Oecologia (1990) 84: 526-533 Oecologia �9 Springer-Verlag 1990

The population biology of Bromus tectorum in forests: effect of disturbance, grazing, and litter on seedling establishment and reproduction Elizabeth A. Pierson* and Richard N. M a c k

Department of Botany, Washington State University, Pullman, WA 99164, USA

Received November 22, 1989 / Accepted in revised form May 31, 1990

Summary. The effect of tree canopy, understory, herbi- vores, and litter depth on seedling establishment, surviv- al, and reproduction of the alien grass, Bromus tectorum (cheatgrass), was examined in a series of experiments in four forest habitat types in western North America. Higher recruitment, survival, and reproduction on clear- cuts, which would be expected if the overstory alone is limiting the distribution of cheatgrass in forests, were not observed. Removing the understory in an otherwise undisturbed Pinus ponderosa forest did, however, in- crease the emergence of B. tectorum, but plants in these experimentally-created openings were more vulnerable to grazing by small mammals. In contrast, removing the sparse understory in an Abies forest neither enhanced recruitment nor increased the incidence of grazing of B. tectorum seedlings. Regardless of the forest habitat, most grazed plants died before maturity; even fewer grazed plants produced seeds. Litter depth influenced both recruitment and biomass production: both the rate of germination and the size of resultant seedlings were lower on thick litter (6 cm) compared to results on thin litter (1.5 cm). In the more open Pinus ponderosa and Pseudotsuga menziesii forests, cheatgrass colonization may often occur in openings in the understory alone. Colonization in the more shady A. grandis and Thuja plieata forests is unlikely, however, unless the opening extends through both the understory and the overstory. As a result, cheatgrass is unlikely to increase in any of these forests unless the scale and incidence of distur- bance increases substantially.

Key words: Bromus tectorum - Litter - Grazing - Forest disturbance - Herbivory

Many vascular plants require local disturbance for suc- cessful establishment in mature communities (Platt 1975; Grubb 1977; Gross 1980). Such disturbance is most

* Present address and address for offprint requests: Department of Plant Pathology, University of Arizona, Tucson, AZ 85721, USA

commonly associated with a reduction in the influence of neighbors. The link between some alteration of the community and the entry of alien plants seems especially strong because the vast majority of naturalized species world-wide are weeds (sensu Baker 1974) that occupy sites where the original vegetation has been altered (Sal- isbury 1961 ; Harper 1965). For example, the naturalized flora of probably any forested region is largely confined to the clear-cuts and forest habitats converted to arable fields (Mack 1985 and references therein; Crawley 1986), suggesting that naturalized species may commonly re- quire increased irradiance before they can become natur- alized in a forest site.

Is shade alone the factor that precludes naturaliza- tions in forests to the extent seen in other, more open habitats ? Many wide-ranging alien species do have high net photosynthetic rates (Patterson 1979; Baruch et al. 1985), and their distribution in both home and new ranges includes only altered, i.e., logged, forest sites (Forcella and Harvey 1983; Geldenhuys et al. 1986). But standing dead vegetation and litter represent further components requiring experimental manipulation be- cause these materials may alternatively increase or retard plant establishment (Sydes and Grime 1981 a, b; Gold- berg and Werner 1983).

Ironically, opening the canopy may reduce an in- vader's competition while increasing its apparency and quality to herbivores (Root 1973), but such losses to grazing may be offset by the enhanced vegetative growth the plant achieves in the absence of neighbors. Grazing can also exacerbate the impact of competition by slow- ing growth and increasing mortality (Lee and Bazzaz 1980; Parker and Salzman 1985). Whether a colonizing plant, such as a potential invader, will persist in a dis- turbed site in a forest may well depend on its response to the interaction between disturbance and herbivory.

In western North America, the range of the Eurasian grass Bromus tectorum (cheatgrass) has expanded con- currently in steppe with the disturbance caused by farm- ing and livestock (Mack 1981; Mack and Thompson 1982). In contrast, cheatgrass is uncommon in the re-

527

gion 's und i s tu rbed forests, a l though it is f requent ly found in clear-cuts, pastures and b u r n e d s tands wi th in forest zones ( U p d h y a y a et al. 1986). The cur ren t distri- b u t i o n of cheatgrass in forest zones suggests tha t open- ings in the tree canopy, the forest unders to ry , the litter layer, or all three may be necessary for its successful co loniza t ion . In addi t ion , grazing may also l imit the spread of cheatgrass in to forests because herb ivory main ly by cricetids can be subs tan t ia l a long the ecotone between Ponderosa p ine forests and steppe (Mack et al. unpub l i shed data).

In this study, we examined the role of herb ivory as well as the effects of var ious levels of physical distur- bance on exper imenta l ly - in t roduced popu la t i ons of B. tectorum in four forest habi tats . Ou r objectives were to quant i fy the effects o f 1) the forest overstory, 2) herbi- vory in openings in the unders tory , and 3) litter depth on seedling es tabl ishment , survival, and subsequen t re- p roduc t i on of B. tectorum.

