Direct and Indirect Effects of Host Plants: Implications for Reintroduction of an Endangered Hemiparasitic Plant (Castilleja levisecta)

2008·6.5 MB·English

by  Beth A Lawrence

Madrono, Vol. 55, No. 2, pp. 151-158, 2008
DIRECT AND INDIRECT EFFECTS OF HOST PLANTS: IMPLICATIONS FOR
REINTRODUCTION OF AN ENDANGERED HEMIPARASITIC PLANT
{CASTILLEJA LEVISECTA)
Beth A. Lawrence*' and Thomas N. Kaye'-^
OR
' Oregon State University, Department ofBotany and Plant Pathology, Corvallis, 97330
SW OR
-Institute for Applied Ecology, 563 Jefferson St., Corvallis, 97333
Abstract
Rare, parasitic plants pose an interesting challenge to restoration practitioners. In addition to the
problems associated with small population size, rare parasites may also be limited by their host
requirements. Weexamined how the performance ofa rare PacificNorthwest hemiparasite, Castilleja
levisecta, was affected by the availability ofdifferent host combinations in the greenhouse and in the
field. Castillejalevisectaindividualsweregrown with twoindividuals ofthegrassFestucaroemeri, two
individuals of the aster Eriophyllum kinatum, one individual of each of these species (a "mixed"
treatment), or without any host. We did not find support for the complimentary diet hypothesis,
which predicts that parasites grown with multiple host species perform better than individuals grown
alone or with a single host. In the greenhouse, C levisecta individuals grown in the mixed treatment
hadgreaterstemgrowth than thoseplantedwithF. roemeri, butdidnot differfrom E. lanatum orno-
hosttreatments. Inthefield, voleactivityhadindirecteffectson C. levisectasurvivalmediated through
host species: vole tunneling and C. levisecta mortality were strongly associated with host treatments
including E. lanatum. Vole tunneling and C levisecta mortality were strongly associated with host
treatmentsincludingE. lanatum. Field survivalofno-hostand F. roemeritreatmentsweresignificantly
higher than those grown with E. lanatum. Our results emphasize the importance of basing
conservation decisions on experimental research conducted under conditions similar to those ofthe
intended application, as greenhouse results were a poor predictor of field performance. For
restoration of endangered hemiparasitic plants, we recommend planting with hosts that are not
attractive to herbivores.
Key Words: Castilleja levisecta, complimentary diet hypothesis, host-use, rare hemiparasite,
reintroduction.
Parasitic plants are a dynamic component of United States, to evaluate its host preferences in
many plant communities capable of altering support of recovery actions.
productivity (Marvier 1998b; Matthies 1997), Although facultative hemiparasites are photo-
competitive interactions (Gibson and Watkinson synthetic and do not require a host plant, they
1991; Matthies 1996), and community structure often form haustoria (i.e., physical connections
(Gibson and Watkinson 1992; Press 1998). with other root systems) through which nutrients,
Although many parasitic plants are agricultural water, and secondary compounds are obtained
pests, some are ofconservation concern and pose from the host (Kuijt 1969; Press 1989). In natural
an interesting challenge to restoration practition- systems, unattached mature facultative parasites
ers (Marvier and Smith 1997). In addition to the are uncommon, and attachment to a host
diversity of obstacles typically encountered dur- generally stimulates parasite fitness and growth
ing reintroduction, rare parasites may also be (Kuijt 1969). Most members of the genus
limited by host requirements. Uncertainties asso- Castilleja are considered generalist hemiparasites,
ciated with parasite host specificity and the capable of parasitizing multiple host species
availability and quality of hosts at restoration (Dobbins and Kuijt 1973; Heckard 1962). How-
sites are likely to impede parasitic plant reintro- ever, the degree to which a host stimulates
duction efforts (Marvier and Smith 1997). hemiparasite fitness varies considerably among
Therefore, successful management of rare para- host species (Chuang and Heckard 1971; Gibson
sites necessitates consideration of their unique and Watkinson 1992; Marvier 1998b; Matthies
biology. We conducted greenhouse and field 1996, 1997; Seel and Press 1993). Interactions
experiments with Castilleja levisecta Greenman between plant parasites and host species can have
(golden paintbrush), a rare hemiparasite endemic direct and indirect effects both on host and
parasite performance, as well as their pollinators
to the prairies of the Pacific Northwest of the
(Adler et al. 2001), and herbivores (Adler 2002,
2003; Adler et al. 2001; Marko 1996; Marvier
*Current address: University ofWisconsin-Madison, 1996). Parasitic plants can acquire secondary
Botany Department, 430 Lincoln Drive, Madison, WI compounds from host species (Govier et al. 1967;
53706; email: [email protected] Schneider and Stermitz 1990; Stermitz and Harris
MADRONO
152 [Vol. 55
1987), which in turn can alterspecies interactions. (Caplow 2004; Wayne 2004), but host-mediated
For example, acquisition of alkaloids from the effects of herbivory on the species has not
host Lupinus a/bus directly reduced insect herbiv- previously been evaluated.
ory of Castilleja indivisa, and indirectly increased Here, we use greenhouse and field studies to
pollination (Adler et al. 2001). test the complimentary diet hypothesis and
In the field, hemiparasitic plants often parasit- examine how host-interactions affect herbivory
ize several hosts simultaneously (Gibson and by rodents under field conditions, as well as
Watkinson 1989; Matthies 1996). Generalist provide recommendations for future C. levisecta
hemiparasites may perform better on a mixed recovery efforts.
diet relative to a homogenous diet due to
improved nutrient balance and/or dilution of Methods
toxic secondary plant compounds (Marvier
1998a). Many taxa benefit from multiple food Study species
sources, including some insects (Bernays et al.
