Pickart et al 1998
Yellow Bush Lupine nonnative grasses, including Vulpia bromoides, Hol-
cus lanatus (velvet grass), Bromus spp. (brome), and
Aira spp. (European hairgrass), were significantly re-
Invasion in Northern duced in those restoration plots from which litter and
duff was removed. Native species increased signifi-
cantly in vegetation types that were less influenced by
California Coastal Dunes lupine. By the third year, soil variables differed
among vegetation types but not by treatment. Bush lu-
I. Ecological Impacts pine seedling emergence was higher, however, in
plots receiving the litter and duff removal treatment.
Based on these results, we conclude that bush lupine
and Manual Restoration invasion results in both direct soil enrichment and in-
direct enrichment as a result of the associated en-
croachment of other nonnative species, particularly
Techniques grasses. Although treatment did not affect soil nutri-
ents during the period of this study, it did reduce es-
tablishment of nonnative grasses and recruitment of
Andrea J. Pickart1 new bush lupine seedlings. Restoration should there-
fore include litter and duff removal. In areas that are
Linda M. Miller2 heavily influenced by lupine and contain few native
Thomas E. Duebendorfer3 propagules, revegetation is also required.
Abstract
Introduction
We studied the ecological effects of the invasion of
coastal dunes by Lupinus arboreus (yellow bush lu-
pine), an introduced species, and used the results to
develop manual restoration techniques on the North
T he coastal dunes of Humboldt County, California,
have been extensively altered by invasive species.
The two species most responsible are Lupinus arboreus
Spit of Humboldt Bay. Vegetation and soil data were Sims (yellow bush lupine) and Ammophila arenaria (L.)
collected in five vegetation types representing points Link (European beachgrass). Since these two species
along a continuum of bush lupine’s invasive influ- were introduced in the early 1900s, they have come to
ence. We collected data on the number and size of dominate 83% of the 1077 ha of vegetated foredunes in
shrubs, vegetation cover, and soil nutrients. One set of Humboldt County (Pickart & Sawyer 1998).
plots was subjected to two restoration treatments: Yellow bush lupine is a large shrub up to 2 m in
removal of lupine shrubs only, or removal of all non- height; it is generally restricted to sandy soils from Ven-
native vegetation and removal of litter and duff. tura County, California, northward to at least Vancou-
Treatments were repeated annually for four years, and ver Island, Washington (Hitchcock & Cronquist 1973;
emerging lupine seedlings were monitored for three Horn 1993). It is native to dunes in the central and
years. Prior to treatment, ammonium and nitrate were southern portion of its range, but the demarcation be-
found to increase along the lupine continuum, but or- tween native and naturalized populations in the north
ganic matter decreased at the extreme lupine end. Yel- remains cloudy (Sholars 1993) despite Davy’s (1902) ob-
low bush lupine was not the most significant variable servation that the species was not found north of Point
affecting variation in soil nutrients. After four years, Reyes. The introduction of yellow bush lupine to the
Humboldt Bay region in 1908, and its subsequent spread
on the North Spit, were documented by Miller (1988).
Changes in species composition as the result of yel-
low bush lupine invasion in Humboldt County have
1 TheNature Conservancy, Lanphere-Christensen Dunes Pre- been inferred by vegetation classification (Parker 1974;
serve, 6800 Lanphere Road, Arcata, CA 95521, U.S.A.
2 Center for Natural Lands Management, 6800 Lanphere Road,
Duebendorfer 1990; LaBanca 1993). The native foredune
vegetation of northern California consists of low-grow-
Arcata, CA 95521, U.S.A.
3 Humboldt State University, Department of Biology, Arcata, ing, herbaceous to suffrutescent plants that form a mat-
CA 95521, U.S.A. like layer of vegetation, classified as the Sand-verbena–
beach bursage vegetation series (Sawyer & Keeler-Wolf
© 1998 Society for Ecological Restoration 1995). Known colloquially as “dune mat” (Fig. 1), this
MARCH 1998 Restoration Ecology Vol. 6 No. 1, pp. 59–68 59
Yellow Bush Lupine Invasion I
Figure 1. Dune mat vegeta-
tion type (Artemisia phase of
the Sand-verbena–beach bur-
sage vegetation series) at the
study site. Artemisia pycno-
cephala (coastal sagewort) is
the species most visible in the
photograph, accompanied by
Solidago spathulata (dune gold-
enrod) in the foreground.
vegetation type is variable in cover and frequently con- shown to create nitrogen-rich resource patches that fa-
tains large amounts of open sand. Two associations of cilitate the invasion of exotic annual weeds by creating
this vegetation type have been described (Pickart & “points of entry” (Alpert & Mooney 1996; Maron & Con-
Sawyer 1998). The Artemisia phase is distinguished by nors 1996). Once ecosystem-level changes have occurred,
the presence of Artemisia pycnocephala DC. (coastal sage- the removal of an invading species may not be suffi-
wort), whereas the Lathyrus phase is characterized by cient to return an ecosystem to its pre-invasion state
Lathyrus littoralis (Nutt.) Endl. (beach pea). The Yellow (Hobbs & Humphries 1995).
bush lupine vegetation series (Sawyer & Keeler-Wolf The purpose of our study was two-fold. First, we
1995), also referred to as “lupine scrub,” is dominated wished to document the ecosystem effects of yellow
by a near-continuous canopy of yellow bush lupine, bush lupine by measuring its contribution, relative to
with Baccharis pilularis DC. (coyote brush) locally abun- other vegetation variables, to available soil nitrogen
dant (Fig. 2). and organic matter. Our second objective was to de-
Invading plant species, in addition to their direct, velop a restoration strategy designed to reverse ob-
negative effects on native species and plant communi- served ecosystem effects, thereby increasing the chance
ties, can also alter ecosystem-level properties such as of long-term restoration success.
productivity, nutrient cycling, and soil characteristics
(Vitousek 1986; Ramakrishnan & Vitousek 1989). Changes
Methods
in productivity can occur as the result of both the intro-
duction of a new life form or the addition of a new bio- Study Site
logical process, such as nitrogen fixation (Vitousek et al.
1987; Vitousek 1990). Prior to invasion by yellow bush The study site was located on the 16-ha Samoa Dunes
lupine, northern California foredunes were both lack- Endangered Plant Protection Area owned by the U.S.
ing in shrub species and deficient in nitrogen and other Bureau of Land Management and located at the south-
macronutrients (Barbour et al. 1985). A nitrogen-fixing ern end of the North Spit of Humboldt Bay, northern
species invading a nitrogen-limited community not only California. The site contains a mosaic of vegetation
has a clear competitive advantage but may release ni- types representing a continuum of yellow bush lupine
trogen into the soil, making it available to other species. invasion from undisturbed, semi-stable dunes (Artemi-
At Bodega Bay, California, yellow bush lupine has been sia phase of dune mat) to completely stabilized, lupine-
60 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Figure 2. Lupine scrub vege-
tation type (Yellow bush lu-
pine vegetation series) at the
study site, characterized by a
near-continuous canopy of
yellow bush lupine shrubs.
dominated dunes (lupine scrub). Upland dune vegeta- mogeneity. Plots were delineated by placing wooden
tion on the site was previously classified by means of stakes around the perimeter at 1-m intervals.