Materials and methods

Field sites

Field studies were conducted in the Pinus ponderosa-Symphoricar- pos albus, Pseudotsuga menziesii-Physocarpus malvaceus, Abies grandis-Paehistima myrsinites, and Thuja plicata-Pachistima myrsin- ires habitat types (sensu Daubenmire and Daubenmire 1968) in eastern Washington and northern Idaho (USA). The Pinus and Pseudotsuga sites each consisted of two stands : one stand represen- tative of the mature forest vegetation supported by the habitat type and an adjacent stand that had been logged about 20 years earlier. General features of these habitat types and the forested sites used in this study have been previously described (Pierson and Mack 1990). The field sites will be referred to hereafter as the Pinus, Pseudotsuga, Abies, and Thuja sites.

Light measurements

Measurements of Photosynthetic Photon Flux Density (PPFD) were made along a 20 m transect at 1 m intervals within the undis- turbed stand at each site using a LiCor 190S quantum sensor. Irradiance was measured at 100 cm and 2 cm above the soil surface at approximately noon on cloudless days in October. Percentage of the light attenuated by the understory was estimated from the difference in PPFD measured at 100 cm and at 2 cm, divided by PPFD at t00 cm. Mean PPFD measurements at each height, the percentage of light attenuated by the understory, and the arc- sine transformed percent data were compared among sites using a one-way analysis of variance. Multiple comparisons of treatment means were made using Protected Least Significant Difference (LSD) tests (Steel and Torrie 1980) (P<0.05).

Effect o f forest overstory

Bromus tectorum seeds for this experiment were collected in July, 1981 from the Festuca idahoensis-Symphoricarpos albus habitat type, the most likely source of seeds disseminating into the Region's forests (see Pierson 1988 for details of the collection procedure). Germination capacity was 99% and 97% before the 1982 and 1983 field seasons (where a field season is defined as September through the following June), respectively.

The experiment to assess the role of the forest overstory was conducted in the Pinus and Pseudotsuga sites and employed both

the logged (i.e., a clear-cut area of approximately 1 ha) and the adjacent undisturbed stand at each site. Despite the logging, the understory of both the clear-cut and the adjacent undisturbed for- est were typical of mature communities in these habitat types.

Four 4 x 5 dm plots were established in both forested and clear- cut stands at each site, On 15 September 1982 and 5 September 1983, 1000 cheatgrass seeds were broadcast sown into each of these plots (final sowing density=5000 seeds-m-Z). Plots were estab- lished in different locations in the stands each September.

The time of emergence as well as the location of seedlings within the plots were recorded by mapping the location of emergent seedlings, following procedures described by Mack and Pyke (1983). Only the interior half of each plot was mapped. Plots were mapped at least four times during a field season: in November before the first snowfall, in late winter after snowmelt, in late April during flowering, and at harvest in late June. All surviving plants were harvested in late June. Each plant was clipped at the soil surface, oven dried (60 ~ C for 48 h), and weighed to the nearest rag. The number of seeds produced by each plant was also recorded.

Recruitment, survivorship, and reproduction among popula- tions for each treatment were compared. The number of recruits within a plot was expressed on a per m 2 basis. Survivorship was expressed as the proportion of individuals in the total emergent population that was still alive at June harvest. The proportion of the total population to produce at least one seed (percent par- ents) was also calculated. These demographic parameters were compared across field seasons (i.e., from September through the following June), sites, and treatments using a three-way analysis of variance and Protected LSD tests. Analysis of variance and protected LSD tests were also performed on arcsine transformed percent survivorship, percent parents, and net reproduction data and on log linear transformed mean biomass data_

Effects o f disturbance and grazing

A randomized complete block 2 x 2 x 2 x 2 factorial design was used to test the importance of understory disturbance and grazing on emergence, survivorship, and fitness of cheatgrass within for- ested stands at the Pinus and Abies sites across two field seasons. Treatment plots (1 m 2) consisted of all combinations of understory disturbance, no disturbance, grazing (unprotected) and no grazing (protected) treatments. These plots were arranged in a randomized complete block design, and the experiment consisted of four blocks per site.

Disturbed plots consisted of clipping all understory vegetation (up to 3 m tall) that was either growing within or overhanging the plot and removing all undecomposed litter-procedures that ap- proximated the damage caused by the traverse of the sites by vehi- cles or livestock trampling, or both. A soil sterilant, VAPAM R (Stauffer Chemical Company), was applied to these plots in August 1985; the plots were covered with black plastic for two days, and then left uncovered for the following 19 days to allow dissipation of the sterilant before the seeds were sown. An additional plot that had been disturbed but not treated with VAPAM was included with every block in 1985. To simulate a more natural disturbance, no VAPAM was applied in 1986. No vegetation or litter was re- moved in undisturbed plots in either year.

A hardware cloth (1.3 cm gauge) exclosure (65 x 65 x 60 cm tall) was installed over plots that were to be protected from small mammalian grazers. Each exclosure was fastened along its lower perimeter to a 30 cm wide hardware cloth fence that was buried 15 cm underground. An 8 cm wide collar of aluminum flashing was also attached at right angles at the top of the exclosure to prevent the entry of small mammals; the top of the exclosure was left open. The aluminum flashing and hardware cloth were covered by white latex paint to minimize metallic leachate. The hardware cloth exclosure was removed during each mapping session.

We controlled for the effect of the exclosure itself by designat- ing unprotected plots (hereafter referred to as exclosm'e-controt

528

plots). The exclosure frames over these plots were open from 0-15 cm above the soil surface and lacked the buried hardware cloth fence, thereby allowing the entry of small herbivores. Disturbed and undisturbed plots without exclosures were also included with each block. In 1986, disturbed plots without exclosures were mapped to determine whether the exclosure-control had an effect on the incidence of herbivory in disturbed plots.

