Ó 2003 The American Genetic Association

Journal of Heredity 2003:94(5):407–415 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. 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 cultivation and selection. The earliest records of avocado consumption come from plant remains found in cave dwellings in the Tehuaca´n area of Puebla State in central Mexico (Smith 1966). These date back to about 7000–8000 B.C. Similar evidence from other caves nearby (Smith 1966) and from the Oaxaca Valley (Smith 1969) suggests that consumption and possibly selection were taking place from 4000–2800 B.C. By the time of the Spanish Conquest in the early to mid-sixteenth century, local peoples from Mexico and as far as Peru had long been cultivating avocado (Popenoe 1963; Storey et al. 1986). Linguistic evidence (Gama-Campillo and Go´mez-Pompa 1992) also suggests that the avocado had been used by indigenous populations for a considerable time before the advent of European discoverers. Botanical explorations in Central America over the past century have sought to catalogue the great diversity of forms encountered, including wild Persea species and especially local P. americana selections (cultigens). Three main centers of

diversity were distinguished in highland Mexico, highland Guatemala, and lowland (coastal) Guatemala to Costa Rica, respectively, each characterized by its own locally adapted type of avocado. Taxonomic evidence (Bergh and Ellstrand 1986; Popenoe 1941) suggests that the three avocado types are distinct botanical races of P. americana. Today they are commonly known as the Mexican, Guatemalan, and (inappropriately) West Indian races. The distinction of three races is not a recent observation but was noted as long ago as 1653 by Father Bernabe´ Cobo and various other chroniclers (Gama-Campillo and Go´mez-Pompa 1992; Popenoe 1963). Three Botanical Races The differences between the three races relate primarily to the ecological preferences of the tree (temperature, humidity) and fruit characteristics (skin texture and color, oil content), Guatemalan avocados being somewhat intermediate between the other two (Bergh and Ellstrand 1986; Knight 1999; Williams 1976). Mexican race avocados are characterized by

407

Journal of Heredity 2003:94(5)

good adaptation to a Mediterranean climate, relatively good cold tolerance, anise-scented leaves, and fruits covered by a thin, purplish-black skin. Guatemalan race avocados are somewhat cold tolerant and are conspicuous primarily because of their thick, tough skin, which remains green until maturity. West Indian genotypes are adapted to humid tropical conditions and are hence very cold sensitive but show tolerance of soil salinity and other adverse edaphic conditions. The lower oil content of the fruit flesh confers a slightly watery, almost sweet taste not found in the other two botanical races. The apparent distinctiveness of the three races is consistent with the notion that very little exchange has taken place between respective centers of diversity until comparatively recently (Williams 1976), with geographic isolation accounting for much of the interracial variation. It is unclear which or how many wild species or cultigens were the progenitors of the extant avocado races. Ethnobotanical accounts (Storey et al. 1986; Williams 1976) suggest that topographic and climatic barriers and the bulky size of the avocado seed prevented the races from coming together until after 1513, when Balboa discovered the Pacific Ocean. Nonetheless, the centers of origin postulated for the botanical races are in relatively close proximity, and those of the Mexican and Guatemalan races overlap to some degree in the Guatemalan and other Central American highlands. Indeed, numerous collecting accounts by Eugenio Schieber and George Zentmyer, especially in the 1970s and 1980s (e.g., Schieber and Zentmyer 1973, 1981; Schieber et al. 1983; Zentmyer and Schieber 1982) document this overlap and instances of suspected racial introgression. Of the three racial ecotypes, the West Indian race migrated the farthest from its putative center of origin on the Pacific coast of Guatemala (Storey et al. 1986) and did so much earlier on, reaching Peru about 3,000–4,000 years ago (Williams 1976). Spanish and other European seafarers introduced West Indian avocados to the Philippines and later (starting about 1823) to Hawaii. Ironically, the West Indian race reached neither the east coast of Central America nor the West Indies until the sixteenth century. Breeding Although some form of selection and improvement has been practiced since ancient times (Smith 1966; Williams 1976), practically nothing is known until the end of the nineteenth century. Much of the early breeding efforts in Florida and California centered on private individuals and nursery owners using avocado seeds collected in Cuba and Mexico, 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, breeding in California has focused on creating Guatemalan 3 Mexican hybrids, while Florida has concentrated on producing new Guatemalan West Indian hybrid varieties. Even today, the precise pedigree of a cultivar is rarely known, because the flowering habit of avocado makes controlled (hand) pollination impracticable: despite prolific

