Journal of Heredity 2003:94(5):407–415 Ó 2003 The American Genetic Association

DOI: 10.1093/jhered/esg076

Microsatellite Markers in Avocado

(Persea americana Mill.): Genealogical
Relationships Among Cultivated
Avocado Genotypes
V. E. T. M. ASHWORTH AND M. T. CLEGG
From the Department of Botany and Plant Sciences, University of California, Riverside, CA 92521.

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

Address correspondence to Dr. Ashworth at the address above.

Abstract
Twenty-five microsatellite markers uniquely differentiated 35 avocado cultivars and two wild relatives. Average
heterozygosity was high (60.7%), ranging from 32% in P. steyermarkii to 84% in Fuerte and Bacon. In a subset of 15
cultivars, heterozygosity averaged 63.5% for microsatellites, compared to 41.8% for restriction fragment length
polymorphisms (RFLPs). A neighbor-joining tree, according to average shared allele distances, consisted of three clusters
likely corresponding to the botanical races of avocado and intermediate clusters uniting genotypes of presumably racially
hybrid origin. Several results were at odds with existing botanical assignments that are sometimes rendered difficult by
incomplete pedigree information, the complexity of the hybrid status (multiple backcrossing), or both. For example, cv.
Harvest clustered with the Guatemalan race cultivars, yet it is derived from the Guatemalan 3 Mexican hybrid cv. Gwen.
Persea schiedeana grouped with cv. Bacon. The rootstock G875 emerged as the most divergent genotype in our data set.
Considerable diversity was found particularly among accessions from Guatemala, including G810 (West Indian race), G6
(Mexican race), G755A (hybrid Guatemalan 3 P. schiedeana), and G875 (probably not P. americana). Low bootstrap support,
even upon exclusion of (known) hybrid genotypes from the data matrix, suggests the existence of ancient hybridization or
that the botanical races originated more recently than previously thought.

Relationships within avocado reflect a long history of human diversity were distinguished in highland Mexico, highland
cultivation and selection. The earliest records of avocado Guatemala, and lowland (coastal) Guatemala to Costa Rica,
consumption come from plant remains found in cave respectively, each characterized by its own locally adapted
dwellings in the Tehuacán area of Puebla State in central type of avocado. Taxonomic evidence (Bergh and Ellstrand
Mexico (Smith 1966). These date back to about 7000–8000 1986; Popenoe 1941) suggests that the three avocado types
B.C. Similar evidence from other caves nearby (Smith 1966) are distinct botanical races of P. americana. Today they are
and from the Oaxaca Valley (Smith 1969) suggests that commonly known as the Mexican, Guatemalan, and (in-
consumption and possibly selection were taking place from appropriately) West Indian races. The distinction of three
4000–2800 B.C. By the time of the Spanish Conquest in the races is not a recent observation but was noted as long ago as
early to mid-sixteenth century, local peoples from Mexico 1653 by Father Bernabé Cobo and various other chroniclers
and as far as Peru had long been cultivating avocado (Gama-Campillo and Gómez-Pompa 1992; Popenoe 1963).
(Popenoe 1963; Storey et al. 1986). Linguistic evidence
(Gama-Campillo and Gómez-Pompa 1992) also suggests
Three Botanical Races
that the avocado had been used by indigenous populations
for a considerable time before the advent of European The differences between the three races relate primarily to
discoverers. the ecological preferences of the tree (temperature, humidity)
Botanical explorations in Central America over the past and fruit characteristics (skin texture and color, oil content),
century have sought to catalogue the great diversity of forms Guatemalan avocados being somewhat intermediate be-
encountered, including wild Persea species and especially local tween the other two (Bergh and Ellstrand 1986; Knight 1999;
P. americana selections (cultigens). Three main centers of Williams 1976). Mexican race avocados are characterized by

407
Journal of Heredity 2003:94(5)

good adaptation to a Mediterranean climate, relatively good flowering (with an estimated 106 flowers per tree; Robinson
cold tolerance, anise-scented leaves, and fruits covered by 1926), a large proportion of flowers are shed, and only a tiny
a thin, purplish-black skin. Guatemalan race avocados are fraction actually sets fruit. Bee pollination inside cages or
somewhat cold tolerant and are conspicuous primarily nets is conventionally used in breeding situations to perform
because of their thick, tough skin, which remains green until controlled cross- or self-pollination (Bergh 1969). However,
maturity. West Indian genotypes are adapted to humid this method is not reliable, because bees can introduce
tropical conditions and are hence very cold sensitive but extraneous pollen on their bodies (Menge J, personal
show tolerance of soil salinity and other adverse edaphic communication), do not thrive on pure avocado pollen
conditions. The lower oil content of the fruit flesh confers (Williams 1976), and may not be the only pollinators inside
a slightly watery, almost sweet taste not found in the other a cage. Pedigrees that claim a given paternal origin thus may
two botanical races. not always be accurate.
The apparent distinctiveness of the three races is The complex legacy of ancient and recent avocado
consistent with the notion that very little exchange has improvement has left us with a profusion of genotypes of
taken place between respective centers of diversity until uncertain affinities and with diffuse racial boundaries. Several
comparatively recently (Williams 1976), with geographic researchers (Bergh et al. 1973; Chao CC and Devanand PS,
isolation accounting for much of the interracial variation. It personal communication; Davis et al. 1998; Furnier et al.
is unclear which or how many wild species or cultigens were 1990; Mhameed et al. 1997; Rhodes et al. 1971; Torres et al.