TWINSPAN, a multivariate classification and ordina- In each plot we tallied the number of yellow bush lu-
tion program, to describe vegetation (Duebendorfer pine individuals by size class ( 15 cm, 15–50 cm, 50
1990; Pickart et al. 1990). We selected five vegetation cm) in order to calculate the density of lupines per
types representing points along the yellow bush lupine square meter. One soil core was collected form a ran-
gradient ranging from dune mat (lupine absent) to lu- dom location in each plot. First, litter was cleared from
pine scrub (maximum lupine cover) (Figs. 1 & 2). Inter- the soil surface, and then the top 20 cm of soil were col-
mediate vegetation types were identified in the field lected by means of an 8-cm auger. Soil was analyzed for
with a key developed for this purpose (Appendix); they ammonium by potassium chloride extraction plus steam
included mat-lupine, lupine-mat, and lupine-grass (Figs. distillation, for nitrate by potassium chloride extraction
3–5). The lupine-grass type was characterized by the and nitrate electrodes, and for organic matter content
presence of abundant, annual, nonnative grasses, in- by loss-on-ignition.
cluding Bromus hordeaceus L. (soft chess), B. diandrus A second sample was designed to permit a more ac-
Roth (ripgut grass), Vulpia bromoides (L.) S.F. Gray, Aira curate assessment of the correlation between soil and
praecox L. (European hairgrass), and Aira caryophyllea L. vegetation variables. We located 30 plots in three of the
(silver European hairgrass). five vegetation types, representing the middle and end-
points of the vegetation continuum (dune mat, lupine-
mat, and lupine scrub). Each plot was centered around
a randomly placed soil core within the vegetation type
Experimental Design
and consisted of three nested quadrats of 0.06 m2, 0.6
Two separate samples of vegetation and soil variables m2 and 1.6 m2. Cover within each nested subplot was
were collected. The first sample consisted of 42 vari- visually estimated for the following vegetation vari-
able-sized plots randomly located in the mat-lupine, ables: yellow bush lupine, native species, nonnative
lupine-mat, lupine-grass, and lupine scrub vegetation forbs, nonnative grasses, and litter and duff. The 0.6-m2
types. Plot size ranged from 15 to 80 m2 as a function of plot size was later selected for use based on minimal
observed vegetation variability. In general, lupine-grass variances. Soil cores were collected and analyzed as in
and lupine scrub plots were smaller due to greater ho- the preceding sampling design.
MARCH 1998 Restoration Ecology 61
Yellow Bush Lupine Invasion I
Figure 3. Mat-lupine vegeta-
tion type, characterized by the
presence of native dune mat
species such as Eriogonum lati-
folium (beach buckwheat),
right foreground, with rela-
tively low yellow bush lupine
influence (right background).
Restoration treatments were tested in the first sample emergence ceased, and lupine seedlings were removed as
of vegetation and soil plots described above. Prior to they emerged. Soil sampling was repeated three years
treatment, we estimated cover for the following vegeta- after plots were established.
tion classes: yellow bush lupine, native species, nonna-
tive forbs, and nonnative grasses. Two treatments were
Results
used: (1) removal of yellow bush lupine only and (2) re-
moval of all nonnative species in addition to the litter Analysis of variance (ANOVA) revealed that all three
and duff layer. In the lupine-grass and lupine scrub soil variables—nitrate, ammonium, and organic mat-
types, characterized by high lupine cover, only the sec- ter—differed significantly among vegetation types in
ond treatment was applied, based on past observations the first sample (p 0.0009, 0.0031, and 0.0001, respec-
that removal of lupine only from severely degraded ar- tively). Data for the dune mat vegetation type were ob-
eas does not result in vegetation changes. There were tained from the second sample, because dune mat was
five replicates and three controls per treatment, resulting not present in the first sample. Tukey multiple compar-
in 13 plots for dune mat and mat-lupine (two treatments isons were used to locate significant differences (Table
plus controls) and eight plots for lupine-mat, lupine- 1). Results differed for each soil variable, with organic
grass, and lupine scrub (one treatment plus controls). matter exhibiting the greatest number of significant dif-
Treatments were applied in the spring, following soil ferences among types. Vegetation types at or near the
and vegetation sampling. In lupine removal plots, yel- ends of the continuum (dune mat, mat-lupine, and lu-
low bush lupine shrubs were removed manually with pine scrub) were significantly different from one an-
hand tools. In litter and duff removal plots we also raked other for all three soil variables. In general, levels of all
the surface clean of other herbaceous weeds, litter, and three soil variables increased with the increasing influ-
duff. A buffer area around all plots was cleared of yel- ence of yellow bush lupine, although not all differences
low bush lupine, and dispersal barriers were erected were significant. One exception was organic matter,
where needed to prevent new dispersal of lupine seeds. which was lower in lupine scrub than in lupine-grass.
Treatments were repeated annually for four additional As expected, the density of large lupine shrubs in-
years. Vegetation was monitored prior to treatment and creased with the progression along the lupine-vegeta-
annually thereafter. Yellow bush lupine seedling emer- tion continuum (Table 2). The number of smaller indi-
gence was monitored monthly for three years until viduals decreased at the lupine end of the continuum,
62 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Figure 4. Lupine-mat vegeta-
tion type, with moderately
high yellow bush lupine influ-
ence, retains native species
such as Abronia latifolia (sand-
verbena) and Artemisia pycno-
cephala (coastal sagewort),
both visible in the right fore-
ground.
however, presumably because mature lupine cover Yellow bush lupine seedlings were removed from plots
suppressed seed germination and/or emergence. Cor- during recruitment monitoring, so cover values for yel-
relation analysis was used to explore the relationship low bush lupine were not analyzed. A repeated mea-
between density of yellow bush lupine individuals and sures analysis, with year as the within-subject factor,
levels of ammonium (r 0.404, p 0.008), nitrate (r was performed for each vegetation response variable to
0.304, p 0.001), and organic matter (r 0.398, p identify significant changes in cover over time. For the
0.009) in the first sample. mat-lupine and lupine-mat vegetation types, the effect
Data from the 0.6-m2 plots in the second sample were of treatment was analyzed as the between-subject fac-
used to perform multiple and step-wise regressions on tor. Results (Figs. 6–8) demonstrated a fairly continuous
each soil variable (Table 3). In the step-wise equation increase in native plants over time in the mat-lupine
for organic matter, four vegetation variables exclusive and lupine-mat types (p 0.0001), a small but statisti-
of yellow bush described virtually 100% of the variation cally significant reduction in nonnative forbs between
described in the multiple regression (r 0.905, p the first and second years in the mat-lupine type (p
0.0004). In the nitrate stepwise equation, only two vari- 0.002), and a decrease in nonnative grasses in all four
ables, litter and duff and nonnative forbs, were re- vegetation types (p 0.001). In the mat-lupine and lu-
quired to explain all but 0.4% of the variation described pine-mat types, nonnative grasses first increased and
by the regression (r 0.803, p 0.0013). Yellow bush then decreased below pre-treatment levels. The duff re-
lupine entered at the second step in the ammonium re- moval treatment reduced nonnative forbs and grasses
gression, accounting for an additional 21% (after non- in both the mat-lupine and lupine-mat types (p 0.021).
native grasses) of the total variation described by the Because vegetation cover in control plots was not
multiple regression (r 0.769, p 0.004). measured in 1992, controls were not included in the
ANOVA, but t tests were used to compare mean cover
in control plots by vegetation type for the three vegeta-
tion variables between 1988 and 1991. No significant
Effects of Treatment on Species Composition
differences were detected between years (p 0.05), con-
Changes in mean cover by year and treatment for the firming that changes detected in treated plots were the
three response variables (native species, nonnative result of treatments rather than regional vegetation
forbs, and nonnative grasses) are shown in Figures 6–8. changes over time.
MARCH 1998 Restoration Ecology 63
Yellow Bush Lupine Invasion I
Figure 5. Lupine-grass veg-
etation type, characterized
by high yellow bush lupine
influence and nonnative
grasses (center foreground).
Effects of Treatment on Soil Variables both highest in the lupine scrub type. Organic matter
We used two-way ANOVA of the third-year soil data (Ta- was unique in that lupine scrub, at the end of the lu-
ble 4) to determine whether treatment affected soil vari- pine-vegetation continuum, was characterized by lower
ables. A separate ANOVA was used for each soil variable, values than lupine-grass. As expected, the density of
with vegetation type and treatment as factors. All three large yellow bush lupine shrubs increased with pro-
soil variables showed significant differences among vege- gression along the lupine-vegetation continuum. These
tation type (p 0.05), but there was no difference between results suggested a simple linear relationship. Despite
treated and control plots, indicating that treatments did the fact that both lupine density and soil nutrients in-
not change available nitrogen or organic matter. creased along the lupine continuum, however, there
was not a high correlation between them, implying that
lupine abundance is not solely or primarily controlling
Effects of Treatment on Yellow Bush Lupine Recruitment nutrient levels.