408

flowering (with an estimated 106 flowers per tree; Robinson 1926), a large proportion of flowers are shed, and only a tiny fraction actually sets fruit. Bee pollination inside cages or nets is conventionally used in breeding situations to perform controlled cross- or self-pollination (Bergh 1969). However, this method is not reliable, because bees can introduce extraneous pollen on their bodies (Menge J, personal communication), do not thrive on pure avocado pollen (Williams 1976), and may not be the only pollinators inside a cage. Pedigrees that claim a given paternal origin thus may not always be accurate. The complex legacy of ancient and recent avocado improvement has left us with a profusion of genotypes of uncertain affinities and with diffuse racial boundaries. Several researchers (Bergh et al. 1973; Chao CC and Devanand PS, personal communication; Davis et al. 1998; Furnier et al. 1990; Mhameed et al. 1997; Rhodes et al. 1971; Torres et al. 1978) have used morphological and molecular tools to dissect relationships between the main cultivars, the three botanical races, and wild Persea relatives. However, their findings are not statistically evaluated, and often no effort is made to subject the three botanical races to closer scrutiny. This article reexamines the complex relationships in avocado against the backdrop of a more thorough account of the breeding history and in the face of missing or inaccurate pedigree information. It is the first to use a large number of microsatellite markers to infer genealogical alignments among 35 cultivars, rootstocks, and wild relatives of P. americana and to apply a measure of statistical support (bootstrap values). We also compare our results with those of Davis et al. (1998) who employed restriction fragment length polymorphism (RFLP) markers for a similar set of genotypes. Microsatellite Markers We are working with microsatellites because they are versatile genetic markers that combine the useful properties of high variability, codominant inheritance, and good 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 movement (see Jarne and Lagoda 1996; Queller et al. 1993; and for reviews see Goldstein and Pollock 1997; Goldstein and Schlo¨tterer 1999; and Sunnucks 2000). Additional benefits derive from their relative abundance and even distribution across the genome (Hamada et al. 1982; Stallings et al. 1991; Weissenbach et al. 1992; but see also Roots and Baker 2002; Schmidt and Heslop-Harrison 1996).

Materials and Methods Plant Material Most plant material was taken from trees in the germplasm and breeding blocks maintained at the University of California South Coast Research and Extension Center, Irvine, or from collections at UC Riverside Agricultural

Ashworth and Clegg  Microsatellite Marker in Avocado

Operations. Leaf material (fresh or stored at
808C) was ground in liquid nitrogen, and DNA was extracted with the Qiagen DNeasyÒ Plant Mini Kit (Qiagen, Valencia, CA). 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, Guatemalan, and West Indian) or a hybrid combination thereof, in accordance with information from the Variety Database of the University of California at Riverside Web site (Arpaia 2002) and from articles in the California Avocado Society Yearbook (McCormac 2002). We distinguish between rootstocks and cultivars, because they reflect the outcome of contrasting breeding strategies: cultivars are selected for good fruit-bearing properties, whereas rootstock genotypes must be adapted to soil-specific conditions, such as tolerance of root rot (Phytophthora cinnamomi) and salinity, but their fruit characteristics are of less relevance. Marker Development and Data Manipulation: Because of the paucity of DNA sequences in the DNA sequence databases, we chose to develop our own markers based on a microsatellite-enriched DNA library of cv. Hass. In brief, we sequenced clones from the microsatelliteenriched library on a LI-COR DNA 4200 Long Read Sequencer (LI-COR, Lincoln, NE), using 6% acrylamide gels to check for the presence of a suitable repeat. Primers were designed with use of Oligo Primer Analysis Software version 4.01 (National Biosciences, Inc., Plymouth, MN), achieving PCR product sizes of 100–400 bp. PCR products for a panel of avocado genotypes were visualized first by using radioactivity (incorporated as 32P-dCTP; NEN EasyTides, PerkinElmer Life Sciences) to verify the interpretability of a microsatellite locus. Informative loci were applied to fragment analysis (ABI GeneScan) on an ABI 377 Automated DNA Sequencer (PerkinElmer-Applied Biosystems) using forward primers labeled with the fluorescent dyes 6-FAM, VIC, or NED. All reactions were performed on a Stratagene Robocycler (Stratagene Inc., La Jolla, CA). The following settings were used: 2 min at 958C, 30 cycles of 958C for 1 min, 50–688C (depending on the primer) for 1 min, and 728C for 1 min, with a final extension of 45 min at 728C. A more detailed account of the methodology of marker development and experimental procedures, including primer sequences, are given elsewhere (Ashworth VETM et al., unpublished data). This study uses a total of 25 microsatellite loci. Genetic distances were derived from calculations of average shared alleles, as described in Davis et al. 1998. The distance matrix was imported into PAUP (Swofford 2002), and genealogical relationships were depicted as an unrooted tree with use of the Neighbor Joining algorithm. Bootstrap values (1,000 replications) were calculated in PHYLIP version 3.57c (Felsenstein 1989), for which the data set was converted to a gene frequency format and four accessions were omitted because of missing data. The average shared alleles data were also used to generate a distance matrix for the RFLP data set that consisted of 14 probes, each digested by three different