The seeds of B. teetorum for this experiment were collected in July, 1985 from the Festuca idahoensis-Symphoricarpos albus habi ta t type. Seed germination capacity was 92% and 87% before the 1985 and 1986 field seasons, respectively. Seeds were sown in the center 30 x 30 cm area within the 65 x 65 cm exclosure on 25 September 1985 and 5 September 1986. Each of these 900 c m 2

areas received 450 seeds sown in a broadcast manner, resulting in a final sowing density of 5000 seeds, m - 2.

Recruitment and survival data were collected by mapping indi- viduals in the same manner as described in the forest overstory experiment. At each census, plants damaged by herbivory were noted; the recovery or death of grazed plants was recorded at the next census. June harvest was conducted in the same manner as described above.

Recruitment and survivorship were compared among grazing treatments, disturbance treatments, field seasons, and years using a four-way analysis of variance and Protected LSD tests. The same general statistical procedure was used to compare the incidence of grazing, percent parents (proportion of the total populat ion to produce at least one seed), mean biomass, and mean seed number, within exclosure-control plots. Analysis of variance and Protected LSD tests were also performed on arcs• transformed percent survivorship, percent parents, and net reproduction data and on log linear t ransformed mean biomass data.

Glasshouse studies" the effect o f litter

Litter treatments were assigned to plots using a randomized com- plete block design. Each cedar flat (60 x 40 x 20 cm deep) constitut- ed a block and was parti t ioned into three 20 x 40 cm plots that had been filled with a standard greenhouse soil mix (Pierson ].988). A total of eight blocks was used.

Seeds used in this experiment were from the same collection used in the 1986 field experiments. Forty seeds were broadcast sown into each plot on 1 October 1986 (sowing density = 500 seeds. m-2) , and the number of emergent plants was recorded weekly for eight weeks. In June 1987, plants were harvested, dried (60 ~ C for 72 h), and weighed. The total number of recruits and the mean plant biomass per plot were calculated. Total recruitment, mean plant biomass, and log linear transformed biomass data were com- pared among litter treatments using a one-way analysis of variance and Protected LSD tests.

Results

Forest light levels

T h e h i g h e s t l eve l s o f P P F D b o t h a b o v e ( a t 100 c m ) a n d b e l o w ( a t 2 c m ) t h e f o r e s t u n d e r s t o r y w e r e r e c o r d e d in

t h e Pinus site. I n t e r m e d i a t e v a l u e s w e r e r e c o r d e d i n t h e

Table 1. Photosynthetic pho ton flux density (PPFD) (gmol 'm -2- s - * + 1 SE) along a 20 m transect in the Pinus, Pseudotsuga, Abies and Thuja sites in October 1985, October 1987, October 1985, and October 1987, respectively. The mean percentage of light attenuat- ed by the understory (difference in PPFD measured at 100 cm and at 2 cm/PPFD at 100 cm)_+ ]̀ SE is also given. For each mea- surement, values with the same superscript are not significantly different (P < 0.05). PPFD in full sunlight was 1450 Bmol-m-2"s 1 on 23 October 1985

We examined in a glasshouse study the effect of different depths of forest litter on seedling emergence, survivorship, and fitness. Site PPFD PPFD Average litter depth in the field was first estimated by measuring (100 cm) (2 cm) litter depth at 1 m intervals along two parallel 20 m transects in the undisturbed Pinus ponderosa-Symphoricarpos albus stand at Pinus 463 _+ 62" 223 _+ 29 a Smoot Hill; litter depth ranged from 1 to 15 cm and averaged Pseudotsuga 340_+].3 b 139+_22 b approximately 6-t-2 cm (_+ 1 SE). Based on this survey, three litter Abies 98 _+ 7 ~ 81 _+ 5 ~ treatments using Ponderosa pine needles were implemented: no Thuja 65 +_ 5 c 59 + 3 c litter, low-litter ( d e p t h = L 5 cm), and high-litter ( d e p t h = 6 cm).

% Attenuation

68 _+ 4" 3 6 + 5 b 15+_5 ~

5+_6 c

Table 2. Total number of individuals of B. tectorum censused from September to the following June, mean recruitment density (plants. m - z + 1 SD), estimated percentage of introduced seed that became established within plots, the total number alive at June harvest, mean-percentage of the total populat ion alive in June ( + ]. SD), the total number of parents among the June population, the mean percentage of the June survivors that were parents ( + 1 SD), and the mean size of an individual at harvest. Data are shown for Pinus and Pseudotsuga sites in which the overstory is either present (P) or absent (A) for two field seasons (1982-83 and 1983-84). Within each field season, values with the same superscript are not significantly different (P < 0.05)

Site Total Mean recruits Percent Alive Percent Total Percent Mean plant censused per m 2 establ, in June survival parents parents biomass (rag)

1982--83

Pinus(P) 887 4107 + 744 a 82 552 63 • 9" 169 31 _+4 a 7.3 _+ 0.5a Pinus(A) 764 3475 • ]`39" 70 371 48 + 4 b 30 8 + 4 c 3 .2+0.5 b