1994), gastropods (Pennings et al. 1993), and Castilleja levisecta (Orobanchaceae, formerly
reptiles (Bjorndal 1991). Therefore, we propose classified in Scrophulariaceae) is a short-lived (5-
that providing multiple nutrient sources increases 6 yr), multi-stemmed, perennial endemic to the
individual fitness (the complimentary diet hy- native grasslands of the Western Pacific North-
pothesis). We test this hypothesis by comparing west United States. It is an out-crossing species
the size and survival ofa rare hemiparasitic plant primarily pollinated by Bombus spp. and is
in mixed host, single host, and no-host plantings. known only to reproduce by seed (Kaye and
Castilleja levisecta is a federally threatened Lawrence 2003; Wentworth 2001). The eleven
species and is currently restricted to eleven remaining C levisecta populations are concen-
populations in the Pacific Northwest. The species trated in the San Juan Archipelago of the Puget
is extinct in the southern portion of its historic Trough eco-region, and are found on sandy, well
range, including the Willamette Valley, Oregon. drained soils of glacial origin (Chappell and
Federal recovery criteria for C. levisecta call for Caplow 2004). Despite the rarity of this species,
the existence of20 populations composed of 1000 the remaining populations maintain unusually
flowering individuals (USFWS 2000). However, high levels of genetic diversity compared with
the species has limited capacity for natural other endemic species and members of the
dispersal and colonization of new sites, necessi- Orobanchaceae (Godt et al. 2005).
tating ex situ conservation techniques to meet
recovery goals. Thus, a strategic reintroduction Greenhouse experiment
plan has been prepared to support the long-term
viability of C. levisecta and requires the estab- To test for differences in C. levisecta perfor-
lishment of new populations within its historic mance when grown with different host combina-
range (Caplow 2004). Although several studies tions, we randomly assigned individuals to one of
C
have investigated levisecta host use (Pearson four host treatments, including no-host (control),
and Dunwiddie 2006; Wayne 2004; Wentworth two E. lanatum (Asteraceae) individuals, two F.
2001), clarification of its host dynamics in a roemeri (Poaceae) individuals, or one individual
restoration context is necessary before large scale ofeach ofthese host species ("mixed"). We used
reintroduction efforts are pursued. plant material from two C levisecta source
WA
While C. levisecta does not require a host to populations located on Whidbey Island,
reproduce in a greenhouse environment and does (Ebey's Landing: 48°13'35"N, 122°46'00"W and
not appear to be host specific (Wentworth 2001), Forbes Point: 48°16'15"N, 122°37'35"W). Ap-
evidence suggests planting C. levisecta in the field proximately twenty host treatment replicates
with a perennial host increases size and repro- from each ofthese source populations were used
ductive output (Pearson and Dunwiddie 2006; to test our hypotheses, for a total of39 replicates
Wayne 2004). Greenhouse observations suggest per host treatment (n = 156). Eriophyllum
that C. levisecta can form haustorial connections lanatum and F. roemeri were used as host plants
C
with several perennial prairie species [e.g., Leu- because levisecta forms haustorial connections
canthemum vulgare Lam., Eriophyllum lanatum with these native perennials (Wayne 2004, Beth
(Pursh) Forbes, Festuca roemeri (Pavlick) Alex- Lawrence, pers. obs.), they are common at
eev, and Fragaria vesca L.], and with itselfwhen southern extant populations (Chappell and Ca-
grown alone (Kaye 2001; Wentworth 2001). Field plow 2004), and are likely to be present at
C
experiments indicate that outplanting levisecta reintroduction sites.
with F. roemeri increases the number of inflores- Castilleja levisecta capsuleswere collected from
cences produced compared to no-host controls, 17 maternal plants from each of the two source
although host presence did not affect field populations in August 2003 to provide seeds for
survival rates (Wayne 2004). In addition, C. this experiment and were germinated using the
levisecta is frequently eaten by small mammals methods outlined in Lawrence and Kaye (2005).
