PLS221                                             Instructor: Carlos F. Quiros

Apiaceae: Apium graveolens, celery

List of references (Includes reading assignments)


Origin and distribution

Domestication: horticultural varieties

Apium species

Interspecific/intergeneric hybridization


Floral biology: protandry

Inheritance of most importance traits

Male sterility

Disease resistance

Fast vs slow bolting

Quality: pithiness, hollow petioles, petiole color, stringiness

Chromosome markers: isozymes, crown proteins, DNA-base markers

Biotechnological applications



Apium graveolens, diploid species 2n=2x=22.

There are three horticultural varieties cultivated for this species:

var.: dulce, celery, grown for its succulent petioles.

var. rapaceum, celeriac, grown for its enlarged root/hypocotyl

var. secalinum, smallage, grown for its leaves.

Celery was first used as a medicinal plant, condiment or for ornamentation in the 16th century by Greeks, Romans Egyptians and Chinese. It was used to combat malaise after drunkenness (hangover) and also as an aphrodisiac (later it was found to contain androsterone).  Early celery types had hollow petioles.

Celery is rich in fiber and in chemical compounds, some of which are health-promoting compounds, others are allergens. Recent work Tsia and Tan (2000) indicates that celery promote excretion of cholesterol in rats. Therefore, its consumption might lower the cholesterol levels in the blood of people suffering from hypercholesterolaemia. The detrimental compounds of celery include furanocomarins, which include psoralens. These produce skin rush and other reactions when susceptible people are exposed to celery plants, especially in the presence of light. High concentration of these compounds is not a problem in most varieties, however it varies depending on environment and also virus infestation. Plants infested with virus accumulate high content of furanocumarins, which have been found to inhibit infestation by the late blight fungus Septoria apicola. Additionally, celery contains profilin a minor allergen that can produce strong allergic reactions in a few people (Scheurer et al. 1999).

Celery was domesticated in Italy as a vegetable in 17th century, resulting in selections with solid stems. At that time the three different existing botanical varieties were developed. China might have been another center of domestication.

A. graveolens is a Mediterranean species, but spreads from Sweden to India.

Species of the genus Apium are widely distributed throughout the world, including the Mediterranean, Australia, New Zealand, So. Africa, and So. America. Approximately 20 wild species of Apium have been reported. The south cone of South. America is a rich focus of wild species. Most of these have not been studied or tested for desirable traits, or even for crossing affinities with celery.

Weedy celery types are found throughout the California coast. These most likely originate from escapes to cultivation previous to the 1950's since they not have any of the traits characterizing our modern varieties.

Germplasm: The germplasm available for celery breeding is limited, although a recent IPGRI tally estimates a total of 1270 throughout the world. Germplasm centers for Apium graveolens are located in New Delhi, India; Gatersleben, Germany; Bari, Italy; Prague, Czechoslovakia and Geneva, New York, in the United States (Toll and van Sloten 1982). The USDA Northeast Regional Plant Introduction Station (NE-9) at Geneva, NY maintains a collection of approximately 100 accessions, mostly cultivated types and land races of A. graveolens. Very few wild species are represented in the collection. In addition to these collections, the Department of Vegetable Crops at the University of California maintains one of the largest working collection of Apium germplasm, consisting of approximately 300 accessions. It includes all the cultivated types of A. graveolens and a few accessions of the wild species A. chilense, A. panul, A. prostratum, A. annuum and A. nodiflorum.

The bulk of the wild species are not represented in any germplasm collection, and so they are in urgent need of collection in their centers of distribution. Many are found only as herbarium specimens. Their taxogenetic relationships to cultivated celery is largely undetermined. Furthermore, some may be extinguished or close to extinction such as A. fernandizianum, native to the island Juan Fernandez off the coast of Chile.

Interspecific and intergeneric hybridization

Few attempts of interspecific hybridization have been made in celery because of the unavailability of wild species. Ochoa and Quiros (1989) succeeded in hybridizing celery to A. chilense and A. panul. The hybridization between these three species can be achieved in both directions regardless of which species is used as female.

The celery x A. panul, and celery x A. chilense F1 hybrids

 are very vigorous but display a pollen fertility range from 0 to 20%. These two wild species are important as sources for late blight (Septoria apicola f.s. apii) resistance. The sterility seems to be due to chromosomal inversions. The sterility of the hybrids precludes obtaining F2 progeny, although backcross progeny to celery are readily obtained. After two backcross, the petiole characteristics of celery is partially recovered. On the other hand, A. chilense x A. panul F1 hybrids were fully fertile, and did not show any evidence of chromosomal rearrangements. This indicates that the two wild species have the same chromosomal constitution and are closely related to each other.