Pseudo tsuga(P) 957 4283 • ]̀ 096 a 86 604 63 +_ 4 ~ 82 14 • 3 ~ 3.3 • 0.4 b Pseudotsuga(A) 689 4050 • 2]. 3 a 8 ]. 382 56 • 3 ~b 8 ]. 23 • 4 b 3.3 + 0.3 b

1983-84

Pinus(P) 600 2218 _ 558 b 44 t 20 21 • 7 ~ 70 58 • 15 a 6.8 • 0.4 a Pinus(A) 295 922-/-214 ~ 18 90 31 +_ 8 "b 39 33 -}-6 b 4.3 +0.2 ab

Pseudotsuga(P) 1088 4193_+561" 84 458 42___19" 8]. 18 "}- 8 b 2.9_0 .5 b Pseudotsuga(A) 420 2458 _+ 214 b 49 188 4]` • 33" 86 46 + 20 ~ b 5.4 + 0.3 a b

Pseudotsuga site; lowest values were found in the Abies and Thuja sites (Table I). Higher P P F D levels at 100 cm in the Pinus and Pseudotsuga sites confirm that the tree canopy in these stands is more open than in the Abies and Thuja stands. Percentage of the light at tenuat- ed by the understory was greater in the Pinus and Pseu- dotsuga sites than in the Abies and Thuja sites. The un- derstory in the undisturbed Pinus and Pseudotsuga stands intercepts a greater percentage of photosynthet i- cally active radiation during October, a major period of seedling recruitment and growth, than the understory of the Abies and Thuja stands.

Effect of the forest overstory

The effect of the tree canopy on the recruitment of B. tectorum varied between field seasons (Table 2). During the 1982-83 field season, no difference between the treat- ments was detected in either the Pinus or the Pseudotsuga site. Recrui tment during the 1983-84 field season was, however, greater under an overstory than in the adjacent clear-cuts.

Survival of cheatgrass also varied between field sea- sons on these same sites (Table 2). For example, survival was greater in plots under a tree canopy than in clear- cuts during the 1982-83 field season. In contrast, there was no difference in the propor t ion of the populat ion alive at harvest between t reatment plots during the 1983- 84 field season.

In both sites, the absence of the tree canopy did not coincide with the product ion of larger plants and greater reproduction for B. tectorum (Table 2). In the Pinus site, generally larger plants and more reproducing individuals were found under an intact forest canopy. These plants, however, were nonetheless very small ( < 1 0 mg dry weight) and produced few seeds. In the Pseudotsuga site, there were no differences in plant size or number of reproducing individuals among the clear-cut and dis- turbed stands.

Effect of disturbance and grazing

Recruitment. A significant site x disturbance interaction (ANOVA) indicated that the role of disturbance in af- fecting recruitment was site dependent. Prior disturbance of the understory always led to a significantly (P<0 .05) higher level of recruitment on the Pinus site than oc- curred without disturbance; disturbance did not, how- ever, appear to play a significant role on the Abies site (Table 3). The role of grazing was more consistent: re- cruitment was significantly greater in plots protected f rom grazing (exclosures) compared to those left unpro- tected (exclosure-control plots) for both sites, both field seasons. Recrui tment was also higher in 1986 than in 1987.

Survivorship. The role of disturbance on survival appears to be a function of both the variability of the environ- ment f rom field season-to-field season and intrinsic fea-

529

Table 3. Mean number of recruits• SE (n=4) of B. tectorum for disturbed and undisturbed treatments and for plots protected (exclosures) and unprotected from grazing (exclosure-controls) in the Pinus and Abies sites for two field seasons. Analysis of variance (ANOVA) table shows the source, degrees of freedom, mean square error, F statistic, and level of significance

Year Pinus recruits Abies recruits

Disturbed Undisturbed Disturbed Undisturbed

1985 1986 Exclosure 289 + 17 146 • 8 226 • 19 177 • 14 Exclosurecontro1182• 142• 166• 159•

1986-87 Exclosure 319• 189• 283• 296• Exclosurecontro1294• 177• 258• 206•

ANOVA table df MS F Signif. Source

Field Season I 71958 48.56 0.0001

Site 1 280 0.19 0.6476

Block 3 3 048 3.06 0.0364

Disturbance Treatment 1 68906 51.93 0.0001

Exclosure Treatment 1 28815 21.71 0.0001

Site x Disturbance 1 27 805 20.95 0.0001 Treatment

Site x Exclosure Treatment 1 506 0.38 0.5396

Disturbance x Exclosure 1 2139 1.61 0.2101 Treatment

Field Season x Site 1 2209 1.66 0.2029

Field Seaon x Dis- 1 540 0.41 0.5262 turbance Treatment

Field Season x Exclosure 1 380 0.29 0.5948 Treatment

tures of each site (Table 4). This interaction is suggested by the 0.054 probabil i ty of a site x disturbance interac- tion and the 0.053 probabil i ty of a field season x distur- bance interaction. Although these probabilities of inter- action just exceed the 0.05 significance level, they suggest that interactions among field season, site, and distur- bance may play a role in survival. Analyses of the effect of disturbance on percent survival by field season and site suggest that percent survival in disturbed plots was significantly (P<0 .05) higher than in undisturbed plots on the Pinus site during both field seasons; disturbance played no detectable role on survival on the Abies site in either field season. Curiously, protecting plants f rom po- tential grazing had no significant effect on percent sur- vival at either site, during either field season (Table 4).

