Light and heavy turfgrass seeds differ in germination percentage and mean germination thermal time.
Variation in seed weight or size exists among species (Werker, 1997), among cultivars within species (Churchill et al., 1997), and among seed lots within cultivars (Christians et al., 1979). Variation also exists among seeds within a population, either because of variation among seeds from individual plants (Weis, 1982) and/or because of variation among seeds from the same plant (Briggs, 1991; Werker, 1997). The relationship between seed weight or size and various parameters of seed performance has often been studied among seed lots of different seed weights (e.g., Christians et al., 1979; Newell and Bludau, 1993). Among seed lots, many interfering genetic and developmental factors may, however, be confounded with seed size and, hence, affect the relations between seed size and seed performance, whereas individual seeds within a seed lot have a more common background (Perry, 1980). Within a seed lot, environmental conditions may primarily affect the number of seeds per plant and not the size of individual seeds on the plant (McGinley, 1993), and the fraction of "under-sized" seeds may be low and remain relatively constant over a wide seed yield range (Roy et al., 1994). Nevertheless, within any sample of seeds, there is often a large variation in seed weight (Arnott, 1969; Naylor, 1980), e.g., because of floret position, tiller order (Briggs, 1991), harvest time, or variations in flowering time (Nordestgaard, 1983). Besides, there may be a considerable fraction of unfilled spikelets within a seed lot of grass species (Naylor, 1972). By presenting averages of seed weights or sizes for seed lots, much of the variation between individual seeds within a population is obscured (Kane and Cavers, 1992).
Germination percentage and germination speed are both considered sensitive indicators of seed vigor (Ellis and Roberts, 1980), and many papers are concerned with the effect of seed weight or size on these indicators of performance. A positive relationship between seed weight or size and germination percentage has been found within seed lots in a number of studies (Brown, 1977; Macchia and Magnani, 1982; Veronesi et al., 1983; Klinga, 1986a, 1986b). In some studies, medium size seeds had higher germination percentage than both smaller and larger seeds (McKersie et al., 1981; Meharwade et al., 1992), and in other studies, no relationship has been found between seed weight or size and germination percentage (Hayes, 1975; Nordestgaard, 1983). In certain studies, a negative relationship has been found between seed weight or size and germination percentage (Macchia and Magnani, 1982; Milberg et al., 1996).
Studies of the relationship between seed weight or size and germination speed have also provided various conclusions. In some studies, heavy seeds germinated faster than lighter or smaller seeds (Weis, 1982; Charlton, 1989). In one study, medium weight seeds germinated faster than both lighter and heavier seeds in one population whereas in another population the medium weight seeds had the slowest germination (Milberg et al., 1996). In some cases, there was no relationship between seed weight or size and germination speed (Macchia and Magnani, 1982), and in certain studies, heavy seeds germinated more slowly than light seeds (McKersie et al., 1981; Weis, 1982).
Thus, various patterns have been found in studies of the effect of seed weight or size on germination percentage and germination speed. Most studies, however, have been performed on the basis of very few seed weight or size fractions from the seed lot, often only two (Hayes, 1975) or three (Brown, 1977). The use of more fractions is likely to detect a more consistent relationship between seed weight or size and germination. Only very few studies have attempted to describe the relationship between seed weight or size and seed performance within seed lots (Aiken and Springer, 1995), whereas most studies have only included pair-wise comparisons of the performance of the seed weight or size fractions (e.g., Macchia and Magnani, 1982; Klinga, 1986b). To get a better understanding of the relationship between seed weight or size and seed performance, it is necessary to include a considerable number of seed weight or size fractions from the same seed lot, and the data must be subjected to an analysis which can describe the relationship sufficiently over the whole range of seed weights or seed sizes within the seed lot. The aim of this study was to (i) investigate and describe the potential relationship between seed weight and germination within seed lots of red rescue, perennial ryegrass, and Kentucky bluegrass, (ii) study the relationship between temperature and germination of seeds of different weight, and (iii) determine the potential effect of removing a certain fraction of the seeds on the overall germination of the remaining seeds in a seed lot.
MATERIALS AND METHODS
Seeds from one seed lot of red rescue cultivar Symphony, perennial ryegrass cultivar Taya, and Kentucky bluegrass cultivar Andante were used. The seed lots were chosen because they had relatively high germination percentage (91, 98, and 89%, respectively) and fast germination in a previous screening experiment (data not shown), comprising 20, 19, and 16 seed lots of the three species, respectively. The seed lots were commercially produced and harvested in Denmark in year 2000. All three seed lots originated from one single field, measuring 8.8 ha for red rescue, 9.8 ha for perennial ryegrass, and 12.4 ha for Kentucky bluegrass. A sample of 250 g was drawn according to standard sampling procedures (ISTA, 1999) from bulk seed lots, which weighed 4950, 10 150, and 8780 kg for the three species, respectively. The bulk seed lots had been cleaned to a purity of 97.5, 99.6, and 92.0%, respectively.
A sample of 100 g of each seed lot was separated in seed weight fractions with an air separator (Damas, model LAST1, Vester Aaby, Denmark), which can separate seeds into three fractions with a vertical air stream. By regulating the airflow, our intension was to obtain three seed weight fractions of equal total weight but with different average TSW. Since the air separator can only divide a sample into three fractions, each of the three seed weight fractions was separated again into three subfractions of equal total weight by regulating the airflow. The two-step grading procedure thus produced a total of nine seed weight fractions of equal weight but with different average TSW values. The grading procedure was chosen as a realistic industrial method to separate seed fractions by weight and is an alternative to weighing individual seeds.