restriction enzymes (EcoRI, EcoRV, and HindIII). For the distance matrix, shared alleles for a given locus (probe) were averaged for each of the three digests.

Results A total of 25 microsatellite loci (10 dinucleotide and 15 trinucleotide) uniquely distinguished 37 avocado genotypes. Allele number per locus ranged from 3 to 21 and averaged 10.4 alleles/locus. Heterozygosity ranged from 32% (8 of 25 loci) in P. steyermarkii to 84% (21 of 25 loci) in Fuerte and Bacon. Two unrooted Neighbor Joining phenograms are presented, one generated from average shared alleles for all 37 genotypes (Figure 1) and the other from a consensus tree of 1,000 resampled data sets containing bootstrap values, with four accessions omitted (Figure 2). Both phenograms combine a core of six Mexican race genotypes (Duke 7, Duke 9, G6, Thomas, Topa Topa, UC2001, and Walter Hole) into the same cluster with a bootstrap support of 382. In Figure 1 two additional accessions (a rogue tree and Toro Canyon) attach to the Mexican cluster. The Guatemalan race avocados Linda, Nimlioh, and Nabal associate into a cluster that also includes CRI-71 and Harvest. A high bootstrap value (923) unites Linda and Nimlioh, and a moderately high value (436) supports the cohesion of these two genotypes with Nabal and CRI-71. The West Indian cluster is tentative and assigned on the basis of cv. Arue. In both phenograms Arue associates with P. steyermarkii and G810. Many of the cultivars of putatively Mexican 3 Guatemalan hybrid status cluster in an intermediate position on the phenogram. A compact cluster in Figure 1 comprises Hass, Gwen, and seven cultivars having Gwen as their maternal parent, as well as OA184 and Pinkerton. This cluster is also present in Figure 2 but is joined by cv. Sir Prize. Genotype G755A, a hybrid between P. schiedeana and an unspecified Guatemalan race avocado (Bergh and Ellstrand 1986), appears between the clusters that include its putative parents. In both figures, P. schiedeana Nees and P. steyermarkii C. K. Allen, both wild relatives of cultivated avocado, are very divergent from one another. Always close to P. schiedeana is a cluster composed of Bacon, Zutano, and Ettinger. A bootstrap of 570 supports the association between Bacon and P. schiedeana. The most divergent genotype (based on the fewest alleles shared with other genotypes) is G875, an accession collected in Guatemala. The extended genetic distance of this genotype was further suggested by amplification failure at several microsatellite loci, implying DNA sequence divergence at the primer annealing sites. The association of G875 with P. schiedeana may be an artifact of long-branch attraction. Very different placements on the two phenograms were observed for Dusa, Fuerte, Velvick, and Sir Prize. In Figure 1 these four genotypes unite into the same assemblage near the Guatemalan cluster, whereas they are scattered in Figure 2: Dusa and Fuerte attach near the Mexican cluster, Velvick attaches with the putative West Indian cluster, and Sir Prize 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 5-552 Arue Bacon