Quiros (1990) reported successful hybridization between celery and A. prostratum, which is resistant to leaf miner (Lyriomiza trifolii) (Trumble et al. 1990), beet army worm Spodoptera exigua (Diawara et al. 1994) and Western yellow mosaic, a viral disease. The F1 hybrids are quite vigorous, but have a pollen fertility of only 25%. There is no evidence of meiotic abnormalities or chromosomal behavior in the hybrids. A. nodiflorum, another species resistant to Septoria and leaf miners has been hybridized to celery by Pink et al.(1983). No information is available on the characteristics of the resulting hybrids. Later attempts to hybridize these species have failed.

Crossing relationships between celery and three wild species


A phylogenetic tree based on molecular markers for the nuclear genome and chloroplast was constructed by Huestis (1992). All the cultivated types cluster together. The wild species are separated in a second main cluster.

For intergeneric crosses there are at least two independent reports on hybridization between parsley Petroselinum crispum and celery. Madjarova and Bubarova (1978) used three cultivars of celery and two of parsley. The parsley ‘Lister’ x celery ‘Pioneer’ cross resulted in a new parsley cultivar known as ‘Festival 68’. They also reported new forms of leaf celery from these crosses, characterized by higher vitamin C, carotene, essential oil and amino acid content, and an improved celeriac line.

The second report, by Honma and Lacy (1980) had the objective of transferring late blight resistance from parsley to celery. However, the level of resistance in the hybrid derivatives was weak. This may have occurred because parsley is susceptible to a different species of the pathogen, Septoria petroselini. For the crossing experiment, they used green stem color from parsley as marker for hybrid detection, which is dominant over yellow stems present in the celery parent. ‘Golden Spartan’ a yellow celery variety, was allowed to outcross with parsley. Three green seedlings were found among 1000 yellow seedlings germinating from the open pollinated seed collected from the celery parent. Later attempts to repeat these experiments have failed.


The celery genome consists of 11 large chromosomes, nine submetacentric, one metacentric and one telocentric, described by Murata and Orton, (1984). The genome size of celery is fairly large compared to carrot, approximately 3.5 pg per 2C nucleus or 3x109 bp per haploid nucleus (E. Earle, 1995, personal communication). Cytologically, a single pair of chromosomes in the genome organizes a nucleolus, containing the 18S-25S rRNA gene family. It is formed by a tandem repeat unit of 9.3 Kbp, which is invariable in all three A. graveolens cultivated types. There is, however, variation for unit size and restriction sites in the Apium wild species tested.

In addition to the 18S-25S genes, very few repetitive coding sequences have been observed after RFLP mapping based on a celery cDNA library clones (Huestis et al. 1993). The proportion of single-copy coding sequences in the celery has been estimated to range between 59% and 78%. This observation strongly suggests that celery is a true diploid species whose genome has not suffered extensive duplications for coding genes. The amount of total repetitive DNA in the celery genome has not been determined, but it is expected to be similar or higher than observed in carrots.

The level of polymorphism in A. graveolens is relatively low. cDNA clones tested in celery and celeriac for five restriction enzymes generated only 23% polymorphic loci. A similar level of polymorphism was detected for RAPD markers (Yang and Quiros 1993). The most likely mechanism for RFLPs in celery seems to be due to base deletion and insertion more than to base substitution.

Pollination control

Celery flowers are very small, which precludes the removal of individual anthers for emasculation. Furthermore, the different developmental stages of the flowers in the umbels make it difficult to control pollination. The standard hybridization technique in celery consists in selecting flower buds of the same size, eliminating older and younger flowers. The umbelets are covered with glycine paper bags for 5 to 10 days, when the stigmas become receptive. At this time, pollen or umbelets shedding pollen from selected male parents are rubbed onto the receptive stigmas of the female parent. An improvement of this technique consist of washing away the anthers from open, unreceptive flowers with a water stream, covering them with the paper bags and pollinating them a few days later when stigmas become receptive. The advantage of this technique is the lower rate of accidental selfing caused by the failure of the anthers to abscise before stigma receptivity (Ochoa et al 1986).

