Ecological Applications, 12(1), 2002, pp. 40–51
q 2002 by the Ecological Society of America
BIOTIC AND ABIOTIC LIMITS TO THE SPREAD OF EXOTIC
REVEGETATION SPECIES
JENNIFER WILLIAMSON1
AND
SUSAN HARRISON
Department of Environmental Science and Policy, University of California, Davis, California 95616 USA
Abstract. Natural habitats vary in the degree to which they are invaded by exotic
species, but it is unclear whether they differ in the mechanisms underlying the spatial spread
of a given exotic species. To compare the dynamics of invasion in highly invaded nonserpentine oak woodlands and less-invaded serpentine habitats, we used an historical ‘‘experiment’’ consisting of the introduction of several nonnative species for post-mining revegetation, supplemented by a pot experiment and a factorial field experiment.
Three species showed significant declines in abundance on transects from revegetated
zones into natural habitats, indicating that these species had spread into the natural habitats
from revegetated zones. Dactylis glomerata and Trifolium hirtum were found up to 95 m
into oak woodland, 35 m into serpentine meadows, and 0–25 m into serpentine seeps and
chaparral, while Elytrigia pontica was found up to 45 m into serpentine seeps. The pot
experiment showed that this pattern of distribution for Dactylis was not caused or limited
by variation in soil properties. The field experiment showed that Dactylis invasion in both
oak woodland and serpentine meadow habitats was limited by disturbance and seed supply.
Dactylis success was negatively correlated with species richness in oak woodlands, but
positively correlated with richness in serpentine meadows, suggesting that the relationships
between diversity, invasion, and underlying habitat suitability differed between these habitats.
Our results show that in harsh serpentine soils, the spread of recently introduced exotic
species is slower than in more fertile and more invaded oak woodlands. However, disturbance and propagule addition are equally important in promoting the spread of invaders
in both environments.
Key words: Dactylis glomerata; disturbance; diversity; Elytrigia pontica; exotic plants; invasion;
oak woodlands; revegetation; serpentine habitats; spatial spread of invasive plants; Trifolium hirtum.
INTRODUCTION
tats at a 4000-ha site surrounding a gold mine. The
habitats included one relatively heavily invaded community (oak woodland), two intermediately invaded
communties (serpentine meadow and seep), and one
very little invaded community (serpentine chaparral).
By analyzing the present spatial distribution of the six
species and by conducting factorial experiments with
one of them (Dactylis glomerata) in two habitats (oak
woodlands and serpentine meadows), we asked whether these habitats differed in terms of the roles of dispersal limitation, habitat suitability, disturbance, and
the species richness of the background plant community in determining the spread of certain invaders.
Variation in invasibility among communities is often
attributed to the same suite of abiotic variables that
shape native community composition. For example, the
richness or cover of exotics may be correlated with soil
nitrogen (e.g., Huenneke et al. 1990, Knops et al. 1997,
Stohlgren et al. 1999a, b), soil carbon content (Stohlgren et al. 1999a, b), light levels (Knops et al. 1997),
or water availability (Higgins et al. 1999, Stohlgren et
al. 1999a). In general, abiotically harsh or unproductive
environments tend to be less invaded than more productive ones (e.g., Crawley 1987, Hobbs 1989, Mack
1989, Rejmanek 1989). The ultimate limits to the
spread of particular invasions may also be abiotic fac-
Many studies have analyzed the effects of biotic and
abiotic environmental factors on the process of biological invasion. However, relatively few have experimentally compared the dynamics of invasion in communities that differ in their invasibility, i.e., their susceptibility to invasion and dominance by exotics.
Therefore, we still do not know to what degree the
large-scale differences in invasibility among communities may reflect dynamic differences in the mechanisms determining the rates of spread and establishment of invaders. Scale is an inherent difficulty in trying to make such comparisons; mechanism-oriented experimental studies tend to be short-term and small in
spatial scale, while studies at the scale of multiple communities tend to be correlational and nonmechanistic.
In this study, we took advantage of a large-scale
‘‘experiment’’ initiated 16 yr ago, when six nonnative
plants were used to revegetate roadsides, pipelines, and
other disturbed areas adjacent to several natural habiManuscript received 2 October 2000; revised 19 March 2001;
accepted 2 April 2001.
1 Present address: LSA Associates, Inc., One Park Plaza,
Suite 500, Irvine, California 92614 USA.
E-mail: [email protected]
40
February 2002
SPREAD OF EXOTIC REVEGETATION SPECIES
tors, as has been shown in studies in which the species
of interest was transplanted beyond its current range
and failed to survive the conditions (Kruckeberg 1986,
Pierson and Mack 1990). Transplant experiments also
serve to test for the role of dispersal limitation, which
is indicated by successful establishment beyond the
current range (Rice 1989, Mack 1996).
Disturbance, usually defined as the removal or reduction of aboveground biomass, has been shown
abundantly to play an important role in plant invasions
(e.g., Crawley 1987, Hobbs 1989, Rejmanek 1989,
Burke and Grime 1996). Whether natural or humancaused, disturbance tends to facilitate invasion by increasing the availability of light, water, nutrients, and
‘‘safe sites’’ for germination (Grubb 1977, Grime 1979,
Hobbs and Huenneke 1992, Bergelson et al. 1993).
Less well known is whether the importance of disturbance in promoting invasion varies systematically
among communities or whether some of the differences
in invasibility among communities are caused by variation in disturbance regimes. Some evidence suggests
that disturbance may have a greater effect on invader
success in communities that are more fertile (Hobbs
and Atkins 1988).
Ecological theory has often predicted an inverse relationship between invasion success and the diversity
of the invaded community, because resources such as
space and nutrients will be more completely used in
species-rich communities (Elton 1958, MacArthur and
Wilson 1967, Pimm 1991). Support for this idea has
been found in both correlative and manipulative studies
at small spatial scales such as 1-m2 plots (e.g., Knops
et al. 1997, Tilman 1997, Stachowicz et al. 1999). In
contrast, an observational study across multiple biomes
found a positive correlation between exotic species
richness and native species richness, both of which
were positively correlated with soil fertility (Stohlgren
et al. 1999a). The diversity–invasibility relationship is
likely to be strongly scale-dependent, with negative
effects only likely to be seen within relatively homogeneous localities (Tilman 1999). However, little work
has examined variation among communities in the
strength and direction of the diversity–invasibility relationship.
Lowland grasslands and oak woodlands of California, USA, have experienced particularly severe invasion and ecosystem alteration in the past 250 yr, with
perennial bunchgrasses being replaced by exotic annual
grasses and forbs (Burcham 1957, Heady 1977, Mack
1989). One of the most important refugia for native
species in the region is serpentine soil, a nutrient-poor
substrate derived from mafic and ultramafic (iron- and
magnesium-rich) rocks that are found in zones where
faulting and uplift of the earth’s oceanic crust has occurred (Kruckeberg 1984). In our study area, shallow
rocky serpentine soils support serpentine chaparral,
while deeper alluvial and colluvial serpentine soils support meadows, and serpentine areas with late-summer
41
moisture support the ‘‘serpentine seep’’ community.