Incidence of defoliation and mortality due to grazing

The effect of disturbance on the incidence of defoliation and subsequent mortal i ty by grazing varied among field

530

Table 4. The mean percentage of the total population of B. tectorum alive at the June harvest+ 1 SE (n=4) for disturbed and undis- turbed treatments and for plots protected (with exclosures) and unprotected from grazing (with exclosure-controls) in the Pinus and Abies sites for two field seasons. Components of the analysis of variance table are as explained in Table 3

Year Pinus Abies

% Survivorship % Survivorship

Disturbed Undisturbed Disturbed Undisturbed

1985-86

Exclosure 96_+2 83_+4 86_+7 82_+7 Exclosure control 94 + 2 86 + 6 84 • 9 69 • 8

1986-87 Exclosure 77_+5 31_+9 43+11 17_+6 Exclosure control 63__12 31+-9 17+-5 22+-10

ANOVA table df MS F Signif. Source

Field Season 1 3 5 9 2 4 162.41 0.0001

Site 1 4940 22.33 0.0001

Block 3 519 2.35 0.0838

Disturbance Treatment 1 4667 21.10 0.0001

Exclosure Treatment 1 601 2.72 0.1056

Site x Disturbance Treatment 1 859 3.88 0.0543

Site x Exclosure Treatment 1 140 0.63 0.4298

Disturbance x Exclosure 1 383 1.73 0.1938 Treatment

Field Season x Site 1 98 4.46 0.0397

Field Season x Dis- 1 872 3.94 0.0525 turbance Treatment

Field Season x Exclosure 1 122 0.56 0.4595 Treatment

seasons and among sites as suggested by the significant (P<0 .05) field season x disturbance and site x distur- bance interactions (Table 5). Incidence of defoliation was significantly higher in disturbed plots than undisturbed plots in the Pinus site during both field seasons. Al- though the incidence of grazing was very low in undis- turbed plots, grazing was always fatal in these plots. In contrast, only 36-65% of the grazed individuals in the disturbed plots died before harvest. Herbivory was negligible in the Abies site during 1985 86. In the field season ending in 1987, the incidence of defoliation was comparable in disturbed and undisturbed plots, al- though mos t defoliated plants in both treatments died before harvest.

Incidence of herbivory during 1986-87 was not sig- nificantly different between the disturbed plots which lacked exclosures and the exclosure-control/disturbed plots (P<0 .05) (Data not shown). This result suggests that the open-bo t tom exclosures did not significantly affect the incidence of herbivory in disturbed plots at either site.

Table g. Percentage of total population of B. tectorum in plots unprotected from grazing that had defoliation damage _+ 1 SE. Mean percentage of these grazed individuals that eventually died ___ 1 SE is shown in parentheses. Data are shown for disturbed and undisturbed treatments in the Pinus and Abies sites during two field seasons. Components of the analysis of variance table are as explained in Table 3

Site Disturbed Undisturbed

Pinus

1985-86 33 + 6 (36 _ 6) 2__. 2 (100) 1986-87 13 • 3 (65 + 13) 1 • 1 (100)

Abies

1985-86 4+ 1 (100) 3+1 (100) 1986-87 22-t-2 (95_+ 3) 19_+ 1 (89 + 2)

ANOVA table df MS F Signif. Source

Field Season 1 72 2.02 0.1693 Site 1 i 0.03 0.8606 Block 3 62 1.74 0.1876 Disturbance Treatment 1 1205 33.82 0.0001 Site x Disturbance 1 848 23.80 0.0001 Treatment Field Season x Site 1 1375 38.56 0.0001 Field Season x Dis- 1 171 4.81 0.0391 turbance Treatment

Plant size and reproduction

The effect of disturbance on the size and reproduction of B. tectorum varied between sites (Table 6). In the Pin- us site, the percentage of surviving plants that produced seeds was greater in disturbed plots than undisturbed plots. As many as 86% of surviving individuals in dis- turbed (no VAPAM) plots produced at least one seed, while only 42% of surviving individuals in undisturbed plots produced seed. Furthermore, plants in disturbed (no VAPAM) plots were four-fold larger and produced more seeds than plants in undisturbed plots. Average seed product ion in disturbed plots was 4.8 seeds; one plant produced 36 seeds. In undisturbed plots average seed product ion was <1.0 ; the max imum number of seeds produced by a parent was 8.

Disturbance in the Abies site produced no significant effect on plant size or on the number of seeds produced by a parent (Table 6). Except for five grazed individuals that produced seeds in the disturbed plots in the Pinus site in 1986, few grazed plants produced seeds.

At both sites during 1985-86, plants in disturbed plots treated with V A P A M were significantly (P<0 .05) larger and produced more seeds than disturbed plots without V A P A M (Table 6).

The effect o f litter

The mean number of recruits differed substantially among litter treatments in the glasshouse: mean recruit-

Table 6. Total number of B. tectorum alive at June harvest, total number of parents, the mean percentage of surviving plants that produced seeds (percent parents), mean above-ground biomass (mg dry weight) of harvested plants, mean number of seeds produced per individual • 1 SE. Data are shown for disturbed and undisturbed treatment plots in the Pinus and Abies sites for 1985-86 and 1986 87. Values are also shown for disturbed/VAPAM treated sites or 1985-86. For each field season, values with different superscripts are significantly different (P< 0.05)

531

Site Total alive Total parents Percent parents Mean biomass Mean seed no.