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

the progenitors of the extant avocado races. Ethnobotanical 1978) have used morphological and molecular tools to
accounts (Storey et al. 1986; Williams 1976) suggest that dissect relationships between the main cultivars, the three
topographic and climatic barriers and the bulky size of the botanical races, and wild Persea relatives. However, their
avocado seed prevented the races from coming together until findings are not statistically evaluated, and often no effort is
after 1513, when Balboa discovered the Pacific Ocean. made to subject the three botanical races to closer scrutiny.
Nonetheless, the centers of origin postulated for the This article reexamines the complex relationships in
botanical races are in relatively close proximity, and those avocado against the backdrop of a more thorough account of
of the Mexican and Guatemalan races overlap to some the breeding history and in the face of missing or inaccurate
degree in the Guatemalan and other Central American pedigree information. It is the first to use a large number
highlands. Indeed, numerous collecting accounts by Eugenio of microsatellite markers to infer genealogical alignments
Schieber and George Zentmyer, especially in the 1970s and among 35 cultivars, rootstocks, and wild relatives of P.
1980s (e.g., Schieber and Zentmyer 1973, 1981; Schieber et americana and to apply a measure of statistical support
al. 1983; Zentmyer and Schieber 1982) document this (bootstrap values). We also compare our results with those of
overlap and instances of suspected racial introgression. Of Davis et al. (1998) who employed restriction fragment length
the three racial ecotypes, the West Indian race migrated the polymorphism (RFLP) markers for a similar set of
farthest from its putative center of origin on the Pacific coast genotypes.
of Guatemala (Storey et al. 1986) and did so much earlier on,
reaching Peru about 3,000–4,000 years ago (Williams 1976).
Microsatellite Markers
Spanish and other European seafarers introduced West
Indian avocados to the Philippines and later (starting about We are working with microsatellites because they are
1823) to Hawaii. Ironically, the West Indian race reached versatile genetic markers that combine the useful properties
neither the east coast of Central America nor the West Indies of high variability, codominant inheritance, and good
until the sixteenth century. reproducibility (Litt and Luty 1989; Smeets et al. 1989;
Tautz 1989; Weber and May 1989). Their codominance
makes them suitable for tracing paternity and tracking pollen
Breeding
movement (see Jarne and Lagoda 1996; Queller et al. 1993;
Although some form of selection and improvement has been and for reviews see Goldstein and Pollock 1997; Goldstein
practiced since ancient times (Smith 1966; Williams 1976), and Schlötterer 1999; and Sunnucks 2000). Additional
practically nothing is known until the end of the nineteenth benefits derive from their relative abundance and even
century. Much of the early breeding efforts in Florida and distribution across the genome (Hamada et al. 1982; Stallings
California centered on private individuals and nursery et al. 1991; Weissenbach et al. 1992; but see also Roots and
owners using avocado seeds collected in Cuba and Mexico, Baker 2002; Schmidt and Heslop-Harrison 1996).
respectively (Knight 1999). Clonal propagation by grafting
onto a rootstock was not developed until 1900 by George
Cellons (Knight 1999). Since the mid-twentieth century, Materials and Methods
breeding in California has focused on creating Guatemalan
Plant Material
3 Mexican hybrids, while Florida has concentrated on
producing new Guatemalan West Indian hybrid varieties. Most plant material was taken from trees in the germplasm
Even today, the precise pedigree of a cultivar is rarely and breeding blocks maintained at the University of
known, because the flowering habit of avocado makes California South Coast Research and Extension Center,
controlled (hand) pollination impracticable: despite prolific Irvine, or from collections at UC Riverside Agricultural

408
 Ashworth and Clegg  Microsatellite Marker in Avocado

Operations. Leaf material (fresh or stored at
808C) was restriction enzymes (EcoRI, EcoRV, and HindIII). For the
ground in liquid nitrogen, and DNA was extracted with the distance matrix, shared alleles for a given locus (probe) were
Qiagen DNeasyÒ Plant Mini Kit (Qiagen, Valencia, CA). averaged for each of the three digests.
Table 1 lists the origins and salient information of the 37
genotypes used in this study. Wherever possible, cultivars are
assigned to one of the three botanical races (Mexican,
Results
Guatemalan, and West Indian) or a hybrid combination A total of 25 microsatellite loci (10 dinucleotide and 15
thereof, in accordance with information from the Variety trinucleotide) uniquely distinguished 37 avocado genotypes.
Database of the University of California at Riverside Web Allele number per locus ranged from 3 to 21 and averaged
site (Arpaia 2002) and from articles in the California Avocado 10.4 alleles/locus. Heterozygosity ranged from 32% (8 of 25
Society Yearbook (McCormac 2002). We distinguish between loci) in P. steyermarkii to 84% (21 of 25 loci) in Fuerte and
rootstocks and cultivars, because they reflect the outcome of Bacon.
contrasting breeding strategies: cultivars are selected for Two unrooted Neighbor Joining phenograms are
good fruit-bearing properties, whereas rootstock genotypes presented, one generated from average shared alleles for all
must be adapted to soil-specific conditions, such as tolerance 37 genotypes (Figure 1) and the other from a consensus tree
of root rot (Phytophthora cinnamomi) and salinity, but their fruit of 1,000 resampled data sets containing bootstrap values,
characteristics are of less relevance. with four accessions omitted (Figure 2). Both phenograms