Adjacent soils derived from sedimentary rock, in contrast, support blue oak (Quercus douglassi) woodland
with a heavily invaded grassland understory. Previous
work at our study site (Harrison 1999a) and elsewhere
in California (Murphy and Ehrlich 1989, Huenneke et
al. 1990) confirms that exotic species richness and
abundance are considerably lower in serpentine than
nonserpentine habitats.
In this study, we first tested the hypothesis that the
extent of spatial spread by exotics would vary among
the three serpentine and one nonserpentine habitat just
described, by performing a field survey to take advantage of the long-term, large-scale revegetation ‘‘experiment.’’ We analyzed the current abundances of
these species as functions of habitat, distance from
point of introduction, and background richness of the
community, in addition to other environmental variables. We used a pot experiment on one species ( Dactylis glomerata) to test whether the results of the field
survey were confounded by spatial variation in soil
quality. Finally, we used a field experiment to test the
hypothesis that disturbance and propagule addition promote the success of Dactylis glomerata more strongly
in a heavily invaded habitat (oak woodland) than in a
less-invaded habitat (serpentine meadow).
METHODS
Study site
This study was conducted at the Homestake Mine/
Donald and Sylvia McLaughlin UC Natural Reserve
(Yolo, Napa, and Lake Counties, California, USA; 388
N, 1228 W; 525–800 m). The climate is Mediterranean
with 80–150 cm annual precipitation, winter temperatures that often drop below freezing, and summer temperatures that regularly exceed 408C (D’Appolonia
1982, Major 1988). Major vegetation types include
blue oak woodland, nonserpentine annual grasslands,
and serpentine chaparral and meadows (D’Appolonia
1982). Annual plants typically germinate after the first
fall rains, grow in winter and spring, and senesce in
May–July. Soils and geology have been mapped by
D’Appolonia (1982) and Fox et al. (1973). Vegetation
and plant communities have been described by Kruckeberg (1984) and Callizo (1992), among others. Plant
species names follow Hickman (1993).
Blue oak woodlands and nonserpentine grasslands
are found on loams of the Bressa, Dibble, and Yolo
soil series, formed of sedimentary parent materials
(D’Appolonia 1982). These woodlands consist of primarily Quercus douglassi in an otherwise open savanna. Most of the understory consists of exotic annual
grasses such as Avena fatua, Bromus hordeaceus, Lolium perenne, Taeniatherum caput-medusae, and others.
Native herbs include Clarkia purpurea, Vulpia microstachys, Navarretia pubescens, Trifolium wildenovii, and Micropus californicus. Open grasslands, some
42
JENNIFER WILLIAMSON AND SUSAN HARRISON
of which may be the product of 19th- and 20th-century
clearing, average 60% exotic species at the 1-m 2 scale
(Harrison 1999a).
Serpentine meadows are found mainly on soils of
the Montara and Henneke series, weathered from serpentine and related parent materials (gabbro, greenstone, peridotite) (D’Appolonia 1982). These soils contain more clay and retain more water than the soils that
support serpentine chaparral. Serpentine meadows are
dominated by forbs rather than grasses. Common native
species in serpentine meadows include Hemizonia congesta ssp. luzulifolia, Vulpia microstachys, Brodiaea
elegans, Epilobium brachycarpum, and Nassella pulchra. These meadows harbor relatively few species endemic to serpentine, but support many natives that are
now scarce on richer soils because of invasions (Huenneke et al. 1990). However, serpentine meadows have
also undergone invasion: in a previous study, 20% of
species found in 1-m2 quadrats were exotics (Harrison
1999a).
Serpentine outcrops in the study region support numerous spring-seeps and small streams that, in contrast
to the rest of the landscape, remain moist into the late
summer. These serpentine seeps support several endemics, including Senecio clevelandii, Helianthus exilis, Mimulus nudatus, Delphinium uliginosum, and Astragalus clevelandii, that are considered uncommon or
rare by the California Native Plant Society (Harrison
et al. 2000).
Serpentine chaparral, found on rocky serpentine outcrops, is dominated by the endemic or near-endemic
shrubs Quercus durata (leather oak), Arctostaphylos
viscida (whiteleaf manzanita), Cupressus macnabiana
(McNab cypress), and Ceanothus jepsonii (musk
brush); and the generalists Adenostomata fasciculatum
(chamise), Heteromeles arbutifolia (toyon), and Pinus
sabiniana (gray pine; Kruckeberg 1984, Callizo 1992,
Harrison 1997). The sparse herb layer includes numerous species endemic to serpentine soils, and almost
no exotic species (Harrison 1999b).
Revegetation history
The Homestake–McLaughlin gold mine was built beginning in 1983. Roughly 80% of the 4000-ha property
has remained undeveloped, but is traversed by roads
and pipelines. Other disturbances adjacent to natural
habitats include a reservoir, two waste rock piles, and
a tailings pond. Starting in 1983, these disturbed areas
were revegetated by adding seeds, fertilizer, and straw
mulch (Homestake Mining Company 1997). The two
main sites used in this study are a slurry pipeline ( ;7
km) and an aqueduct (1.5 km) that cross multiple habitats and were first revegetated in 1984–1986.
The revegetation species included two perennial
grasses, Dactylis glomerata and Elytrigia pontica; an
annual legume, Trifolium hirtum; and three annual
grasses: Bromus madritensis ssp. madritensis, Bromus
hordeaceus, and Lolium multiflorum. The three annual
Ecological Applications
Vol. 12, No. 1
grasses were introduced to California .100 yr ago and
are widespread components of the California annual
grassland. Dactylis glomerata (orchard grass), Trifolium hirtum (rose clover), and Elytrigia pontica (wheatgrass) are often used for revegetation of disturbed areas
and as forage crops in the United States and elsewhere
and have been the subjects of numerous range improvement studies (e.g., Anderson et al. 1999, Mitchell
et al. 1999, Acharya et al. 2000). Dactylis glomerata
is native to Europe, North Africa, and Asia. In the
Californian climate, this species germinates in winter,
grows in spring, and is dormant in summer; mature
plants are long-lived (Christie and McElroy 1995). Trifolium hirtum is a native to the Mediterranean region
and Asia Minor that is known for its ability to persist
in harsh soils (Hoveland and Evers 1995). Elytrigia
pontica is a cool-season perennial that is native to
southern Europe and parts of Asia (Asay 1995).
Three of the revegetation species (Dactylis glomerata, Trifolium hirtum, and Elytrigia pontica) are not
common in the background grassland community. In
several other grassland studies at this site, in which
.100 plots were sampled that were not near revegetated areas, these three species were never found (Harrison 1999a, Safford and Harrison 2001). The other
three revegetation species (Bromus madritensis, B. hordeaceus, and Lolium multiflorum) are common and
ubiquitous throughout California (Heady 1977) and at
our site (Harrison 1999a, Safford and Harrison 2001).
We therefore expected to see declines in abundances
of the first three species, but not the last three, with
increasing distance from revegetated areas.