2008] LAWRENCE AND KAYE: CASTILLEJA LEVISECTA HOST USE 153
wOenre1 pDlaencteemdbeirnto200ce3l,l Cfl.atsleviinsecataweglelr-mdirnaainntesd a5n°nCuaalndmi1n7°iCm,urmesapencdtimvaelxyi(mWuRmCtCemp2e0r05a)t.ures of
medium amended with slow release micro- and We randomly planted host-parasite replicates
macro- nutrients and were placed in a greenhouse (each pot was a replicate) into the center of a
with 400 watt high pressure sodium lights and 1 m^ plot within a 10 X 15-m grid fenced to
temperature fluctuatingevery 12 hr (12°C /18°C). exclude deer. Deer are frequent herbivores ofthe
A randomized block design was implemented to species and threaten extant populations; fences to
assign host treatments to C. levisecta individuals, exclude deer have been built at two ofthe extant
with source population and maternal line serving C. levisecta populations (Beth Lawrence, pers.
as the blocking factors. Two maternal lines from obs.). Castilleja levisecta individuals and hosts
Ebey's Landing and three from Forbe's Point were dormant at the time of outplanting, and
were assigned to two blocks, because extra plants senesced material was removed. A balanced
from these maternal lines were available. Plants design could not be executed in the field because
were repotted into 3.8 L pots with their assigned some greenhouse plants died during the previous
host treatment on 28 January 2004. Castilleja summer. However, at least 22 replicates of each
levisecta individuals and potential hosts were ofthe fourhost treatments weretransplanted into
planted in a triangle with all plants 10 cm apart; the field (no-host, n = 39; F. roemeri, n = 31;
C. levisecta individuals assigned the no-host lanatum, n = 26; mixed, n = 22).
treatment were planted in the center of the pot. Field transplants were monitored in early June
We used a no-host control rather than planting 2005 because surveys conducted in a companion
three C. levisecta individuals together because we 2004 field study revealed that transplants were at
had a limited number of plants. Eriophyllum theirmaximum size and peak inflorescence at this
lanatum plants were rooted cuttings from Will- time (Lawrence 2005). Vole abundance was
amette Valley genetic stock provided by Heritage unusually high throughout the Pacific Northwest
Seedling Co., Salem, OR. We used F. roemeri during the 2005 growing season and all surviving
individuals grown from Willamette Valley seed C. levisecta individuals at the field site were
that were one year old when paired with C. subjected to herbivory, most likely from grey-
levisecta. We attempted to equalize above- and tailed voles (Microtus canicaudus) (Beth Law-
below-ground biomass of provided hosts by rence, pers. obs.). Stem length and/or number, as
trimming them with shears. Plants were random- well as flower and/or seed production were not
ized on greenhouse benches and fertilized bi- reliable measures of C. levisecta performance, as
weekly with a liquid 15-30-15 fertilizer to herbivory appeared to stimulate resprouting,
encourage growth and establishment. alter plant morphology, and prevent individuals
We recorded total stem length, stem number, from flowering (Beth Lawrence, pers. obs.).
and number of flowers produced by each C. Therefore, we used C. levisecta survival as the
levisecta individual in May 2004, approximately response variable for the field component of our
15 wk after potting the hemiparasites and hosts study. Vole tunneling was also very frequent,
together. Flowering had finished at this time, so indicating herbivore pressure occurred in the root
our measurements are considered estimates of zone as well as above ground. Tunnels were
total fiower production. Plants were moved to a unevenly distributed throughout the study area,
shade-house in June 2004 and received supple- so herbivore pressure by voles was measured as
mental water throughout the summer. presence or absence of tunnels within 15 cm of
the transplant root crown.
Field experiment
Statistical analyses
To test our host and herbivore hypotheses
under field conditions, we transplanted the same We used multivariate analysis of variance
potted plants with hosts used in the greenhouse (MANOVA) with a Wilks' lambda multivariate
study to an upland prairie on 1 December 2004. F test to simultaneously test for differences in C.
Our field site was located at Pigeon Butte, Finley levisecta greenhouse response variables (stem
National Wildlife Refuge, OR (44°23'54"N, length, stem number, and flower number) among
IMANOVA
123°19'11"W), in habitat likely to be used for host treatments. Prior to analysis,
future C. levisecta recovery efforts in the Wil- stem number was log-transformed to improve
lamette Valley. The site had a high diversity of homoscedasticity. Castilleja levisecta source pop-
native perennials and abundant non-native pas- ulation and maternal effects were used as
ture grasses (e.g., Festuca arundinacea Schreb. blockingfactors in this analysis becausedifferential
and Arrhenatherum elatius (L.) P. Beauv. ex J. growth among populations and individuals from
Presl & C. Presl). It was situated on the shoulder different maternal lines has been observed in this
ofa butte at 150 m elevation, dominated by silty- species (Kaye 2001). However, we focus our
clay-loam soils. Average annual precipitation in analysis on host treatment effects. Following
this region is approximately 115 cm, with average MANOVA, univariate ANOVAs were conducted
MADRONO
154 [Vol. 55
Table 1. Results from Univariate ANOVAs Testingthe Effectof Host Treatment (host), Source
Population (source), Maternal Line (matline), and the Host Treatment*Source Population
Interaction (host*source) on the Number of C. Levisecta Flowers Produced (Flower #), Number
OF Stems (Stem #), and Total Stem Length (Stem Length) inthe Greenhouse. Significant effects at a =
0.05 denoted with *.