The total weight of each seed weight fraction was determined and the proportion of the total seed lot was calculated. TSW was determined for each of the nine seed weight fractions and for a nongraded control sample by counting and weighing eight replicates of 100 seeds. TSW is calculated following officially recommended procedures (ISTA, 1999). Seed moisture content was determined in a nongraded seed sample by drying two replicates of 5 g of each seed lot for 1 h at 130[degrees]C (ISTA, 1999). TSW for all seed weight fractions was corrected by calculating the seed dry weight plus 100 g [kg.sup.-1] moisture content.
Germination Test in the Laboratory
Germination tests were performed from March to May 2001. For each species, four replicates of 100 seeds from each seed weight fraction and from the nongraded control sample were counted and weighed (the same samples as used for the determination of TSW). Each seed sample of known TSW was then germinated in a standard laboratory test at 15/25[degrees]C darkness/light for 16/8 h per day (ISTA, 1999) in two germination incubators (Termaks KBP 6395 LL, Norway). Seeds were germinated on the top of paper in a "mini Jacobsen apparatus" (Copenhagen Tank) (ISTA, 1999). Before the germination test, the filter paper was soaked in a 0.2% (w/v) solution of KN[O.sub.3] (ISTA, 1999). The germination boxes were daily rearranged to avoid effects of potential temperature differences or gradients in the incubators.
Germination was recorded once or twice daily for 12 d for red fescue and perennial ryegrass and for 21 d for Kentucky bluegrass. To avoid a "saw tooth" cumulative germination curve under the diurnally alternating germination temperatures (Brown and Mayer, 1988), the two daily counts were performed at such intervals that the average temperature between each counting was constant. A seed was regarded as germinated when the radicle had protruded at least 1 mm. Only seeds with normal radicles were counted. Germinated seeds were removed from the germination test.
The germination experiment was repealed at a lower temperature regime at 5/15[degrees]C darkness/light for 16/8 h per day. In this experiment, germination was recorded once daily for 25 d for red rescue, 21 d for perennial ryegrass, and 60 d for Kentucky bluegrass.
For each species, the relationship between TSW and the cumulative weight proportion of the seed lot [Cum(TSW)] was determined on the basis of TSW values of the nine seed weight fractions. The relationship was calculated by probit analysis with the probit procedure of the SAS package (SAS, 2000):
 Cum(TSW) = [PHI][TSW - [[mu].sub.TSW])/[S.sub.TSW]]
where [PHI] is the cumulative distribution function of the standard normal variable, presented as percent, [[mu].sub.TSW] is the average TSW and [S.sub.TSW] is the standard deviation in TSW within the seed lot. The coefficient of variation (CV) was calculated as the standard deviation in percent of the average TSW and was used as an expression of uniformity of seed weight within a seed lot.
Final germination percentage (GP) was calculated as the cumulative number of germinated seeds with normal radicles:
 GP = [SIGMA]n
where n is number of seeds that had germinated at each counting.
Mean germination time (MGT) in days was calculated according to Ellis and Roberts (1980):
 MGT = [SIGMA](Dn)/[SIGMA]n
where n is number of seeds germinated on day D, and D is number of days counted from the beginning of the germination test. Differences in GP and MGT between the control sample and the weighted mean of the nine seed weight fractions were tested for each species and temperature, using t tests of the glm procedure of the SAS package (SAS, 2000).
Nonlinear relationships between TSW and germination data were analyzed, using the procedure nlin of the SAS package (SAS, 2000). For each species and temperature regime, a nonlinear function was fitted to the data for the relationship between TSW and GP for seed weight fractions:
 GP = a - b x [e.sup.-cTXSW]
where parameter a is the asymptotic maximal GP. Parameter b is the difference between GP and the intercept with the y axis, i.e., the difference between maximal GP and the theoretical GP if TSW is zero. Parameter c expresses the rate of increase in GP with increasing TSW.
For each species and temperature regime, a nonlinear function was fitted to the data for the relationship between TSW and MGT for seed weight fractions:
 MGT = a + b x [e.sup.-cXTSW]
where parameter a is the asymptotic minimal MGT. Parameter b is the difference between the minimal MGT and the intercept with the y axis, i.e., the difference between minimal MGT and the theoretical MGT if TSW is zero. Parameter c expresses the rate of decrease in MGT with increasing TSW.
When the seed weight fraction with the lightest seeds was excluded from the data, the relationships between TSW and GP and between TSW and MGT appeared to be linear, and the data was subjected to linear regression in two models including temperature regime, TSW, and the interaction between temperature regime and TSW:
 GP = [a.sub.i] + [b.sub.i] x TSW
 MGT = [a.sub.i] + [b.sub.i] x TSW
where parameter [a.sub.i] is the intercept with the y axis and [b.sub.i] is the slope for temperature regime i. Species were analyzed separately since the variance was not homogeneous among species.
