The efficacy of dispersal: Intuitively, all species that occupy habitats undergoing succession are destined to be replaced in their existing transient habitats and must in time disperse elsewhere to survive. In perennial species dispersal away from the parental site also decreases the annual potential for sibling competition, which oftentimes may be more severe because of similarities in resource use among siblings than is competition with non-sib conspecifics (e.g. Ellstrand and Antonovics 1985; McCall et al. 1989), though numerous examples to the contrary also exist (e.g. Smith 1977; Williams et al. 1983; Kelley 1989; McCall et al. 1989). In species of Solidago diaspore morphology was selected for dispersal effectiveness and dispersal was at least in some respects optimized with the scale of environmental heterogeneity (Werner and Platt 1976; Platt and Weis 1977). Generally, though, the dispersal potential of a species can be increased by altering the pattern of resources allocated to respective diaspore parts through natural selection. In herbaceous wind-dispersed species this alteration can be reflected by reductions in fruit mass, by increasing the drag efficiency of structures such as pappus bristles, by increasing the relative allocation of resources to the dispersal structures (i.e. increasing the pappus bristle mass to fruit mass ratio), by releasing the diaspore at a higher height, or by some combination of these (Harper 1977).

This said, the commonly realized dispersal distances versus potential maximum dispersal distances under the most optimal of conditions are typically disparate values. The distance over which a wind-borne diaspore is dispersed is a function of launch height, rate of descent in still air, and wind velocity, which includes temporal, vertical, and horizontal components (Sheldon and Burrows 1973; Horn et al. 2001). For those species that produce pogonochores (wind-dispersed plumed seeds or fruits) such as oftentimes occurs in the Asteraceae, presumably most, if not all, are viable candidates for long-distance dispersal because variations in wind velocity and direction are greater than is the within-species variation in their rate of descent in still air (Horn et al. 2001).

The pappus as the primary structure affecting dispersal: The question of whether Asteraceae fruits are effectively dispersed over long distances has been debated for the past century (e.g. Small 1918a,b; Carlquist 1966; Sheldon and Burrows 1973; Matlack 1987). The primary structure affecting dispersal success in many genera of the Asteraceae, including the asters, is the pappus (Rowlee; 1893; Small 1918a,b; Carlquist 1967; Sheldon and Burrows 1973). The aster pappus is primitive, setose scabrid or setose denticulate, and represents the fusion of uniseriate rows of cells with the obtuse terminal cell of each row being free and projecting outwards as a lateral cilium for a distance which is less than the diameter of the seta (Small 1918a). The relative efficiency of the pappus as a dispersal structure is easily evaluated by measuring the rate of descent of the propagule through still air. The slower the rate of descent, the greater will be the relative effect of any upward or horizontal air currents on trajectory, and consequently the greater the distance over which dispersal will occur (Sheldon and Burrows 1973).

That the aerodynamics of pogonochores can be adequately modeled by considering the plume as a single long cylinder with a characteristic diameter and angle of attack has been established (Greene and Johnson 1990). However, small differences in the design of the plume may have significant consequences on the aerodynamics of a pogonochore (e.g. Augspurger and Franson 1987; Matlack 1987; Sacchi 1987). In the current study we address dispersal potential among species of commonly co-occurring asters in the genera Doellingeria, Eurybia, Oclemena, and Symphyotrichum. Presumably a high degree of functional convergence exists among not only these genera, but also less closely related, though co-occurring, genera such as Asclepias, Eupatorium, Lilium, Solidago, and Vernonia. During the course of this study we acknowledge that the one-seeded, dry, indehiscent fruit of the Asteraceae, though technically a cypsela, is typically referred to as an achene or seed. Specifically, achenes of the aster species are compared on the basis of mass allocation to respective parts (dispersal structures such as the pappus bristles versus the achene per se) and of their respective rates of descent in still air. Inasmuch as previous studies have hypothesized that selection acts to produce larger seeds in habitats with decreased moisture availability or through successional time (Salisbury 1942; Stebbins 1971; Baker 1972), we question whether the less weedy aster species exhibit a dispersal strategy that mirrors that of the weedier aster species.