Field survey
To test our hypothesis that the four habitats differed
in the extent of recent invasion by the revegetation
species, we carried out stratified random sampling
across the study site. Our assumption was that any revegetation species that showed significant negative relationships between abundance and distance from the
revegetated areas was in the process of invading from
these sites into natural habitats, while species that did
not show negative distance effects were probably widespread in the study region before the revegetation. We
chose 5–7 sites of each habitat type abutting revegetated areas throughout the reserve and ran transects
from the revegetated areas into the undisturbed habitat.
Transects were 50 m long in all serpentine habitats and
100 m in oak woodland sites because preliminary observations suggested that invasion was more extensive
in oak woodland. However, in comparisons of habitat
types, we used only the data from the first 50 m of oak
woodland transects in order to avoid unequal variances.
Transects were a minimum of 50 m apart, and 1–6
transects were located at each site. Transects were
straight and perpendicular to revegetated areas in oak
woodlands and serpentine meadows. Serpentine seep
transects approximately followed the seep courses; we
February 2002
SPREAD OF EXOTIC REVEGETATION SPECIES
only used seeps approximately perpendicular to revegetated areas. Transects in serpentine chaparral followed
the open areas between shrubs in a direction roughly
perpendicular to the revegetated area. (We did not have
to use this procedure in oak woodland because trees
are widely spaced.)
At 10-m intervals we placed a 1-m2 quadrat in which
we recorded the cover of each revegetation species on
a nine-class scale (Daubenmire 1959), the identities of
all other species present, presence or absence of soilexposing disturbances such as gopher mounds, and
shade cover on a three-class scale. We recorded the
slope and aspect of each site and classified sites into
three categories: cool (.10% slope, 3308–908 aspect),
warm (.10% slope, 1508–2708 aspect), or neutral (all
others). A total of 490 quadrats were sampled throughout the study region.
To test our hypothesis that rates of spread vary
among habitats, we analyzed cover values for the revegetation species with respect to the distance, habitat,
and the habitat 3 distance interaction. We first analyzed
cover values with respect to a large range of environmental factors that could potentially confound the effects of interest. These factors, including slope, shade,
disturbance, and grade (the slope of the habitat relative
to the revegetated area), and their interactions with
distance, did not significantly affect cover of the target
species (P . 0.15) and were not included in final models. Our preliminary analyses also showed that percentage of foliar cover, percentage of bare ground, and
frequency of gopher disturbance did not vary significantly with distance within habitat types.
The most conservative way to view the transect data
was to consider site as the experimental unit, instead
of individual transects, and to take averages at each
distance of all transects within a site. We used a repeated measures nested ANOVA (or ‘‘split-plot’’ ANOVA) for statistical analysis, with the nested site term
MS as the denominator for the F statistic for the effect
of habitat (JMP 3.2.1; SAS Institute 1997). The ‘‘within
effects’’ (distance and the habitat 3 distance interaction) were then tested against the remaining residual
error, a conservative approach that decreases the likelihood of Type I error (Potvin 1993). Distance was
treated as a categorical variable because treating it as
a covariate assumes a linear relationship, an assumption that was not fulfilled by the data. After examining
several transformations, we used the logit transformation on the midpoints of the recorded cover classes
[logit P 5 (ln(P 1 0.01))/(12(P 1 0.01))], which gave
the best distribution of residuals and variance. The results were robust to arcsine square-root transformation
and nontransformation.
To examine relationships between total quadrat species richness (not including the revegetation species of
concern) and cover of each invading species, we ran
separate correlation analyses between these variables
within each habitat type (Statview 5.0.1; SAS Institute
43
1998). We also examined correlations between quadrat
species richness and distance from revegetated areas.
Pot experiment
We used a pot experiment to determine whether the
patterns of spatial distribution observed in the field
survey were caused or influenced by variation in soil
suitability. We used Dactylis glomerata because it was
the most ubiquitous of the three species that showed
significant distance effects. We grew Dactylis in soil
collected from behind and beyond the observed invasion front (maximum extent of invasion) along transects throughout the study area. Specifically, in the
serpentine meadow and oak woodland sites, we collected soil from three zones: zone 1, where Dactylis
had been planted (‘‘revegetation zone’’); zone 2, adjacent undisturbed habitat where Dactylis had spread
and established (‘‘invaded zone’’); and zone 3, natural
habitat 10 m past the Dactylis plant farthest from the
revegetation zone (‘‘uninvaded zone’’). In serpentine
chaparral only zones 1 and 3 were used because we
observed no invasion in this habitat. We did not use
serpentine seeps since their distinctive hydrology is
difficult to reproduce in a pot experiment.
For each habitat and zone combination, soil was collected from 10 sites throughout the study region. Before
collecting each sample, we scraped away the top 10–
15 mm of plant litter and detritus and collected to a
depth of ;20 cm. Soils were air-dried, sieved (6 mm),
and placed in tubular ‘‘cone-tainers,’’ 3.8 cm diameter
3 21 cm depth (Stuewe and Sons, Corvallis, Oregon,
USA). Ten Dactylis seeds collected in the field from
revegetated areas were planted into each pot in August
1998. Pots were placed in outdoor racks beneath a shadecloth and watered at 2-d intervals with distilled water.
In the third week, all but three seedlings were removed,
and in the second month all but the center seedling was
removed. Plants were harvested in June 1999, and
height and live tiller number for each plant were recorded. Roots and shoots were separated, washed, and
oven-dried at 608C for 72 h, and weighed to obtain
total plant biomass and root/shoot ratios.
Separate analyses were done for each of the four
measurements, using split-plot ANOVA models (JMP
3.2.1; SAS Institute 1997). The nested block effect MS
was used as the error term for the habitat effect, and
the residual error was the MSdenominator for zone and the
habitat 3 zone interaction. Transformations were not
necessary for the response variables to fit model assumptions. Because the design was not completely balanced (there was no ‘‘invaded’’ zone in serpentine
chaparral habitats), two separate models were required
for each variable: one with the invaded zone excluded
and one with serpentine chaparral excluded. Where
main effects or habitat 3 zone interactions were significant, Tukey-Kramer HSD pairwise comparison tests
were used with a 5 0.05 (JMP 3.2.1; SAS Institute
Ecological Applications
Vol. 12, No. 1
JENNIFER WILLIAMSON AND SUSAN HARRISON
44
TABLE 1. Summary of community characteristics at the Homestake Mine/Donald and Sylvia McLaughlin UC Natural
Reserve, California, USA.
Background community
Focal species
1-m2 Scale
Habitat
Regional
richness
Oak woodland
Serpentine meadows
Serpentine seeps
Serpentine chaparral
255
100
79
203
Richness
5.2
5.9
5.6
3.5
6
6
6
6
0.2
0.2
0.3
0.2
Cover
of exotics (%)
55.8
37.2
24.6
6.2
6
6
6
6
2.8
2.6
3.2
1.9
Dactylis glomerata
Cover
of natives (%)
44.2
62.8
75.4
93.8
6
6
6
6
2.8
2.6
3.2
1.9
Percent
cover
Maximum
distance (m)
6
6
6
6
95
35
25
5
8.8
6.2
3.3
1.4
1.5
1.5
1.2
0.6
Notes: The table presents total species diversity from a regional species list (UCNRS 2000); other data are from the field
survey in this study (mean 6 1 SE). For comparability between habitats, data from 0–50 m transect distances were used to
calculate these values, except for the maximum spread of Dactylis and Trifolium in oak woodland (where transects were
extended to 100 m for the purpose of recording maximum distances only). Revegetated zones were not used for calculations
of species richness, cover of natives, or cover of exotics.
