| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Usda-Ars
b Program in Statistics, Washington Sate Univ., Pullman WA 99164-6402
* Corresponding author (rcjohnson{at}wsu.edu)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Among the factors that have a strong bearing on the genetic quality of accessions maintained in gene banks are the initial germplasm collecting process, the need for initial and periodic seed regeneration of accessions, and consideration of differences in mating systems (Crossa and Vencovsky, 1994, 1997; Vencovsky and Crossa, 1999). The potential for random genetic drift is a major concern in the relatively small populations associated with regeneration. The effective population size (Ne) rather than the census population size (Nc) is the key parameter in genetic drift, and is defined as the size of an ideal population that would have the same amount of random genetic drift as the actual census population (Wright, 1931; Crow and Kimura, 1970). There are a number of factors relating to germplasm conservation that affect Ne (Frankel et al., 1995). Normally, Ne is lower than Nc. A major cause of this is the differential production of gametes among parents; that is, variation in potential fecundity (Heywood, 1986). However, if mating is controlled so that every individual in a population contributes two gametes to the next generation, then Ne is essentially doubled relative to Nc (Crow and Kimura, 1970). Other regeneration schemes to maximize Ne include various types of controlled polycrosses (Breese, 1989). When male and female gametes are uncontrolled and seeds are bulk harvested, variation in fecundity can substantially reduce Ne compared with Nc.
Thus, whenever possible, germplasm managers should control mating to maximize Ne. But because of the need to regenerate large numbers of accessions with limited resources, controlled mating is often difficult. In any case, it is important for germplasm managers to understand the degree that variation in fecundity among plants can lower Ne during regeneration. With this information, the extent that Ne is reduced relative to Nc can be estimated and different methods of sampling considered that would minimize potential genetic drift. There are, however, very few estimates of Ne available, especially generic to germplasm regeneration. In addition, there is a need to develop and implement cost effective sampling methods that maximize Ne. The objectives of this study were to estimate Ne/Nc in three model wind pollinated perennial grass species and to recommend cost effective sampling methodology to maximize Ne/Nc during seed regeneration.
|
MATERIAL AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
 | [1] |
and µ is the standard deviation among plants and the mean family size. Thus, when each parent contributes equally to the gamete pool, 2 is zero and Ne = Nc. Under these conditions genetic drift is equal to the traditional binomial sampling model (Heywood, 1986).
Variation in seed production per plant normally reflects variation in pollen and ovule output. When the variance and mean number of seeds sampled per plant is unknown then an estimate of 2/µ2 can be obtained as follows:
 | [2] |
Statistical Model
Consider the following definitions:
 | [3] |
 | [4] |
 | [5] |
Using the above definitions, we developed an ANOVA model (Table 1) to examine how sampling can be used to minimize variation in fecundity and thereby minimize reductions in Ne/Nc. The between or among plant component (Table 1) is equivalent to 2 in Eq. [2] above, which is the sum of the among and within plant variance for seed number in a given population. Total seeds per plant is the number of seeds per inflorescence times the number of inflorescences per plant. Both of these factors are included in the within plant component for variance in seed number per plant. Thus, if every plant naturally produced the same number of inflorescences per plant and the same number of seeds per inflorescence, seed number among plants would not vary and Ne would equal Nc. This would not be expected to occur naturally, but artificial sampling can be completed so that the number of seeds harvested per plant is equal for each plant. Equalizing the number of seeds harvested per plant eliminates the variation in seed number among plants. Although assembling such a regeneration sample would not control the variation in pollen production occurring before seed development, it would eliminate the maternal variation in gamete production, and Ne would be increased compared with harvesting all seeds from whole plants. Instead of sampling an equal number of seeds per plant, an equal number of inflorescences could be sampled from each plant. Even though seeds per plant sampled would still vary among plants as seeds per inflorescence varies, a major source of variation; that is, inflorescence number per plant, is eliminated. As a result, the variation in seed number among plants would be reduced along with the maternal variation to the gamete pool. Although Ne would not be as high as when seeds per plant are fully equalized, Ne would nonetheless increase compared with whole plant sampling.
|
The experiment consisted of a randomized complete block with two replications established at both Pullman and Central Ferry, WA. Each plot consisted of 30 plants, and there were 18 plots at each location, for a total of 1080 plants. The Pullman location (46°43'55'' N, 117°9'25'' W) is about 800 m in elevation and plants were grown under dryland conditions. Central Ferry (46°40'13'' N, 117°45'8'' W) is located approximately 50 km from Pullman in the Snake River Canyon at about 200 m in elevation. Plants at Central Ferry were grown under irrigation as described previously (Johnson and Li, 1999). Associated with its lower elevation and more westerly location toward the Columbia Basin desert, temperatures at Central Ferry average about 3.5°C higher and precipitation about 200 mm lower per year than at Pullman. The soil at Pullman was a fine-silty, mixed, mesic, Pachic Ultic Haploxeroll and the soil at Central Ferry a fine-silty, mixed, mesic Natrixeroll.