Methods
Specimen collection: Mature capitula of the nine aster species, Doellingeria umbellata (Aster umbellatus var. umbellatus; open field), Eurybia macrophylla (A. macrophyllus; woodlot margin), Oclemena nemoralis (A. nemoralis; shoreline), Symphyotrichum lanceolatum (A. lanceolatus subsp. lanceolatus; open field), Symphyotrichum lateriflorum (A. lateriflorus; open field), Symphyotrichum novae-angliae (A. novae-angliae; open field), Symphyotrichum pilosum (A. pilosus; open field), Symphyotrichum puniceum (A. puniceus; open field), and Symphyotrichum urophyllum (A. urophyllus; open field), were harvested from groupings of stems (likely clones) during the fall of 2003. A single species, O. nemoralis, grew on the shore of Toad Lake, Muskoka District, Ontario, Canada, whereas the remaining natural populations were located in native areas on the grounds of the Slippery Rock University of Pennsylvania campus, Slippery Rock, Pennsylvania, USA.

Although seven of the species occurred in the same open field they occupied distinct regions within the habitat as S. puniceum is an obligate wetland species, D. umbellata, S. lanceolatum, and S. novae-angliae are facultative wetland species, S. pilosum and S. urophyllum are upland species, and the status of S. lateriflorum has not been decided. The woodland margin species E. macrophylla is an upland species, whereas the shoreline species O. nemoralis is a facultative wetland species (United States Department of the Army 1987). Each of S. lateriflorum, S. novae-angliae, and S. pilosum is weedy in both Canada and the United States (Anonymous 1990; Mulligan 1992; Chmielewski and Semple 2001a,b, 2003), S. lanceolatum is a weed species in Canada (Mulligan 1992; Chmielewski and Semple 2001a) but not the United States (Anonymous 1990), and none of D. umbellata, E. macrophylla, O. nemoralis, S. puniceum, or S. urophyllum is weedy in Canada or the United States.

Data collection: For each species mature capitula were harvested when natural dispersal was just beginning to occur. At the time of harvest capitula were placed in manila envelopes and returned to the laboratory where they were allowed to air dry. To determine their rate of fall in still air individual achenes were released at the top of a 120 cm length of glass tubing (22 mm internal diameter), and the time taken to fall through 1 m of tubing was measured with a digital stopwatch. The timing of descent did not begin until the achene had traveled 20 cm through the glass tube to allow the achene to first reach terminal velocity (Greene and Johnson 1990). A Cahn C-33 microbalance (+2μg) was used to determine total propagule mass for individual achenes. The length of the pappus bristles (hair) was measured with an ocular micrometer and the number of bristles was counted for each achene. Although Greene and Johnson (1990) measured the mean angle of attack of a hair (qh) by calculating cos-1 of the ratio of pappus radius to hair length, we directly measured the angle of attack of the pappus bristles (the angle from the horizontal to the outermost bristle) with an ocular protractor for each achene. The diameter of the pappus bristles was determined as the average of 25 pappus bristle measurements on a single achene of each species. The pappus bristles were then removed and the achene was re-weighed. The difference between initial achene mass (m) and achene mass less the pappus bristles represented that part of total achene mass allocated to the plume.

Statistical analysis: Formulae used throughout the study and the terms used to define variables relating to seed aerodynamics followed Greene and Johnson (1990). In addition to those variables mentioned above, the following were also utilized during the course of the study: total projected area of the plume (Ap), plume loading (m/Ap), plan area of an imaginary disk of the plumed achene (Ad), solidity (Ap/Ad), drag force (m · g where g is gravitational acceleration and equivalent to 9.81 ms-2), Reynold’s number (where the product of terminal velocity (vf) and mean pappus bristle diameter (dh) are divided by n, the kinematic viscosity of air, which is equal to 1.5X10-5 m2s-1 ), and drag coefficient (CD). Derived quantities were calculated from measured variables, and the theoretical relationship between the square root of plume loading and vf(ρC'D/2g)0.5 was illustrated through the use of Mathematica (Wolfram 2003).