RESPONSE EFFECT DF MS F P
1 IVJWCl -ff~ J 474 0.84 0yj.'4-T.71
source 1 18313 32.32 <0.001*
matHne 32 877 1.55 0.048*
host*source 3 38 0.067 0.98
Stem # host 3 0.24 2.87 0.039*
source 1 2.61 31.76 <0.001*
matHne 32 0.097 1.19 0.25
host*source 3 0.0069 0.084 0.97
Stem Length host 3 7112 3.69 0.014*
source 1 197 0.10 0.75
mathne 32 3689 1.9 0.0067*
host*source 3 1879 0.97 0.41
on the three greenhouse response variables. Signif- vealed that individuals grown with mixed hosts
icantANOVAswere followed byTukey's HSD for had a greater number of stems and total stem
pairwise comparisons among host treatments. length compared to those grown with F. roemeri,
We used binary logistic regression to test for but did not differ from those grown without a
differences among host treatments in C. levisecta host or with E. lanatum.
field survival and vole tunnel presence. Signifi-
cance was measured by drop in deviance (DEV) Field experiment
with a chi-square distribution. Dunn-Sidak cor-
rections were used to adjust alpha levels for all Field survival of C. levisecta differed among
pair-wise comparisons among host treatments. host treatments (DEV^^gi = 44.65, P < 0.001),
We used linear regression to determine if the but neither source population (DEVi gi = 0.089,
proportion of transplants with vole tunnels was P = 0.77) nor maternal line (DEV32,8i = 34.43, P
associated with the proportion of C levisecta = 0.40) accounted for a significant portion ofthe
individuals surviving. Finally, we calculated an residual deviance. A higher proportion ofno-host
odds ratio to compare C. levisecta survival when C. levisecta individuals survived compared to
planted with E. lanatum versus survival when not those planted with either E. lanatum or mixed
planted with this species (i.e., alone or with F. hosts, but did not differ from plants with F.
roemeri). All analyses were conducted using S- roemeri hosts (Fig. 1). Also, C. levisecta planted
PLUS v. 6.2 (Insightful 2000). with F. roemeri hosts had significantly higher
survival than those planted with E. lanatum
Results (Fig. 1).
Rodent tunnel presence near transplant root
Greenhouse experiment crowns differed significantly among host treat-
ments (DEV3,n4 = 50.17, P < 0.001). Castilleja
MANOVA
According to analyses, Castilleja levisecta individuals planted with F. roemeri or
levisecta greenhouse performance differed among without a host had fewer rodent holes near their
host treatments (Wilkes = 0.84, F3,ii6 = 2.37, P root crowns compared to those planted with
= 0.014) as well as source populations (Wilkes = either E. lanatum or mixed hosts (Fig. 1). In
0.55, Fi,n6 = 31.47, P < 0.001), but no addition, we measured a strong inverse relation-
differences were observed among maternal lines shipbetween C. levisectasurvivalandthepresence
(Wilkes = 0.42, Fgsjie = 1.19, P = 0.13). Host oftunneling within the vicinity ofthe root crown
treatment effects were consistent among C (Fi,2 = 23.07, P = 0.04, R^ = 0.92) (Fig. 2). The
levisecta individuals from the two source popu- odds ofa C. levisecta transplant surviving in the
lations used in this study, as the interaction field when planted without an E. lanatum host
between source population and host treatment were 11.25 (95% C.I. = 4.29, 28.78) times greater
was not significant (Wilkes = 0.93, F3 = 0.92, than when co-planted with an E. lanatum host.
P = 0.51). While C. levisecta stem number and
total stem length differed among host treatments, Discussion
the number offlowers did not (Table 1). Univar-
iate ANOVA statistics for the three response We did not find support for the compHmentary
variables are presented in Table 1. Post-hoc pair diet hypothesis, which predicts that individuals
wise comparisons of univariate ANOVAs re- with multiple nutritional sources will perform
2008] LAWRENCE AND KAYE: CASTILLEJA LEVISECTA HOST USE 155
0 1 a ab c cb 0.8 R2=0.92, P=0.04
0.7
.£ 0.8
1 0.6
I
°' c 0.5
1 1
2o. 0.2 MC3 0.4
0 lortio 0.3
nohost F.roemeri Elanatum mixed 8-
Q. 0.2
0.1
ii) 0
a a b b
0.2 0.4 06 0.8
tunni 0.8
proportionneartunnel
« 0.6-
B Fig. 2. Scatterplot and trendline from linear regres-
r)tion 0.4- 1 1 sion of the average proportion of C. levisecta trans-
QO. 0.2- plants within 15 cm of a vole tunnel and average
a. transplant survival for each host treatment (no host =
0- nohost F.roemeri Elanatum mixed , F. roemeri — *, mixed = #, E. lanatum = O).