In order to induce seed stalk development during the vegetative phase, the plants require a period of vernalization, from 6 to 10 weeks at 5 to 8oC when they are at least 4 weeks old (Honma 1959).  Due to a wide range of response to the cold treatment, it is often difficult to synchronize crosses since plants will flower at different times. However, pollen can be stored for 6 to 8 months at -10oC in silica gel or calcium chloride, and with a viability decline of approximately 20 to 40%, thus providing flexibility to perform crosses most of the time.

For selfing, the plant or selected umbels are caged in cloth bags. These are shaken several times during the day to promote pollen release. House flies (Musca domestica) can also be introduced weekly into the bags for pollination.

Inheritance of most important traits:

The inheritance of some phenotypic traits has been determined in celery. These include mainly morphological characters and several disease resistance genes. These traits have economic importance and a description of their inheritance is summarized below:

Hollow petioles: This character segregates as a dominant, monogenic trait (Emsweller 1933, Townsend et al 1946). Quiros (1993) proposed Ho as the symbol for this gene. Hollow petioles are widespread in celeriacs and in some smallage accessions. Solid stems most likely were selected and fixed for the domestication of celery stalks.

Flowering behavior: Annual habit is a partially dominant, monogenic trait determined by the gene Hb. This locus forms a linkage group with two isozyme loci and gene A, coding for anthocyanin pigmentation in the plant (Quiros et al 1987). Bouwkamp and Honma (1970) reported that early bolting, was dominant over slow bolting and determined by a single gene denominated Vr. It is possible that Hb and Vr are allelic. In contrast to the simple genetic determination of annual vs biennial habit, resistance to bolting in biennial types is a complex trait, most likely polygenic and affected by environment.

Fusarium resistance: Resistance to fusarium yellows is partially dominant and determined by two loci (Orton et al 1984). The resistance based on these two genes has a quantitative nature, where the dominant allele Fu1 from celeriac contributes the largest effect. The dominant allele Fu2 of the second locus, is found in tolerant celery varieties. This allele has a small contribution to resistance. (Quiros 1987). Based on a series of F2 and backcross progenies the following genotypes are postulated for disease reaction:



















In order to maximize resistance, when possible the breeder should aim to fix both loci in homozygous dominant condition.

Virus resistance: Celery mosaic virus resistance has been found in a feral accession of A. graveolens.  Resistance is monogenic and recessive (cmv). Molecular markers have been identified allowing marker assisted selection.

Male-sterility: Only a single genetic male sterile have been reported in celery. It is recessive and determined by a single gene named ms-1 (Quiros et al 1986). It was found as an spontaneous mutant in a weedy Iranian accession. Therefore, substantial breeding activity will be necessary to transfer it into a useful genetic background. Male sterility is due to tapetal degeneration in the anthers. Nectar production in the male sterile mutants is not impaired, therefore their flowers still attract pollinators (Quiros et al 1986). A series of molecular markers have been found around this locus, which might be useful for selection of sterile plants at seedling stage in segregating populations. Cytoplasmic male-sterility has been reported by Dawson (1993), after finding this trait in unidentified wild celery plants growing in the UK.

Stem color: Anthocyanin pigmentation is determined by single dominant gene designated A (Townsend et al 1946, Arus and Orton, 1984, Quiros et al 1987). This gene has been found to be tightly linked, approximately 2cM, to the isozyme locus Aco-1, coding for the enzyme aconitase (Huestis and Quiros 1993). Yellow celery (y) is recessive to green and it is determined by a single gene (Townsend et al. 1946). This trait often has been used as a marker for hybrid identification.

Leaf shape: Bouwkamp and Honma (1970) reported that deeply toothed leaf is recessive and determine by the gene dt.

Cultivar development

Celery breeding has been done exclusively at the diploid level. Polyploidy and aneuploidy have not been exploited in celery breeding and genetics. Most celery varieties are open pollinated. Existing commercial celery hybrids are few and are not as popular as the existing open pollinated varieties.

Basically, there are two types of celery varieties, self-blanching or yellow, and green or Pascal celery. In North America, almost exclusively green celery varieties are commercially grown, whereas in Europe and in many other parts of the world, mostly self-blanching varieties are produced. According to Vilmorin (1950), in 1887 the French varieties ‘Paris Golden Self-Blanching’ and ‘Pascal Celery’ a green celery line selected from the former, were introduced to North America under the names of ‘White Plume’ for the self-blanching type and ‘Giant Pascal’ for the green type. These two are the chief progenitors of our modern varieties, with little genetic contribution from accessions of other origins (Quiros 1993).