G3M G3M WI? G3M

OA184 P. schiedeana

Seedling of Gwen; UCRBP, B-type; Hass-like fruit Seedling of Gwen; UCRBP, B-type; Hass-like fruit From Society Islands, 1932; fruit 20–30 oz, skin rough, seed large, A-type Chance seedling originating in Buena Park, California, 1928; fruit ovoid, 7–12 oz; skin green, thin, and smooth; flesh very pale yellow-green; B-type; excellent frost tolerance; tree tall and slender Seedling of Gwen; UCRBP; fruit Hass-like, pear to oval-pear, 6–9 oz, green and black when hard, black when soft; seed very small to medium Seedling of Gwen; UCRBP, 1997; fruit Hass-like, 8–14 oz; skin green and black when hard, black when soft, smoother than Hass; seed medium; B-type Rootstock; collection from Costa Rica (CASY 75:26) Rootstock; probably progeny of Duke (CASY 61:17) Rootstock; irradiated seedling of Duke Seedling of Duke 7; from South Africa; CASY 78:124 Seedling of Fuerte (1947 Israel, 1954 USA); fruit pyriform, 6–12 oz; skin green, rough, ‘‘Fuerte-type’’; seed large, coat adheres to flesh; flesh yellowish; tree with strong central leader, drooping laterals; B-type From Atlixco, Mexico, 1911; fruit pyriform, 16 oz; skin dark green with small raised pale spots, thin; seed medium; alternate bearing; tree open, spreading, tall Rootstock; collection from near Antigua, Guatemala; fruit small, skin purplish black (CASY 72: 243–248; 67:87ff, 93ff) Rootstock; collection from Coba´n, Guatemala Rootstock; collection from Guatemala (CASY 75: 26) Rootstock; collection from Guatemala (CASY 75: 26) Seedling of Gwen, UCRBP; fruit 7–15 oz; skin green or black when hard, black when soft, thick and pebbly; seed medium to large, tight in cavity; A-type Seedling of Hass 3 Thille, UCRBP; fruit pyriform to round, 10 oz; skin green, moderately thick, rough; seed small; A-type Seedling of Gwen; UCRBP; fruit Hass-like, more round-oval; skin black Originating in La Habra Heights, California, 1926; fruit pyriform, 7–10 oz; skin turning dark on tree, black when soft, pebbled, leathery; seed small, tight in cavity; flesh creamy; A-type; tree starts bearing second year Seedling of Gwen, UCRBP; fruit Hass-like with flat shoulder, 10–18 oz; skin black when mature, medium thick; seed small to medium; A-type; leaves darker than Hass; tree upright, canopy density greater than Hass Originating in Antigua, Guatemala, 1914; fruit round to oblong, 16–48 oz; skin dull-purple, smooth, medium thick; seed small, tight in cavity; flesh yellow; tree low and spreading; regular bearing; B-type Originating in Antigua, Guatemala, 1917; fruit nearly spherical, 12–17 oz; skin green, smooth; seed medium–small, tight in cavity; flesh yellow; marked alternate bearing; B-type Originating in Antigua, Guatemala, 1917; fruit round, 28–40 oz; seed medium, tight in cavity; skin thick, black and rough; alternate bearing; B-type UCRBP; fruit Hass-like Wild species from Guatemala/southern Mexico

P. steyermarkii

Wild species from Guatemala; sometimes treated as variety of P. Americana

Pinkerton

Seedling of Hass’ Rincon (1974); fruit pyriform, 8–14 oz, skin green, leathery; seed small, separates well from flesh; tree habit low and spreading; A-type Tree of unknown parentage, probably rootstock; leaves anise-scented Fruit lopsided oval pear-shaped; skin black, not thick or pebbly; B-type; cold tolerant; seedling of HX48 (a Hass progeny genotype) Rootstock; fruit ovate, low quality; skin black, thin, and smooth; seed large; B-type; survivor tree showing resistance to Phytophthora cinnamomi root rot Originating in Ojai, California, 1907; fruit oblique pyriform, 6–10 oz, poor eating quality; skin black, glossy, and smooth; seed used for grafting of rootstocks; A-type Rootstock; survivor tree showing resistance to Phytophthora cinnamomi root rot

BL516 (Marvel) BL667 (Nobel) CRI-71 Duke 7 Duke 9 Dusa Ettinger Fuerte G6 G755A G810 G875 GEM Gwen Harvest Hass Lamb/Hass Linda Nabal Nimlioh

Rogue Sir Prize Thomas Topa Topa Toro Canyon

410

G3M G3M G M M M G3M G3M M G 3 P. schiedeana G3M G3M G3M G G3M G G G G3M Thought to be progenitor of modern varieties Thought to be progenitor of the Guatemalan race G3M M M M M M

Ashworth and Clegg  Microsatellite Marker in Avocado Table 1.