Fig. 1. i) Castilleja levisecta field survival by host
ltroecaattmeednwti,thiiin) 1P5rocpmorotfiornodeonfttCu.nnleelvsi.seHcotsatttrraenastpmleannttss creolmapteitveittioonF.maryoeemxeprliaihnowsthsy. FA.ltreorenmaetrivieilsy,a proooort
notsharing acommon letterwere significantlydifferent host in pots, as pots with F. roemeri were
(P < 0.05) after Dunn-Sidak corrections. generally more root bound than other host
treatments (Beth Lawrence, pers. obs.). This is
better than those provided with a limited diet. consistent with our previous work that found C
Mixed hosts improved some measures (i.e., stem levisecta grown in pots with F. roemeri were
number and total stem length) of C. levisecta smaller and flowered less frequently in the second
greenhouse performance compared to those growing season compared to those potted with E.
paired with F. roemeri, but did not confer an lanatum (Kaye 2001). This explanation is more
advantage overno-host orE. lanatum treatments. plausible, as C. levisecta performed similarly
Likewise, mixed hosts did not promote C. among all other treatments, but did poorly when
levisecta field survival. In fact, no-host controls paired just with F. roemeri. Our greenhouse
had greater field survival than both mixed and E. results may also have been confounded by our
lanatum treatments (Fig. 1). The complementary judicious use offertilizer, as attachment to hosts
diet hypothesis may not be the most appropriate may not confer fitness benefits in the presence of
theory to apply to hemiparasite nutrition, as this abundant nutrients.
hypothesis has primarily been tested in animal Vole activity had strong indirect effects on C
systems. Other studies addressing hemiparasite levisecta field survival mediated by host species.
fitness using multiple hosts have also found Populations of the grey-tailed vole (Microtus
mixed results. During greenhouse studies, Mel- canicaudus) were larger than average in the
ampyrum arvense did not benefit from mixed Willamette Valley during the 2005 field season
hosts (a legume and a grass) (Matthies 1996), due to a mild winter in 2004-05, increasing
though Castilleja wightiigrowth and reproductive herbivore pressure on C. levisecta transplants and
output were improved by simultaneous attach- impacting the region's grass seed crop. Nine
ment to two host species (a legume and an aster) Oregon counties were declared agricultural disas-
(Marvier 1998a). However, greenhouse studies ter areas by the U.S. Department of Agriculture
may oversimplify field dynamics and should be due to large crop losses from voles (A.P. 2005).
extrapolated to the field with caution. For Although population sizesin 2005 were atypically
example, we observed strong indirect effects of large, voles are ubiquitous in Pacific Northwest
herbivory mediated by host species in the field, prairies and are major herbivores contributing to
which has important consequences for C. levi- grassland dynamics (Wilson and Carey 2001).
secta recovery efforts. Further, global warming may increase the
While we observed improved C. levisecta stem frequency of mild winters in the Pacific North-
performance when grown with mixed hosts west (Leung and Ghan 1999) and result in greater
relative to F. roemeri in the greenhouse, the regularity of vole outbreaks. Selective herbivory
number offlowers produced did notdifferamong by voles in other grassland systems has been
host treatments. Mixed hosts may have improved shown to dramatically alter species composition
C levisecta nutrition by providing complimenta- and diversity (Batzli and Pitelka 1970; Howe and
ry resources, thereby improving stem growth Lane 2004). Using exclusion experiments, Howe
MADRONO
156 [Vol. 55
and Lane (2004) observed that meadow voles nutrients to hemi-parasites during periods of
eliminated otherwise common plants due to critical environmental stress (Kuijt 1969; Press
preferential herbivory. 1989). However, under horticultural growing
Castilleja levisecta field survival also did not conditions with ample water, nutrients, and light,
support the complimentary diet hypothesis, C. levisecta individuals produced abundant bio-
possibly as a result of indirect effects from mass and had high reproductive output without
herbivore activity. While herbivory was evident hosts. In our field study, no-host C. levisecta
on all surviving C. levisecta individuals at the individuals had the highest proportion surviving
study site, vole tunneling and field mortaUty were (a- = 0.78), although at the time of monitoring
strongly associated with host treatments that these plants had yet to experience summer
included E. lanatum, whose roots may have been drought conditions typical ofthe region. Natural
particularly palatable to voles. Castilleja levisecta populations of C. levisecta emerge in early
plantspaired with two E. lanatum individuals had March, flower in May, and senesce in July in
higher field mortality than those planted with a response to dry conditions (Caplow 2004).
single E. lanatum individual (mixed host), al- Summer drought is a strong selective force
though these effects were not strictly additive resulting in substantial C. levisecta transplant
(Fig. 1). The mechanism contributing to high mortality, as field survival is typically high the
mortality of C. levisecta individuals associated first growing season, but is generally reduced the
with E. lanatum is unclear, but root system second growing season (Lawrence 2005; Pearson
disturbance, direct grazing of C. levisecta roots, and Dunwiddie 2006; Swenerton 2003; Wayne
or the indirect effect of reduced host vigor/ 2004). Results from a companion common
survival likely contributed to this observation. garden experiment indicate that planting a
Meanwhile, C. levisecta individuals planted with- perennial host with C. levisecta transplants
out a host or with F. roemeri had much higher improves second year survival (Lawrence 2005).
survival rates and less rodent tunneling. This Second year survival at a site where individuals
indicates that voles did notjust target potting soil were planted with F. roemeri was particularly
or areas with low root density to tunnel in, but high {x = 0.75), compared to the average
were specifically attracted to E. lanatum. Foliage proportion surviving at the other nine common
and roots of plants in the genus Eriophyllum gardens {x = 0.21), thatwerenot provided ahost.