For plot establishment, seeds of each accession were germinated in water-saturated vermiculite in a growth chamber at 25°C. Seedlings were transplanted into cardboard bands (50- by 50- by 80-mm height) containing a steam sterilized soil mix consisting of one-third each sand, peat moss and Spofford soil (a fine-silty, mixed, mesic Natrixeroll). The plants, contained within the cardboard plant bands and metal flats, were initially established and maintained in a greenhouse for about one month and then placed in a lath house for about 10 d prior to field transplanting.
The field plot areas were cultivated prior to transplanting and 50 kg ha-1 N and 15 kg ha-1 S applied. The plot areas were cultivated again to incorporate the fertilizer and rolled to smooth the soil surface prior to transplanting. Soil tests indicated adequate concentrations of P and K.
For transplanting, plants were placed in a furrow approximately 0.25 m wide and 0.1 m deep formed by a V-blade attached to a lawn tractor. The exterior bands were removed as the plants were placed in the furrow and plant roots were covered with soil. Each of the 30 plants per plot were spaced 0.5 m apart and plots were spaced 1.8 m apart. The plots were incorporated into the grass regeneration nurseries at the WRPIS, and each plot was isolated by at least 25 m from other accessions of the same species. [This isolation distance was shown by Johnson et. al. (1996) to keep average genetic contamination by pollen to less than 5% in Bromus inermis Leyss.] Transplanting was completed at Central Ferry during the third week of April, and at Pullman during the second week of May 1997. Plots were maintained over the summer of 1997 and cut before seed development.
In 1998, individual plants from each plot were harvested by hand when the maximum number of seeds were judged to have reached physiological maturity. Since plant development varied within accessions, this usually required several harvest times for each plot. For one accession of each species, an inflorescence sample was taken from each plant. The inflorescence sample consisted of selecting two representative panicles or spikes from each plant and using the resulting seeds for a separate calculation of Ne/Nc. The contribution of the inflorescence samples was added to the total seeds per plant for calculation of mean seed number per plant, the standard deviation of seeds among plants, and Ne/Nc for whole plants. Seeds were threshed and cleaned separately for each sample, and total seed weight and 100-seed weight determined. The total sample seed weight and 100-seed weight was used to determine seeds per whole plant or per inflorescence sample and the standard deviations of seed number among plants within each plot of 30 plants calculated. Estimates of Ne/Nc were made as described above for each plot. Data were analyzed over locations by means of SAS general linear models (GLM) release 6.12, assuming fixed effects with treatment differences declared at P < 0.05 (SAS Institute, 1985). The variance was partitioned into locations, block within locations, species, accessions (within species), and associated interactions. For the analysis of the inflorescence samples, the same general model was used excluding accession effects. If a treatment difference was significant, the LSD at <0.05 was used for multiple comparisons.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
than at Central Ferry 
. Although other factors may have contributed to location differences, there was a high incidence of rust (Puccinia spp.) on L. perenne, and to a lesser extent on F. pratensis observed at the irrigated Central Ferry site. At the dryland Pullman site, rust was less severe. Plants of P. spicata did not show disease symptoms at either location, and location means for P. spicata did not differ between Central Ferry and Pullman for treatment factors (P < 0.05). There were no obvious differences in rust incidence among accessions within Lolium and Festuca. There were significant differences among species, and accessions within species, for mean seeds per plant, standard deviations for seed number among plants, and Ne/Nc (P < 0.05). Since the interactions with location were not significant, the accessions were averaged over locations (Table 2). Although seed number per plant was highly variable, it differed significantly within F. pratensis accessions, but not among L. perenne or P. spicata accessions. Differences in standard deviations were detected for F. pratensis and L. perenne accessions, but not P. spicata accessions (Table 2). For Ne/Nc, differences among accessions within species were only observed for L. perenne. For the L. perenne accessions W6 9344 and W6 9359, Ne/Nc values were less than 0.40, showing strong reductions in Ne associated with high variation in seed number among plants. Overall, F. pratensis had significantly higher seed numbers per plant, and standard deviations among plants, than L. perenne or P. spicata, but L. perenne and P. spicata did not differ (Table 2). The differences among species for Ne/Nc were always significant (P < 0.05) with P. spicata having the highest and L. perenne the lowest values (Table 2). As far as we know, these are the first estimates of Ne in regeneration populations of these species. The relatively low Ne/Nc emphasizes the strong potential for genetic drift in small regeneration populations of outcrossing grass germplasm.
|
. Thus inflorescence samples always had a higher Ne/Nc than the whole plant samples as s2/z2 was reduced. Although substantial variation in seed number per inflorescence still exists, a major source of variation in seed number among plants; that is, inflorescence number per plant, was eliminated. Since inflorescence sampling does not control variation in pollen production among plants before seed development, the values from this method represent improvements in Ne/Nc only associated with the contribution of maternal gametes in the population. For whole plant seed sampling, however, the variation in seed number among plants represents the variation in both ovule and pollen production on Ne/Nc.
|
We assumed a fixation index F of zero, which is expected for the species in this study (Brown, 1979; Johnson, 1998). In some cases, a degree of selfing does occur in normally self incompatible species resulting in an F > 0. The effect of this is to increase the denominator in Eq. [1] leading to lower Ne/Nc. For example, if F = 0.25 then Ne/Nc would have declined to 0.47 compared with the experimental wide average of 0.52 when F = 0. Thus, a 25% increase in F would have led to just a 10% decreases in Ne/Nc.