1997) to determine which treatments were significantly
different.
Factorial field experiment
To test whether Dactylis glomerata invasion into serpentine meadow and oak woodland was limited by habitat suitability, disturbance, and/or dispersal, we used
a field experiment with a split-plot design including
three treatments: (a) presence or absence of disturbance, (b) seeding (at two levels: 50, 100 seeds) and
not seeding of Dactylis, and (c) main plots (hereafter
‘‘plots’’) near and far from established populations of
Dactylis. This experiment was conducted in two habitats, serpentine meadow and oak woodland, with 20
plots in each habitat. Because of the physical layout of
the serpentine and nonserpentine habitats adjacent to
revegetated areas, it was not feasible to spatially intersperse the serpentine meadow and oak woodland
plots. However, the serpentine meadow plots (centered
at 388509440 N, 1228229440 W) were adjacent to the
oak woodland plots (388509500 N, 1228229140 W), and
all experimental plots were at the same approximate
elevation (700 m). Soil analyses (Harrison 1999 a) and
transect data suggest the plots were representative of
their habitat types in terms of species composition,
average cover, and soil chemistry. Habitat, i.e., community type, was included in models as a two-level
fixed factor.
Within each habitat, there were 10 plots situated
within 5 m of the original revegetated areas (‘‘near’’
treatment) and 10 plots located 5 m beyond where 90%
of Dactylis plants were observed based on transect data
(‘‘far’’ treatment): 30 m and 65 m from the revegetated
area for the serpentine meadow and the oak woodland
plots, respectively. No ‘‘far’’ plots were placed ,10 m
from any Dactylis plant, and all plots were .30 m
apart. Each of the 20 plots (;3 m2) per habitat consisted
of six experimental subplots (25 3 25 cm) separated
from each other by 75-cm buffer zones. Within each
plot, half of the subplots were randomly selected for
the disturbance treatment, which consisted of clipping
and removing aboveground biomass and cultivating the
soil to a depth of ;2–3 cm to simulate the effects of
gopher mounds, animal trails, etc. Dactylis seeds collected from revegetation areas were added to four of
the six subplots in sets of 50 and 100, with two subplots
remaining as controls to determine background seed
presence. (The original design included three seeding
treatments: 100 Dactylis, 50 Dactylis and 50 Trifolium,
and 100 Trifolium. However, Trifolium seeds failed to
germinate in this experiment due to inadequate scarification.) The experiment was set up in December 1998,
immediately after the first winter rains. From February
to June 1999, we conducted monthly counts of all Dactylis seedlings present in each subplot and marked their
locations to determine the total number of Dactylis
seedlings germinating per subplot over the course of
the growing season. We estimated the cover of litter
and photosynthetically active (green) cover in April,
May, and June 1999 and counted the number of species
in each subplot in June. Surviving seedlings were harvested in late June after the surrounding annual vegetation had died. Number and heights of tillers were
measured. Plants were oven-dried at 608C for 72 h and
weighed. Tiller heights were summed to provide a
rough estimate of total leaf area per plant, which was
divided by plant dry mass to obtain an estimate of
specific leaf area (SLA) for each plant. Specific leaf
area, defined as leaf area per unit of dry leaf mass, has
been used as a descriptor of growth strategy because
of its purported links with photosynthetic capacity, leaf
nitrogen, water-use efficiency, and relative growth rate
and leaf longevity (Bjorkman and Holmgren 1963,
Pierce et al. 1994, Garnier et al. 1997).
A nested split-plot ANOVA (JMP 3.2.1; SAS Institute 1997), using only data from noncontrol (seeded)
subplots, was used to determine the effects of habitat,
distance from revegetated area, and disturbance on germination success (total number of plants germinated/
number of seeds sown). Germination success was arcsine square-root transformed to fit ANOVA model assumptions. A random plot effect was included in the
model, nested with habitat and distance, and the MS for
this term was used to test habitat and distance effects.
SPREAD OF EXOTIC REVEGETATION SPECIES
February 2002
TABLE 1.
Extended.
Focal Species
Trifolium hirtum
Elytrigia pontica
Percent
cover
Maximum
distance (m)
6.0 6 1.6
5.7 6 1.7
0.7 6 0.7
0
95
35
5
0
Percent
cover
Maximum
distance (m)
6
6
6
6
15
15
45
15
1.0
0.9
3.9
0.5
0.6
0.6
1.4
0.4
(Propagule addition was not included because there was
zero germination in all far, unseeded control plots; this
made it impossible to define the relative strength of the
treatment effect.) The same model was used for percentage of survival (also arcsine square-root transformed), using only data from subplots with total germination .0. Percentage of survival was calculated as
the number of Dactylis plants surviving at the time of
harvest divided by the total number that germinated in
that subplot. Unlike percentage of germination, which
could have been influenced by differences in the background seed bank, percentage of survival was a useful
measurement of habitat suitability between plots near
and distant to revegetation areas. Subplot averages of
plant height, plant biomass, specific leaf area, and tiller
number were log transformed to fit ANOVA model
assumptions, except for SLA, which was logit transformed [logit SLA 5 (ln(SLA 1 0.01))/(1 2 (SLA 1
0.01))] for the best distribution of residuals and variance.
The response variables (percentage of germination,
percentage of survival, plant biomass, tiller number,
and estimated SLA) were regressed against subplot species richness to determine whether Dactylis success
was correlated with species richness (intercept simple
regression model, Statview 5.0.1; SAS Institute 1998).
This was done separately for the oak woodland and
serpentine meadow habitats. To determine whether this
relationship was affected by the disturbance treatment,
we also performed separate regressions for disturbed
and undisturbed subplots in each habitat. All response
variables were transformed as described above; species
richness was normally distributed and did not require
transformation.
RESULTS
Field survey
Oak woodlands were the most invaded habitat observed in this study, with an average of 56% exotic
species in all 1-m2 quadrats .10 m from disturbed
areas. Comparable numbers for other habitats were:
serpentine meadows, 37% exotic; serpentine seeps,
25%; serpentine chaparral, 6% (Table 1).
As we hypothesized, the abundances of the three
revegetation species that are not otherwise common in
45
grasslands in our area (Dactylis glomerata, Trifolium
hirtum, and Elytrigia pontica) showed significant decreases in cover with increasing distance from the revegetated areas (Fig. 1; ANOVA, P , 0.001). In contrast, and also in agreement with our hypothesis, the
abundances of the common and widespread annual
grasses Bromus hordeaceus, Bromus madritensis, and
Lolium multiflorum did not vary significantly with distance from revegetated areas (ANOVA, P . 0.50).