Emergence of yellow bush lupine seedlings was ex- Multiple and step-wise regression analysis of the sec-
tremely high in the year following treatment, then de- ond data set confirmed that vegetation variables other
creased dramatically (Fig. 9). Recruitment was higher in than yellow bush lupine may be influencing soil vari-
litter and duff removal treatments than those in which ables. Ammonium, the product of symbiotic bacteria in
only lupine was removed. New emergence ceased by the root nodules of yellow bush lupine (Holton et al.
the fourth year. 1991), was the only stepwise equation entered by lupine
as a vegetation variable. Nearly all of the variation in
nitrate was accounted for by litter and duff and nonna-
tive forbs. The source of the litter and duff could not be
Discussion
determined, but it is probable that lupine contributed
Ecological Impacts of Invasion
significantly because it is large, fast-growing, and short-
lived (Davidson & Barbour 1997). Nonnative forbs and
The initial ANOVAs and multiple comparisons per- grasses accounted for 91% of the variation in organic
formed on pre-treatment data demonstrated that the ni- matter. The influence of nonnative grasses on organic
trate, ammonium, and organic matter varied with re- matter can be explained by their fibrous root systems,
spect to vegetation type. Nitrate and ammonium were which provide organic matter that is easily incorpo-
64 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Table 1. Means of soil variables (nitrate, ammonium, and Table 3. Results of multiple (total r) and step-wise regressions
organic matter) by vegetation type, and results of Tukey (p 0.005) of vegetation variables (as independent variables)
multiple comparisons showing differences among vegetation on soil variables (as dependent variables).
types.*
Variable r
Position Along
Tukey Vegetation Lupine Organic matter
Mean Grouping Type Continuum Nonnative forbs 0.581
Nonnative grasses 0.821
Nitrate (ppm) Litter and duff 0.843
5.89 A Dune mat 1 Native species 0.904
8.46 AB Mat-lupine 2 Total r 0.905
9.00 ABC Lupine-mat 3
11.25 B Lupine-grass 4 Nitrate
13.00 C Lupine scrub 5 Litter and duff 0.769
Ammonium (ppm) Nonnative forbs 0.800
4.62 A Mat-lupine 2 Total r 0.803
4.89 A Dune mat 1 Ammonium
6.31 AB Lupine-mat 3 Nonnative grasses 0.620
7.00 AB Lupine-grass 4 Bush lupine 0.767
9.25 B Lupine scrub 5 Total r 0.769
Organic matter (%)
5.88 A Dune mat 1
8.46 A Mat-lupine 2
9.00 B Lupine-mat 3 yellow bush lupine invasion on soils, changes may oc-
11.25 C Lupine scrub 5 cur indirectly by the facilitation of colonization by non-
13.00 D Lupine-grass 4
native grasses and forbs that further enrich the soil.
*Significant differences (p 0.05) are identified by different letters. The position The role of soil fertility in plant invasions has been
of each vegetation type along the lupine influence continuum is indicated along examined in several recent studies. Burke and Grime
a scale from 1, least influence, to 5, most influence.
rated into the soil at shallow depths (Hausenbuiller
1975). This would also explain why the lupine-grass
vegetation type was higher in organic matter than lu-
pine scrub.
These findings suggest that the invasion of yellow
bush lupine into a nitrogen-deficient dune environment
creates complex changes in soil and vegetation. Lupine
directly results in soil enrichment, particularly of am-
monium, during both growth and decay. A similar phe-
nomenon was described by Vitousek (1986, 1990), who
studied ecosystem changes resulting from the invasion
of Myrica faya Ait., a nitrogen-fixing tree, into young
volcanic substrates in Hawaii. Nitrogen fixation by
Myrica was found to alter both the quantity and avail-
ability of nitrogen. In addition to the direct affects of
Table 2. Mean density of nonseedling yellow bush lupine
individuals by size class and vegetation type.*
Position
Along Lupine Mean Lupines/m2 Mean Lupines/m2
Vegetation Type Continuum (SD) 15–50 cm (SD) 50 cm n
Mat-lupine 1 0.19 (0.11) 0.09 (0.04) 13
Lupine-mat 2 0.25 (0.25) 0.28 (0.12) 13
Lupine-grass 3 0.16 (0.16) 0.43 (0.20) 8 Figure 6. Changes in mean cover ( SE) of native species, non-
Lupine scrub 4 0.03 (0.05) 0.52 (0.39) 8 native forbs, and nonnative grasses in the mat-lupine vegeta-
*The position of each vegetation type along the lupine continuum is indicated tion type, lupine removal treatment (a) and litter and duff re-
along a scale from 1, least abundant, to 4, most abundant. moval treatment (b) over the four years of the study.
MARCH 1998 Restoration Ecology 65
Yellow Bush Lupine Invasion I
Figure 8. Changes in mean cover ( SE) of native species,
Figure 7. Changes in mean cover ( SE) of native species, nonnative forbs, and nonnative grasses in the lupine-grass
nonnative forbs, and nonnative grasses in the lupine-mat veg- vegetation type (litter and duff removal treatment) (a) and the
etation type, lupine removal treatment (a) and litter and duff lupine scrub vegetation type (litter and duff removal treat-
removal treatment (b) over the four years of the study. ment) (b) over the four years of the study.
(1996) demonstrated the importance of fertility changes grasses through enhanced soil productivity (Maron &
in predicting plant invasions in a nutrient-limited eco- Connors 1996). Zink et al. (1996) found that disturbance
system in the United Kingdom. In the dune system at caused by a pipeline placed through several intact
Bodega Bay, California, yellow bush lupine, a putative southern California plant communities resulted in the
native, was responsible for the invasion of nonnative proliferation of exotic annual plants, which in turn
Table 4. Mean and standard deviation of soil variables (ammonium, nitrate, and organic matter) by
vegetation type and treatment in the third year of the study (n 5 per treatment type; n 3 per control).
Vegetation Type
Mat-lupine Lupine-mat Lupine-grass Lupine Scrub
Treatment Mean (s) Mean (s) Mean (s) Mean (s)
Ammonium (ppm)
Lupine removal only 2.26 (1.72) 2.40 (0.73) — — — —
Lupine plus duff removal 1.94 (0.65) 2.86 (1.84) 2.96 (1.31) 3.62 (1.19)
Control 1.70 (0.18) 2.52 (1.07) 2.19 (0.83) 3.96 (2.41)
Nitrate (ppm)
Lupine removal only 4.07 (1.16) 4.46 (1.01) — — — —
Lupine plus duff removal 3.74 (0.58) 3.66 (0.54) 4.70 (0.83) 5.89 (1.38)
Control 3.43 (0.25) 6.38 (0.46) 3.84 (1.86) 5.90 (1.65)
Organic matter (%)
Lupine removal only 0.54 (0.13) 0.89 (0.56) — — — —
Lupine plus duff removal 0.56 (0.22) 0.63 (0.30) 1.17 (0.33) 1.05 (0.38)
Control 0.45 (0.17) 0.60 (0.10) 1.24 (0.20) 1.36 (0.13)
66 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Nonnative forbs underwent little change during the
4-year period of the study, although even the minor
changes that occurred as a result of the litter and duff
removal treatment in the mat-lupine type were statisti-
cally significant. Nonnative forb-cover values were ini-
tially low, and reductions may not have been essential
to restoration.
After three years, treated plots did not differ signifi-
cantly from controls in levels of available nitrogen and
organic matter. The lack of effect of treatment on soil
variables implies that a reduction in nitrogen and/or
organic matter is not a prerequisite for the restoration of
lupine-influenced dunes, despite the fact that soils un-
Figure 9. Yellow bush lupine seedling emergence (mean seed- derlying native vegetation are deficient in nitrogen. But
lings/m2) by vegetation per treatment type (error bar denotes there are several caveats to this conclusion. First, be-
SE). 1, mat-lupine, lupine removal; 2, mat-lupine, duff re- cause this sample represents a single slice in time, it is
moval; 3, lupine-mat, lupine removal; 4, lupine-mat, duff re- possible that soil variables initially increased after re-
moval; 5, lupine-grass; 6, lupine scrub. moval of vegetation from plots. A subsequent decline
would then be masked, and the net result would be in-
distinguishable from control plots. It is also possible
caused unstable litter and increased mineralization, fa- that soil changes are lagging behind vegetation changes
voring the persistence of weedy over indigenous spe- and may be more noticeable in the future.
cies. Monitoring of yellow bush lupine recruitment dem-
onstrated that seedling emergence is stimulated by re-
moval of the litter and duff layer. But if treatment is
Restoration
continued for at least three years this can be considered
Restoration treatments resulted in a decrease in nonna- a benefit, because the seedbank is presumably being de-
tive grasses and sometimes forbs, and/or an increase in pleted. Lupine seeds are characterized by a hard seed
native species cover over a 4-year period. Only those coat (Murdoch & Ellis 1992) and, without the distur-
vegetation types less strongly influenced by yellow bance or temperature fluctuations associated with re-
bush lupine (mat-lupine and lupine-mat) experienced moval of litter and duff, may remain in the soil and con-
significant increases in native cover. The increase in na- tinue to emerge for a longer period.