Greater variation in seed number among plants and fewer seeds per plant would be expected when plants are under either biotic or abiotic stress. In this study the prevalence of rust in L. perenne and to a lesser extent F. pratensis appeared to increase variation in seed number among plants and reduce seed production, especially at Central Ferry. The extent of rust in 1998 was unexpected on the basis of previous experience. But as a result, careful monitoring for stem rust and control through fungicides has been implemented to promote plant health during regeneration. Differences in seed production could also arise from naturally occurring genetic variation within and among accessions in response to environment, vernalization requirements, and different stresses.
How may germplasm managers offset limitations in resources to provide the best regeneration sample possible for the maintenance of diversity? It should be remembered that as long as seed held in long-term storage can be drawn from an original, or close to original sample with high viability, there would be no need for regeneration and no genetic drift would occur. But at some point regeneration is inevitable, and a decision must be made concerning the basic plant population number per accession for a regeneration program. This should be in terms of Ne rather than Nc, recognizing that in a random mating diploid population, theory indicates that 1/(2Ne) heterozygotes will be lost per generation (Crow and Kimura, 1970). Once the target Ne is established the manager may control mating to maximize Ne or simply increase Nc to compensate for the estimated reductions in Ne associated with variation in fecundity.
To maximize Ne with the fewest plants possible, mating can be controlled so that each individual in the population contributes two gametes to the next generation. With this sampling system Ne = 2N - 1 so Ne is essentially doubled compared with Nc (Crow and Kimura, 1970). To accomplish this in self-incompatible grasses, paired inflorescences from different plants could be bagged together to obtain a reciprocal cross. If the progeny from the crosses are kept separate, a viable seed from each parent can be used for future regeneration. But the expense of bagging, separate plant harvests, seed cleaning, storage, and the potential for differential poor seed set and seed quality of bagged plants often makes this approach impractical.
At the WRPIS, about 600 accessions are regenerated each year and this has barely kept pace with the average number of incoming accessions. If the target Ne is 50, corresponding to an expected erosion in hetrozygosity of 1.0% per regeneration, 30 000 separate harvests, seed cleaning, and seed counts would be needed before equal numbers of seeds per plant could be assembled for the 600 accessions. Paired crossing or controlled polycrosses described by Breese (1989) that require individual plant harvests are not possible with current resources. Bulk sampling with uncontrolled mating, but with increased Nc to approximate the desired Ne, may be the most cost effective option. Considerable mechanization for planting and care of nurseries is available making increases in plant populations per accession a feasible alternative to controlled crossing. On the basis of the Ne/Nc average of 0.52 (Table 2), it would take 96 plants to obtain a Ne of 50. Sampling a constant inflorescence number per plant would further reduce the Nc needed to attain the desired Ne.
Research is needed concerning the optimum number of inflorescences to be sampled per plant, and this is underway at the WRPIS for wide range of species and accessions. Because of the known relationship between the variance among individuals and among means of individuals (mean s2 = 
) (Steel and Torrie, 1980), increasing the number of inflorescences m sampled per plant would decrease mean s2within plants. Since s2within plants is a component of total s2among (Table 1), a reduction in s2among would be expected with increased m. As a result, s2/z2 in Eq. [2] would decrease and Ne/Nc for the maternal gamete variation would increase. When seeds per plant are equal, the maternal gamete variation is zero and Ne is increased by about one-third compared with whole plant sampling (Breese, 1989). In other words, without controlling the variation in pollen production among plants this would be the upper limit of the expected gain in Ne associated with methods that equalize seeds per plant.
On the basis of the average Ne/Nc of 0.52, the WRPIS has set a minimum regeneration target population of 100 plants per accession. In addition, if a constant number of inflorescences per plant is sampled, Ne/Nc will consistently increase compared with bulking seeds from whole plants. When available resources prevent the application of controlled crosses to maximize Ne, selecting an appropriate population size based on Ne/Nc estimates, along with sampling a constant number of inflorescences, is recommended as a cost effective approach to grass seed regeneration.
|
ACKNOWLEDGMENTS |
|---|
Received for publication January 29, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. C. Johnson, V. L. Bradley, and M. A. Evans Inflorescence Sampling Improves Effective Population Size of Grasses Crop Sci., July 1, 2004; 44(4): 1450 - 1455. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
i is the mean number of seeds per inflorescence on the ith plant, and