Based on historical accounts by Beattie (1944), Munger and Newhall (1952), Guzman et al (1973), and Hill, E., (personal communication), and on the presence of biochemical (Quiros et al 1987) and molecular markers (Yang and Quiros 1993), the possible origins of the main North American varieties have been charted (adapted from Quiros 1993). The three cultivated types, celery, celeriac, and smallage occupy independent branches, however they share 68% of the molecular markers, which implies considerable homology among them. The 21 celery cultivars are grouped in three clusters: A, B, and C according to their origin.

Biotechnological Applications

A number of research areas with great potential are also under development in celery that include biochemical and molecular markers, tissue culture and transgenics.

Biochemical and molecular markers: In addition to the few phenotypic markers, four classes of biochemical/molecular markers are available in celery: isozymes (Arus and Orton 1984, Quiros et al. 1989), crown proteins (Quiros et al. 1987), restriction fragment length polymorphisms (RFLPs) (Huestis and Quiros 1993, Yang and Quiros 1994) and random amplified polymorphic DNA (RAPDs) (Yang and Quiros 1993 and 1994).

Most of these markers have been used to develop a linkage map in an F2 progeny resulting from crossing celery by celeriac. The present map developed by Yang and Quiros (1994) has 135 loci (33 RFLPs, 128 RAPDs, 5 isozymes, the Fusarium disease resistance gene Fu1, and annual habit gene Hb) on 11 major linkage groups (A1 - A11) plus nine small linkage groups (A12 - A20). The total coverage of the present map is 803 cM, and the average distance between two markers is 6.4 cM. The existence of a few unlinked markers and more than 11 linkage groups indicates that the current map of 803 cM is still sparsely populated, considering the large genome size of celery.

Two important traits, Fusarium resistance (Fu1) and annual habit (Hb), are located in groups A2 and A7, respectively. The closest marker to Fu1 is RAPD marker O7-900/800 with a distance of 12.6 cM, which is too large to be used as a tag for Fusarium resistance.

Isozymes, crown storage protein and RAPD markers have been useful for variety identification and for following their origin and pedigree. A few SSR loci are also available (Acquadro et al 2006)

Also, these markers have been found useful for outcrossing determinations (Orton and Arus 1984) and for hybrid identification after intra and interspecific crosses (Quiros et al 1987).

Tissue culture: Celery is amenable to tissue and cell culture. It has a high embryogenic capacity and can be regenerated from protoplasts, cells in suspension and tissue explants. An extensive review of this field, including recommendations of optimal media for in vitro culture are published by Browers and Orton (1986). These techniques may have application for micropropagation and mass vegetative propagation of parental lines for F1 hybrid seed production. Another example is the use of somaclonal variation for selection of Fusarium resistant plants (Heath-Pagliuso et al 1988).

Attempts to produce haploids through anther culture have failed in celery. Initial stages of microspore division were obtained by anther culture, resulting in eight and 16 cell embryoid like structures. These structures, however, failed to develop into plants (Quiros and Ochoa, unpublished). However,  recently a seed company has been able to produce doubled haploids via anther culture but the protocol has not been published.

Celery has been transformed successfully using Agrobacterium tumefaciens, resulting in stable, kanamycin resistant transgenic plants (Catlin et al 1988). Kanamycin resistance behaved like a monogenic, dominant trait, segregating 3:1 in the selfed progenies of the transgenic plants. Transgenic celery plants have not been field tested. For a more recent report on transformation see Song t al 2007. 


These developments in biotechnology place celery at the same level of sophistication of other horticultural crops. Immediate celery improvement most likely will continue toward the direction of disease resistance, mainly for Fusarium, Septoria and viruses. The availability of a well developed map will make possible the use of marker based selection for disease resistance. Similarly, transformation techniques open up the possibility to engineer virus resistant plants. Other areas of research activity will be the use of wild species as sources of useful traits and their introgression into celery by marker assisted selection. Development of male sterile lines for F1 hybrid seed production resulting in the synthesis of dominant alleles for multiple disease resistant varieties certainly will continue to be an active research field in the years to come.

Return to syllabus

    Last modified: May 11, 2010
    © Carlos F Quiros, 1998