contain sesquiterpene lactones (Bohlmann et al. Pearson and Dunwiddie (2006) observed greater
1981), an extremely diverse group ofcompounds C. levisecta flower production when grown with
that may be desirable to herbivores due to anti- E. lanatum compared with F. roemeri, but field
fungal, anti-bacterial, anti-tumourgenic, or anti- survival was greater with F. roemeri. Another
inflammatory properties (Pieman 1986), and may fieldexperiment also observed greater C. levisecta
have contributed to increased vole tunneling in survival when outplanted with F. roemeri relative
the vicinity ofE. lanatum. to no-host and E. lanatum treatments (S. Reich-
Although we have provided evidence that vole ard 2005, University of Washington, pers.
activity mediated C. levisecta survival through comm.). Castilleja hispida Benth., which can
host species in the field (Fig. 2), an alternative hybridize with C levisecta, also had higher field
process could be responsible for the observed survival when planted with F. roemeri than when
field patterns. Due to a malfunction of the planted with no host (Schmidt 1998). These
automatic watering system, we observed differ- observations suggest that planting a perennial
ential survival of the potted plants during the host with C levisecta in the field is beneficial, and
2004 summer in the shadehouse. Survival was may allow the parasite to take advantage ofhost
greater among no-host (100%) and F. roemeri roots to exploit nutrients and water from a larger
(79.5%), than among E. lanatum (66.7%) and volume of soil during periods of environmental
mixed hosts (54.4%), similar to the pattern of stress. Further, this suggests that C. levisecta
differential survivalwe observed in the field. Thus survival is higher when planted with F. roemeri
it is possible that our field survival rates were than E. lanatum, and this may be due, at least in
only spuriously correlated with vole activity and part, to preferential vole herbivory ofE. lanatum.
that survival of C. levisecta is directly influenced Other native perennial species, including le-
by host treatment, rather than indirectly via gumes and showy angiosperms that can attract
herbivore activity. pollinators, may also be appropriate hosts for C.
Results from our greenhouse and field studies levisecta. Leguminous hosts are commonly better
suggest that planting C. levisecta with a host may hemiparasite hosts than grass species because of
not be absolutely necessary, but mayconfer some theircapacity to fix nitrogen (Adler 2003; Gibson
advantages to field plantings. Although our and Watkinson 1991; Matthies 1997; Seel and
findings are likely context dependent, no-host Press 1993). Additionally, alkaloid uptake from
controlsperformed as well or betterthan all other leguminous hosts can confer resistance to herbiv-
host treatments in both greenhouse and field ory (Adler 2002), and increase pollinator visita-
environments. Host plants can provide water and tion (Adleret al. 2001). Although themycorrhizal
2008] LAWRENCE AND KAYE: CASTILLEJA LEVISECTA HOST USE 157
status of C. levisecta has not been investigated, support. Bruce McCune and Manuela Huso provided
many hemiparasites in the Orobanchaceae are statistical guidance. Drafts ofthis work were reviewed
considered non-mycorrhizal (Harley and Harley by Matt Blakely-Smith, Dave Pyke, and Aaron Liston,
1987). The mycorrhizal status of the host plant as well as two anonymous reviewers. This research was
however, can influence the performance of the fBuonndneidebTyeampNlaettiovenPAlawnatrSdocifeotry oPflaOnrtegSoynstfeimealtdigcrsa,nt,a
hemiparasite. Studies have shown that hemipar- National Science Foundation Graduate Research Fel-
asites attached to mycorrhizal hosts have greater lowship, and USFWS contract 101813M465.
biomass and flower production than those
growing with non-mycorrhizal hosts (Davies Literature Cited
and Graves 1998; Salonen et al. 2001). We
suggest that new C levisecta potential host A.P. 2005. Voles head for Oregon vineyards, orchards.
Available at: http://www.wtopnews.com/index.
species and mycorrhizal inoculation of hosts php?nid=220&sid=594057; accessed 15 November
should be examined experimentally in the field 2005.
to further examine the complimentary diet Adler, L. S. 2002. Host effects on herbivory and
hypothesis and improve the success oflarge-scale pollination in a hemiparasitic plant. Ecology 83:
reintroductions ofthis endangered species. 2700-2710.
. 2003. Host species affects herbivory, pollina-
Implications for practice tion, and reproduction in experiments with para-
sitic Castilleja. Ecology 84:2083-2091.