tive cover observed in mat-lupine and lupine-mat may These results have led to a restoration protocol for
have been caused by the release of nutrients associated dunes invaded by yellow bush lupine. In addition to
with dead lupine roots in addition to competitive re- the removal of lupine, other nonnatives (especially non-
lease. The lack of change in native species cover in veg- native grasses) and litter and duff should be cleared
etation types more heavily influenced by lupine was from the site, even in newly invaded areas, to discour-
most likely due to the absence of remnant native plants age recolonization of lupine and other weeds. Treat-
or nearby sources for dispersal. ment must be repeated for at least three years in order
Annual, nonnative grasses decreased in all four vege- to deplete the weedy and lupine seedbanks. In areas
tation types. It has been previously observed that where yellow bush lupine has become heavily estab-
grasses are frequently the species to respond and domi- lished, revegetation with natives will be necessary if a
nate—to the detriment of broad-leaved plants—under source of propagules is lacking.
nutrient enhancement (Hobbs & Huenneke 1992). The
decline of grasses in the more lupine-influenced vegeta-
Acknowledgments
tion types (lupine-grass and lupine scrub) was the most
dramatic. In the lupine-mat type, grasses initially in- Funding for this study was provided by Louisiana-
creased following treatment, probably due to competi- Pacific Corporation, Simpson Timber Company, The
tive release. In the litter and duff removal treatments Nature Conservancy, the U.S. Bureau of Land Manage-
grasses eventually declined, but in the lupine-only ment, and the California Department of Fish and Game
treatment grasses never returned to pre-treatment lev- Endangered Plant Program. We thank J. Sawyer for his
els. Removal of the litter and duff layer has similarly contributions to the study, M. Barbour for facilitating
been shown to be effective in preventing recolonization soils testing, and A. Buell for her review. Comments by
of sand dunes by weedy grasses on the Great Lakes Associate Editor Edith Allen and two anonymous re-
(Choi & Pavlovic 1994). viewers were very helpful.
MARCH 1998 Restoration Ecology 67
Yellow Bush Lupine Invasion I
LITERATURE CITED Maron, J. L., and P. G. Connors. 1996. A native nitrogen-fixing
shrub facilitates weed invasion. Oecologia 105:302–312.
Alpert, P., and H. A. Mooney. 1996. Resource heterogeneity by Miller, L. M. 1988. How yellow bush lupine came to Humboldt
shrubs and topography on coastal sand dunes. Vegetatio Bay. Fremontia 16:6–7.
122:83–93. Murdoch, A. J., and R. H. Ellis. 1992. Longevity, viability and dor-
Barbour, M. G., T. M. DeJong, and B. M. Pavlik. 1985. Marine mancy. Pages 193–230 in M. Fenner, editor. Seeds: the ecol-
beach and dune plant communities. Pages 296–322 in B. F. ogy of regeneration in plant communities. CAB Interna-
Chabot and H. A. Mooney, editors. Physiological ecology of tional, Wallingford, United Kingdom.
North American plant communities. Chapman and Hall, Parker, J. 1974. Coastal dune systems between Mad River and Lit-
New York. tle River, Humboldt County, California. M.A. thesis. Depart-
Burke, M. J., and J. P. Grime. 1996. An experimental study of plant ment of Biology, Humboldt State University, Arcata, Califor-
community invasibility. Ecology 77:776–790. nia.
Choi, Y. D., and N. D. Pavlovic. 1994. Comparison of fire, herbi- Pickart, A. J., and J. O. Sawyer. 1998. Ecology and restoration of
cide, and sod removal to control exotic vegetation. Natural northern California coastal dunes. California Native Plant
Areas Journal 14:217–218. Society. Sacramento, California.
Davidson, E. D., and M. G. Barbour. 1997. Germination, establish- Pickart, A. J., T. E. Duebendorfer, and L. M. Miller. 1990. An inte-
ment, and demography of coastal bush lupine (Lupinus ar- grated approach to enhancing rare plant populations
boreus) at Bodega Head, California. Ecology 58:592–600. through habitat restoration. III. Restoration of altered coastal
Davy, J. B. 1902. Stock ranges of northwestern California: notes on dunes. Pages 488–500 in H. G. Hughes and T. M. Bonnicksen,
the grasses and forage plants and range conditions. Bulletin editors. Restoration ‘89: the new management challenge. So-
no. 12 of the Bureau of Plant Industry, U.S. Department of ciety for Ecological Restoration, Madison, Wisconsin.
Agriculture. Government Printing Office, Washington, D.C. Ramakrishnan, P. S., and P. M. Vitousek. 1989. Ecosystem-level
Duebendorfer, T. E. 1990. An integrated approach to enhancing processes and the consequences of biological invasions.
rare plant populations through habitat restoration. II. Habi- Pages 281–300 in J. A. Drake, H. A. Mooney, F. Di Castri,
tat characterization through classification of dune vegeta- R. H. Groves, F. J. Kruger, M. Rejmanek, and M. Williamson,
tion. Pages 478–487 in T. M. Bonnicksen and H. G. Hughes, editors. Biological invasions: a global perspective. Wiley,
editors. Restoration ‘89: the new management challenge. So- New York.
ciety of Ecological Restoration, Madison, Wisconsin. Sawyer, J. O., and T. Keeler-Wolf. 1995. A manual of California
Hausenbuiller, R. L. 1975. Soil science, principles and practices. vegetation. California Native Plant Society, Sacramento, Cal-
William C. Brown, Dubuque, Iowa. ifornia.
Hitchcock, C. L., and A. Cronquist. 1973. Flora of the Pacific Sholars, T. 1993. Lupines. Pages 622–636 in J. C. Hickman, editor.
Northwest: an illustrated manual. University of Washington The Jepson manual: higher plants of California. University of
Press, Seattle. California Press, Berkeley.
Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, Vitousek, P. M. 1986. Biological invasions and ecosystem proper-
and invasion: implications for conservation. Conservation ties: can species make a difference? Pages 163–176 in H. A.
Biology 6:324–337. Mooney and J. A. Drake, editors. Ecology of biological inva-
Hobbs, R. J., and S. E. Humphries. 1995. An integrated approach sions of North America and Hawaii. Springer-Verlag, New
to the ecology and management of plant invasions. Conser- York.
vation Biology 9:761–770. Vitousek, P. M. 1990. Biological invasions and ecosystem pro-
Holton, B., M. G. Barbour, and S. N. Martens. 1991. Some aspects cesses: towards an integration of population biology and
of the nitrogen cycle in a California strand ecosystem. Madroño ecosystem studies. Oikos 57:7–13.
38:170–184. Vitousek, P. M., L. R. Walker, L. D. Whiteaker, D. Mueller-Dom-
Horn, E. L. 1993. Coastal wildflowers of the Pacific Northwest. bois, and P. A. Matson. 1987. Biological invasion by Myrica
Mountain Press Publishing Company, Missoula, Montana. faya alters ecosystem development in Hawaii. Science 238:
LaBanca, T. 1993. Vegetation changes at Clam Beach coastal 802–804.
dunes, Humboldt County, California. M.A. thesis. Depart- Zink, T. A., M. F. Allen, B. Heindl-Tenhunen, and E. B. Allen.
ment of Biology, Humboldt State University, Arcata, Califor- 1996. The effect of a disturbance corridor on an ecological re-
nia. serve. Restoration Ecology 3:304–310.
Appendix. Vegetation types studied.
A. Lupine absent Dune mat (Fig. 1)
A. Lupine present B
B. Total plant cover 80%; yellow bush lupine cover 25%; dune mat species
present Mat-lupine (Fig. 2)
B. Total plant cover 80%; yellow bush lupine cover 25%; dune mat species
present or absent C
C. Dune mat species (except Solidago) 25% Lupine-mat (Fig. 3)
C. Dune mat species (except Solidago) 25% D
D. Yellow bush lupine cover 75%; nonnative grasses 25% Lupine-grass (Fig. 4)
D. Yellow bush lupine cover 75% Lupine scrub (Fig. 5)
68 Restoration Ecology MARCH 1998
cus lanatus (velvet grass), Bromus spp. (brome), and
Aira spp. (European hairgrass), were significantly re-
Invasion in Northern duced in those restoration plots from which litter and
duff was removed. Native species increased signifi-
cantly in vegetation types that were less influenced by
California Coastal Dunes lupine. By the third year, soil variables differed
among vegetation types but not by treatment. Bush lu-
I. Ecological Impacts pine seedling emergence was higher, however, in
plots receiving the litter and duff removal treatment.