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

combine a core of six Mexican race genotypes (Duke 7, Duke
9, G6, Thomas, Topa Topa, UC2001, and Walter Hole) into
Marker Development and Data Manipulation:
the same cluster with a bootstrap support of 382. In Figure 1
Because of the paucity of DNA sequences in the DNA two additional accessions (a rogue tree and Toro Canyon)
sequence databases, we chose to develop our own markers attach to the Mexican cluster. The Guatemalan race
based on a microsatellite-enriched DNA library of cv. Hass. avocados Linda, Nimlioh, and Nabal associate into a cluster
In brief, we sequenced clones from the microsatellite- that also includes CRI-71 and Harvest. A high bootstrap
enriched library on a LI-COR DNA 4200 Long Read value (923) unites Linda and Nimlioh, and a moderately high
Sequencer (LI-COR, Lincoln, NE), using 6% acrylamide gels value (436) supports the cohesion of these two genotypes
to check for the presence of a suitable repeat. Primers were with Nabal and CRI-71. The West Indian cluster is tentative
designed with use of Oligo Primer Analysis Software version and assigned on the basis of cv. Arue. In both phenograms
4.01 (National Biosciences, Inc., Plymouth, MN), achieving Arue associates with P. steyermarkii and G810.
PCR product sizes of 100–400 bp. PCR products for a panel Many of the cultivars of putatively Mexican 3 Guate-
of avocado genotypes were visualized first by using malan hybrid status cluster in an intermediate position on the
radioactivity (incorporated as 32P-dCTP; NEN EasyTides, phenogram. A compact cluster in Figure 1 comprises Hass,
PerkinElmer Life Sciences) to verify the interpretability of Gwen, and seven cultivars having Gwen as their maternal
a microsatellite locus. Informative loci were applied to parent, as well as OA184 and Pinkerton. This cluster is also
fragment analysis (ABI GeneScan) on an ABI 377 present in Figure 2 but is joined by cv. Sir Prize. Genotype
Automated DNA Sequencer (PerkinElmer-Applied Biosys- G755A, a hybrid between P. schiedeana and an unspecified
tems) using forward primers labeled with the fluorescent Guatemalan race avocado (Bergh and Ellstrand 1986),
dyes 6-FAM, VIC, or NED. All reactions were performed on appears between the clusters that include its putative parents.
a Stratagene Robocycler (Stratagene Inc., La Jolla, CA). The In both figures, P. schiedeana Nees and P. steyermarkii C. K.
following settings were used: 2 min at 958C, 30 cycles of 958C Allen, both wild relatives of cultivated avocado, are very
for 1 min, 50–688C (depending on the primer) for 1 min, and divergent from one another. Always close to P. schiedeana is
728C for 1 min, with a final extension of 45 min at 728C. A a cluster composed of Bacon, Zutano, and Ettinger. A
more detailed account of the methodology of marker bootstrap of 570 supports the association between Bacon
development and experimental procedures, including primer and P. schiedeana. The most divergent genotype (based on the
sequences, are given elsewhere (Ashworth VETM et al., fewest alleles shared with other genotypes) is G875, an
unpublished data). accession collected in Guatemala. The extended genetic
This study uses a total of 25 microsatellite loci. Genetic distance of this genotype was further suggested by
distances were derived from calculations of average shared amplification failure at several microsatellite loci, implying
alleles, as described in Davis et al. 1998. The distance matrix DNA sequence divergence at the primer annealing sites. The
was imported into PAUP (Swofford 2002), and genealogical association of G875 with P. schiedeana may be an artifact of
relationships were depicted as an unrooted tree with use of long-branch attraction.
the Neighbor Joining algorithm. Bootstrap values (1,000 Very different placements on the two phenograms were
replications) were calculated in PHYLIP version 3.57c observed for Dusa, Fuerte, Velvick, and Sir Prize. In Figure 1
(Felsenstein 1989), for which the data set was converted to these four genotypes unite into the same assemblage near the
a gene frequency format and four accessions were omitted Guatemalan cluster, whereas they are scattered in Figure 2:
because of missing data. The average shared alleles data were Dusa and Fuerte attach near the Mexican cluster, Velvick
also used to generate a distance matrix for the RFLP data set attaches with the putative West Indian cluster, and Sir Prize
that consisted of 14 probes, each digested by three different is embedded in the cluster of Gwen progeny genotypes.