, R. Karban, and S. Y. Strauss. 2001. Direct
• Conservation decisions should be based on and indirect effects ofalkaloids on plant fitness via
experimental research conducted under condi- herbivory and pollination. Ecology 82:2032-2044.
tions similar to those ofthe intended applica- Batzli, G. O. and F. a. Pitelka. 1970. Influence of
tion; C. levisecta greenhouse performance was mlaenad.doEwcolmooguyse51:p1o0p2u7l-a1t0i3o9n.s on Cahfornia grass-
a poor predictor of field survival. Extrapola- Bernays, E. a., K. L. Bright, N. Gonzalez, and J.
tion of greenhouse results to natural systems Angel. 1994. Dietary mixing in a generalist
can oversimplify the complex biotic interactions herbivore: tests of two hypotheses. Ecology 75:
thatspeciesareexposedtointhefield,andworse, 1997-2006.
suggest inappropriate management actions. Bjorndal, K. a. 1991. Diet mixing: nonadditive
• Greenhouse propagation ofendangered hemi- interations of diet items in an omnivorous fresh-
parasites like C. levisecta may not require a water turtle. Ecology 72:1234-1241.
host, but growth and survival after field BOHLMANN, F., C. Zdero, J. Jakuovic, H. Robin-
planting may be improved by planting with son, AND R. M. King. 1981. Eriolanolides,
additional species. See (Lawrence and Kaye eudesmanolidesandrearranged sesquiterpenefrom
Eriophyllum species. Phytochemistry 20:2239-2244.
2005) for details on propagation techniques Caplow, F. 2004. Reintroduction plan for golden
for C. levisecta. paintbrush {Castilleja levisecta). Washington Nat-
• Failure to find support for the complimentary ural Heritage Program, Department of Natural
diet hypothesis with C. levisecta suggests that Resources, Olympia, WA.
outplanting rare hemi-parasites with multiple Chappell, C. B. and F. Caplow. 2004. Site charac-
hosts may not be necessary. teristics of golden paintbrush populations. Wash-
• We recommend against planting hemiparasites ington Natural Heritage Program, Department of
with hosts that are attractive to herbivores Natural Resources, Olympia, WA.
when and where these animals are present. We Chuang, T. I. AND L. R. Heckard. 1971. Observa-
suspect planting C. levisecta with a perennial tions on root-parasitism in Cordylanthus (Scroph-
ulariaceae). American Journal of Botany
host will increase future field performance and
58:218-228.
recommend using F. roemeri over E. lanatum Davies, D. M. and J. D. Graves. 1998. Interactions
as a host for C. levisecta recovery efforts. between arbuscular mycorrhizal fungi and the
• Herbivore management should be an integral hemiparasitic angiosperm Rhinanthus minor during
part ofrare hemiparasite recovery and manage- co-infection of a host. New Phytologist 139:
ment. Herbivore control may involve the same 555-563.
actions as prairie habitat management, such as Dobbins, D. R. and J. Kuijt. 1973. Studies on the
mowing or burning to reduce the accumulation haustorium ofCastilleja (Scrophulariaceae). I. The
ofthatch. Largefencescanbeerectedtoexclude upper haustorium. Canadian Journal of Botany
51:917-922.
ungulate browsers from an outplanting, while
small wire cages dug into the ground can Gibson, C. C. and A. R. Watkinson. 1989. The host
range and selectivity of a parasitic plant: Rhi-
prevent rodent grazing ofindividual plants. nanthus minor L. Oecologia 78:401^06.
AND . 1991. Host selectivity and the
Acknowledgments mediation ofcompetition by the root hemiparasite
Rhinanthus minor. Oecologia 86:81-87.
We would like to thank the USFWS Castilleja AND 1992. Theroleofthehemiparasitic
.
levisectaRecoveryTeam, InstituteforApplied Ecology, annual Rhinanthus minor in determining grassland
Jock Bealle, and Amy Bartow, for advice and logistical community structure. Oecologia 89:62-68.
MADRONO
158 [Vol. 55
GODT, M., F. Caplow, and J. L. Hamrick. 2005. Pearson, S. F. and P. Dunwiddie. 2006. Experimen-
Allozyme diversity in the federally threatened tal seeding and outplanting of golden paintbrush
golden paintbrush, Castilleja levisecta (Scrophular- {Castilleja levisecta) at Glacial Heritage, Mima
iaceae). Conservation Genetics 6:87-99. Mounds, and Rocky Prairie, Thurston County,
GoviER, R. N., M. D. Nelson, and J. S. Pate. 1967. WA. U.S. FishandWildlifeService,Olympia,WA.
Hemiparasitic nutrition in angiosperms. I. The Pennings, S. C, M. T. Nadeau, and V. J. Paul.
transfer of organic compounds from host to 1993. Selectivity and growth of the generalist
Odontites verna (Bell.) Dum. (Scrophulariaceae). herbivore Dolabella auricularia feeding upon com-
New Phytologist 66:285-297. plementary resources. Ecology 74:879-890.
Harley,J. L. ANDE. L. Harley. 1987. Acheck-Hstof PiCMAN, A. K. 1986. Biological activities ofsesquiter-
mycorrhiza in the British flora. New Phytologist pene lactones. Biochemical Systematics and Ecol-
105:1-102. ogy 14:255-281.
Heckard, L. R. 1962. Root parasitism in Castilleja. Press, M. C. 1989. Autotrophy and heterotrophy in
Botanical Gazette 124:21. root hemiparasites. Trendsin Ecology&Evolution
Howe, H. F. and D. Lane. 2004. Vole-driven 4:258-263.
succession in experimental wet-prairie restorations. 1998. Dracula or Robin Hood? A functional
Ecological Applications 14:1295-1305. rol.e of root hemiparasites in nutrient poor
Insightful. 2000. S-PLUS 2000 for Windows. In- ecosystems. Oikos 82:609-611.