Based on these results, we conclude that bush lupine
and Manual Restoration invasion results in both direct soil enrichment and in-
direct enrichment as a result of the associated en-
croachment of other nonnative species, particularly
Techniques grasses. Although treatment did not affect soil nutri-
ents during the period of this study, it did reduce es-
tablishment of nonnative grasses and recruitment of
Andrea J. Pickart1 new bush lupine seedlings. Restoration should there-
fore include litter and duff removal. In areas that are
Linda M. Miller2 heavily influenced by lupine and contain few native
Thomas E. Duebendorfer3 propagules, revegetation is also required.
Abstract
Introduction
We studied the ecological effects of the invasion of
coastal dunes by Lupinus arboreus (yellow bush lu-
pine), an introduced species, and used the results to
develop manual restoration techniques on the North
T he coastal dunes of Humboldt County, California,
have been extensively altered by invasive species.
The two species most responsible are Lupinus arboreus
Spit of Humboldt Bay. Vegetation and soil data were Sims (yellow bush lupine) and Ammophila arenaria (L.)
collected in five vegetation types representing points Link (European beachgrass). Since these two species
along a continuum of bush lupine’s invasive influ- were introduced in the early 1900s, they have come to
ence. We collected data on the number and size of dominate 83% of the 1077 ha of vegetated foredunes in
shrubs, vegetation cover, and soil nutrients. One set of Humboldt County (Pickart & Sawyer 1998).
plots was subjected to two restoration treatments: Yellow bush lupine is a large shrub up to 2 m in
removal of lupine shrubs only, or removal of all non- height; it is generally restricted to sandy soils from Ven-
native vegetation and removal of litter and duff. tura County, California, northward to at least Vancou-
Treatments were repeated annually for four years, and ver Island, Washington (Hitchcock & Cronquist 1973;
emerging lupine seedlings were monitored for three Horn 1993). It is native to dunes in the central and
years. Prior to treatment, ammonium and nitrate were southern portion of its range, but the demarcation be-
found to increase along the lupine continuum, but or- tween native and naturalized populations in the north
ganic matter decreased at the extreme lupine end. Yel- remains cloudy (Sholars 1993) despite Davy’s (1902) ob-
low bush lupine was not the most significant variable servation that the species was not found north of Point
affecting variation in soil nutrients. After four years, Reyes. The introduction of yellow bush lupine to the
Humboldt Bay region in 1908, and its subsequent spread
on the North Spit, were documented by Miller (1988).
Changes in species composition as the result of yel-
low bush lupine invasion in Humboldt County have
1 TheNature Conservancy, Lanphere-Christensen Dunes Pre- been inferred by vegetation classification (Parker 1974;
serve, 6800 Lanphere Road, Arcata, CA 95521, U.S.A.
2 Center for Natural Lands Management, 6800 Lanphere Road,
Duebendorfer 1990; LaBanca 1993). The native foredune
vegetation of northern California consists of low-grow-
Arcata, CA 95521, U.S.A.
3 Humboldt State University, Department of Biology, Arcata, ing, herbaceous to suffrutescent plants that form a mat-
CA 95521, U.S.A. like layer of vegetation, classified as the Sand-verbena–
beach bursage vegetation series (Sawyer & Keeler-Wolf
© 1998 Society for Ecological Restoration 1995). Known colloquially as “dune mat” (Fig. 1), this
MARCH 1998 Restoration Ecology Vol. 6 No. 1, pp. 59–68 59
Yellow Bush Lupine Invasion I
Figure 1. Dune mat vegeta-
tion type (Artemisia phase of
the Sand-verbena–beach bur-
sage vegetation series) at the
study site. Artemisia pycno-
cephala (coastal sagewort) is
the species most visible in the
photograph, accompanied by
Solidago spathulata (dune gold-
enrod) in the foreground.
vegetation type is variable in cover and frequently con- shown to create nitrogen-rich resource patches that fa-
tains large amounts of open sand. Two associations of cilitate the invasion of exotic annual weeds by creating
this vegetation type have been described (Pickart & “points of entry” (Alpert & Mooney 1996; Maron & Con-
Sawyer 1998). The Artemisia phase is distinguished by nors 1996). Once ecosystem-level changes have occurred,
the presence of Artemisia pycnocephala DC. (coastal sage- the removal of an invading species may not be suffi-
wort), whereas the Lathyrus phase is characterized by cient to return an ecosystem to its pre-invasion state
Lathyrus littoralis (Nutt.) Endl. (beach pea). The Yellow (Hobbs & Humphries 1995).
bush lupine vegetation series (Sawyer & Keeler-Wolf The purpose of our study was two-fold. First, we
1995), also referred to as “lupine scrub,” is dominated wished to document the ecosystem effects of yellow
by a near-continuous canopy of yellow bush lupine, bush lupine by measuring its contribution, relative to
with Baccharis pilularis DC. (coyote brush) locally abun- other vegetation variables, to available soil nitrogen
dant (Fig. 2). and organic matter. Our second objective was to de-
Invading plant species, in addition to their direct, velop a restoration strategy designed to reverse ob-
negative effects on native species and plant communi- served ecosystem effects, thereby increasing the chance
ties, can also alter ecosystem-level properties such as of long-term restoration success.
productivity, nutrient cycling, and soil characteristics
(Vitousek 1986; Ramakrishnan & Vitousek 1989). Changes
Methods
in productivity can occur as the result of both the intro-
duction of a new life form or the addition of a new bio- Study Site
logical process, such as nitrogen fixation (Vitousek et al.
1987; Vitousek 1990). Prior to invasion by yellow bush The study site was located on the 16-ha Samoa Dunes
lupine, northern California foredunes were both lack- Endangered Plant Protection Area owned by the U.S.
ing in shrub species and deficient in nitrogen and other Bureau of Land Management and located at the south-
macronutrients (Barbour et al. 1985). A nitrogen-fixing ern end of the North Spit of Humboldt Bay, northern
species invading a nitrogen-limited community not only California. The site contains a mosaic of vegetation
has a clear competitive advantage but may release ni- types representing a continuum of yellow bush lupine
trogen into the soil, making it available to other species. invasion from undisturbed, semi-stable dunes (Artemi-
At Bodega Bay, California, yellow bush lupine has been sia phase of dune mat) to completely stabilized, lupine-
60 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Figure 2. Lupine scrub vege-
tation type (Yellow bush lu-
pine vegetation series) at the
study site, characterized by a
near-continuous canopy of
yellow bush lupine shrubs.
dominated dunes (lupine scrub). Upland dune vegeta- mogeneity. Plots were delineated by placing wooden
tion on the site was previously classified by means of stakes around the perimeter at 1-m intervals.
TWINSPAN, a multivariate classification and ordina- In each plot we tallied the number of yellow bush lu-
tion program, to describe vegetation (Duebendorfer pine individuals by size class ( 15 cm, 15–50 cm, 50
1990; Pickart et al. 1990). We selected five vegetation cm) in order to calculate the density of lupines per
types representing points along the yellow bush lupine square meter. One soil core was collected form a ran-
gradient ranging from dune mat (lupine absent) to lu- dom location in each plot. First, litter was cleared from
pine scrub (maximum lupine cover) (Figs. 1 & 2). Inter- the soil surface, and then the top 20 cm of soil were col-
mediate vegetation types were identified in the field lected by means of an 8-cm auger. Soil was analyzed for
with a key developed for this purpose (Appendix); they ammonium by potassium chloride extraction plus steam
included mat-lupine, lupine-mat, and lupine-grass (Figs. distillation, for nitrate by potassium chloride extraction
3–5). The lupine-grass type was characterized by the and nitrate electrodes, and for organic matter content
presence of abundant, annual, nonnative grasses, in- by loss-on-ignition.
cluding Bromus hordeaceus L. (soft chess), B. diandrus A second sample was designed to permit a more ac-
Roth (ripgut grass), Vulpia bromoides (L.) S.F. Gray, Aira curate assessment of the correlation between soil and
praecox L. (European hairgrass), and Aira caryophyllea L. vegetation variables. We located 30 plots in three of the
(silver European hairgrass). five vegetation types, representing the middle and end-
points of the vegetation continuum (dune mat, lupine-
mat, and lupine scrub). Each plot was centered around
a randomly placed soil core within the vegetation type
Experimental Design
and consisted of three nested quadrats of 0.06 m2, 0.6
Two separate samples of vegetation and soil variables m2 and 1.6 m2. Cover within each nested subplot was
were collected. The first sample consisted of 42 vari- visually estimated for the following vegetation vari-
able-sized plots randomly located in the mat-lupine, ables: yellow bush lupine, native species, nonnative
lupine-mat, lupine-grass, and lupine scrub vegetation forbs, nonnative grasses, and litter and duff. The 0.6-m2
types. Plot size ranged from 15 to 80 m2 as a function of plot size was later selected for use based on minimal
observed vegetation variability. In general, lupine-grass variances. Soil cores were collected and analyzed as in
and lupine scrub plots were smaller due to greater ho- the preceding sampling design.