To study the relationship between TSW and thermal time to germination, the base temperature for germination ([T.sub.b]) was determined for each species by regressing the reciprocal of MGT, i.e., the germination rate, against temperature, using 11.7 and 21.7[degrees]C as mean daily temperature (T) at the two temperature regimes, respectively. The base temperature was estimated as the intercept with the x axis, i.e., the temperature at which germination rate is zero (Roberts, 1988). The estimated base temperature for each species was used for calculation of the mean germination thermal time (MGTT) for each seed weight fraction:
 MGTT = MGT x (T - [T.sub.b])
After exclusion of the seed weight fraction with the lightest seeds, the relationship between MGTT and TSW was studied within each species by linear regression, using the following model:
 MGTT = [a.sub.i] + [b.sub.i] x TSW
where parameter [a.sub.i] is the intercept with the y axis and [b.sub.i] is the slope for temperature regime i.
Since GP can only be in the range from 0 to 100% and MGT cannot be zero or negative, the nonlinear and linear functions are only defined within the possible ranges.
The relationship between the proportion (weight percent) of seeds removed from the seed lot, by successively excluding the lightest seeds, and the GP and MGT, respectively, was determined by calculating the weighted mean values when successively excluding from zero to eight of the nine seed weight fractions.
TSW Distribution Within Seed Lots
Weighted mean TSW of all seed weight fractions of graded seeds were 0.819, 1.739, and 0.309 g for red rescue, perennial ryegrass, and Kentucky bluegrass, respectively. These TSW values were slightly higher than for the nongraded control samples, which were 0.790, 1.676, and 0.274 g, respectively. When fitting a cumulative normal distribution to the TSW distribution by the probit analysis (Fig. 1), there was significant lack of fit for all three species (P < 0.001, [R.sup.2]-values ranging from 0.87-0.96), indicating that TSW is not normally distributed within the seed lots. However, it is obvious from Fig. 1 that the normal distribution describes the data relatively well in the range from 20 to 80% of the seed lot, whereas there seems to be systematic deviation at the tails. When the seed weight fraction with the lowest TSW was excluded, there was no longer significant lack of fit in perennial ryegrass (P = 0.144) and in Kentucky bluegrass (P = 0.086). By also excluding the fraction with the highest TSW within each species, there was no significant lack of fit in red fescue (P = 0.082), perennial ryegrass (P = 0.721) or Kentucky bluegrass (P = 0.255). The coefficient of variation (CV), in TSW based on the probit analysis, was 21.6, 16.8, and 37.4% for the three species, respectively, indicating that in relative terms, TSW is most heterogeneous within the seed lot of Kentucky bluegrass and least heterogeneous within the seed lot of perennial ryegrass. However, since the species are represented by only one seed lot each, other distributions of TSW may be found in other seed lots.
[FIGURE 1 OMITTED]
Variation in Germination
Germination of samples from the various seed weight fractions with different TSW showed a considerable variation in GP as well as MGT (Table 1, Fig. 2 and 3). There was no effect of temperature on GP within the seed lots. On the other hand, mean MGT was considerably higher and there was a larger range of MGT within a seed lot at the lower temperature than at the higher temperature, particularly for Kentucky bluegrass.
At both germination temperature regimes, the weighted mean value of GP and MGT for the nine seed weight fractions was generally very consistent with the values of the nongraded control sample, with the exception of GP for Kentucky bluegrass at both temperatures and for red fescue at 5/15[degrees]C, which were significantly higher than for the nongraded control (Table 1).
Nonlinear Functions Describing the Relationship between TSW and Germination
The fitted nonlinear functions described quite well the relationship between TSW and GP (Fig. 2) and between TSW and MGT (Fig. 3), although there was in some cases a considerable deviation between observed and predicted values. In all cases, convergence was obtained m the iteration estimation procedure. Parameter estimates, confidence limits and overall standard error for the fitted functions are provided for TSW and GP (Table 2) and for TSW and MGT (Table 3). Overall standard errors for the fitted relationships ranged from 1.4 to 3.1% for GP (Table 2) and from 0.1 to 0.7 d for MGT (Table 3), which is reasonable compared to the observed range in GP and MGT (Fig. 2 and 3). For both GP and MGT, the seeds from the seed weight fraction with lightest seeds had a large impact on the shape of the fitted functions and particularly parameters b and c (Fig. 2 and 3, Table 2 and 3).
For the relationship between TSW and GP, parameter a was similar for all three species and both temperatures (Table 2), i.e., the heaviest seeds generally reached the same high asymptotic final GP which is consistent with observed maximum GP (Table 1). Together with parameter c, parameter b defines the shape of the function in the relevant TSW range, and in general there were wide confidence limits for parameters b and c. Parameter c, which indicates the increase in GP with increasing TSW, was significantly higher for Kentucky bluegrass than for red fescue at both temperature regimes and higher than for perennial ryegrass at 5/15[degrees]C (Table 2). For perennial ryegrass, there was a poor relationship between TSW and GP for the heaviest seeds and a large influence from the lightest seeds at 15/25[degrees]C (Fig. 2), leading to convergence with a poor fit with very wide confidence limits for parameters b and c (Table 2). For the relationship between TSW and MGT, there were also wide confidence limits for parameters b and c (Table 3). In red fescue and Kentucky bluegrass, parameter c was lower at 5/15[degrees]C than at 15/25[degrees]C, indicating a more slow decrease in MGT with increasing TSW at the low temperature regime (Table 3). However, the relative difference in MGT between the lightest seeds and the heaviest seeds was very similar at the two temperature regimes. Hence, the light seeds of red fescue took 13% longer to germinate than the heavy seeds at both the low and high temperature, whereas light seeds of Kentucky bluegrass took 23 and 26% longer to germinate than the heavy seeds at the low and high temperature, respectively. For perennial ryegrass, the light seeds took 6 and 15% longer to germinate than heavy seeds at the two temperatures, respectively.