409
Journal of Heredity 2003:94(5)

Table 1. Salient characteristics and origin, where known, of 37 avocado genotypes used in this study (flowering is either A-type
or B-type, with pollination being optimal among opposite flowering types)

Genotype Characteristics and origin Botanical race

5-186 Seedling of Gwen; UCRBP, B-type; Hass-like fruit G3M
5-552 Seedling of Gwen; UCRBP, B-type; Hass-like fruit G3M
Arue From Society Islands, 1932; fruit 20–30 oz, skin rough, seed large, A-type WI?
Bacon Chance seedling originating in Buena Park, California, 1928; fruit ovoid, 7–12 oz; G3M
skin green, thin, and smooth; flesh very pale yellow-green; B-type; excellent
frost tolerance; tree tall and slender
BL516 (Marvel) Seedling of Gwen; UCRBP; fruit Hass-like, pear to oval-pear, 6–9 oz, green and G3M
black when hard, black when soft; seed very small to medium
BL667 (Nobel) Seedling of Gwen; UCRBP, 1997; fruit Hass-like, 8–14 oz; skin green and black G3M
when hard, black when soft, smoother than Hass; seed medium; B-type
CRI-71 Rootstock; collection from Costa Rica (CASY 75:26) G
Duke 7 Rootstock; probably progeny of Duke (CASY 61:17) M
Duke 9 Rootstock; irradiated seedling of Duke M
Dusa Seedling of Duke 7; from South Africa; CASY 78:124 M
Ettinger Seedling of Fuerte (1947 Israel, 1954 USA); fruit pyriform, 6–12 oz; skin green, G3M

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

rough, ‘‘Fuerte-type’’; seed large, coat adheres to flesh; flesh yellowish; tree with
strong central leader, drooping laterals; B-type
Fuerte From Atlixco, Mexico, 1911; fruit pyriform, 16 oz; skin dark green with small G3M
raised pale spots, thin; seed medium; alternate bearing; tree open, spreading, tall
G6 Rootstock; collection from near Antigua, Guatemala; fruit small, skin purplish M
black (CASY 72: 243–248; 67:87ff, 93ff)
G755A Rootstock; collection from Cobán, Guatemala G 3 P. schiedeana
G810 Rootstock; collection from Guatemala (CASY 75: 26)
G875 Rootstock; collection from Guatemala (CASY 75: 26)
GEM Seedling of Gwen, UCRBP; fruit 7–15 oz; skin green or black when hard, black G3M
when soft, thick and pebbly; seed medium to large, tight in cavity; A-type
Gwen Seedling of Hass 3 Thille, UCRBP; fruit pyriform to round, 10 oz; skin green, G3M
moderately thick, rough; seed small; A-type
Harvest Seedling of Gwen; UCRBP; fruit Hass-like, more round-oval; skin black G3M
Hass Originating in La Habra Heights, California, 1926; fruit pyriform, 7–10 oz; skin G
turning dark on tree, black when soft, pebbled, leathery; seed small, tight in
cavity; flesh creamy; A-type; tree starts bearing second year
Lamb/Hass Seedling of Gwen, UCRBP; fruit Hass-like with flat shoulder, 10–18 oz; skin G3M
black when mature, medium thick; seed small to medium; A-type; leaves darker
than Hass; tree upright, canopy density greater than Hass
Linda Originating in Antigua, Guatemala, 1914; fruit round to oblong, 16–48 oz; skin G
dull-purple, smooth, medium thick; seed small, tight in cavity; flesh yellow; tree
low and spreading; regular bearing; B-type
Nabal Originating in Antigua, Guatemala, 1917; fruit nearly spherical, 12–17 oz; skin G
green, smooth; seed medium–small, tight in cavity; flesh yellow; marked
alternate bearing; B-type
Nimlioh Originating in Antigua, Guatemala, 1917; fruit round, 28–40 oz; seed medium, G
tight in cavity; skin thick, black and rough; alternate bearing; B-type
OA184 UCRBP; fruit Hass-like G3M
P. schiedeana Wild species from Guatemala/southern Mexico Thought to be
progenitor of
modern varieties
P. steyermarkii Wild species from Guatemala; sometimes treated as variety of P. Americana Thought to be
progenitor of the
Guatemalan race
Pinkerton Seedling of Hass’ Rincon (1974); fruit pyriform, 8–14 oz, skin green, leathery; G3M
seed small, separates well from flesh; tree habit low and spreading; A-type
Rogue Tree of unknown parentage, probably rootstock; leaves anise-scented M
Sir Prize Fruit lopsided oval pear-shaped; skin black, not thick or pebbly; B-type; cold M
tolerant; seedling of HX48 (a Hass progeny genotype)
Thomas Rootstock; fruit ovate, low quality; skin black, thin, and smooth; seed large; M
B-type; survivor tree showing resistance to Phytophthora cinnamomi root rot
Topa Topa Originating in Ojai, California, 1907; fruit oblique pyriform, 6–10 oz, poor M
eating quality; skin black, glossy, and smooth; seed used for grafting of
rootstocks; A-type
Toro Canyon Rootstock; survivor tree showing resistance to Phytophthora cinnamomi root rot M

410
 Ashworth and Clegg  Microsatellite Marker in Avocado

Table 1. Continued

Genotype Characteristics and origin Botanical race

UC2001 Rootstock; seedling of Duke 7 M
Velvick Bud wood of seedling resistant to Phytophthora cinnamomi; G
rootstock from Australia
Walter Hole Fruit pyriform, 3–5 oz; skin purplish-black, thin, and smooth; seed small; M
flavor fair
Zutano Originating in Fallbrook, California, 1926; fruit pyriform, 8–12 oz; skin very M
thin, pale green; flesh watery; seed medium; cold tolerant; tree upright; B-type

Data are taken from the Variety Database of the University of California Riverside Web site (http://ucavo.ucr.edu) or (for rootstock varieties) from articles
in CASY (California Avocado Society Yearbook). UCRBP ¼ UC Riverside Breeding Program.