MARCH 1998 Restoration Ecology 61
Yellow Bush Lupine Invasion I
Figure 3. Mat-lupine vegeta-
tion type, characterized by the
presence of native dune mat
species such as Eriogonum lati-
folium (beach buckwheat),
right foreground, with rela-
tively low yellow bush lupine
influence (right background).
Restoration treatments were tested in the first sample emergence ceased, and lupine seedlings were removed as
of vegetation and soil plots described above. Prior to they emerged. Soil sampling was repeated three years
treatment, we estimated cover for the following vegeta- after plots were established.
tion classes: yellow bush lupine, native species, nonna-
tive forbs, and nonnative grasses. Two treatments were
Results
used: (1) removal of yellow bush lupine only and (2) re-
moval of all nonnative species in addition to the litter Analysis of variance (ANOVA) revealed that all three
and duff layer. In the lupine-grass and lupine scrub soil variables—nitrate, ammonium, and organic mat-
types, characterized by high lupine cover, only the sec- ter—differed significantly among vegetation types in
ond treatment was applied, based on past observations the first sample (p 0.0009, 0.0031, and 0.0001, respec-
that removal of lupine only from severely degraded ar- tively). Data for the dune mat vegetation type were ob-
eas does not result in vegetation changes. There were tained from the second sample, because dune mat was
five replicates and three controls per treatment, resulting not present in the first sample. Tukey multiple compar-
in 13 plots for dune mat and mat-lupine (two treatments isons were used to locate significant differences (Table
plus controls) and eight plots for lupine-mat, lupine- 1). Results differed for each soil variable, with organic
grass, and lupine scrub (one treatment plus controls). matter exhibiting the greatest number of significant dif-
Treatments were applied in the spring, following soil ferences among types. Vegetation types at or near the
and vegetation sampling. In lupine removal plots, yel- ends of the continuum (dune mat, mat-lupine, and lu-
low bush lupine shrubs were removed manually with pine scrub) were significantly different from one an-
hand tools. In litter and duff removal plots we also raked other for all three soil variables. In general, levels of all
the surface clean of other herbaceous weeds, litter, and three soil variables increased with the increasing influ-
duff. A buffer area around all plots was cleared of yel- ence of yellow bush lupine, although not all differences
low bush lupine, and dispersal barriers were erected were significant. One exception was organic matter,
where needed to prevent new dispersal of lupine seeds. which was lower in lupine scrub than in lupine-grass.
Treatments were repeated annually for four additional As expected, the density of large lupine shrubs in-
years. Vegetation was monitored prior to treatment and creased with the progression along the lupine-vegeta-
annually thereafter. Yellow bush lupine seedling emer- tion continuum (Table 2). The number of smaller indi-
gence was monitored monthly for three years until viduals decreased at the lupine end of the continuum,
62 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Figure 4. Lupine-mat vegeta-
tion type, with moderately
high yellow bush lupine influ-
ence, retains native species
such as Abronia latifolia (sand-
verbena) and Artemisia pycno-
cephala (coastal sagewort),
both visible in the right fore-
ground.
however, presumably because mature lupine cover Yellow bush lupine seedlings were removed from plots
suppressed seed germination and/or emergence. Cor- during recruitment monitoring, so cover values for yel-
relation analysis was used to explore the relationship low bush lupine were not analyzed. A repeated mea-
between density of yellow bush lupine individuals and sures analysis, with year as the within-subject factor,
levels of ammonium (r 0.404, p 0.008), nitrate (r was performed for each vegetation response variable to
0.304, p 0.001), and organic matter (r 0.398, p identify significant changes in cover over time. For the
0.009) in the first sample. mat-lupine and lupine-mat vegetation types, the effect
Data from the 0.6-m2 plots in the second sample were of treatment was analyzed as the between-subject fac-
used to perform multiple and step-wise regressions on tor. Results (Figs. 6–8) demonstrated a fairly continuous
each soil variable (Table 3). In the step-wise equation increase in native plants over time in the mat-lupine
for organic matter, four vegetation variables exclusive and lupine-mat types (p 0.0001), a small but statisti-
of yellow bush described virtually 100% of the variation cally significant reduction in nonnative forbs between
described in the multiple regression (r 0.905, p the first and second years in the mat-lupine type (p
0.0004). In the nitrate stepwise equation, only two vari- 0.002), and a decrease in nonnative grasses in all four
ables, litter and duff and nonnative forbs, were re- vegetation types (p 0.001). In the mat-lupine and lu-
quired to explain all but 0.4% of the variation described pine-mat types, nonnative grasses first increased and
by the regression (r 0.803, p 0.0013). Yellow bush then decreased below pre-treatment levels. The duff re-
lupine entered at the second step in the ammonium re- moval treatment reduced nonnative forbs and grasses
gression, accounting for an additional 21% (after non- in both the mat-lupine and lupine-mat types (p 0.021).
native grasses) of the total variation described by the Because vegetation cover in control plots was not
multiple regression (r 0.769, p 0.004). measured in 1992, controls were not included in the
ANOVA, but t tests were used to compare mean cover
in control plots by vegetation type for the three vegeta-
tion variables between 1988 and 1991. No significant
Effects of Treatment on Species Composition
differences were detected between years (p 0.05), con-
Changes in mean cover by year and treatment for the firming that changes detected in treated plots were the
three response variables (native species, nonnative result of treatments rather than regional vegetation
forbs, and nonnative grasses) are shown in Figures 6–8. changes over time.
MARCH 1998 Restoration Ecology 63
Yellow Bush Lupine Invasion I
Figure 5. Lupine-grass veg-
etation type, characterized
by high yellow bush lupine
influence and nonnative
grasses (center foreground).
Effects of Treatment on Soil Variables both highest in the lupine scrub type. Organic matter
We used two-way ANOVA of the third-year soil data (Ta- was unique in that lupine scrub, at the end of the lu-
ble 4) to determine whether treatment affected soil vari- pine-vegetation continuum, was characterized by lower
ables. A separate ANOVA was used for each soil variable, values than lupine-grass. As expected, the density of
with vegetation type and treatment as factors. All three large yellow bush lupine shrubs increased with pro-
soil variables showed significant differences among vege- gression along the lupine-vegetation continuum. These
tation type (p 0.05), but there was no difference between results suggested a simple linear relationship. Despite
treated and control plots, indicating that treatments did the fact that both lupine density and soil nutrients in-
not change available nitrogen or organic matter. creased along the lupine continuum, however, there
was not a high correlation between them, implying that
lupine abundance is not solely or primarily controlling
Effects of Treatment on Yellow Bush Lupine Recruitment nutrient levels.