Continued

Genotype

Characteristics and origin

Botanical race

UC2001 Velvick

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

M G

Walter Hole Zutano

M M

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.

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)

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 microsatellite data broadly agree with those from the RFLP data of Davis et al. (1998) when analyzed for a subset of 15 cultivars that were common to both data sets (phenogram not shown). However, in the RFLP phenogram Bacon and Zutano are embedded in the Mexican cluster and removed from the vicinity of P. schiedeana. Because of the lower allelic diversity of the RFLP data, compared to the microsatellite data (on average 3.8 alleles/locus; see Table 2), resolution of the Guatemalan, West Indian, and Guatemalan 3 Mexican genotypes is poor. Heterozygosity averaged 41.8% (compared to 63.5% for microsatellite loci), with values for individual genotypes ranging from 14.3% (Nimlioh) to 78.6% (Zutano and G755A; Table 2).

Discussion Differences between the genealogies of Figures 1 and 2 are modest and perfectly consistent with the unequal number of

412

constituent accessions and the low bootstrap values. Indeed, bootstrap support is weak throughout, with only six branches supported by values in excess of 500 (of 1,000) replicates. We hypothesize that weak bootstraps result from the large number of hybrid genotypes in our data set. Hybrids unite alleles from two genetically distinct sources, thereby decreasing the genetic distance between those sources. To test the influence of hybrids on bootstrap values, we performed a separate analysis on a data set from which (known) hybrids were removed. However, the bootstrap values remained low, suggesting that additional factors may be at play, including the possibility of ancient hybridization or a more recent date for racial differentiation than previously thought. In the genealogies of Figures 1 and 2, we have circled clusters that are most likely to correspond to the three botanical races of avocado, based on phenotypic, ecological, and historical information. However, numerous additional clusters remain that presumably represent instances of

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)

Arue Nabal Linda Nimlioh Fuerte Gwen Hass Bacon Zutano Pinkerton Duke 7 Thomas Topa Topa P. schiedeana G755A Total no. alleles/locus