1986-86

Pinus

Disturbed/VAPAM 274 256 93 • 7" 57.8 + 4.0 ~ 9.1 + 0.9 a Disturbed/no VAPAM 224 192 86 + 6 ~ 24.8 -}- 3.1 b 4.8 -t- 0.7 8

Undisturbed 124 52 42 • 12 b 5.4 • 0.1 c 0.7 • 0.3 c

Abies

Disturbed/VAPAM 190 56 30 • 5 b 5.6 _+ 0.3 c 0.6 • 0.2 ~ Disturbed/no VAPAM 165 37 23• b 3.2• c 0.3 • c Undisturbed 123 6 5 • 2 ~ 1.9 _+ 0.2 ~ 0.1 • 0.1 ~

1986-87

Pinus

Disturbed Undisturbed

Abies

Disturbed Undisturbed

207 159 72 + 7 a 33.2 + 2.3 a 4.2 _ 0.7 a 60 11 19• b 8.3+3.1 b 0.9 0.4 b

86 10 13 + 4 be 4.0-t- 1.8 e 0 . 2 ! 0 . 1 b

46 0 0 ~ 1.1 __ 0.1 c 0 b

ment was 23-t-2, 22-t-2 and 14_+2 for the no-litter, low- litter and high-litter t reatments, respectively. Means for recrui tment in the no-li t ter and low-litter t reatments were no t significantly different bu t were bo th significant- ly (P < 0.05) greater than mean recrui tment in the high- litter t reatment. In the low-litter t reatment , all recruits germinated on the soil below the litter. Mos t recruits in the high-litter t rea tment germinated in the litter itself, while the remainder germinated on the mineral soil.

Lit ter also delayed the t iming o f emergence. N o emergence was detected within the first week after seeds were sown. By the second week, 82% o f the total recruit- men t was completed in plots wi thou t litter c o m p a r e d to 65% and 11% o f the total emergence in low- and high-litter plots, respectively. Recru i tment ceased by the

third week in bo th the no-lit ter and low-litter t reatments, while little more than ha l f the recrui tment in the high- litter t rea tment had been completed by tha t time. N o fur ther recrui tment was detected in the high-litter plots after the sixth week (Fig. 1). In all plots more than 97% o f the plants tha t emerged survived to harvest.

Average final b iomass per individual also differed a m o n g litter treatments. Average plant weights for no- litter and low-litter t reatments (7.3 + 1.1 and 5.3__0.7, respectively) were significantly greater than mean weights for the high-litter t rea tment (3 .4+0.2) . In the highqi t ter t reatment , average weight was greater for plants tha t germinated on the under ly ing mineral soil than plants tha t germinated in the litter ( 3 . 4 + 0 . 2 and 0.9 _+ 0.1, respectively).

~0o

z so

6o

~ 4o IJA

0 i i E i i i

1 2 3 4 5 6

W E E K

Fig. 1. Mean percentage of the total recruitment of B. tectorum that occurred by each date. Census dates are the number of weeks since seeds were sown on 1 October 1986. Data are shown for no litter (-e-), low litter (-tz-), and high litter (-A-) depth treat- ments. Standard errors are <0.1%

Discussion

These results reveal tha t the colonizat ion o f the alien grass Bromus tectorum in forest habi ta t types can coin- cide with a reduct ion in the canopy ; success in coloniza- t ion varies, however, with the al terat ion o f different layers o f the canopy and is often modif ied by other forces. The somewha t surprising lack o f a significant increase in establishment in the clear-cuts on any o f the sites is perhaps the strongest evidence that the pauci ty o f the grass in the forest zones is no t dictated by the tree overs tory alone. For example, an opening or gap in the tree canopy in the low elevation Pinus and Pseu- dotsuga forests was insufficient to allow B. tectorum to persist. But cheatgrass could become established even

532

under a mature, although comparatively open, canopy of P. ponderosa, if the understory was disturbed. Remov- al of the shrubby understory in the low-elevation Pinus sites increases the PPFD available at the soil surface. In contrast, disturbance of the sparse understory in the mature Abies forest may be of little or no significance as its removal does not significantly increase light avail- ability because most shade in these forests is provided by the tree canopy. The influence of higher incident irra- diation is not restricted to just a photosynthetic response by the understory plants, including B. tectorum. It can lead to higher daily temperatures and a decrease in the duration of snow cover. Because the emergence of B. tectorum during winter is restricted to periods in which soil surfaces are free of snow and air temperatures are at or above 0 ~ C (Mack and Pyke 1984), these microcli- matic changes likely result in an increase in the number of days suitable for establishment and growth.

The greater size and fitness of cheatgrass in plots treated with VAPAM are intriguing results because re- lease from belowground competition has been often ob- served to cause a corresponding increase in the perfor- mance of forest herbs (Horn 1985 and references there- in). For example, McCune (1986) found that trenching produced a several fold increase in the cover of several pernnial herbs in a seral Abies grandis-Pseudotsuga men- ziesii stand in Montana. Results from a VAPAM treat- ment are, however, more equivocal since plant response could be a function of competitive release, green manur- ing, death of root pathogens, or all three. Nevertheless, these results give further credence to the notion that the limitation of cheatgrass in forests has multiple facets.