Linear Functions Describing the Relationship between TSW and Germination
When the observations for the seed weight fraction with lightest seeds were excluded, the relationship between TSW and germination could be adequately described by linear functions (Fig. 2 and 3). TSW positively affected the GP in all three species (P [less than or equal to] 0.010) (Fig. 2, Table 4), but temperature did not affect neither the slope (P - 0.174-.409) nor the intercept (P = 0.135-.345) of the relationship between TSW and GP. TSW affected MGT negatively in all three species (P [less than or equal to] 0.029), although not significantly in red fescue at 15/25[degrees]C (P = 0.241) (Fig. 3, Table 5). In all three species, the intercept of the relationship between TSW and MGT was significantly higher at 5/15[degrees]C than at 15/25[degrees]C (P [less than] 0.001). Temperature did not affect the slope in perennial ryegrass (P = 0.852) but in red fescue and Kentucky bluegrass (P [less than] 0.001) with more negative slopes at the lower temperature (Table 5).
The estimated base temperatures were 5.2, 5.4, and 5.3[degrees]C in red rescue, perennial ryegrass, and Kentucky bluegrass, respectively. When expressed on a thermal time scale, the average MGTT for all seed weight fractions of red rescue was 67.5 degree days at 5/15[degrees]C and 67.6 degree days 15/25[degrees]C. The corresponding values were 52.3 degree days and 52.7 degree days for perennial ryegrass and 113.2 degree days and 113.6 degree days for Kentucky bluegrass, indicating that within each species very similar thermal times were required for germination at the two temperature regimes. In all species, there was a significant negative relationship between MGTT and TSW (P [less than] 0.002) (Fig. 4, Table 6), suggesting that TSW affect the required thermal time for germination. Neither the slope (P = 0.116-0.403) nor the intercept (P = 0.111-0.403) of the relationship between MGTT and TSW was affected by temperature, indicating that there was no interaction between the effect of TSW and temperature.
The Effect of Removing the Lightest Seeds
By successively removing the lightest seeds from the seed lot, the GP increased gradually and MGT decreased gradually (Fig. 5). By removing the seeds in the fraction with the lightest seeds, i.e., 9 to 13% of the seed lot, GP increased 2.7% for red fescue, 0.4 to 0.7% for perennial ryegrass, and 6.8 to 7.1% for Kentucky bluegrass with a similar increase at both temperature regimes. Exclusion of 21 to 27% of the lightest seeds increased GP 3.4 to 3.5%, 0.6 to 0.7%, and 7.5 to 8.2% for the three species, respectively. Exclusion of the 9 to 13% of the lightest seeds decreased MGT 0.0 to 0.1 d for red rescue, 0.0 to 0.1 d for perennial ryegrass, and 0.1 to 0.2 d for Kentucky bluegrass, with the largest increase obtained at the lower temperature regime. Exclusion of 44 to 50% of the lightest seeds decreased MGT 0.1 to 0.2, 0.1 to 0.2, and 0.4 to 0.9 d for the three species, respectively.
Seed Weight Distribution Within Seed Lots
The distribution of individual seed weights within a seed lot may be skewed toward small or large seeds (Carleton and Cooper, 1972), but it is often near normal and can be described by the mean and standard deviation (Naylor, 1976; Brown, 1977; Naylor, 1980; Veronesi et al., 1983). However, a normal distribution may be more easily detected when weighing individual seeds than when separating seeds into a number of fractions. In the present study, the air separator was set at air speed levels giving approximately the same weight proportion in each fraction. Therefore, the fractions with the extreme seed weights possibly cover a larger range of seed weights than the fractions of medium weights, and cause the relatively imprecise fit of the curves at the extreme values. The tendency toward left-skewness in all three species (Fig. 1) may be related to a larger proportion of immature or empty seeds in the seed weight fraction with the lightest seeds, causing a considerably lower mean TSW in this fraction. This is consistent with findings in blackgrass (Alopecurus myosuroides Huds.), where spikelet weight was not normally distributed because of a large proportion of empty seeds without caryopsis whereas caryopsis weight was normally distributed (Naylor, 1972). A separation of the seeds into a higher number of fractions may have given a better description of the seed weight distribution and possibly have revealed a normal distribution in the present study when excluding empty seeds.
Relationship between TSW and Germination
The present results shows that within the applied seed lots of red rescue, perennial ryegrass, and Kentucky bluegrass, there is a distinct relationship between seed weight and germination. As each species was only represented by one seed lot it is not possible to make generalizations beyond the three cultivars and specific seed lots used in the study. For instance, it cannot be concluded whether the relationship between seed weight and germination may be different for seed lots with lower vigor or lower GP or for seed lots from different cultivars. Since there are some common features of the relationship between seed weight and germination among the species, however, it is likely that the relationship may be of a more general nature among grass species. Thus, despite the three species differing considerably in their range of TSW as well as GP and MGT, the relationships shown in Fig. 2 and 3 exhibit some general features across the species.