Abundances varied among habitats for Dactylis (ANOVA, P , 0.005) and Trifolium (ANOVA, P , 0.001;
Fig. 2). The habitat 3 distance interaction term was
also significant for Dactylis (ANOVA, P , 0.05) and
Trifolium (ANOVA, P , 0.001), suggesting that these
species have different rates of spread in each habitat
type, not just different abundances by habitat type (Fig.
1). The overall effect of habitat type was not significant
for Elytrigia, which was the least common of the six
species (Table 1); however, Elytrigia was more abundant in serpentine seeps than in any other habitat (Tukey-Kramer HSD, a 5 0.05; Fig. 2).
Cover of Dactylis was significantly negatively correlated with species richness in oak woodlands (r
520.271, P , 0.0001) and serpentine meadows (r 5
20.328, P , 0.001); Trifolium cover was negatively
correlated with richness in oak woodlands (r 5 20.166,
P , 0.02), and Elytrigia cover showed a negative correlation with richness in seeps (r 5 20.366, P , 0.01).
Thus, abundances of the invading species showed inverse relationships with species richness, but only in
habitats where these species were abundant. In oak
woodlands, but not the other habitats, there was also
a significant positive correlation between species richness and distance from the revegetated area (r 5 0.210,
P , 0.01).
Pot experiment
To test the hypothesis that revegetated areas, invaded
natural areas, and uninvaded natural areas differed in
the suitability of soil for Dactylis glomerata, we analyzed the pot experiment data in two ways: (1) with
only the data from revegetated and uninvaded habitats
(i.e., the invaded zone excluded) and (2) with all three
zones included and serpentine chaparral removed. In
both analyses, either a habitat effect or habitat 3 zone
effect was significant for biomass, tiller number, and
height (ANOVA, P , 0.01; Fig. 3). Tukey-Kramer
HSD tests (a 5 0.05) of pairwise mean comparisons
of all habitat 3 zone interactions showed no significant
differences between zones of soil collection for any
response variable within any habitat type, except for
tiller number, which was greater in the ‘‘invaded’’ zone
than in the ‘‘revegetation’’ zone soils from oak woodland (Fig. 3). In agreement with our expectation that
soil variation is not the cause of the spatial patterns
observed in our field survey, none of the pot experiment
results indicated lower plant performance in soil from
46
JENNIFER WILLIAMSON AND SUSAN HARRISON
Ecological Applications
Vol. 12, No. 1
uninvaded zones than in soil from revegetated or invaded zones.
There were significant main effects of habitat on
Dactylis biomass (ANOVA, P , 0.001) and tiller number (ANOVA, P , 0.01). Plants grown in oak woodland
soils had higher biomass and tiller counts than plants
grown in serpentine chaparral and serpentine meadow
soils; plant height was also significantly greater in oak
woodland soils than in serpentine chaparral soils (Fig.
4); however, these differences were weaker in the revegetated zones, leading to the significant interaction
terms. Serpentine meadow and serpentine chaparral
soils did not differ significantly from each other for
any response variable.
Factorial field experiment
Germination success of experimentally planted Dactylis seeds was significantly affected by habitat (ANOVA, P , 0.0001) and disturbance (ANOVA, P ,
0.0001; Fig. 5A). There was significantly higher germination in oak woodland than in serpentine meadow
plots, and germination was also significantly higher in
disturbed subplots than undisturbed subplots. Apparent
germination success was also higher in plots located
near previously revegetated areas than in plots far from
revegetated areas (ANOVA, P , 0.01). However, this
distance effect appeared to be a spurious result caused
by the existence of a background seed bank in the
‘‘near’’ plots; there was substantial germination in the
control (nonseeded, disturbed or undisturbed) subplots
near to revegetated areas within both habitat types,
while there was zero germination in the ‘‘far’’ control
plots in either habitat type.
Seedling survival was significantly higher in disturbed subplots than in undisturbed subplots (ANOVA,
P , 0.02), but was not significantly affected by either
habitat or distance from revegetated areas (ANOVA, P
. 0.10; Fig. 5A). Growth measurements of the surviving seedlings (dry biomass, height, and SLA), averaged by subplot, were also significantly affected by
disturbance, but not by habitat or distance. Disturbed
subplots contained seedlings with significantly higher
average biomass (ANOVA, P , 0.01), significantly
lower SLA (ANOVA, P , 0.0001), and lower average
height (ANOVA, P , 0.05; Fig. 5B–C). With all nonsignificant interactions removed from the models, the
effect of habitat was significant for biomass (ANOVA,
P , 0.02) and marginally significant for tiller number
(ANOVA, P 5 0.06). Dactylis seedlings had higher
average biomass and tiller number in oak woodland
than serpentine meadow plots.
FIG. 1. Spatial distributions of revegetation species in different habitats at the Homestake Mine/Donald and Sylvia
McLaughlin UC Natural Reserve, California, USA: (A) Dactylis glomerata; (B) Trifolium hirtum; (C) Elytrigia pontica.
Distance 5 distance from revegetation area. Habitat types
are: OW 5 oak woodland, SM 5 serpentine meadows, SS 5
←
serpentine seeps, SC 5 serpentine chaparral. Blank rectangles
represent areas sampled with a mean of zero; absence of a
bar or rectangle indicates that distance was not sampled (i.e.,
.50 m in SM, SS, and SC communities).
February 2002
SPREAD OF EXOTIC REVEGETATION SPECIES
47
, 0.05). When these data were partitioned by disturbance treatment, this relationship held for disturbed
subplots (R2 5 0.114, P , 0.05) but not for undisturbed
subplots (R2 5 0.080, P . 0.15).
DISCUSSION
Three of the revegetation species, Dactylis glomerata, Trifolium hirtum, and Elytrigia pontica, showed
patterns of declining abundance with distance from revegetated areas. There were no evident gradients in
disturbance frequency, foliar cover, bare ground or other covariates that would explain these patterns. Nor did
the pot experiment on Dactylis glomerata reveal reduced seedling growth in soils from invaded vs. nearby
uninvaded areas. In field plots in serpentine meadow
and oak woodland habitats, the survival and growth of
FIG. 2. Abundances of revegetation species in different
habitats (means 1 1 SE): (A) Dactylis glomerata; (B) Trifolium hirtum; (C) Elytrigia pontica. Means sharing a lowercase
letter are not significantly different (Tukey-Kramer hsd pairwise mean comparisons, family a 5 0.05). Habitat abbreviations and species are as in Fig. 1.
In the oak woodland plots, the biomass of experimentally planted Dactylis seedlings was significantly
negatively correlated with the species richness in a subplot (intercept regression model; R2 5 0.121, P , 0.05).