Our manipulations of forest canopies clarify reasons for the distribution of B. tectorum in forest zones. In the Pinus and Pseudotsuga forests, populations of cheat- grass occur in patches under a forest canopy (Hulbert 1955; personal observations). Along roadways in these two zones, cheatgrass commonly occurs on embank- ments, and it is frequently found in stands in which the understory has been disturbed by livestock. Cheat- grass does not occur in bare patches in the Thuja habitat type unless both the canopy overstory and understory have been removed (Upadhyaya et al. 1986).

As has been observed elsewhere for other understory species (Parker 1982; Parker and Salzman 1985; Ellison 1987), the gain conferred to B. tectorum seedlings that become established in canopy openings may be largely negated by the higher incidence of herbivory they en- counter. This consequence was clearest in the Pinus site in which as many as one third of the plants in the dis- turbed plots were grazed to varying degrees. Cheatgrass appears to be found by herbivores in Abies stands with the same frequency whether or not the understory was disturbed. Removing the sparse understory in these for- ests does not improve the plants' apparency as much as in understories in lower elevation forests. Further- more, small mammalian grazers are not as prevalent in these shady forests (Rickard 1960).

The consequences of grazing also varied among the different forest zones. In the Pinus site, 36-65% of grazed individuals in disturbed plots died before harvest,

while a higher proportion of the grazed plants in the Abies forest died. These contrasting results likely stem from the lower ability of seedlings in the shady Abies forest to replace biomass lost through grazing, thereby pointing to a means by which competition and herbivory operate synergistically (Harper 1977). Based on results in a series of grazing trials with B. tectorum of various ages, Pyke (1987) suggested that rapid root growth is important to cheatgrass seedlings in resisting herbivory by small mammals because much mortality stems from whole plant removal by these herbivores. If, for example, full sunlight is reduced by 90%, equivalent to the irradia- tion within mature Abies and Thuja stands, the rate of root growth would be significantly reduced (Pierson et al. 1990). In a sense, distinctions in the levels of graz- ing among these sites become moot because even if plants survived grazing they seldom produce any seeds, regardless of the site.

The depth of forest litter appears to have a significant effect on cheatgrass recruitment and biomass produc- tion. This effect may be caused by physical or chemical properties of the litter itself (Goldberg and Werner 1983; Winn 1985) or by its ability to maintain a high relative humidity that supports pathogens (Sydes and Grime 1981 b; Burdon 1987). Although these mechanisms may operate singly or in concert, recruitment and production appear to be determined largely by whether a cheatgrass caryopsis falls through the litter layer and germinates in the mineral soil. In steppe communities the suspension of cheatgrass in the litter without contact with the under- lying mineral soil can slow its germination (Young et al. 1971). Furthermore, periodic drying of germinating seeds, which is likely even on forest sites when seeds are suspended in litter, can increase mortality (Young et al. 1971). Litter's influence may be especially impor- tant in Thuja forest where the litter layer is composed of flattened branchlets of T. plieata and is often > 10 cm thick (Alban 1967).

Our results provide a partial explanation for the dis- tribution and current limitation of B. teetorum within the intact coniferous forests in western North America. The establishment of cheatgrass in the low-elevation Pin- us and Pseudotsuga forests can clearly be enhanced through disturbance that leads to the opening of the understory or through litter removal, or both. The grass is unlikely to colonize stands in the Abies and Thuja habitat types without the simultaneous creation of a gap in both the overstory and probably the understory as well. Events do occur, although infrequently from place- to-place, that would provide such disturbances in these forests (e.g., livestock trampling, the cutting of Christ- mas trees, the traverse of vehicles). The persistence of populations established under such circumstances is made tenuous at best, however, by the inevitable closure of the gap in the forest canopies. As a result, cheatgrass is unlikely to spread and persist in these forest habitats unless the scale and incidence of disturbance increases substantially.

Acknowledgments. We thank R.A. Black, R.D. Evans, S.S. Higgins, S.J. Novak, L.S. Pierson III, D.A. Pyke, K.J. Rice, and J.N. Thompson for helpful discussions throughout the course of the

533

study and L.S. Pierson III, G. Radamaker and K.J. Rice for assis- tance at various stages in the execution of the work. This work was supported by a National Science Foundation grant BSR83- 15149 to R.N. Mack and R.A. Black and the research program of Washington State University.

References Alban DH (1967) The influence of western hemlock and western

red cedar on soil properties. Soil Sci Soc Am Proc 33:453-457 Baker HG (1974) The evolution of weeds. Annu Rev Ecol Syst

5 : 1-25 Baruch Z, Ludlow MM, Davis R (1985) Photosynthetic responses

of native and introduced C4 grasses from Venezuelan savannas. Oecologia 67: 388-393

Burdon JJ (1987) Diseases and plant population biology. Cam- bridge University Press, Cambridge, 208 pp

Crawley MJ (1986) What makes a community invasible? In: Gray AJ, Crawley M J, Edwards PJ (eds) Colonization, succession, and stability. Blackwell Scientific Publications, Oxford, pp 429- 453

Daubenmire R, Daubenmire JB (1968) Forest vegetation of eastern Washington and northern Idaho. Wash Agric Exp Station Tech Bull 60. Pullman, 104 pp