For each of the variables for the respective species mean values of total achene mass, percent of total achene mass allocated to the plume, number of pappus bristles in a plume, pappus bristle length, angle of attack of the pappus bristles, terminal velocity, total projected area of the plume, plume loading, plan area of an imaginary disk of the plumed achene, solidity, drag force, Reynold’s number, and drag coefficient were compared with an analysis of variance (PROC GLM) (SAS Institute Inc. 1997). A posteriori comparisons among group means for the respective species were conducted with the Student-Newman-Keuls multiple range test (PROC GLM, SNK option) (SAS Institute Inc. 1997).

Multiple regression with terminal velocity as the dependent variable and the number of pappus bristles in a plume, pappus bristle length, angle of attack of the pappus bristles and achene mass (less the pappus) as the predictor variables, was initiated with PROC REG (SAS Institute Inc. 1997) for each species. The STB option was included as part of the MODEL statement. Because the beta coefficients resulting from this option are measured in standard deviations as opposed to the original units they may be directly compared to one another thereby indicating the relative strength of each of the predictors.

Because measures of the coefficient of variation eliminate variation in the data resulting as a consequence of the magnitude of the data, comparisons within species were made between pairs of achene characteristics (total achene mass, percent of the total achene mass allocated to the plume, number of pappus bristles in a plume, pappus bristle length, and angle of attack of the pappus bristles) per se, as well as with terminal velocity and the total projected area of the plume using the variance ratio test for coefficients of variation (Zar 1984). Because the achene characteristics have a genetic basis and are formed over a relatively short period of time we hypothesized that they would be less variable than total achene mass within a species.

Results
Descriptive statistics for total achene mass, percent of total achene mass allocated to the plume, number of pappus bristles in a plume, pappus bristle length, angle of attack of the pappus bristles, terminal velocity, total projected area of the plume, plume loading, plan area of an imaginary disk of the plumed achene, solidity, drag force, Reynold’s number, and drag coefficient are summarized for each species (Table 1). Similarities between and among species are oftentimes associated with the species’ weed status, though for some characters (angle of attack of the pappus bristles, terminal velocity, projected area of the plume, plan area of an imaginary disk of the plumed seed, and drag coefficient) the generic associations predominate.

Table 1. Summary of: (A) total achene mass (mg) [F=713.47; df=8, 891; P<0.0001]; (B) percent of total achene mass allocated to the plume [F=235.08; df=8, 891; P<0.0001]; (C) number of bristle hairs in a plume [F=231.85; df=8, 891; P<0.0001]; (D) bristle hair length (mm) [F=686.43; df=8, 891; P<0.0001]; (E) angle of attack of the pappus bristles (degrees) [F=58.58; df=8, 891; P<0.0001]; (F) terminal velocity (m·s-1) [F=149.30; df=8, 891; P<0.0001]; (G) projected area of the plume (mm2) [F=610.78; df=8, 891; P<0.0001]; (H) plume loading (mg·mm-2) [F=4.67; df=8, 891; P<0.0001]; (I) plan area of an imaginary disk of the plumed seed (mm2) [F=328.05; df=8, 891; P<0.0001]; (J) solidity [F=15.72; df=8, 891; P<0.0001]; (K) drag force (mg·m·s-2) [F=713.74; df=8, 891; P<0.0001]; (L) Reynold’s number [F=268.23; df=8, 891; P<0.0001]; (M) drag coefficient [F=204.36; df=8, 891; P<0.0001]. Results of interspecific comparisons for variables A-M are presented above between square brackets. Mean values followed by the same letter are not significantly different following a posteriori comparisons with the Student–Newman Keuls multiple range test.