The observations belonging to the seed weight fraction with the lightest seeds are very influential for the shape of the nonlinear relationships (Fig. 2 and 3), and a better separation of seed weights, using more weight fractions, might have defined more clearly the relationship at low seed weight levels. Furthermore, the degree of initial cleaning of the applied commercial seed lots may have an impact on the relationship (Wood et al., 1977). When the lightest seeds were excluded from the analysis, a linear function sufficiently described the relationship between seed weight and germination, and there was also a general effect of TSW on germination among the remaining seeds in the seed lot. However, the light seeds do belong to the seed population, even though some of them may be empty or immature, and a linear function was inadequate for describing the relationship for all seeds in the seed lot. In many earlier studies with a positive effect of heavy or large seeds, seeds from the fraction with the lightest or smallest seeds germinated more poorly than the others, whereas no clear differences were found among the other weight or size fractions (e.g., Brown, 1977; Klinga, 1986a; Klinga, 1986b), indicating that the relationship between seed weight or size and germination was not linear. A few studies have attempted to describe the relationship between seed weight or size and seed performance and have generally found a nonlinear relationship (Weis, 1982; Cuya and Lombardi, 1991; Aiken and Springer, 1995), although a linear relationship has been found in certain cases (Weis, 1982). Together with the present findings, these examples suggest that the relationship between seed weight and germination characteristics within seed lots may often be of a nonlinear nature when whole populations are considered.
TSW, Temperature, and Germination
In the present study, 15/25[degrees]C is considered a near optimal temperature regime, whereas 5/15[degrees]C is in the suboptimal temperature range and more realistic under field conditions. Milberg et al. (1996) stated that differences in germination speed are best detected at optimum temperatures while differences in GP are best detected at suboptimal temperatures. This was not the case in the present study in which the same differences in GP were found at both temperatures, indicating that light and heavy seeds do not respond differently to temperature. Conversely, differences in MGT were more pronounced at 5/15[degrees]C, causing larger differences in MGT between light and heavy seeds in red fescue and Kentucky bluegrass. When expressing time to germination on a thermal time scale, however, there was no difference between the required thermal time for germination at the two temperature regimes and there was no interaction between TSW and temperature.
Individual seeds within a population often have the same base temperature but differ in their thermal time requirements for germination (e.g., Garcia-Huidobro et al., 1982; Covell et al., 1986). In the present study, the base temperature was assumed to be constant for the population. Although the estimation of base temperature was based on only two temperatures, the thermal time model explains well the variation in time to germination at the two temperatures, and in all three species light seeds require a longer thermal time to germinate. This is in contrast to results of Wagenvoort and Bierhuizen (1977) who found that seed size affected neither base temperature nor the temperature sum required for germination. The relationship between TSW and thermal time to germination may not be described sufficiently by a linear function when including all seed weight fractions. However, the linear functions describing the relationship for the majority of the seeds in the seed lots (Fig. 4) clearly suggest that some of the variation in thermal time requirement among individual seeds may be related to variation in seed weight.
The Effect of Removing the Lightest Seeds on Seed Lot Performance
The effect of seed grading on seed lot performance depends on the seed weight distribution within the seed lots and on the relationship between seed weight and germination. For the applied seed lots, exclusion of the lightest seeds generally had a small effect on GP and MGT (Fig. 5). The largest effect was in Kentucky bluegrass, corresponding to the relatively larger variation in TSW and the more pronounced tendency toward light seeds germinating more poorly than heavy seeds in this species (Fig. 2 and 3). The greatest improvement of GP is obtained by excluding the first fraction of the seed lot, whereas exclusion of further fractions only gives a relatively small improvement. The effect of exclusion of the lightest seeds on MGT follows a more linear pattern with MGT decreasing with increasing proportion of excluded seeds.
In Sorghum hybrids, Gowda et al. (1997) found that removing 7.9% of the smallest seeds in a seed lot increased GP from 75.3 to 92.7%. This indicates a very large proportion of very small and nonviable seeds within this seed lot, causing a large potential for improving seed lot performance by seed grading. Conversely, a number of studies did not find any difference in performance between the heaviest or largest seeds and the nongraded control sample, and consequently there was no benefit from grading seeds although the smallest seeds performed poorer than the nongraded seeds (Macchia and Magnani, 1982; Clarke, 1985; Cuya and Lombardi, 1991), This is consistent with the present results; despite the observed relationship between seed weight and germination, there is only a limited potential for improving GP and MGT by removing light seeds from these seed lots. Different seed weight distributions and different relationships may, however, be found in other seed lots or cultivars.
It is likely that a reduced germination time for Kentucky bluegrass, obtained by excluding small seeds from a seed lot, may have a positive effect on the competitive ability of this species during the establishment phase.
Furthermore, since differences in seed vigor are often more pronounced under suboptimal field conditions than in laboratory germination tests (ISTA, 1995), it is possible that the use of heavy seeds may have a larger impact during field establishment. Still, it is clear that grading of the applied seed lot of Kentucky bluegrass cannot reduce MGT to a level, which is similar to that of red fescue and perennial ryegrass. Thus, despite a clear positive relationship between seed weight and important germination characteristics, seed grading is unlikely to improve the establishment properties to a magnitude that can change the competitive balance between the three species during establishment. Besides, the need for removing a large proportion of the seed lots to obtain an effect on germination will possibly cause too large expenses to make this a commercially feasible method in seed technology. It is likely, however, that the effect of seed grading may be larger in seed lots of lower initial GP. This needs further investigation.