When the analyses were partitioned by disturbance
treatments, the undisturbed subplots showed negative
correlations of species richness with Dactylis seedling
biomass (R2 5 0.405, P , 0.02), tiller number (R2 5
0.481, P , 0.01), and height (R2 5 0.332, P , 0.05),
but there were no significant correlations in disturbed
subplots (P . 0.75 for all variables). However, in the
serpentine meadows there was a significant positive
relationship between percentage of survival of Dactylis
seedlings and subplot species richness (R2 5 0.101, P
FIG. 3. Effects of soil collection zone within habitat types
on Dactylis growth in the pot experiment (means 1 95% CI)
by zone of collection (‘‘revegetated,’’ ‘‘invaded,’’ and ‘‘uninvaded’’) within each habitat: (A) biomass; (B) number of
tillers; (C) plant height. Treatments sharing a lowercase letter
are not significantly different (Tukey-Kramer hsd multiple
pairwise comparisons, family a 5 0.05). Habitat abbreviations are as in Fig. 1. ‘‘N/A’’ indicates not applicable.
48
JENNIFER WILLIAMSON AND SUSAN HARRISON
Ecological Applications
Vol. 12, No. 1
serpentine habitats; Elytrigia pontica only spread appreciably in serpentine seeps.
Oak woodland soils were significantly more favorable for Dactylis germination and early growth than
serpentine soils, as shown by the pot experiment. However, the pot experiment showed Dactylis germinated
and grew equally well in serpentine meadow soils as
in serpentine chaparral soils, even though the serpentine meadow communities were considerably more invaded than either chaparral or seep communities. Factors not addressed in the pot experiment, such as water
availability, soil depth, disturbance, or plant perfor-
FIG. 4. Effects of soils from different habitats on Dactylis
growth in the pot experiment (means 1 95% CI): (A) biomass;
(B) number of tillers; (C) plant height. Means sharing a lowercase letter are not significantly different (Tukey-Kramer
HSD pairwise mean comparisons, family a 5 0.05). Habitat
abbreviations are as in Fig. 1.
experimental Dactylis seedlings were equally high just
beyond and within the current distributional limits.
Taken together, these results support our hypothesis
that these three species have spread outward from the
revegetated areas into natural habitats in the past 14–16
yr, and our results show no obvious limits to the continued spread of these species. This agrees with the
observation that these three species are not otherwise
common in our study area (Harrison 1999a, Safford
and Harrison 2001), in contrast to the three revegetation
species that did not show significant distance effects
(Bromus madritensis, B. hordeaceus, and Lolium multiflorum), which are widespread at our site and in California grasslands in general (Heady 1977). The results
also agreed with our hypothesis that the spatial extent
of invasion differed by habitat. Dactylis glomerata and
Trifolium hirtum moved further into oak woodland than
FIG. 5. Effects of disturbance on Dactylis germination and
growth in serpentine meadow (SM) and oak woodland (OW)
in the field experiment (means 1 1 SE) for each response
variable: (A) percentage survival and germination; (B) mean
subplot plant biomass and number of tillers; (C) mean subplot
plant height and estimated specific leaf area.
February 2002
SPREAD OF EXOTIC REVEGETATION SPECIES
mance at a later life stage, may explain the differential
invasion of the three serpentine habitats.
Disturbance and propagule supply are important factors limiting recruitment by Dactylis glomerata, as
shown by the field experiment. The positive effect of
disturbance on Dactylis germination, survival, and
growth agrees with the findings of many other studies
of invasion (e.g., Hobbs 1989, Burke and Grime 1996,
Kotanen 1997). The interaction term between habitat
and disturbance was not significant for any response
variable, indicating that disturbance was equally important in oak woodland and serpentine meadow plots.
Differences in disturbance regimes could explain some
of the differences between habitat types in the spread
of the revegetation species; our survey data illustrated
incidentally that serpentine meadows had the highest
frequency of gopher disturbance (21.1% of otherwise
undisturbed quadrats), compared to oak woodland
(11.8%), serpentine seeps (2.3%), and chaparral (0%).
However, it is also important to note that Dactylis recruitment was substantial in our undisturbed experimental plots, suggesting that disturbance is not absolutely essential for its spread.
Propagule supply into plots distant from established
populations was important in both habitats and suggests
that the current distribution of Dactylis is limited by
dispersal. Seeds added to plots distant from established
populations germinated and grew just as well as seeds
added to plots close to established populations. The
fact that there was zero germination in the distant and
unseeded plots suggests that lack of propagules represents a limitation just beyond the invasion front in
both habitats. Given the addition of seeds to distant
sites, germination success was significantly higher in
oak woodland than serpentine meadow plots, suggesting that though dispersal limits the rate of spread in
both communities, the greater extent of invasion of
Dactylis into oak woodland may reflect an increased
probability of establishment after propagule arrival.
Background species richness was negatively correlated with the abundances of Dactylis glomerata, Trifolium hirtum, and Elytrigia pontica in our field survey,
within the habitats where each species was most abundant. In terms of the diversity–invasibility hypothesis,
these correlational data are not very meaningful, since
the presence of mature individuals of these plants could
be the cause rather than the consequence of low species
richness in the quadrats. But our field experiment provided a more satisfactory test of the effect of background species richness on the invasion success of Dactylis, since small Dactylis seedlings were unlikely to
affect the species richness of the plots. Within oak
woodland plots, Dactylis seedling biomass was significantly negatively affected by species richness, supporting the idea that diversity reduces invasion success.
In support of this interpretation, this relationship was
only found in undisturbed subplots, where species interactions would be expected to be strongest, and in
49
these subplots the same negative relationship was also
significant for plant biomass, tillers, and height. One
plausible explanation for this result is that high species
richness may be associated with more complete utilization of above- and belowground resources.
In the experimental plots in serpentine meadows, in
contrast, the seedling survival of planted Dactylis was
positively correlated with species richness. However,
this relationship existed only in the disturbed subplots
where all cover was removed before seed addition. This
suggests that species richness was simply acting as an
indicator of habitat (e.g., soil) quality. Though serpentine soils are generally nutrient poor and have low Ca/
Mg ratios and high heavy-metal concentrations, there
is natural variation in these factors. At our study site
and in other serpentine communities, levels of N, P,
and/or Ca are positively correlated with the richness of
exotics (Armstrong and Huenneke 1992, Harrison
1999a, b), and nutrient addition can increase exotic
abundance on serpentine soils (Huenneke et al. 1990);
our results are consistent with this pattern.
We can also examine the relationship between richness and invasion, qualitatively, among communities
at a regional scale. In a species list for the study site,
the oak woodland and serpentine chaparral habitats had
255 and 203 species, respectively, while serpentine
meadows and seeps had 100 and 79 species, respectively (UCNRS 2000; Table 1). The result that the most
invaded community (oak woodland) and the least invaded community (serpentine chaparral) had similarly
high levels of total species richness demonstrates that
the negative diversity–invasibility relationship does not
hold up at the landscape scale. In fact, if chaparral is
excluded on the grounds that it shares few species with
the other three communities, there is an apparent positive relationship between richness and invasibility at
the regional level (cf. Higgins et al. 1999, Lonsdale
1999, Stohlgren et al. 1999a).