Emergence of yellow bush lupine seedlings was ex- Multiple and step-wise regression analysis of the sec-
tremely high in the year following treatment, then de- ond data set confirmed that vegetation variables other
creased dramatically (Fig. 9). Recruitment was higher in than yellow bush lupine may be influencing soil vari-
litter and duff removal treatments than those in which ables. Ammonium, the product of symbiotic bacteria in
only lupine was removed. New emergence ceased by the root nodules of yellow bush lupine (Holton et al.
the fourth year. 1991), was the only stepwise equation entered by lupine
as a vegetation variable. Nearly all of the variation in
nitrate was accounted for by litter and duff and nonna-
tive forbs. The source of the litter and duff could not be
Discussion
determined, but it is probable that lupine contributed
Ecological Impacts of Invasion
significantly because it is large, fast-growing, and short-
lived (Davidson & Barbour 1997). Nonnative forbs and
The initial ANOVAs and multiple comparisons per- grasses accounted for 91% of the variation in organic
formed on pre-treatment data demonstrated that the ni- matter. The influence of nonnative grasses on organic
trate, ammonium, and organic matter varied with re- matter can be explained by their fibrous root systems,
spect to vegetation type. Nitrate and ammonium were which provide organic matter that is easily incorpo-
64 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Table 1. Means of soil variables (nitrate, ammonium, and Table 3. Results of multiple (total r) and step-wise regressions
organic matter) by vegetation type, and results of Tukey (p 0.005) of vegetation variables (as independent variables)
multiple comparisons showing differences among vegetation on soil variables (as dependent variables).
types.*
Variable r
Position Along
Tukey Vegetation Lupine Organic matter
Mean Grouping Type Continuum Nonnative forbs 0.581
Nonnative grasses 0.821
Nitrate (ppm) Litter and duff 0.843
5.89 A Dune mat 1 Native species 0.904
8.46 AB Mat-lupine 2 Total r 0.905
9.00 ABC Lupine-mat 3
11.25 B Lupine-grass 4 Nitrate
13.00 C Lupine scrub 5 Litter and duff 0.769
Ammonium (ppm) Nonnative forbs 0.800
4.62 A Mat-lupine 2 Total r 0.803
4.89 A Dune mat 1 Ammonium
6.31 AB Lupine-mat 3 Nonnative grasses 0.620
7.00 AB Lupine-grass 4 Bush lupine 0.767
9.25 B Lupine scrub 5 Total r 0.769
Organic matter (%)
5.88 A Dune mat 1
8.46 A Mat-lupine 2
9.00 B Lupine-mat 3 yellow bush lupine invasion on soils, changes may oc-
11.25 C Lupine scrub 5 cur indirectly by the facilitation of colonization by non-
13.00 D Lupine-grass 4
native grasses and forbs that further enrich the soil.
*Significant differences (p 0.05) are identified by different letters. The position The role of soil fertility in plant invasions has been
of each vegetation type along the lupine influence continuum is indicated along examined in several recent studies. Burke and Grime
a scale from 1, least influence, to 5, most influence.
rated into the soil at shallow depths (Hausenbuiller
1975). This would also explain why the lupine-grass
vegetation type was higher in organic matter than lu-
pine scrub.
These findings suggest that the invasion of yellow
bush lupine into a nitrogen-deficient dune environment
creates complex changes in soil and vegetation. Lupine
directly results in soil enrichment, particularly of am-
monium, during both growth and decay. A similar phe-
nomenon was described by Vitousek (1986, 1990), who
studied ecosystem changes resulting from the invasion
of Myrica faya Ait., a nitrogen-fixing tree, into young
volcanic substrates in Hawaii. Nitrogen fixation by
Myrica was found to alter both the quantity and avail-
ability of nitrogen. In addition to the direct affects of
Table 2. Mean density of nonseedling yellow bush lupine
individuals by size class and vegetation type.*
Position
Along Lupine Mean Lupines/m2 Mean Lupines/m2
Vegetation Type Continuum (SD) 15–50 cm (SD) 50 cm n
Mat-lupine 1 0.19 (0.11) 0.09 (0.04) 13
Lupine-mat 2 0.25 (0.25) 0.28 (0.12) 13
Lupine-grass 3 0.16 (0.16) 0.43 (0.20) 8 Figure 6. Changes in mean cover ( SE) of native species, non-
Lupine scrub 4 0.03 (0.05) 0.52 (0.39) 8 native forbs, and nonnative grasses in the mat-lupine vegeta-
*The position of each vegetation type along the lupine continuum is indicated tion type, lupine removal treatment (a) and litter and duff re-
along a scale from 1, least abundant, to 4, most abundant. moval treatment (b) over the four years of the study.
MARCH 1998 Restoration Ecology 65
Yellow Bush Lupine Invasion I
Figure 8. Changes in mean cover ( SE) of native species,
Figure 7. Changes in mean cover ( SE) of native species, nonnative forbs, and nonnative grasses in the lupine-grass
nonnative forbs, and nonnative grasses in the lupine-mat veg- vegetation type (litter and duff removal treatment) (a) and the
etation type, lupine removal treatment (a) and litter and duff lupine scrub vegetation type (litter and duff removal treat-
removal treatment (b) over the four years of the study. ment) (b) over the four years of the study.
(1996) demonstrated the importance of fertility changes grasses through enhanced soil productivity (Maron &
in predicting plant invasions in a nutrient-limited eco- Connors 1996). Zink et al. (1996) found that disturbance
system in the United Kingdom. In the dune system at caused by a pipeline placed through several intact
Bodega Bay, California, yellow bush lupine, a putative southern California plant communities resulted in the
native, was responsible for the invasion of nonnative proliferation of exotic annual plants, which in turn
Table 4. Mean and standard deviation of soil variables (ammonium, nitrate, and organic matter) by
vegetation type and treatment in the third year of the study (n 5 per treatment type; n 3 per control).
Vegetation Type
Mat-lupine Lupine-mat Lupine-grass Lupine Scrub
Treatment Mean (s) Mean (s) Mean (s) Mean (s)
Ammonium (ppm)
Lupine removal only 2.26 (1.72) 2.40 (0.73) — — — —
Lupine plus duff removal 1.94 (0.65) 2.86 (1.84) 2.96 (1.31) 3.62 (1.19)
Control 1.70 (0.18) 2.52 (1.07) 2.19 (0.83) 3.96 (2.41)
Nitrate (ppm)
Lupine removal only 4.07 (1.16) 4.46 (1.01) — — — —
Lupine plus duff removal 3.74 (0.58) 3.66 (0.54) 4.70 (0.83) 5.89 (1.38)
Control 3.43 (0.25) 6.38 (0.46) 3.84 (1.86) 5.90 (1.65)
Organic matter (%)
Lupine removal only 0.54 (0.13) 0.89 (0.56) — — — —
Lupine plus duff removal 0.56 (0.22) 0.63 (0.30) 1.17 (0.33) 1.05 (0.38)
Control 0.45 (0.17) 0.60 (0.10) 1.24 (0.20) 1.36 (0.13)
66 Restoration Ecology MARCH 1998
Yellow Bush Lupine Invasion I
Nonnative forbs underwent little change during the
4-year period of the study, although even the minor
changes that occurred as a result of the litter and duff
removal treatment in the mat-lupine type were statisti-
cally significant. Nonnative forb-cover values were ini-
tially low, and reductions may not have been essential
to restoration.
After three years, treated plots did not differ signifi-
cantly from controls in levels of available nitrogen and
organic matter. The lack of effect of treatment on soil
variables implies that a reduction in nitrogen and/or
organic matter is not a prerequisite for the restoration of
lupine-influenced dunes, despite the fact that soils un-
Figure 9. Yellow bush lupine seedling emergence (mean seed- derlying native vegetation are deficient in nitrogen. But
lings/m2) by vegetation per treatment type (error bar denotes there are several caveats to this conclusion. First, be-
SE). 1, mat-lupine, lupine removal; 2, mat-lupine, duff re- cause this sample represents a single slice in time, it is
moval; 3, lupine-mat, lupine removal; 4, lupine-mat, duff re- possible that soil variables initially increased after re-
moval; 5, lupine-grass; 6, lupine scrub. moval of vegetation from plots. A subsequent decline
would then be masked, and the net result would be in-
distinguishable from control plots. It is also possible
caused unstable litter and increased mineralization, fa- that soil changes are lagging behind vegetation changes
voring the persistence of weedy over indigenous spe- and may be more noticeable in the future.
cies. Monitoring of yellow bush lupine recruitment dem-
onstrated that seedling emergence is stimulated by re-
moval of the litter and duff layer. But if treatment is
Restoration
continued for at least three years this can be considered
Restoration treatments resulted in a decrease in nonna- a benefit, because the seedbank is presumably being de-
tive grasses and sometimes forbs, and/or an increase in pleted. Lupine seeds are characterized by a hard seed
native species cover over a 4-year period. Only those coat (Murdoch & Ellis 1992) and, without the distur-
vegetation types less strongly influenced by yellow bance or temperature fluctuations associated with re-
bush lupine (mat-lupine and lupine-mat) experienced moval of litter and duff, may remain in the soil and con-
significant increases in native cover. The increase in na- tinue to emerge for a longer period.