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

Figure 1. Neighbor-joining consensus tree of 1,000 bootstrap replicates generated from allele frequencies at 25 microsatellite
loci for 33 avocado genotypes. Many bootstrap values are low, reflecting the large number of hybrid cultivars in the data set.
Dotted lines surround genotype assemblages belonging to the three botanical races of avocado—Guatemalan, Mexican, and West
Indian. The West Indian cluster is assigned based on cv. Arue, whose ancestry is presumed to be West Indian. Intermediate clusters
unite genotypes of various hybrid origins.

411
Journal of Heredity 2003:94(5)

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

Figure 2. Neighbor-joining tree of genetic distances generated from average shared alleles at 25 microsatellite loci for 37
avocado genotypes. Clusters corresponding to the three botanical races (Guatemalan, Mexican, and West Indian) are encircled by
dotted lines. Additional clusters are formed by genotypes of hybrid origin. A cluster not present in Figure 1 is composed of cvs.
Dusa, Fuerte, Sir Prize, and Velvick.

Genealogical relationships inferred from our micro- constituent accessions and the low bootstrap values. Indeed,
satellite data broadly agree with those from the RFLP data of bootstrap support is weak throughout, with only six
Davis et al. (1998) when analyzed for a subset of 15 cultivars branches supported by values in excess of 500 (of 1,000)
that were common to both data sets (phenogram not replicates. We hypothesize that weak bootstraps result from
shown). However, in the RFLP phenogram Bacon and the large number of hybrid genotypes in our data set.
Zutano are embedded in the Mexican cluster and removed Hybrids unite alleles from two genetically distinct sources,
from the vicinity of P. schiedeana. Because of the lower allelic thereby decreasing the genetic distance between those
diversity of the RFLP data, compared to the microsatellite sources. To test the influence of hybrids on bootstrap
data (on average 3.8 alleles/locus; see Table 2), resolution of values, we performed a separate analysis on a data set from
the Guatemalan, West Indian, and Guatemalan 3 Mexican which (known) hybrids were removed. However, the
genotypes is poor. Heterozygosity averaged 41.8% (com- bootstrap values remained low, suggesting that additional
pared to 63.5% for microsatellite loci), with values for factors may be at play, including the possibility of ancient
individual genotypes ranging from 14.3% (Nimlioh) to hybridization or a more recent date for racial differentiation
78.6% (Zutano and G755A; Table 2). than previously thought.
In the genealogies of Figures 1 and 2, we have circled
clusters that are most likely to correspond to the three
Discussion botanical races of avocado, based on phenotypic, ecological,
Differences between the genealogies of Figures 1 and 2 are and historical information. However, numerous additional
modest and perfectly consistent with the unequal number of clusters remain that presumably represent instances of

412
 Ashworth and Clegg  Microsatellite Marker in Avocado

Table 2. Allelic diversity at 14 RFLP loci (a–n) for 15 avocado cultivars and one wild relative (P. schiedeana)

a b c d e f g h i j k l m n H1 H2
Arue WI 1 1 2 1 1 1 1 1 2 1 1 1 2 1 21.4 50.0
Nabal G 1 1 2 1 1 1 1 1 2 1 1 2 1 2 28.6 68.0
Linda G 1 1 1 1 1 2 1 1 1 1 2 2 2 1 28.6 48.0
Nimlioh G 1 1 1 1 1 2 1 1 2 1 1 1 1 1 14.3 68.0
Fuerte 3 2 2 2 1 1 1 1 1 2 1 2 2 1 2 50.0 84.0
Gwen 3 1 1 1 1 1 1 1 1 2 1 1 2 2 2 28.6 64.0
Hass 3 1 1 2 1 2 2 1 1 2 1 2 2 1 2 50.0 72.0
Bacon 3 2 1 2 2 2 2 1 2 2 2 1 1 2 1 64.3 84.0
Zutano M 2 2 2 2 2 2 1 2 2 1 1 2 2 2 78.6 0.0
Pinkerton 3 1 1 2 1 1 2 1 2 2 2 2 2 2 2 64.3 4.0
Duke 7 M 2 2 2 2 1 2 1 1 2 2 1 1 1 2 57.15 6.0
Thomas M 1 2 2 1 2 2 1 2 2 1 1 1 1 2 50.07 2.0
Topa Topa M 1 2 1 1 2 1 1 1 2 1 1 1 1 1 21.4 50.0
P. schiedeana 1 1 1 1 1 2 1 2 1 1 2 2 — — 33.3 76.0
G755A 3 2 2 2 1 2 2 2 2 2 1 2 2 2 1 78.6 80.0