Abbreviations: GP, germination percentage; MGT, mean germination time; MGTT, mean germination thermal time; TSW, 1000-seed weight.
Table 1. Final germination percentage and mean germination time (MGT) for nongraded control and for graded seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass at two temperature regimes. Final germination Control Seed weight fractions Mean Mean ([double Species Temperature ([dagger]) dagger]) [degrees]C % Red fescue 5/15 87.5 90.7 * 15/25 90.8 92.1NS ([paragraph]) Perennial ryegrass 5/15 97.0 97.4NS 15/25 98.0 97.8NS Kentucky bluegrass 5/15 79.0 86.0 * 15/25 76.5 86.8 * Final germination MGT Seed weight fraction Control Minimum Maximum Mean Species ([section]) ([section]) ([dagger]) % d Red fescue 64 100 10.4 65 99 4.0 Perennial ryegrass 89 100 8.5 90 100 3.3 Kentucky bluegrass 30 100 17.5 33 99 6.9 MGT Seed weight fractions Mean ([double Minimum Maximum Species dagger]) ([section]) ([section]) d Red fescue 10.4NS 9.7 11.7 4.1NS 3.8 4.7 Perennial ryegrass 8.4NS 8.1 9.0 3.2NS 3.1 3.9 Kentucky bluegrass 17.7NS 15.6 21.0 7.0NS 6.3 8.5 ([dagger)] Mean values for control samples is based on four replicates of 100 seeds. [(double dagger]) Mean for seed weight fractions is weighted mean of four replicates of 100 seeds from nine fractions. Level of significance indicates difference between control and weighted mean for nine fractions. ([section]) Minimum and maximum values are based on single replicates of 100 seeds. ([paragraph]) NS = not significant. Table 2. Estimates and confidence limits for parameters in nonlinear functions (Eq. ) describing the relationship between 1000-seed weight (TSW, g) and germination percentage (GP, %) at two temperature regimes for seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass. Parameter a Approx. 95% Species Temperature Estimate confidence limits [degrees]C Red fescue 5/15 96.3 93.4 99.3 15/25 99.6 96.8 102.3 Perennial ryegrass 5/15 99.1 96.7 101.5 15/25 98.4 97.9 98.9 Kentucky bluegrass 5/15 97.2 95.3 99.0 15/25 96.9 95.4 98.4 Parameter b Approx. 95% Species Estimate confidence limits Red fescue 682.0 -169.0 457.6 124.7 Perennial ryegrass 105.2 -175.0 8.05[e.sup.8] -1.69[e.sup.10] Kentucky bluegrass 620.9 406.8 718.4 451.0 Parameter b Parameter c Approx. 95% Approx. 95% Species confidence limits Estimate confidence limits Red fescue 1532.9 6.7 3.9 790.5 5.5 3.9 Perennial ryegrass 385.4 2.6 -0.1 1.85[e.sup.10] 16.4 3.1 Kentucky bluegrass 834.9 16.8 14.3 985.7 18.0 15.3 Parameter c Approx. 95% Standard Species confidence limits error % Red fescue 9.5 3.1 7.1 2.3 Perennial ryegrass 5.2 1.5 35.9 1.4 Kentucky bluegrass 19.4 3.0 20.7 2.6 Table 3. Estimates and confidence limits for parameters in nonlinear functions (Eq. ) describing the relationship between 1000-seed weight (TSW, g) and mean germination time (MGT, days) at two temperature regimes for seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass. Parameter a Approx. 95% Species Temperature Estimate confidence limits [degrees]C Red fescue 5/15 9.0 5.1 12.9 15/25 4.0 3.8 4.1 Perennial ryegrass 5/15 8.2 8.0 8.5 15/25 3.1 3.0 3.2 Kentucky bluegrass 5/15 3.2 -66.3 72.6 15/25 5.9 5.2 6.6 Parameter b Parameter c Approx. 95% Species Estimate confidence limits Estimate Red fescue 4.4 2.8 6.0 1.5 7.0 -3.7 17.6 5.2 Perennial ryegrass 16.7 -37.0 70.5 2.9 11.2 -12.1 34.4 2.7 Kentucky bluegrass 18.5 -48.9 85.8 0.8 4.0 3.1 4.9 4.6 Parameter c Approx. 95% Standard Species confidence limits error d Red fescue -2.4 5.3 0.3 1.8 8.6 0.1 Perennial ryegrass -0.2 6.1 0.2 0.7 4.7 0.1 Kentucky bluegrass -3.0 4.5 0.7 1.6 7.6 0.2 Table 4. Estimates for parameters in linear functions (Eq. ) describing the relationship between 1000-seed weight (TSW, g) and germination percentage (GP, %) at two temperature regimes for seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass. Data for the seed weight fraction with the lightest seeds are excluded from the linear regressions. P values refer to a test of a slope different from zero. Species Temperature Function [degrees]C Red fescue 5/15 and 15/25 GP = 76.7 + 20.4 x TSW Perennial ryegrass 5/15 and 15/25 GP = 93.5 + 2.6 x TSW Kentucky bluegrass 5/15 and 15/25 GP = 82.1 + 33.9 x TSW Standard P value, Species error, slope slope % Red fescue 3.3 <0.001 Perennial ryegrass 1.0 0.010 Kentucky bluegrass 4.1 <0.001 Table 