tive cover observed in mat-lupine and lupine-mat may These results have led to a restoration protocol for
have been caused by the release of nutrients associated dunes invaded by yellow bush lupine. In addition to
with dead lupine roots in addition to competitive re- the removal of lupine, other nonnatives (especially non-
lease. The lack of change in native species cover in veg- native grasses) and litter and duff should be cleared
etation types more heavily influenced by lupine was from the site, even in newly invaded areas, to discour-
most likely due to the absence of remnant native plants age recolonization of lupine and other weeds. Treat-
or nearby sources for dispersal. ment must be repeated for at least three years in order
Annual, nonnative grasses decreased in all four vege- to deplete the weedy and lupine seedbanks. In areas
tation types. It has been previously observed that where yellow bush lupine has become heavily estab-
grasses are frequently the species to respond and domi- lished, revegetation with natives will be necessary if a
nate—to the detriment of broad-leaved plants—under source of propagules is lacking.
nutrient enhancement (Hobbs & Huenneke 1992). The
decline of grasses in the more lupine-influenced vegeta-
Acknowledgments
tion types (lupine-grass and lupine scrub) was the most
dramatic. In the lupine-mat type, grasses initially in- Funding for this study was provided by Louisiana-
creased following treatment, probably due to competi- Pacific Corporation, Simpson Timber Company, The
tive release. In the litter and duff removal treatments Nature Conservancy, the U.S. Bureau of Land Manage-
grasses eventually declined, but in the lupine-only ment, and the California Department of Fish and Game
treatment grasses never returned to pre-treatment lev- Endangered Plant Program. We thank J. Sawyer for his
els. Removal of the litter and duff layer has similarly contributions to the study, M. Barbour for facilitating
been shown to be effective in preventing recolonization soils testing, and A. Buell for her review. Comments by
of sand dunes by weedy grasses on the Great Lakes Associate Editor Edith Allen and two anonymous re-
(Choi & Pavlovic 1994). viewers were very helpful.
MARCH 1998 Restoration Ecology 67
Yellow Bush Lupine Invasion I
LITERATURE CITED Maron, J. L., and P. G. Connors. 1996. A native nitrogen-fixing
shrub facilitates weed invasion. Oecologia 105:302–312.
Alpert, P., and H. A. Mooney. 1996. Resource heterogeneity by Miller, L. M. 1988. How yellow bush lupine came to Humboldt
shrubs and topography on coastal sand dunes. Vegetatio Bay. Fremontia 16:6–7.
122:83–93. Murdoch, A. J., and R. H. Ellis. 1992. Longevity, viability and dor-
Barbour, M. G., T. M. DeJong, and B. M. Pavlik. 1985. Marine mancy. Pages 193–230 in M. Fenner, editor. Seeds: the ecol-
beach and dune plant communities. Pages 296–322 in B. F. ogy of regeneration in plant communities. CAB Interna-
Chabot and H. A. Mooney, editors. Physiological ecology of tional, Wallingford, United Kingdom.
North American plant communities. Chapman and Hall, Parker, J. 1974. Coastal dune systems between Mad River and Lit-
New York. tle River, Humboldt County, California. M.A. thesis. Depart-
Burke, M. J., and J. P. Grime. 1996. An experimental study of plant ment of Biology, Humboldt State University, Arcata, Califor-
community invasibility. Ecology 77:776–790. nia.
Choi, Y. D., and N. D. Pavlovic. 1994. Comparison of fire, herbi- Pickart, A. J., and J. O. Sawyer. 1998. Ecology and restoration of
cide, and sod removal to control exotic vegetation. Natural northern California coastal dunes. California Native Plant
Areas Journal 14:217–218. Society. Sacramento, California.
Davidson, E. D., and M. G. Barbour. 1997. Germination, establish- Pickart, A. J., T. E. Duebendorfer, and L. M. Miller. 1990. An inte-
ment, and demography of coastal bush lupine (Lupinus ar- grated approach to enhancing rare plant populations
boreus) at Bodega Head, California. Ecology 58:592–600. through habitat restoration. III. Restoration of altered coastal
Davy, J. B. 1902. Stock ranges of northwestern California: notes on dunes. Pages 488–500 in H. G. Hughes and T. M. Bonnicksen,
the grasses and forage plants and range conditions. Bulletin editors. Restoration ‘89: the new management challenge. So-
no. 12 of the Bureau of Plant Industry, U.S. Department of ciety for Ecological Restoration, Madison, Wisconsin.
Agriculture. Government Printing Office, Washington, D.C. Ramakrishnan, P. S., and P. M. Vitousek. 1989. Ecosystem-level
Duebendorfer, T. E. 1990. An integrated approach to enhancing processes and the consequences of biological invasions.
rare plant populations through habitat restoration. II. Habi- Pages 281–300 in J. A. Drake, H. A. Mooney, F. Di Castri,
tat characterization through classification of dune vegeta- R. H. Groves, F. J. Kruger, M. Rejmanek, and M. Williamson,
tion. Pages 478–487 in T. M. Bonnicksen and H. G. Hughes, editors. Biological invasions: a global perspective. Wiley,
editors. Restoration ‘89: the new management challenge. So- New York.
ciety of Ecological Restoration, Madison, Wisconsin. Sawyer, J. O., and T. Keeler-Wolf. 1995. A manual of California
Hausenbuiller, R. L. 1975. Soil science, principles and practices. vegetation. California Native Plant Society, Sacramento, Cal-
William C. Brown, Dubuque, Iowa. ifornia.
Hitchcock, C. L., and A. Cronquist. 1973. Flora of the Pacific Sholars, T. 1993. Lupines. Pages 622–636 in J. C. Hickman, editor.
Northwest: an illustrated manual. University of Washington The Jepson manual: higher plants of California. University of
Press, Seattle. California Press, Berkeley.
Hobbs, R. J., and L. F. Huenneke. 1992. Disturbance, diversity, Vitousek, P. M. 1986. Biological invasions and ecosystem proper-
and invasion: implications for conservation. Conservation ties: can species make a difference? Pages 163–176 in H. A.
Biology 6:324–337. Mooney and J. A. Drake, editors. Ecology of biological inva-
Hobbs, R. J., and S. E. Humphries. 1995. An integrated approach sions of North America and Hawaii. Springer-Verlag, New
to the ecology and management of plant invasions. Conser- York.
vation Biology 9:761–770. Vitousek, P. M. 1990. Biological invasions and ecosystem pro-
Holton, B., M. G. Barbour, and S. N. Martens. 1991. Some aspects cesses: towards an integration of population biology and
of the nitrogen cycle in a California strand ecosystem. Madroño ecosystem studies. Oikos 57:7–13.
38:170–184. Vitousek, P. M., L. R. Walker, L. D. Whiteaker, D. Mueller-Dom-
Horn, E. L. 1993. Coastal wildflowers of the Pacific Northwest. bois, and P. A. Matson. 1987. Biological invasion by Myrica
Mountain Press Publishing Company, Missoula, Montana. faya alters ecosystem development in Hawaii. Science 238:
LaBanca, T. 1993. Vegetation changes at Clam Beach coastal 802–804.
dunes, Humboldt County, California. M.A. thesis. Depart- Zink, T. A., M. F. Allen, B. Heindl-Tenhunen, and E. B. Allen.
ment of Biology, Humboldt State University, Arcata, Califor- 1996. The effect of a disturbance corridor on an ecological re-
nia. serve. Restoration Ecology 3:304–310.
Appendix. Vegetation types studied.
A. Lupine absent Dune mat (Fig. 1)
A. Lupine present B
B. Total plant cover 80%; yellow bush lupine cover 25%; dune mat species
present Mat-lupine (Fig. 2)
B. Total plant cover 80%; yellow bush lupine cover 25%; dune mat species
present or absent C
C. Dune mat species (except Solidago) 25% Lupine-mat (Fig. 3)
C. Dune mat species (except Solidago) 25% D
D. Yellow bush lupine cover 75%; nonnative grasses 25% Lupine-grass (Fig. 4)
D. Yellow bush lupine cover 75% Lupine scrub (Fig. 5)
68 Restoration Ecology MARCH 1998