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

Total no. alleles/locus 3 4 4 2 4 6 4 3 5 3 4 4 4 3 41.8 63.5

Heterozygous loci are annotated by 2; homozygous loci by 1; average heterozygosity of a given cultivar (columns H1 and H2) is expressed as the percentage
of loci at which the genotype is heterozygous out of the total number possible (14 RFLP ¼ H1; 25 microsatellites ¼ H2). The abbreviations WI, G, and M
refer to the three botanical races, and 3 denotes hybridity (all hybrids are G 3 M, except G755A, which is a G 3 P. schiedeana hybrid).

hybrid intermediacy. Intermediacy itself is likely to be Also of Guatemalan 3 Mexican ancestry is cv. Sir Prize,
a continuum reflected in the complexity of the hybrid status yet it is sometimes designated a Mexican avocado. In neither
(multiple backcrossing). A mismatch between phenotypic phenogram does it associate with the Mexican cultivars
and molecular characters is also expected when there is but instead with either a variable cluster near the Guate-
segregation for phenotypic characters (see Bergh 1966, 1967; malan 3 Mexican assemblage (Figure 1) or with its grand-
Bergh and Whitsell 1974, 1975; and Storey et al. 1984 parent (Hass) within the Guatemalan 3 Mexican cluster
for illustrated examples). For instance, the progeny of a (Figure 2).
Guatemalan 3 Mexican hybrid segregating for Guatemalan Cv. Dusa is deemed to be a Mexican variety, having
phenotypic traits will be assigned to the Guatemalan botani- originated from Duke 7. Its placement on the phenogram
cal race, even though its molecular background remains in Figure 2 is indeed adjacent to the Mexican cluster, but
Guatemalan 3 Mexican. no such proximity is seen in Figure 1. Both genealogies
The following examples illustrate the varied and at times agree, however, in placing Dusa adjacent to Fuerte, a
unpredictable relationships between phenotype and geno- Guatemalan 3 Mexican hybrid (Popenoe 1926).
type. The two main criteria used to characterize a cultivar in Cv. Zutano and cv. Bacon are cited as being of Mexican
terms of its botanical origins are its pedigree, if known, and and Guatemalan 3 Mexican origin, respectively. In our data
its fruit and cultural characteristics. For the most part, our set the two cultivars emerge as very similar (further sup-
genealogical data corroborated phenotypic criteria (especially ported by their strong cold tolerance) and attach to the
anise-scented leaves and fruit properties), in placing most genealogy between the Mexican cultivars and the wild
rootstock genotypes in the Mexican cluster. This was true of relative, P. schiedeana. The affiliation with P. schiedeana is
rootstock genotype Thomas, a so-called rescue variety of intriguing and merits further attention.
unknown pedigree. Thomas was discovered in a severely Surprising are also some of the alignments of the wild
root rot-infested field, where it showed superior disease species and local selections in our data set. P. steyermarkii, the
tolerance, compared to the other genotypes present. In- postulated predecessor of the Guatemalan race of avocado
terestingly, rootstock genotype Spencer, another rescue (Schieber and Zentmyer 1980), associates not with the
variety, emerged as identical to Fuerte at all 25 microsatellite Guatemalan cluster but with cv. Arue. Although the origins
loci, although tree and fruit morphology are said to differ. of Arue are shrouded in mystery (it was discovered in the
Conversely, an apparent conflict between phenotypic and Society Islands in the southern Pacific), historical trade
genotypic data can help adjust pedigree information. For routes and its tropical habitat suggest that it is a West Indian
example, cv. Harvest is designated a Guatemalan 3 Mexican cultivar. The two other members of the putative West Indian
hybrid, having Gwen as its maternal parent, yet it clusters cluster are rootstock G810 and, in Figure 2, cv. Velvick.
with the (pure) Guatemalan cultivars. This suggests that the Velvick has been cited variously as a West Indian or
pollen parent must be sought among Guatemalan race Guatemalan cultivar. Our data sheds no further light on this
genotypes. All other Gwen progeny genotypes cluster with ambiguity. Rootstock G810 is a collection from Guatemala.
their maternal parent, Hass (parent of Gwen), and two Although artifactual clustering cannot be ruled out, it is
further cultivars derived from Hass (Pinkerton) or showing conceivable that G810 originated in lowland Guatemala, the
Hass-like characteristics (OA184). center of diversity of the West Indian race.

413
Journal of Heredity 2003:94(5)

Guatemala is also home to the most divergent of all 37 AvocadoWebSite%20folder/AvocadoWebSite/AvocadoVarieties/

