M. 1960. Hybridization betweenLycopersicon esculentum and Solanumpennellii. Phylogenetic and cytogeneticsignificance. Proc. Natl. Acad. Sci.U.S.A. 46: 78–82.

—. 1963. Barriers to interbreeding inLycopersicon peruvianum. Evolution17: 216–232.

—. 1973. Potential genetic resources in tomatospecies: Clues from observations innative habitats. Pp. 255–269 in A. M.Srb (editor), Genes, Enzymes, andPopulations. Plenum, New York.

—. 1979. Biosystematic studies in Lycopersiconand closely related species of Solanum.Pp. 667–677 in J. G. Hawkes, R. N.Lester & A. D. Skelding (editors), TheBiology and Taxonomy of Solanaceae.Linn. Soc. Symp. Ser. 7. Academic Press,New York.

—. 1982a. The potential of exotic germplasmfor tomato improvement. Pp. 1–28 in I.K. Vasil, W. R. Scowcroft & K. J. Frey(editors), Plant Improvement andSomatic Cell Genetics. Academic Press,New York.

—. 1982b. Genetic relationships between self-incompatibility and floral traits in thetomato species. Biol. Zentralbl. 101:185–198.

—. 1986a. Germplasm resources in the wildtomato species. Acta Hort. 190: 39–47.

—. 1986b. Reproductive isolation in theLycopersicon peruvianum complex. Pp.477–495 in W. G. D’Arcy (editor),Solanaceae, Biology and Systematics.Columbia Univ. Press, New York.

—. 1987. Genetic resources in Lycopersicon. Pl.Biol. 4: 17–26.

—. 1988. Tomato-like nightshades: Affinities,autoecology, and breeders opportuni-ties. Econ. Bot. 42: 145–154.

— & M. Holle. 1990. Andean Lycopersicon escu-lentum var. cerasiforme: Genetic varia-tion and its evolutionary significance.Econ. Bot. 44: 69–78.


257

— & R. Lamm. 1955. Biosystematic studies onthe status of Lycopersicon chilense.Amer. J. Bot. 42: 663–675.

— & S. D. Tanksley. 1981. Genetic variation inSolanum pennellii: Comparison withtwo other sympatric tomato species. Pl.Syst. Evol. 139: 11–45.

—, E. Kesicki, J. F. Fobes & M. Holle. 1976.Genetic and biosystematic studies ontwo new sibling species of Lycopersiconfrom Interandean Peru. Theor. Appl.Genet. 47: 55–68.

—, J. F. Fobes & M. Holle. 1977. Genetic variationin Lycopersicon pimpinellifolium:Evidence of evolutionary change in mat-ing systems. Pl. Syst. Evol. 127: 139–170.

—, M. Holle & R. W. Thorp. 1978. Rates of cross-pollination in Lycopersicon pimpinellifoli-um: Impact of genetic variation in floralcharacters. Pl. Syst. Evol. 129: 31–44.

—, J. F. Fobes & S. D. Tanksley. 1979. Evolution ofmating systems in Lycopersicum hirsu-tum as deduced from genetic variationin electrophoretic and morphologicalcharacters. Pl. Syst. Evol. 132: 279–298.

—, J. W. Deverna, R. T. Chetelat & M. A. Stevens.1987. Potential contributions of widecrosses to improvement of processingtomatoes. Acta Hort. 200: 45–55.

—, H. Laterrot & J. Philouze. 1990. A revised keyfor the Lycopersicon species. TomatoGenet. Coop. Rep. 40: 31.

Rohlf, F. J.1992. NTSYS-pc, NumericalTaxonomy andMultivariate System. Exeter, New York.

— & R. R. Sokal. 1981. Comparing numerical tax-onomic studies. Syst. Zool. 30: 459–490.

SAS. 1995. JMP Software, Vers. 3.1. SAS InstituteInc., Cary, North Carolina.

—. 1998. SAS Proprietary Software, Vers. 7. SASInstitute Inc., Cary, North Carolina.

Sokal, R. R. 1986. Phenetic taxonomy: Theory andmethods. Ann. Rev. Ecol. Syst. 17: 423–442.

Spooner, D. M., G. J. Anderson & R. K. Jansen.1993. Chloroplast DNA evidence forthe interrelationships of tomatoes,potatoes, and pepinos (Solanaceae).Amer. J. Bot. 80: 676–688.

Stevens, M. A. & C. M. Rick. 1986. Genetics andbreeding. Pp. 35–109 in J. G. Atherton& J. Rudich (editors), The Tomato Crop:A Scientific Basis for Improvement.Chapman and Hall, London.

Swofford, D. L. 1998. PAUP*: PhylogeneticAnalysis Using Parsimony, Vers. 4d64.Sinauer, Sunderland, Massachusetts.

Tanksley, S. D. & F. Loaiza-Figueroa. 1985.Gametophytic self incompatibility is con-trolled by a single locus on chromosome1 in Lycopersicon peruvianum. Proc.Natl. Acad. Sci. U.S.A. 82: 5093–5096.

— & S. R. McCouch. 1997. Seed banks and molec-ular maps: Unlocking genetic potentialfrom the wild. Science 277: 1063–1066.

Taylor, I. B. 1986. Biosystematics of the tomato.Pp. 1–34 in J. G. Atherton & J. Rudich(editors), The Tomato Crop: A ScientificBasis for Improvement. Chapman andHall, London.

Villagran, C., M. T. K. Arroyo & C. Marticorena.1983. Efectos de la desertización en ladistribución de la flora andina de Chile.Revista Chilena Hist. Nat. 56: 137–157.

morphological characterization and relationships of … and... · morphological characterization...

Documents