WI G G G 3 3 3 3 M 3 M M M 3

a

b

c

d

e

f

g

h

i

j

k

l

m

n

H1

H2

1 1 1 1 2 1 1 2 2 1 2 1 1 1 2

1 1 1 1 2 1 1 1 2 1 2 2 2 1 2

2 2 1 1 2 1 2 2 2 2 2 2 1 1 2

1 1 1 1 1 1 1 2 2 1 2 1 1 1 1

1 1 1 1 1 1 2 2 2 1 1 2 2 1 2

1 1 2 2 1 1 2 2 2 2 2 2 1 2 2

1 1 1 1 1 1 1 1 1 1 1 1 1 1 2

1 1 1 1 1 1 1 2 2 2 1 2 1 2 2

2 2 1 2 2 2 2 2 2 2 2 2 2 1 2

1 1 1 1 1 1 1 2 1 2 2 1 1 1 1

1 1 2 1 2 1 2 1 1 2 1 1 1 2 2

1 2 2 1 2 2 2 1 2 2 1 1 1 2 2

2 1 2 1 1 2 1 2 2 2 1 1 1 — 2

1 2 1 1 2 2 2 1 2 2 2 2 1 — 1

21.4 28.6 28.6 14.3 50.0 28.6 50.0 64.3 78.6 64.3 57.15 50.07 21.4 33.3 78.6

50.0 68.0 48.0 68.0 84.0 64.0 72.0 84.0 0.0 4.0 6.0 2.0 50.0 76.0 80.0

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 a continuum reflected in the complexity of the hybrid status (multiple backcrossing). A mismatch between phenotypic and molecular characters is also expected when there is segregation for phenotypic characters (see Bergh 1966, 1967; Bergh and Whitsell 1974, 1975; and Storey et al. 1984 for illustrated examples). For instance, the progeny of a Guatemalan 3 Mexican hybrid segregating for Guatemalan phenotypic traits will be assigned to the Guatemalan botanical race, even though its molecular background remains Guatemalan 3 Mexican. The following examples illustrate the varied and at times unpredictable relationships between phenotype and genotype. The two main criteria used to characterize a cultivar in terms of its botanical origins are its pedigree, if known, and its fruit and cultural characteristics. For the most part, our genealogical data corroborated phenotypic criteria (especially anise-scented leaves and fruit properties), in placing most rootstock genotypes in the Mexican cluster. This was true of rootstock genotype Thomas, a so-called rescue variety of unknown pedigree. Thomas was discovered in a severely root rot-infested field, where it showed superior disease tolerance, compared to the other genotypes present. Interestingly, rootstock genotype Spencer, another rescue variety, emerged as identical to Fuerte at all 25 microsatellite loci, although tree and fruit morphology are said to differ. Conversely, an apparent conflict between phenotypic and genotypic data can help adjust pedigree information. For example, cv. Harvest is designated a Guatemalan 3 Mexican hybrid, having Gwen as its maternal parent, yet it clusters with the (pure) Guatemalan cultivars. This suggests that the pollen parent must be sought among Guatemalan race genotypes. All other Gwen progeny genotypes cluster with their maternal parent, Hass (parent of Gwen), and two further cultivars derived from Hass (Pinkerton) or showing Hass-like characteristics (OA184).

Also of Guatemalan 3 Mexican ancestry is cv. Sir Prize, yet it is sometimes designated a Mexican avocado. In neither phenogram does it associate with the Mexican cultivars but instead with either a variable cluster near the Guatemalan 3 Mexican assemblage (Figure 1) or with its grandparent (Hass) within the Guatemalan 3 Mexican cluster (Figure 2). Cv. Dusa is deemed to be a Mexican variety, having originated from Duke 7. Its placement on the phenogram in Figure 2 is indeed adjacent to the Mexican cluster, but no such proximity is seen in Figure 1. Both genealogies agree, however, in placing Dusa adjacent to Fuerte, a Guatemalan 3 Mexican hybrid (Popenoe 1926). Cv. Zutano and cv. Bacon are cited as being of Mexican and Guatemalan 3 Mexican origin, respectively. In our data set the two cultivars emerge as very similar (further supported by their strong cold tolerance) and attach to the genealogy between the Mexican cultivars and the wild relative, P. schiedeana. The affiliation with P. schiedeana is intriguing and merits further attention. Surprising are also some of the alignments of the wild species and local selections in our data set. P. steyermarkii, the postulated predecessor of the Guatemalan race of avocado (Schieber and Zentmyer 1980), associates not with the Guatemalan cluster but with cv. Arue. Although the origins of Arue are shrouded in mystery (it was discovered in the Society Islands in the southern Pacific), historical trade routes and its tropical habitat suggest that it is a West Indian cultivar. The two other members of the putative West Indian cluster are rootstock G810 and, in Figure 2, cv. Velvick. Velvick has been cited variously as a West Indian or Guatemalan cultivar. Our data sheds no further light on this ambiguity. Rootstock G810 is a collection from Guatemala. Although artifactual clustering cannot be ruled out, it is conceivable that G810 originated in lowland Guatemala, the 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 genotypes in this data set, rootstock G875. It associates with P. schiedeana, but its branch is extremely long. Irrespective of its placement, G875 is unlikely to belong to P. americana, given that it is more divergent than either of the other two Persea relatives. Possibly, G875 will replace P. schiedeana as the most distant species within subgenus Persea (Furnier et al. 1990; Kopp 1966; Williams 1976). The third Guatemalan rootstock, G6, is firmly embedded within the Mexican cluster. Thus, the three rootstocks G810, G875, and G6 are genetically dissimilar despite their common geographic origin. More information is needed on their phenotypic and cultural characteristics. Conclusions The pronounced phenotypic variability and plasticity of avocado attest to its checkered history of breeding and selection. Low bootstrap support may result from frequent hybridization events—both recently and before written records were available. Care is needed when distinguishing between geographic origin (collection site) and botanical race assignment, because descriptions such as ‘‘Guatemalan cultivar’’ do not make this distinction clear. Moreover, confusion is introduced when a race-specific trait (e.g., green or black skin) undergoes segregation. Indeed, our data— though preliminary—do not suggest that the botanical races are as well differentiated as hitherto assumed. Many more accessions and microsatellite markers will be needed to examine more thoroughly the validity and origins of the three botanical races and the relationships among wild relatives and the countless cultivars and rootstocks in cultivation today. Valuable insights into crop domestication can be gained from large-scale genotyping studies (e.g., Matsuoka et al. 2002). In view of the complex genealogies of most extant avocado accessions, it will be important to explore additional methods of inferring historical relationships, including gene sequences, and to capture the diversity present in wild populations and cultigens. Inevitably, the manipulation of data necessary to generate a matrix of genetic distances, as applied to this study, ignores the potential information inherent in allele frequencies. Methods that use allele frequencies to infer population structure (Pritchard et al. 2000) will be useful to formulate less subjective boundaries among the three avocado races.