Even when scale is controlled for, comparisons of
species richness between habitat types are not particularly useful in terms of predicting community resistance to invasion. Our field survey data show that serpentine chaparral has much lower average species richness at the quadrat (1 m2) level, though it is by far the
least invaded community (Table 1). Our results are consistent with the contention that Elton’s (1958) diversity–invasibility hypothesis is not applicable across
community types where differences in abiotic factors
are likely to play a much more important role.
We did not detect factors that would slow the continued spread of Dactylis glomerata into the habitats
we studied. The relatively modest expansion of Dactylis, Trifolium hirtum, and Elytrigia pontica in a 16yr period (15–35 m in serpentine habitats, 95 m in oak
woodland) suggests these species may not be a highpriority threat to the invaded habitats in the near future.
Nonetheless, there is some cause for concern. Once
established, Dactylis is long-lived and quite competi-
50
JENNIFER WILLIAMSON AND SUSAN HARRISON
tive (Christie and McElroy 1995), and Trifolium may
alter communities through nitrogen fixation. Serpentine
meadows are important for conservation since they
comprise most of the remaining native-dominated
grasslands in California (Murphy and Ehrlich 1989),
and serpentine seeps support small and extinctionprone populations of several rare and endemic plants
(Harrison et al. 2000).
It is sometimes thought that in abiotically harsher
and less invasible environments, there is less reason to
be concerned that increases in disturbance or propagule
supply will promote invasions. However, we found that
disturbance and propagule addition had comparable effects on the establishment of Dactylis in more-invaded
oak woodlands and less-invaded serpentine meadows.
We therefore suggest that managers should be cautious
about disturbing even relatively harsh environments,
and about adding even relatively nonaggressive exotics
such as our study species, which may spread slowly
but be capable of dominating the community once they
are established.
ACKNOWLEDGMENTS
We are indebted to many people for advice and support
throughout this project. Joe Callizo of the Wantrup Wildlife
Sanctuary (Land Trust of Napa County) provided hospitality
and expert plant identification. Andrew Burt, Martha Hoopes,
and Gary Huxel provided invaluable field assistance and advice. Kevin Rice, Emilio Laca, Brian Inouye, and Neil Willits
advised us on statistical methodology and experimental design. Dean Enderlin and Lorraine Van Kekerix provided information on Homestake’s revegetation program. Cini Brown,
Vic Claasen, Jason Hoeksema, Nicole Jurjavcic, Amy Wolf,
and Jessica Utts contributed advice and valuable discussion
throughout this study. Homestake Mining Company and the
UC Natural Reserve System made it possible for us to use
the McLaughlin Reserve. Comments from Debra Peters and
two anonymous reviewers improved the manuscript. This project was supported by the California Native Plant Society; a
U.C. Davis Presidents’ Undergraduate Fellowship; the Howard Hughes Medical Institution SHARP grant; and NSF grant
DEB-9903421 to S.H.
LITERATURE CITED
Acharya, S. N., L. M. Rode, K. J. Chang, and K. Jakober.
2000. Selection for improved forage digestibility and yield
in winter-hardy orchardgrass (Dactylis glomerata L.). Canadian Journal of Plant Science 80:222.
Anderson, M. W., P. J. Cunningham, K. F. M. Reed, and A.
Byron. 1999. Perennial grasses of Mediterranean origin
offer advantages for central western Victorian sheep pasture. Australian Journal of Experimental Agriculture 39:
275–284.
Armstrong, J. K., and L. F. Huenneke. 1992. Spatial and
temporal variation in species composition in California
grasslands: the interaction of drought and substratum. Pages 213–233 in A. J. M. Baker, J. Proctor, and R. D. Reeves,
editors. The vegetation of ultramafic (serpentine) soils. Intercept, Andover, UK.
Asay, K. H. 1995. Wheatgrasses and wildryes: the perennial
Triticeae. Pages 373–394 in R. F. Barnes, D. A. Miller, and
C. J. Nelson, editors. Forages. Iowa State University Press,
Ames, Iowa, USA.
Bergelson, J., J. A. Newman, and E. M. Floresroux. 1993.
Rates of weed spread in spatially heterogeneous environments. Ecology 74:999–1011.
Ecological Applications
Vol. 12, No. 1
Bjorkman, O., and P. Holmgren. 1963. Adaptability of the
photosynthetic apparatus to light intensity in ecotypes from
exposed and shaded habitats. Physiologia Plantarum 16:
889–914.
Burcham, L. T. 1957. California range land. California Department of Natural Resources, Sacramento, California,
USA.
Burke, M. J. W., and J. P. Grime. 1996. An experimental
study of plant community invasibility. Ecology 77:776–
790.
Callizo, J. 1992. Serpentine habitats for the rare plants of
Lake Napa and Yolo counties California. Pages 35–51 in
A. J. M. Baker, J. Proctor, and R. D. Reeves, editors. The
vegetation of ultramafic (serpentine) soils. Intercept, Andover, UK.
Christie, B. R., and A. R. McElroy. 1995. Orchardgrass. Pages 325–334 in R. F. Barnes, D. A. Miller, and C. J. Nelson,
editors. Forages. Iowa State University Press, Ames, Iowa,
USA.
Crawley, M. J. 1987. What makes a community invasible?
Pages 429–454 in A. J. Gray, M. J. Crawley, and P. J.
Edwards, editors. Colonization, succession, and stability:
the 26th Symposium of the British Ecological Society held
jointly with the Linnean Society of London. Blackwell Scientific, Boston, Massachusetts, USA.
D’Appolonia. 1982. McLaughlin project: proposed gold
mine and mineral extraction facility. Environmental report.
D’Appolonia, San Francisco, California, USA.
Daubenmire, R. F. 1959. Canopy coverage method of vegetation analysis. Northwest Science 33:43–64.
Elton, C. S. 1958. The ecology of invasions by animals and
plants. Methuen, London, UK.
Fox, K. F, J. D. Sims, J. A. Barlow, and E. J. Helley. 1973.
Preliminary geologic map of eastern Sonoma and western
Napa counties, California. United States Geological Survey, Denver, Colorado, USA.
Garnier, E., P. Cordonnier, J. L. Guillerm, and L. Sonie. 1997.
Specific leaf area and leaf nitrogen concentration in annual
and perennial grass species growing in Mediterranean oldfields. Oecologia 111:490–498.
Grime, J. P. 1979. Plant strategies and vegetation processes.
John Wiley & Sons, Chichester, UK.
Grubb, P. J. 1977. The maintenance of species richness in
plant communities: the importance of the regeneration
niche. Biological Reviews 52:107–145.
Harrison, S. 1997. How natural habitat patchiness affects the
distribution of diversity in Californian serpentine chaparral.
Ecology 78:1898–1906.
Harrison, S. 1999a. Native and alien species diversity at the
local and regional scales in a grazed California grassland.
Oecologia 121:99–106.
Harrison, S. 1999b. Local and regional diversity in a patchy
landscape: native, alien, and endemic herbs on serpentine.
Ecology 80:70–80.
Harrison, S., J. Maron, and G. Huxel. 2000. Regional turnover and fluctuation in populations of five plants confined
to serpentine seeps. Conservation Biology 14:769–779.