WA.
sightful, Inc., Seattle, Salonen, v., M. Vestberg, AND M. Vauhnkonen.
Kaye, T. N. 2001. Restoration research for golden 2001. The effect ofhost mycorrhizal status on host
paintbrush {Castilleja levisecta), a threatened spe- plant-parasitic plant interactions. Mycorrhiza
cies. Institute for Applied Ecology, Corvallis, OR. 11:95-100.
AND B. A. Lawrence. 2003. Fitness effects of Schmidt, D. 1998. Restoration ofa prairie ecosystem
inbreeding and outbreeding on golden paintbrush
at the Yellow Island preserve and the propagation
{Castilleja levisecta): implications for recovery and of Castilleja hispida by vegetative cutting. M.S.
KuiJTrCo,erivJn.atlr1lo9id6su9,.ctOTiRho.ne.biIonlsotgiytuotfepafroarsitAicppflloiweedrinEgcopllaongtys,. Schntoehfeishdioess,rt,UpnlMiav.netrJ.sailatkynaldoofiFdW.saRbs.yhiSrnotgoettrompnia,rtazSs.eiatti1tc9l9eP,0e.dWiUAcpu.tlaarkies
University ofCalifornia Press, Berkeley, CA.
Lawrence, B. A. 2005. Studies to facilitate the Seel,speWc.iesE..PahnytdocMh.emCi.stPrryes29s.. 1993. Influence ofthe
reintroduction of golden paintbrush {Castilleja
host on three sub-Artie annual facultative root
tlheevsiiAsseN,ctDOa)rTe.tgoNo.nthKeStaWaytieel.lUan2mi0ev0te5tr.esiGtVrya,ollwCeioyrn,vgaOlrClaeissgt,oinlOl.Rej.Ma.fSo.r Sterhmeimtizp,araFs.iRte.s.anNdewG.PHhy.toHlaorgriisst.12159:871.31T-r1a3n8s.fer of
restoration and the garden. Rock Garden Quarter- pyrrolizidine and quinlizidine alkaloids to Castil-
ly 63:128-134. leja (Scrophulariaceae) hemiparasites from com-
Leung, L. R. andS. J. Ghan. 1999. PacificNorthwest positeand legumehostplants. Journal ofChemical
Ecology 13:1917-1925.
climate sensitivity simulated by a regional climate
model driven by a GCM. Part II: 2x C02 Swenerton, K. K. 2003. Soil suitability and site
simulations. Journal ofClimate 12:2031-2053. preparation techniques for Castilleja levisecta
Marksuol,phurMe.a:D.alk1a9l9o6i.d Hienmciorppaorraastiitoinsmanbdy hCearstbiilvloerjea rtheesstiosr.atUinoinveornsiWtyhiodfbWeayshIisnlgantdo,n,WaSsehaittnlget,oWn,A.M.S.
response. M.S. thesis, Colorado State University, USFWS. 2000. Recovery plan for golden paintbrush
Fort Collins, CO. {Castillejalevisecta). U.S. FishandWildlifeService,
Marvier, M. a. 1996. Parasitic plant-host interac- Portland, OR.
tions: plant performance and indirect effects on Wayne, W. C. 2004. Factors affecting the reintroduc-
parasite-feedingherbivores. Ecology 77:1398-1409. tion of golden paintbrush {Castilleja levisecta), a
her.b1i9v9o8rae.rAesmisitxaendcedioeftiampparroavseitsicpeprlfaonrt.maEnccoelaongdy Wtharsehaitnegnteodnp,laSnetattslpee,ciWesA..M.S. thesis, University of
79:1272-1280. Wentworth, J. B. 2001. The demography and
1998b. Parasite impacts on host communities: population dynamics of Castilleja levisecta, a
.
plant parasitism in a California coastal prairie. federally threatened perennial of Puget Sound
Ecology 79:2616-2623. Grasslands. Pp. 49-51 in R. S. Reichard, P.
AND D. L. Smith. 1997. Conservation implica- Dunwiddie, J. Gamon, A. Kruckeberg, and D.
tions of host use for rare parasitic plants. Salstrom (eds.). Conservation of Washington's
Conservation Biology 11:839-848. Native Plants and Ecosystems. Washington Native
Matthies, D. 1996. Interactions between the root Plant Society, Seattle, WA.
hemiparasite Melampyrum arvense and mixtures of Wilson, S. M. and A. B. Carey. 2001. Small
host plants: heterotrophic benefit and parasite- mammals in oak woodlands in the Puget Trough,
mediated competition. Oikos 75:118-124. Washington. Northwest Science 75:342-349.
. 1997. Parasite-host interactions in Castilleja WRCC. 2005. Western Regional Climate Center.
and Orthocarpus. Canadian Journal of Botany Available at: http://www.wrcc.dri.edu/index.html;
75:1252-1260. accessed 15 September 2005.