genotypes in this data set, rootstock G875. It associates with AvocadoVarieties.html.) Accessed July 17, 2002.
P. schiedeana, but its branch is extremely long. Irrespective of Bergh BO, 1966. A Hass open-pollinated progeny set. Calif Avocado Soc
its placement, G875 is unlikely to belong to P. americana, Yearbook 50:64–77.
given that it is more divergent than either of the other two Bergh BO, 1967. Some late-maturing avocado seedlings of various
Persea relatives. Possibly, G875 will replace P. schiedeana as the parentage. Calif Avocado Soc Yearbook 51:131–158.
most distant species within subgenus Persea (Furnier et al. Bergh BO, 1969. Avocado—Persea americana Miller. In: Outlines of
1990; Kopp 1966; Williams 1976). perennial crop breeding in the tropics (Ferwerda FP and Witt F, eds),
The third Guatemalan rootstock, G6, is firmly embedded Miscellaneous Papers 4, Landbouwhogeschool Wageningen. Wageningen,
Netherlands: Veenman & Zoonen; 23–51.
within the Mexican cluster. Thus, the three rootstocks G810,
G875, and G6 are genetically dissimilar despite their common Bergh BO and Ellstrand N, 1986. Taxonomy of the avocado. Calif Avocado
geographic origin. More information is needed on their Soc Yearbook 70:135–145.
phenotypic and cultural characteristics. Bergh BO, Scora RW, and Storey WB, 1973. A comparison of leaf terpenes
in Persea subgenus Persea. Bot Gaz 134:130–134.
Bergh BO and Whitsell RH, 1974. Self-pollinated Hass seedlings. Calif
Conclusions Avocado Soc Yearbook 57:118–126.
The pronounced phenotypic variability and plasticity of Bergh BO and Whitsell RH, 1975. Self-pollinated Fuerte seedlings. Calif

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

avocado attest to its checkered history of breeding and Avocado Soc Yearbook 58:128–134.
selection. Low bootstrap support may result from frequent Davis J, Henderson D, Kobayashi M, Clegg MT, and Clegg MT, 1998.
hybridization events—both recently and before written Genealogical relationships among cultivated avocado as revealed through
records were available. Care is needed when distinguishing RFLP analyses. J Hered 89:319–323.
between geographic origin (collection site) and botanical race Felsenstein J, 1989. PHYLIP—phylogenetic inference package, version 3.2.
assignment, because descriptions such as ‘‘Guatemalan Cladistics 5:164–166.
cultivar’’ do not make this distinction clear. Moreover, Furnier GR, Cummings MP, and Clegg MT, 1990. Evolution of the avocados
confusion is introduced when a race-specific trait (e.g., green as revealed by DNA restriction fragment variation. J Hered 81:183–188.
or black skin) undergoes segregation. Indeed, our data— Gama-Campillo L and Gómez-Pompa A, 1992. An ethnoecological
though preliminary—do not suggest that the botanical races approach for the study of Persea. A case study in the Maya area. In:
are as well differentiated as hitherto assumed. Proceedings of the Second World Avocado Congress, vol. 1, Orange, CA,
Many more accessions and microsatellite markers will be April 21–26, 1991 (Lovatt CJ, ed). Riverside, CA: University of California
and California Avocado Society; 11–17.
needed to examine more thoroughly the validity and origins
of the three botanical races and the relationships among wild Goldstein DB and Pollock PD, 1997. A review of mutation processes and
methods of phylogenetic inference. J Hered 88:335–342.
relatives and the countless cultivars and rootstocks in
cultivation today. Valuable insights into crop domestication Goldstein DB and Schlötterer C (eds), 1999. Microsatellites: evolution and
applications. Oxford: Oxford University Press.
can be gained from large-scale genotyping studies (e.g.,
Matsuoka et al. 2002). In view of the complex genealogies of Hamada H, Petrino MG, and Kakunaga T, 1982. A novel repeated element
with Z-DNA-forming potential is widely found in evolutionarily diverse
most extant avocado accessions, it will be important to
eukaryotic genomes. Proc Natl Acad Sci USA 79:6465–6469.
explore additional methods of inferring historical relation-
ships, including gene sequences, and to capture the diversity Jarne P and Lagoda JL, 1996. Microsatellites, from molecules to populations
and back. Trends Ecol Evol 11:424–429.
present in wild populations and cultigens. Inevitably, the
manipulation of data necessary to generate a matrix of Knight RJ Jr, 1999. Genetic diversity in avocado. In: Proceedings of
Avocado Brainstorming ’99, Riverside, CA, October 27–28, 1999 (Arpaia
genetic distances, as applied to this study, ignores the
ML and Hofshi R, eds). Riverside, CA: California Avocado Commission
potential information inherent in allele frequencies. Methods and University of California; 16–18.
that use allele frequencies to infer population structure
Kopp LE, 1966. A taxonomic revision of the genus Persea in the western
(Pritchard et al. 2000) will be useful to formulate less hemisphere. (Perseae—Lauraceae). Mem N Y Bot Gard 14:1–117.
subjective boundaries among the three avocado races.
Litt M and Luty JA, 1989. A hypervariable microsatellite revealed by in vitro
amplification of a dinucleotide repeat within the cardiac muscle actin gene.
Amer J Hum Genet 44:397–401.
Acknowledgments Matsuoka Y, Vigouroux Y, Goodman MM, Sanchez GJ, Buckler E, and
We thank the California Avocado Commission for financial support; Drs. Doebley J, 2002. A single domestication for maize shown by multilocus
Mary Lu Arpaia and John Menge for valuable discussions; Dr. Thomas Chao microsatellite genotyping. Proc Natl Acad Sci USA 99:6080–6084.
for helpful comments on the manuscript; Eric Focht, Brandon McKee, Paul McCormac J (ed), 2002. California Avocado Society Yearbook Web Site. Available
Robinson, and Dave Stottlemyer for collecting plant material; and Peter at: http://www.avocadosource.com/CAS_Yearbooks/CAS_Yearbooks.
Morrell for advice on analytical software. htm. Accessed July 17, 2002.
Mhameed S, Sharon D, Kaufman D, Lahav E, Hillel J, Degani C, and Lavi
U, 1997. Genetic relationships within avocado (Persea americana Mill.)
References cultivars and between Persea species. Theor Appl Genet 94:279–286.
Arpaia ML, University of California Avocado Web Site. Available at Popenoe W, 1926. The parent Fuerte tree at Atlixco, Mexico. Calif Avocado
http://ucavo.ucr.edu/AvocadoWebSite. (See http://ucavo.ucr.edu/ Soc Yearbook 11:24–34.