5. Estimates for parameters in linear functions (Eq. ) describing the relationship between 1000-seed weight (TSW, g) and mean germination time (MGT, days) at two temperature regimes for seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass. Data for the seed weight fraction with the lightest seeds are excluded from the linear regressions. P values refer to a test of a slope different from zero. Species Temperature Function [degrees]C Red fescue 5/15 MGT = 11.8 - 1.81 x TSW 15/25 MGT = 4.4 - 0.46 x TSW Perennial ryegrass 5/15 MGT = 8.7 - 0.21 x TSW 15/25 MGT = 3.6 - 0.21 x TSW Kentucky bluegrass 5/15 MGT = 21.2 - 11.1 x TSW 15/25 MGT = 8.1 - 3.80 x TSW Standard P value, Species error, slope slope d Red fescue 0.42 <0.001 0.39 0.241 Perennial ryegrass 0.09 0.029 0.09 0.029 Kentucky bluegrass 0.95 <0.001 0.96 <1.001 Table 6. Estimates for parameters in linear functions (Eq. ) describing the relationship between 1000-seed weight (TSW, g) and mean germination thermal time (MGTT, degree days) for seed weight fractions of red fescue, perennial ryegrass, and Kentucky bluegrass. Data for the seed weight fraction with the lightest seeds are excluded from the linear regressions. P values refer to a test of a slope different from zero. Species Temperature Function [degrees]C Red fescue 5/15 and 15/25 MGTT = 75.0 - 9.5 x TSW Perennial ryegrass 5/15 and 15/25 MGTT = 56.4 - 2.5 x TSW Kentucky bluegrass 5/15 and 15/25 MGTT = 134.3 - 67.0 X TSW Standard P value, Species error, slope slope degree days Red fescue 2.1 <0.001 Perennial ryegrass 0.8 0.002 Kentucky bluegrass 5.3 <0.001
We thank the Royal Veterinary and Agricultural University, Danish Centre for Forest, Landscape and Planning, and The Danish Research Agency for the financial support. The supply of seed samples from DLF-Trifolium A/S is gratefully acknowledged. Technical assistance from Olaf Bos and Lars Arne Jensen during the germination tests is greatly appreciated.
Aiken, G.E., and T.L. Springer. 1995. Seed size distribution, germination, and emergence of 6 switchgrass cultivars. J. Range Manage. 48:455458.
Arnott, R.A. 1969. The effect of seed weight and depth of sowing on the emergence and early seedling growth of perennial ryegrass (Lolium perenne). J. Br. Grassl. Soc. 24:104-110.
Briggs, K.G. 1991. Spatial variation in seed size and seed set on spikes of some Canadian spring wheat cultivars. Can. J. Plant Sci. 71:95-103.
Brown, K.R. 1977. Parent seed weight, plant growth, and seeding in 'Grassland Tama' Westerwolds ryegrass. N. Z. J. Exp. Agric. 5:143-146.
Brown, R.F., and D.G. Mayer. 1988. Representing cumulative germination. 2. The use of the Weibull function and other empirically derived curves. Ann. Bot. 61:127-138.
Carleton, A.E., and C.S. Cooper. 1972. Seed size effects upon seedling vigor of three forage legumes. Crop Sci. 12:183-186.
Charlton, J.F.L. 1989. Temperature effects on germination of 'Grassland maku' Lotus and other experimental Lotus selections. Proc. N. Z. Grassl. Assoc. 50:197-201.
Christians, N.E., J.F. Wilkinson, and D.P. Martin. 1979. Variations in the number of seeds per unit weight among turfgrass cultivars. Agron. J. 71:415-418.
Churchill, D.B., D.M. Bilsland. and M.D. Butler. 1997. Selection of woven wire screens for separating Poa pratensis L. and Poa trivialis L. seed. J. Appl. Seed Prod. 15:75-80.
Clarke, B. 1985. Cleaning seeds by fluidization. J. Agric. Engin. Res. 31:231-242.
Covell, S., R.H. Ellis, E.H. Roberts, and R.J. Summerfield. 1986. The influence of temperature on seed germination rate in grain legumes. I. A comparison of chickpea, lentil, soyabean and cowpea at constant temperatures. J. Exp. Bot. 37:705-15.
Cuya, O., and I. Lombardi. 1991. Influencia del tamano de semilla en la germination y crecimiento de plantulas de Schinus molle. Rev. For. Peru 18:17-27.
Ellis, R.H., and E.H. Roberts. 1980. Towards a rational basis for testing seed quality, p. 605-35. In P.D. Hebblethwaite (ed.) Seed production. Bunerworths, London.
Garcia-Huidobro, J., J.L. Monteith, and G.R. Squire. 1982. Time, temperature and germination of pearl millet (Pennisetum typhoides S. & H.). I. Constant temperature. J. Exp. Bot. 33:288-296.
Gowda, B., S J. Gowda, and R. Gowda. 1997. Effect of seed size on recovery and quality in hybrid Sorghum. Curr. Res. Univ. Agric. Sci. Bangalore 26:95-96.