Acknowledgments

AvocadoWebSite%20folder/AvocadoWebSite/AvocadoVarieties/ AvocadoVarieties.html.) Accessed July 17, 2002. Bergh BO, 1966. A Hass open-pollinated progeny set. Calif Avocado Soc Yearbook 50:64–77. Bergh BO, 1967. Some late-maturing avocado seedlings of various parentage. Calif Avocado Soc Yearbook 51:131–158. Bergh BO, 1969. Avocado—Persea americana Miller. In: Outlines of perennial crop breeding in the tropics (Ferwerda FP and Witt F, eds), Miscellaneous Papers 4, Landbouwhogeschool Wageningen. Wageningen, Netherlands: Veenman & Zoonen; 23–51. Bergh BO and Ellstrand N, 1986. Taxonomy of the avocado. Calif Avocado Soc Yearbook 70:135–145. 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 Avocado Soc Yearbook 57:118–126. Bergh BO and Whitsell RH, 1975. Self-pollinated Fuerte seedlings. Calif Avocado Soc Yearbook 58:128–134. Davis J, Henderson D, Kobayashi M, Clegg MT, and Clegg MT, 1998. Genealogical relationships among cultivated avocado as revealed through RFLP analyses. J Hered 89:319–323. Felsenstein J, 1989. PHYLIP—phylogenetic inference package, version 3.2. Cladistics 5:164–166. Furnier GR, Cummings MP, and Clegg MT, 1990. Evolution of the avocados as revealed by DNA restriction fragment variation. J Hered 81:183–188. Gama-Campillo L and Go´mez-Pompa A, 1992. An ethnoecological approach for the study of Persea. A case study in the Maya area. In: Proceedings of the Second World Avocado Congress, vol. 1, Orange, CA, April 21–26, 1991 (Lovatt CJ, ed). Riverside, CA: University of California and California Avocado Society; 11–17. Goldstein DB and Pollock PD, 1997. A review of mutation processes and methods of phylogenetic inference. J Hered 88:335–342. Goldstein DB and Schlo¨tterer C (eds), 1999. Microsatellites: evolution and applications. Oxford: Oxford University Press. Hamada H, Petrino MG, and Kakunaga T, 1982. A novel repeated element with Z-DNA-forming potential is widely found in evolutionarily diverse eukaryotic genomes. Proc Natl Acad Sci USA 79:6465–6469. Jarne P and Lagoda JL, 1996. Microsatellites, from molecules to populations and back. Trends Ecol Evol 11:424–429. Knight RJ Jr, 1999. Genetic diversity in avocado. In: Proceedings of Avocado Brainstorming ’99, Riverside, CA, October 27–28, 1999 (Arpaia ML and Hofshi R, eds). Riverside, CA: California Avocado Commission and University of California; 16–18. Kopp LE, 1966. A taxonomic revision of the genus Persea in the western hemisphere. (Perseae—Lauraceae). Mem N Y Bot Gard 14:1–117. 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. Matsuoka Y, Vigouroux Y, Goodman MM, Sanchez GJ, Buckler E, and Doebley J, 2002. A single domestication for maize shown by multilocus microsatellite genotyping. Proc Natl Acad Sci USA 99:6080–6084.

We thank the California Avocado Commission for financial support; Drs. Mary Lu Arpaia and John Menge for valuable discussions; Dr. Thomas Chao for helpful comments on the manuscript; Eric Focht, Brandon McKee, Paul Robinson, and Dave Stottlemyer for collecting plant material; and Peter Morrell for advice on analytical software.

McCormac J (ed), 2002. California Avocado Society Yearbook Web Site. Available at: http://www.avocadosource.com/CAS_Yearbooks/CAS_Yearbooks. htm. Accessed July 17, 2002.

References

Mhameed S, Sharon D, Kaufman D, Lahav E, Hillel J, Degani C, and Lavi U, 1997. Genetic relationships within avocado (Persea americana Mill.) cultivars and between Persea species. Theor Appl Genet 94:279–286.