Heady, H. F. 1977. Valley grassland. Pages 491–514 in M.
Barbour and J. Major, editors. Terrestrial vegetation of California. Wiley, New York, New York, USA.
Hickman, J. C., editor. 1993. The Jepson manual: higher
plants of California. University of California Press, Berkeley, California, USA.
Higgins, S. I., D. M. Richardson, R. M. Cowling, and T. H.
Trinder-Smith. 1999. Predicting the landscape-scale distribution of alien plants and their threat to plant diversity.
Conservation Biology 13:303–313.
Hobbs, R. J. 1989. The nature and effects of disturbance
relative to invasions. Pages 389–405 in J. A. Drake, H. A.
Mooney, F. di Castri, R. H. Groves, F. J. Kruger, M. Rej-
February 2002
SPREAD OF EXOTIC REVEGETATION SPECIES
manek, and M. Williamson, editors. Biological invasions:
a global perspective. John Wiley & Sons, New York, New
York, USA.
Hobbs, R. J., and L. Atkins. 1988. Effect of disturbance and
nutrient addition on native and introduced annuals in plant
communities in the Western Australian wheatbelt. Australian Journal of Ecology 13:171–179.
Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, and invasion: implications for conservation. Conservation Biology 6:324–337.
Homestake Mining Company. 1997. MacLaughlin Mine: annual monitoring report; July 1, 1996–June 30, 1997. University of California Natural Reserve System, University
of California, Davis, California, USA.
Hoveland, C. S., and G. W. Evers. 1995. Arrowleaf, crimson,
and other annual clovers. Pages 249–260 in R. F. Barnes,
D. A. Miller, and C. J. Nelson, editors. Forages. Iowa State
University Press, Ames, Iowa, USA.
Huenneke, L. F., S. P. Hamburg, R. Koide, H. A. Mooney,
and P. M. Vitousek. 1990. Effects of soil resources on plant
invasion and community structure in Californian serpentine
grassland. Ecology 71:478–491.
Knops, J. M. H., D. Tilman, S. Naeem, and K. M. Howe.
1997. Biodiversity and plant invasions in experimental
grassland plots. Bulletin of the Ecological Society of America 78:125.
Kotanen, P. M. 1997. Effects of experimental soil disturbance
on revegetation by natives and exotics in coastal Californian meadows. Journal of Applied Ecology 34:631–644.
Kruckeberg, A. R. 1984. California serpentines: flora, vegetation, geology, soils, and management problems. University of California Press, Berkeley, California, USA.
Kruckeberg, A. R. 1986. The birth and spread of a plant
population. American Midland Naturalist 116:403–410.
Lonsdale, W. M. 1999. Global patterns of plant invasions and
the concept of invasibility. Ecology 80:1522–1536.
MacArthur, R. H., and E. O. Wilson. 1967. The theory of
island biogeography. Princeton University Press, Princeton,
New Jersey, USA.
Mack, R. N. 1989. Temperate grasslands vulnerable to plant
invasions: characteristics and consequences. Pages 155–
179 in J. A. Drake, H. A. Mooney, F. di Castri, R. H.
Groves, F. J. Kruger, M. Rejmanek, and M. Williamson,
editors. Biological invasions: a global perspective. John
Wiley & Sons, New York, New York, USA.
Mack, R. N. 1996. Predicting the identity and fate of plant
invaders: emergent and emerging approaches. Biological
Conservation 78:107–121.
Major, J. 1988. California climate in relation to vegetation.
Pages 11–74 in M. G. Barbour and J. Major, editors. Terrestrial vegetation of California. California Native Plant
Society, Sacramento, California, USA.
Mitchell, J. P., C. D. Thomsen, W. L. Graves, and C. Shennan.
1999. Cover crops for saline soils. Journal of Agronomy
and Crop Science 183:167–178.
Murphy, D. D., and P. R. Ehrlich. 1989. Conservation biology
51
of California’s remnant native grasslands. Pages 201–211
in L. F. Huenneke and H. Mooney, editors. Grassland structure and function: California annual grassland. Kluwer Academic, Dordrecht, The Netherlands.
Pierce, L. L., S. W. Running, and J. Walker. 1994. Regionalscale relationships of leaf area index to specific leaf area
and leaf nitrogen content. Ecological Applications 4:313–
321.
Pierson, E. A., and R. N. Mack. 1990. The population biology
of Bromus tectorum in forests: distinguishing the opportunity for dispersal from environmental restriction. Oecologia (Berlin) 84:519–525.
Pimm, S. L. 1991. The balance of nature? Ecological issues
in the conservation of species and communities. University
of Chicago Press, Chicago, Illinois, USA.
Potvin, C. 1993. ANOVA: experiments in controlled environments. Pages 46–68 in S. M. Scheiner and J. Gurevitch,
editors. Design and analysis of ecological experiments.
Chapman & Hall, New York, New York, USA.
Rejmanek, M. 1989. Invasibility of plant communities. Pages
369–388 in J. A. Drake, H. A. Mooney, F. di Castri, R. H.
Groves, F. J. Kruger, M. Rejmanek, and M. Williamson,
editors. Biological invasions: a global perspective. John
Wiley & Sons, New York, New York, USA.
Rice, K. J. 1989. Competitive interactions in California annual grasslands. Pages 59–71 in L. F. Huenneke and H.
Mooney, editors. Grassland structure and function: California annual grassland. Kluwer Academic, Dordrecht, The
Netherlands.
Safford, H., and S. Harrison. 2001. Grazing and substrate
interact to affect native vs. exotic diversity in roadside
grasslands. Ecological Applications 11:1112–1122.
SAS Institute. 1997. JMP IN. Version 3.2.1. Statistical computer software. SAS Institute, Cary, North Carolina, USA.
SAS Institute. 1998. Statview. Version 5.0.1. Statistical computer software. SAS Institute, Cary, North Carolina, USA.
Stachowicz, J. J., R. B. Witlatch, and R. W. Osman. 1999.
Species diversity and invasion resistance in a marine ecosystem. Science 286:1577–1579.
Stohlgren, T. J., D. Binkley, G. W. Chong, M. A. Kalkhan,
L. D. Schell, K. A. Bull, Y. Otsuki, G. Newman, M. Bashkin, and Y. Son. 1999a. Exotic plant species invade hot
spots of native plant diversity. Ecological Monographs 69:
25–46.
Stohlgren, T. J., L. D. Schell, and B. Vanden Heuvel. 1999b.
How grazing and soil quality affect native and exotic plant
diversity in Rocky Mountain grasslands. Ecological Applications 9:45–64.
Tilman, D. 1997. Community invasibility, recruitment limitation, and grassland biodiversity. Ecology 78:81–92.
Tilman, D. 1999. The ecological consequences of changes
in biodiversity: a search for general principles. Ecology 80:
1455–1474.
UCNRS [University of California Natural Reserve System].
2000. Natural history of the McLaughlin Reserve: Napa,
Lake and Yolo counties, CA. University of California Natural Reserve System, Davis, California, USA.