Hayes, P. 1975. The influence of seed weight on seedling growth in perennial ryegrass, tall fescue and yorkshire fog. Rec. Agric. Res. 23:3343.
ISTA. 1995. Handbook of Vigour Test Methods. International Seed Testing Association, Zurich. Switzerland.
ISTA. 1999. International Rules for Seed Testing. Seed Sci. Technol. 27:1-333.
Kane, M., and P.C. Cavers. 1992. Patterns of seed weight distribution and germination with time in a weedy biotype of proso millet (Panicum miliaceum). Can. J. Bot. 70:562-567.
Klinga, K. 1986a. Aneuploidy in induced autotetraploid populations of Festuca pratensis, Lolium multiflorum and Lolium perenne. I. The frequency of aneuploids and performance of spaced plants in two autotetraploid populations of Festuca pratensis. Heredity 104: 75-83.
Klinga, K. 1986b. Aneuploidy in induced autotetraploid populations of Festuca pratensis, Lolium multiflorum and Lolium perenne. II. The frequency of aneuploids and performance of spaced plants in Lolium multiforum and Lolium perenne. Heredity 104:121-130.
Macchia, M., and G. Magnani. 1982. [Seed size and its influence on the germination aspects in some forage and horticultural species] Dimensione del seme, germinabilita ed energia germinitiva in alcune specie da orto e da foraggio. Agric. Ital. 111:75-89.
McGinley, M.A. 1993. Variation in reproductive characteristics of Poa pratensis across a successional chronosequence. Texas J. Sci. 45: 107-108.
McKersie, B.D., D.T. Tomes, and S. Yamamoto. 1981. Effect of seed size on germination, seedling vigor, electrolyte leakage, and establishment of bird's-foot trefoil (Lotus coniculatus L.). Can. J. Plant Sci. 61:337-343.
Meharwade, M.N., K. Giriraj, S.D. Shashidhar, B.S. Vyakaranahal, and P.R. Dharmatti. 1992. Influence of achene size-grading on seed quality and germination in sunflower. Karnataka J. Agric. Sci. 5:76-79.
Milberg, P., L. Andersson, C. Elfverson and S. Regner. 1996. Germination characteristics of seeds differing in mass. Seed Sci. Res. 6: 191-197.
Naylor, R.E.L. 1972. Aspects of the population dynamics of the weed Alopeeurux myosuroides Huds. in winter cereal crops. J. Appl. Ecol. 9:127-139.
Naylor, R.E.L. 1976. Changes in the structure of plant populations. J. Appl. Ecol. 13:513-521.
Naylor, R.E.L. 1980. Effects of seed size and emergence time on subsequent growth of perennial ryegrass. New Phytol. 84:313-318.
Newell, A.J., and N.K. Bludau. 1993. Variation in total germination, rate of germination and seed weight among cultivars of Poa pratensis. J. Sports Tuff Res. Inst. 69:83-89.
Nordestgaard, A. 1983. Spiringsundersogelser i graesmarksplanter. Tidsskr. Planteavl 7:445-456.
Perry, D.A. 1980. Seed vigour and seedling establishment. Adv. Res. Technol. Seeds 5:25-40.
Roberts, E.H. 1988. Temperature and seed germination, p. 109-132. In S.P. Long and F.I. Woodward (ed.) Plants and temperature. Society for Experimental Biology, Cambridge, UK.
Roy, S.K., M.P. Rolston, and J.S. Rowarth. 1994. Ryegrass seed yield loss due to under-sized seed. Proc. Agron. Soc. N. Z. 24:95-98.
SAS. 2000. The SAS system for Microsoft Windows. Release 8.01. SAS Inst., Cary, NC.
Veronesi, F., F. Damiani, S. Grando, and M. Falcinelli. 1983. The influence of seed weight on establishment and productivity in the first year in Lolium perenne L. Genet. Agrar. 37:391-402.
Wagenvoort, W.A., and J.F. Bierhuizen. 1977. Some aspects of seed germination in vegetables. II. The effect of temperature fluctuation, depth of sowing, seed size and cultivar, on heat sum and minimum temperature for germination. Sci. Hortic. 6:259-270.
Weis, I.M. 1982. The effects of propagule size on germination and seedling growth in Mirabilis hirsuta. Can. J. Bot. 60:1868-1874.
Werker, E. 1997. Seed Anatomy. Gebr0der Borntraeger, Berlin, Stuttgart.
Wood, D.W., P.C. Longden and R.K. Scott. 1977. Seed size variation; its extent, source and significance in field crops. Seed Sci. Technol. 5:337-352.
Soren U. Larsen * and Christian Andreasen
S.U. Larsen, Forest and Landscape Denmark, Rolighedsvej 23, DK-1958 Frederiksberg C, Denmark; C. Andreasen, Dep. of Agricultural Sciences, Royal Veterinary and Agricultural Univ., Thorvaldsensvej 40, DK-1871 Frederiksberg C, Denmark. Received 15 Mar. 2003. * Corresponding author (email@example.com).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Turfgrass Science|
|Author:||Larsen, Soren U.; Andreasen, Christian|
|Date:||Sep 1, 2004|
|Previous Article:||Simulating potential kernel production in maize hybrid seed fields.|
|Next Article:||Kentucky bluegrass response to potassium and nitrogen fertilization.|