 

SPECIES

A

B

C

D

E

F

G

H

I

J

K

L

M

Symphyotrichum lanceolatum

0.127fg

21.0b

37c

4.0d

48ab

0.36g

2.18d

0.08c

2.34e

1.29ab

1.24fg

0.54f

0.60d

Symphyotrichum lateriflorum

0.109g

17.3d

28d

2.7h

51a

0.53d

0.84f

0.32ab

1.04g

1.16b

1.07g

0.62e

0.43d

Symphyotrichum novae-angliae

0.318d

16.0d

37c

4.5c

49ab

0.50d

3.92c

0.09c

2.92d

1.44a

3.12d

1.16b

5.69c

Symphyotrichum pilosum

0.143f

19.1c

27d

3.3g

48ab

0.38fg

1.07ef

0.16bc

1.65f

0.79c

1.40f

0.44g

0.36d

Composite – weedy species

0.174

18.3

32

3.6

49

0.44

2.00

0.16

1.99

1.17

1.71

0.69

1.77

Doellingeria umbellate

0.775a

6.3f

47b

3.6f

21e

0.80a

3.83c

0.24abc

3.39c

1.16b

7.60a

1.34a

32.34a

Eurybia macrophylla

0.642b

17.2d

40c

5.9b

26d

0.45e

5.11b

0.14bc

8.67a

0.63c

6.29b

0.75d

11.22b

Oclemena nemoralis

0.396c

32.8a

74a

6.2a

34c

0.42f

8.39a

0.05c

8.19b

1.08b

3.88c

0.62e

9.50b

Symphyotrichum puniceum

0.380c

5.3f

27d

3.8e

45b

0.75b

1.41e

0.37a

2.40e

0.75c

3.73c

1.00c

4.94c

Symphyotrichum urophylla

0.235e

12.2e

37c

3.3g

46ab

0.57c

1.47e

0.18bc

1.73f

1.07b

2.30e

0.67e

1.94d

Composite – nonweedy species

0.485

14.7

45

4.6

34

0.60

4.04

0.20

4.88

0.94

4.76

0.88

11.98

That the theoretical relationship between the square root of plume loading [(m/Ap)0.5] and vf(ρC'D/2g)0.5 is defined by a line with a slope of 1.0 (see Greene and Johnson 1990) demonstrates that the drag coefficient of a screen adequately describes the aerodynamics of these plumed seeds (Fig. 1). Individual values in these figures that deviate from the predicted slope are typically a consequence of the angle of attack of the pappus bristles (i.e. large θ) as opposed to atypically large or small values for achene mass, the number of pappus bristles, or the length of the pappus bristles.Based on the rejection of the F-value for each of the multiple regressions (Table 2), the independent variables, number of pappus bristles, pappus bristle length, angle of attack of the pappus bristles, and achene weight (less the weight of the pappus bristles) reliably predicts the dependent variable, terminal velocity.

The respective coefficients of variation, which indicate how well the model fits the data, range from 11.3 to 24.9 for the weedy species and 14.9 to 19.6 for the non-weedy species (Table 2). Using pappus bristle length for Symphyotrichum lanceolatum as the example, the associated beta coefficient may be interpreted as follows: a one standard deviation increase in pappus bristle length corresponds to a 0.46 decrease in terminal velocity (Table 2). Independent variables that are not significant (the coefficients are not significantly different from 0) are bolded (Table 2). Generalizing, the number of pappus bristles has no effect on terminal velocity among the weedy species though does have a significant decreasing effect in the three non-weedy obligate/facultative wetland species, Doellingeria umbellata, Oclemena nemoralis, and S. puniceum. The length of the pappus bristles either has a substantial effect on decreasing terminal velocity in the weedy species, or none at all. The same is true, but to a lesser degree in the non-weedy species. The angle of attack of the pappus bristles has a significant positive effect on terminal velocity among the weedy species and also on the three non-weedy obligate/facultative wetland species, but not the remaining non-weedy species. Achene weight (less the pappus bristles) has a significant positive effect on terminal velocity across all species, regardless of whether they are weedy or not.