Ellison AM (1987) Effects of competition, disturbance, and herbi- vory on Salicornia europaea. Ecology 68: 576-586

Forcella F, Harvey SJ (1983) Eurasian weed infestation in western Montana in relation to vegetation and disturbance. Madrono 30 : 102-109

Geldenhuys CJ, le Roux PJ, Cooper KH (1986) Alien invasions in indigenous evergreen forest. In: Macdonald IAW, Kruger F J, Ferrar AA (eds) The ecology and management of biological invasions in southern Africa. Oxford University Press, Cape Town, pp 119-131

Goldberg DE, Werner PA (1983) The effects of size of opening in vegetation and litter cover on seedling establishment of gold- enrods (Sotidago spp.). Oecologia 60:149-155

Gross KL (1980) Colonization by Verbascum thapsus (Mullein) of an old-field in Michigan: experiments on the effects of vegeta- tion. J Ecol 68:919-928

Grubb PJ (1977) The maintenance of species richness in plant com- munities: the importance of the regeneration niche. Biol Rev 52:107-145

Harper JL (1965) Establishment, aggression, and cohabitation in weedy species. In: Baker HG, Stebbins GL (eds) The genetics of colonizing species. Academic Press, New York pp 245-265

Harper JL (1977) Population biology of plants. Academic Press, London, 892 pp

Horn JC (1985) Responses of understory tree seediings to trench- ing. Am Midl Nat 114:252-258

Hulbert LC (1955) Ecological studies of Bromus tectorum and other annual bromegrasses. Ecol Monogr 25 : 181-213

Lee TD, Bazzaz FA (1980) Effects of defoliation and competition on growth and reproduction in the annual plant Abutilon theophrasti. J Ecol 68:813-821

Mack RN (1981) Invasion of Bromus tectorum L. into western North America: an ecological chronicle. Agro-Ecosyst 7:145- 165

Mack RN (1985) Invading plants: their potential contribution to population biology. In: White J (ed) Studies on plant demogra-

phy. A festschrift for John L. Harper. Academic Press, London, pp 127-142

Mack RN, Pyke DA (1983) The demography of Bromus tectorum: variation in time and space. J Ecol 71 : 69-93

Mack RN, Pyke DA (1984) The demography of Bromus tectorum: the role of microclimate, grazing and disease. J Ecol 72:731-748

Mack RN, Thompson JN (1982) Evolution in steppe with few large, hooved mammals. Am Nat 119:757-773

McCune B (1986) Root competition in a low-elevation Grand fir forest in Montana: a trenching experiment. Northwest Sci 60: 52-54

Parker MA (1982) Association with mature plants protects seed- lings from predation in an arid grassland shrub, Gutierrezia microcephala. Oecologia 53 : 276-280

Parker MA, Salzman AG (1985) Herbivore exclosure and competi- tor removal: effects on juvenile survivorship and growth in the shrub Gutierrezia microcephala. J Ecol 73:903-913

Patterson DT (1979) The effects of shading on the growth and photosynthetic capacity of itchgrass (Rottboellia exaltata). Weed Sci 27:54%553

Pierson EA (1988) Limits to the distribution of Bromus tectorum in forests of eastern Washington and northern Idaho. Ph.D. Dissertation, Washington State University, Pullman, 132 pp

Pierson EA, Mack RN (1990) The population biology of Bromus teetorum in forests : distinguishing the opportunity for dispersal from environmental restriction. Oecologia 84: 519-525

Pierson EA, Mack RN, Black RA (1990) The effect of shading on photosynthesis, growth, and regrowth following defoliation in Brornus tectorum. Oecologia 84:534~543

Platt WJ (1975) The colonization and formation of equilibrium plant species associations on badger disturbances in a tall-grass prairie. Ecol Monogr 45:285-305

Pyke DA (1987) Demographic responses of Bromus tectorum and seedlings of Agropyron spicatum to grazing by small mammals: the influence of grazing frequency and plant age. J Ecol 75:825 835

Rickard WH (1960) The distribution of small mammals in relation to the climax vegetation mosaic in eastern Washington and northern Idaho. Ecology 41 : 99-106

Root RB (1973) Organization of plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica oleracea). Ecol Monogr 43 : 95-120

Salisbury EJ (1961) Weeds and aliens. Collins, London, 384 pp Steel RGD, Torrie JH (1980) Principles and procedures of statis-

tics: a biometric approach (2nd Ed.) McGraw Hill, New York, 633 pp

Sydes C, Grime JP (1981 a) Effects of tree leaf litter on herbaceous vegetation in deciduous woodland: I. field investigations. J Ecol 69 : 237-248

Sydes C, Grime JP (1981 b) Effects of tree leaf litter on herbaceous vegetation in deciduous woodland: II. an experimental investi- gation. J Ecol 69 : 249-262

Upadhyaya MK, Turkington R, McIlvride D (1986) The biology of Canadian weeds. 75. Bromus tectorum L. Can J Plant Sci 66: 689-709

Winn AA (1985) Effects of seed size and microsite on seedling emergence of Prunella vulgaris in four habitats, J Ecol 73:831- 840

Young JA, Evans RA, Kay BL (1971) Germination of caryopses of annual grasses in simulated litter. Agron J 63:551-555