414
 Ashworth and Clegg  Microsatellite Marker in Avocado

Popenoe W, 1941. The avocado—a horticultural problem. Trop Agric 18:3–7. Stallings RL, Ford AF, Nelson D, Torney DC, Hildebrand CE, and Moyzis
Popenoe W, 1963. Early history of the avocado. Calif Avocado Soc RK, 1991. Evolution and distribution of (GT)n repetitive sequences in
Yearbook 47:19–24. mammalian genomes. Genomics 10:807–815.

Pritchard JK, Stephens M, and Donnelly P, 2000. Inference of population Storey WB, Bergh BO, Platt RG, and Miller M, 1984. Observations on
structure using multilocus genotype data. Genetics 155:945–959. a second-generation progeny of a Mexican 3 West Indian cross. Calif
Avocado Soc Yearbook 68:161–165.
Queller DC, Strassmann JE, and Hughes CR, 1993. Microsatellites and
kinship. Trends Ecol Evol 8:285–288. Storey WB, Bergh BO, and Zentmyer GA, 1986. The origin, indigenous
range, and dissemination of the avocado. Calif Avocado Soc Yearbook
Rhodes AM, Malo SE, Campbell CW, and Garner SG, 1971. A numerical 70:127–133.
taxonomic study of the avocado (Persea americana Mill.). J Amer Soc Hort Sci
Sunnucks P, 2000. Efficient genetic markers for population biology. Trends
96:391–395.
Ecol Evol 15:199–203.
Robinson TR, 1926. Avocados for Florida. Proc Fla State Hort Soc 39:
Swofford DL, 2002. PAUP, phylogenetic analysis using parsimony (and
182–191.
other methods), version 4. Sunderland, MS: Sinauer.
Roots EH and Baker RJ, 2002. Distribution and characterization of
Tautz D, 1989. Hypervariability of simple sequences as a general source for
microsatellites in the emu (Dromaius novaehollandiae) genome. J Hered
polymorphic DNA markers. Nucleic Acids Res 17:6463–6471.
93:100–106.
Torres AM, Diedenhofen U, Bergh BO, and Knight RJ, 1978. Enzyme
Schieber E and Zentmyer GA, 1973. Collecting Persea’s in Central America
polymorphisms as genetic markers in the avocado. Amer J Bot 65:131–139.

Downloaded from http://jhered.oxfordjournals.org/ at University of Rochester on May 28, 2014

and Mexico. Calif Avocado Soc Yearbook 56:94–101.
Weber JL and May PE, 1989. Abundant class of human DNA
Schieber E and Zentmyer GA, 1980. The ‘‘Guatemalan Criollos.’’ Calif
polymorphisms which can be typed using the polymerase chain reaction.
Avocado Soc Yearbook 64:85–90. Amer J Hum Genet 44:388–396.
Schieber E and Zentmyer GA, 1981. Exploring for Persea on Volcano Weissenbach J, Gyapay G, Dib C, Vignal A, Morisette J, Millaseau P,
Quetzaltepeque, Guatemala. Calif Avocado Soc Yearbook 65:57–63. Vaysseix G, and Lathrop M, 1992. A second generation linkage map of the
Schieber E, Zentmyer GA, and Coffey MD, 1983. Variability in Mexican human genome. Nature 359:794–801.
avocados (Matuloj) in Guatemala. Calif Avocado Soc Yearbook 67:87–91. Williams LO, 1976. The botany of the avocado and its relatives. In:
Schmidt E and Heslop-Harrison JS, 1996. The physical and genomic Proceedings of the First International Tropical Fruit Short Course. The
organization of microsatellites in sugar beet. Proc Natl Acad Sci USA Avocado, Miami Beach, Florida, November 5–10, 1976 (Sauls JW, Phillips
93:8761–8765. RL, Jackson LK, eds). Gainesville, FL: University of Florida; 9–15.
Smeets AJM, Brunner HG, Ropers HH, and Wieringa B, 1989. Use of Zentmyer GA and Schieber E, 1982. Persea explorations in Honduras. Calif
variable simple sequence motifs as genetic markers: application to study of Avocado Soc Yearbook 66:93–102.
myotonic dystrophy. Hum Genet 83:245–251.
Smith CE Jr, 1966. Archaeological evidence for selection in avocado. Econ
Received December 23, 2002
Bot 20:169–175. Accepted June 17, 2003
Smith CE Jr, 1969. Additional notes on pre-conquest avocados in Mexico.
Econ Bot 23:135–140. Corresponding Editor: James L. Hamrick

415