Table 2. Summary of respective F-values and probabilities associated with the multiple regressions, coefficients of variation (CV), and beta coefficients for each of the species following multiple regression of terminal velocity with the predictor characters number of pappus bristles (Number), pappus bristle length (Length), angle of attack of the pappus bristles (Degrees), and achene weight (less the weight of the pappus bristles) (Weight). Respective coefficients that are not significantly different (alpha=0.01) from 0 are bolded.

 

Species	F	Pr > F	CV	Number	Length	Degrees	Weight
Symphyotrichum lanceolatum	19.42	<0.0001	18.7	-0.01	-0.46	0.40	0.47
Symphyotrichum lateriflorum	10.75	<0.0001	24.9	-0.05	-0.14	0.46	0.31
Symphyotrichum novae-angliae	9.35	<0.0001	11.3	-0.09	-0.10	0.45	0.60
Symphyotrichum pilosum	20.59	<0.0001	16.8	-0.03	-0.68	0.39	0.39
Composite - weedy species	77.61	<0.0001	21.6	0.04	-0.70	0.37	0.75
Doellingeria umbellata	21.63	<0.0001	17.5	-0.42	-0.22	0.23	0.57
Eurybia macrophylla	3.79	0.0066	17.0	-0.14	-0.23	0.19	0.31
Oclemena nemoralis	20.78	<0.0001	14.9	-0.27	-0.01	0.53	0.44
Symphyotrichum puniceum	23.37	<0.0001	19.6	-0.54	-0.13	0.31	0.51
Symphyotrichum urophylla	5.20	0.0008	18.1	-0.14	-0.32	0.19	0.37
Composite - non-weedy species	155.90	<0.0001	23.5	-0.14	-0.45	0.26	0.53

Coefficients of variation for total achene mass, percent of the total achene mass allocated to the plume, number of pappus bristles, pappus bristle length, angle of attack of the pappus bristles, terminal velocity, total projected area of the plume, plume loading, plan area of an imaginary disk of the plumed seed, solidity, drag force, Reynold’s number, and drag coefficient are summarized for each of the species (Table 3).

 

 

Figure 1. Relationship between (m/Ap)0.5 and vf(ρC'D/2g)0.5 for each of the aster species. The depicted line represents a slope of 1.0. Table 3. Summary of coefficients of variation: (A) total achene mass; (B) percent of total achene mass allocated to the plume; (C) number of bristle hairs in a plume; (D) bristle hair length; (E) angle of attack of the pappus bristles; (F) terminal velocity; (G) projected area of the plume; (H) plume loading; (I) plan area of an imaginary disk of the plumed seed; (J) solidity; (K) drag force; (L) Reynold’s number; (M) drag coefficient.

 

SPECIES	A	B	C	D	E	F	G	H	I	J	K	L	M
Symphyotrichum lanceolatum	21.4	28.1	23.8	10.5	31.7	24.7	45.1	127.3	53.9	95.3	21.4	24.7	70.9
Symphyotrichum lateriflorum	23.9	47.0	28.9	14.2	30.7	29.4	51.0	458.4	67.8	89.6	23.9	29.4	87.4
Symphyotrichum novae-angliae	25.6	23.7	28.5	9.9	24.4	13.1	45.1	42.9	48.6	34.8	25.6	13.1	72.3
Symphyotrichum pilosum	21.5	39.4	19.8	10.4	25.0	22.5	35.0	73.1	47.7	83.7	21.5	22.5	52.6
Composite – weedy species	55.3	36.7	29.7	21.8	28.2	28.8	80.4	450.5	65.0	79.8	55.3	45.3	173.5
Doellingeria umbellate	21.4	43.7	26.3	7.6	81.1	23.7	33.2	79.8	28.0	29.0	21.4	23.7	59.8
Eurybia macrophylla	19.0	25.8	26.1	11.8	43.5	17.9	29.1	44.0	29.5	35.5	19.1	17.9	47.2