Arpaia ML, University of California Avocado Web Site. Available at http://ucavo.ucr.edu/AvocadoWebSite. (See http://ucavo.ucr.edu/

Popenoe W, 1926. The parent Fuerte tree at Atlixco, Mexico. Calif Avocado 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. Popenoe W, 1963. Early history of the avocado. Calif Avocado Soc Yearbook 47:19–24. Pritchard JK, Stephens M, and Donnelly P, 2000. Inference of population structure using multilocus genotype data. Genetics 155:945–959. Queller DC, Strassmann JE, and Hughes CR, 1993. Microsatellites and kinship. Trends Ecol Evol 8:285–288. Rhodes AM, Malo SE, Campbell CW, and Garner SG, 1971. A numerical taxonomic study of the avocado (Persea americana Mill.). J Amer Soc Hort Sci 96:391–395. Robinson TR, 1926. Avocados for Florida. Proc Fla State Hort Soc 39: 182–191. Roots EH and Baker RJ, 2002. Distribution and characterization of microsatellites in the emu (Dromaius novaehollandiae) genome. J Hered 93:100–106. Schieber E and Zentmyer GA, 1973. Collecting Persea’s in Central America and Mexico. Calif Avocado Soc Yearbook 56:94–101. Schieber E and Zentmyer GA, 1980. The ‘‘Guatemalan Criollos.’’ Calif Avocado Soc Yearbook 64:85–90. Schieber E and Zentmyer GA, 1981. Exploring for Persea on Volcano Quetzaltepeque, Guatemala. Calif Avocado Soc Yearbook 65:57–63. Schieber E, Zentmyer GA, and Coffey MD, 1983. Variability in Mexican avocados (Matuloj) in Guatemala. Calif Avocado Soc Yearbook 67:87–91. Schmidt E and Heslop-Harrison JS, 1996. The physical and genomic organization of microsatellites in sugar beet. Proc Natl Acad Sci USA 93:8761–8765.

Stallings RL, Ford AF, Nelson D, Torney DC, Hildebrand CE, and Moyzis RK, 1991. Evolution and distribution of (GT)n repetitive sequences in mammalian genomes. Genomics 10:807–815. Storey WB, Bergh BO, Platt RG, and Miller M, 1984. Observations on a second-generation progeny of a Mexican 3 West Indian cross. Calif Avocado Soc Yearbook 68:161–165. Storey WB, Bergh BO, and Zentmyer GA, 1986. The origin, indigenous range, and dissemination of the avocado. Calif Avocado Soc Yearbook 70:127–133. Sunnucks P, 2000. Efficient genetic markers for population biology. Trends Ecol Evol 15:199–203. Swofford DL, 2002. PAUP, phylogenetic analysis using parsimony (and other methods), version 4. Sunderland, MS: Sinauer. Tautz D, 1989. Hypervariability of simple sequences as a general source for polymorphic DNA markers. Nucleic Acids Res 17:6463–6471. Torres AM, Diedenhofen U, Bergh BO, and Knight RJ, 1978. Enzyme polymorphisms as genetic markers in the avocado. Amer J Bot 65:131–139. Weber JL and May PE, 1989. Abundant class of human DNA polymorphisms which can be typed using the polymerase chain reaction. Amer J Hum Genet 44:388–396. Weissenbach J, Gyapay G, Dib C, Vignal A, Morisette J, Millaseau P, Vaysseix G, and Lathrop M, 1992. A second generation linkage map of the human genome. Nature 359:794–801. Williams LO, 1976. The botany of the avocado and its relatives. In: Proceedings of the First International Tropical Fruit Short Course. The Avocado, Miami Beach, Florida, November 5–10, 1976 (Sauls JW, Phillips RL, Jackson LK, eds). Gainesville, FL: University of Florida; 9–15.

Smeets AJM, Brunner HG, Ropers HH, and Wieringa B, 1989. Use of variable simple sequence motifs as genetic markers: application to study of myotonic dystrophy. Hum Genet 83:245–251.

Zentmyer GA and Schieber E, 1982. Persea explorations in Honduras. Calif Avocado Soc Yearbook 66:93–102.

Smith CE Jr, 1966. Archaeological evidence for selection in avocado. Econ Bot 20:169–175.

Received December 23, 2002 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