Genetics of Male Sterility in Gynodioecious Plantago coronqPus . 11 . Nuclear Genetic Variation Hans

Inheritance of male sterility was studied in the gynodioecious species Plantago coronopus using five plants and their descendants from an area of -50 m' from each of four locations. In each location, crosses between these five plants yielded the entire array of possible sex phenotypes. Both nuclear and cytoplasmic genes were involved. Emphasis is placed on the nuclear (restorer) genetics of two cytoplasmic types. For both types, multiple interacting nuclear genes were demonstrated. These genes carried either dominant or recessive restorer alleles. The exact number of genes involved could not be determined, because different genetic models could be proposed for each location and no common genetic solution could be given. At least five genes, three with dominant and two with recessive restorer allele action, were involved with both cytoplasmic types. Segregation patterns of partially male sterile plants suggested that they are due to incomplete dominance at restorer loci. Restorer genes interact in different ways. In most instances models with independent restorer gene action were sufficient to explain the crossing results. However, for one case we propose a model with epistatic restorer gene action. There was a consistent difference in the segregation of male sterility between plants from the two cytoplasmic types. Hermaphrodites of cytoplasmic type 2 hardly segregated male steriles, in contrast to plants with cytoplasmic type 1. A major problem in understanding the gynodioecious breeding system and the occurrence of male steriles is the limited knowledge of the genetics of this system. Theoretical studies show that the existence of the breeding system is dependent on the combined action of the mode of inheritance and the extent of the disadvantage of being male sterile. ROSS ( 1978) surveyed most genetic studies up to 1978 and concluded that monogenic inheritance has hardly been found in naturalpopulations. Inheritance of male sterility was either claimed to be digenic or complex with cytoplasmic effects. CHAFUESWORTH ( 1981) argued that nuclearcytoplasmic models were likely to fit most data. Her suggestion is now justified by detailed genetic studies. Nuclearcytoplasmic inheritance of male sterility in a species is typically thought of in terms of several cytoplasmic types: each is either unrestored (i.e., determining female gender) or restored (i.e., determining hermaphrodite gender) by specific alleles of nuclear genes (the so-called restorer genes). For nuclear genes of this type, one allele is called the restorer allele and the other the sterility allele. The occurrence of cytoplasmic variation within populations has now been shown in several gynodioecious species; e.g., Nemophila menziesii (GANDERS 1978), Origanum vulgare ( KHEYR-POUR 1980, Corresponding author: Hans P. Koelewijn, Netherlands Institute of Ecology, Department of Plant Population Biology, PO Box 40, 6666 ZG Heteren, The Netherlands. E-mail: [email protected] Genetics 139 1759-1775 (April, 1995) 1981 ) , Plantago lanceolata (VAN DAMME and VAN DEL DEN 1982), Thymus vulgaris (BELHASSEN et al. 1991) and P. emonopus ( KOELEWIJN and VAN DAMME 1995). Cytoplasmic variation, however, is only one side of the coin. Restorer gene variation, the reverse side, has proven to be much more difficult to solve for natural populations. It is often stated that there is variation in restorer genes, but that the genetics is complex ( KHEYRPOUR 1981; BELHASSEN et al. 1991). Only few authors have proposed genetic models based on crossing results, but even they state that they cannot fully explain all their crossing results (VAN DAMME 1983; CONNOR and CHARLESWORTH 1989). The involvement of both dominant and recessive restorer alleles is beyond doubt. However, the number of restorer genes involved in restoring male fertility is still unknown for any species. VAN DAMME ( 1983) gave a minimum estimate of five genes for MS1 in P. lanceolata, of which three had dominant and two had recessive restorer allele action. The complexity of the inheritance in studies from natural populations is in contrast with that of most cultivated species. KAUL (1988) states that in the majority of crop plants, male sterility is determined by a single recessive gene. However, these results are often based on inbred lines, selected maintainer lines or on a few plants from known genetic background, which may mask the variation originally present in the parental plants. In studies on natural populations, in which plants from unknown genetic origin are used, this variation might become 1760 H. P. Koelewijn and J. M. M. Va Damme apparent in crossing studies and cause problems interpreting the results. Some insight into the nuclear genetics is required for understanding of gynodioecious breeding systems. Most theoretical models for the maintenance of gynodioecy require variation in restorer genes within sex phenotypes to explain joint polymorphism between cytoplasmic and nuclear genes ( e.g., FRANK 1989) . Evaluation of crossing results in terms of numbers of genes involved and their interaction is therefore indispensable and could provide a guideline for the type of models to be used. In our first article ( KOELEWIJN and VAN DAMME 1995), we gave a description of anther morphology and of the subsequent distinction of three sex phenotypes: male steriles (MS) , partially male steriles (PMS) and hermaphrodites ( H ) . The involvement of cytoplasmic genes in determining male sterility was proven and shown to be limited to only two, henceforward referred to as cytoplasmic type 1 and 2. These two cytoplasmic types were present in most of the populations studied. In this paper we focus our attention on variation in restorer genes. We present detailed nuclear genetic models for four populations and address questions about the number and type of nuclear genes involved in male sterility. In addition we look for differences in restorer gene frequencies among populations and pay some attention to the role of the partially male sterile phenotypes. MATEIUALS AND METHODS In the Netherlands P. coronopus is an annual or short lived perennial, mainly growing in dune grasslands along the coast. It is a wind pollinated and self-compatible species, with an average outcrossing rate of -75% (range 20-100%; WOLFF et al. 1988). Plants used for crosses originated from two locations in each of two sites (Oostvoornse meer, Kwade Hoek) in the Southwest of the Netherlands. From each location three hermaphrodites (H) and two male steriles (MS) were used [see KOELEWJN and VAN DAMME ( 1995) for a description of the sites and experimental procedures]. Each MS was crossed with each of the three H and the three H were crossed in a diallel design. Offspring of these crosses were grown in an experimental garden and a number of these were selected for second and third generation crosses. In a previous paper (KOELEWIJN and VAN DAMME 1995), we showed that there was cytoplasmic variation at all four locations and parental plants were classified according to their sex and cytoplasmic type (hermaphrodites: H1, H2; partial male steriles: PMSl, PMS2; male steriles: MS1 and MS2). Sex type scores were based on 20-50 flowering spikes per plant. Sex-type ratio testing was analyzed with Gtests (SOL% and ROHLF 1981 ) . Unless stated otherwise analyses were based on the categories MS us. not-MS, i.e., H and PMS were grouped together. Each cross received a family number and offspring were coded according to this number. Thus, M S (2801 ) means that the plant is a male sterile, originates from cross 280 and is the first plant used in subsequent crossing. In 1990 a random pollen sample was collected in six populations by tapping pollen of 25 plants into a small tube. The plants were located in a square of -10 X 10 m. Pollen was collected during peak flowering ( i . e . , >80% of the plants were flowering). These pollen samples were used in crosses with MS of known cytoplasmic and nuclear background to look for differences in restorer allele frequencies among populations. In agreement with the literature ( KAUL 1988), we will denote dominant restorer alleles by R and recessives by r. Because cytoplasmic genes are thought to be the oldest on an evolutionary time scale, the nuclear sterility allele is considered to be the wild-type allele and will be indicated by +.