INFLUENCE OF HERBIVORES ON LIFE HISTORY STAGES OF A

DOMINANT NITROGEN-FIXING SHRUB CLUPINUS CHAMISSONIS):

EFFECTS ON SURVIVAL. GROWTH,

ARCHITECTURE, AND REPRODUCTION

by

Peter J. Warner

A thesis submitted to

Sonoma State University

in partial fulfillment of the requirements for the degree of

MASTER OF ARTS in

Eiolog~

Dr. J. Hall Cushman, Chai Copyright 2000 By Peter J. Warner

ii AUTHORIZATION FOR REPRODUCTION OF MASTER'S THESISIPROJECT

I grant pennission for the reproduction of this thesis in its entirety, without further authorization from me, on the condition that the person or agency requesting reproduction absorb the cost and provide proper acknowledgment of the authorship.

DATE: 7'-/{J-:?O(}O

Street Address

City, State. Zip

ill INFLUENCE OF HERBIVORES ON LIFE HISTORY STAGES OF A DOMINANT NITROGEN-FIXING SHRUB (LUPINUS CHAMISSONIS): EFFECTS ON SURVNAL, GROWTH, ARCIDTECTURE, AND REPRODUCTION

Thesis by Peter J. Warner

ABSTRACT Herbivores have diverse impacts on their host plants, affecting recruitment, survival, growth, architecture, fecundity, and other aspects of plant performance. Especially for longer-lived plant species, the effects of any single herbivore species can differ markedly throughout the life of the host plant. Additionally, a number of herbivore taxa may be influential at different life stages of the same plant species. To investigate the effects of herbivory by black-tailed deer (Odocoiietls hemionus columbianus) and snails (Helminthoglypta arrosa and Helix aspersa) on a dominant nitrogen-fixing shrub lupine, Lupinus chamissonis, I established three exclosure experiments, over a two-year period, in a sand dune community on the California coast. The first of these experiments investigated the effects of deer alone on the survival, growth, and architecture of individual plants from seedling through juvenile life stages. The second expeliment addressed the separate and joint effects of deer and snails on seedlings/juveniles. The third experiment assessed the impacts of deer alone on growth and reproduction in mature shrubs. Deer browsing significantly reduced volume in juvenile plants, and also significantly increased the architectural density (branching structure) of juveniles. Since I determined that plant volume was strongly correlated with aboveground dry biomass for lupines of all ages, I conclude that these volume reductions translate into losses of aboveground biomass. Protection from deer was positively correlated with leaf-galling by a cecidomyid fly (Dasineura lupinorwn), suggesting that deer indirectly affect another herbivore species in the system. Although deer did not significantly affect seedling survival, seedlings protected from deer consistently exhibited greater survival in the two separate experiments. Snails did not have a noticeable effect on seedling survival or on the growth of juvenile lupines, but environmental conditions may have contributed to this result. In mature shrubs, herbivory by deer significantly shortened shoots, yet did not substantially reduce biomass. Unprotected plants also produced significantly more inflorescences, raising the possibility that shrubs may compensate or even overcompensate in response to herbivory. . Collectively, these data indicate that mammalian and molluscan herbivores can have widely varying effects on their shared host plant, and that the effects of a vertebrate herbivore species change markedly with the life history stage of a host plant. Given the well-documented effects ofN-fixing species on soil fertility, my results suggest that black-tailed deer, through their effects on a dominant N-fixing shrub, may serve as key regulators of nitrogen availability in a nutrient-deficient dune .Chair: Si e

MA Program: Biology Sonoma State University Date: 7(<<1 0 [J

iv ACKNOWLEDGEMENT

I want to thank wholeheartedly the following Sonoma State University faculty for their support, guidance, and inspiration during the completion of this manuscript: Dr. 1. Hall Cushman, my thesis adviser, research colleague and collaborator, counselor, supporter, and chief motivator; Dr. Nathan E. Rank, for his statistical consultations, research guidance, and review of my thesis; Dr. Chris K. Kjeldsen, for his review of my thesis and longstanding support; and Dr. Elizabeth Dahlhoff, an original thesis committee member who posed challenging and insightful questions during my qualifying examination. I also want to acknowledge the particularly constructive suggestions and comments on my research offered by Dr. John L. Maron, and the natural history information and logistical support provided by Dr. Peter G. Connors and the staff at the Bodega Maline Laboratory. I also appreciate the time and expeltise provided by several ecologists visiting Sonoma State University and the Bodega Marine Laboratory; Drs. Michael Barbour, Eric Berlow, Valerie K. Brown, Sharon Collinge, Carla D' Antonio, Rodolfo DirL.o, Sharon Strauss, and Truman Young. Thank you also to Dr. Brian Jersky of Sonoma State University for his consultation on statistical methods. Fellow graduate students who provided invaluable commentaries on drafts of my thesis manuscript, as well as support and camaraderie, are Denise Cadman, Julia Clothier, Karen Gaffney, Sean McNeil, Darca Morgan, and Trisha Tierney. I particularly want to thank Sean McNeil for his capable assistance in my field work and more critically. in providing support on statistical analyses. Students who provided their time and energy to supporting my field activities were Joy Diaz, Jon­ Paul Hames, Derek Hoak, Jennifer Michaud, Emily Raab, Corrie Robb, Anna Sears, Mark Smith, and Simone Watts. Reviewers of my manuscript drafts included Drs.

1. Hall Cushman, Chris K. Kjeldsen, and Nathan E. Rank, as well as Vanessa K.

v Rashbrook and Etin Siska. Thank you all very much for your wisdom, insights, and contributions. I appreciate the financial support provided for my field work by the Bodega Field Conference, the California Native Plant Society, the National Geographic Society, and Sonoma State University. In closing, my sincere thank you to Professor Stephen J. Barnhart of Santa Rosa Junior College, for inspiting me with his passion for plants, the biological world, and for life itself.

vi TABLE OF CONTENTS

Chapter Page

I. Introduction...... , ...... ,...... 1

II. Study Site and OrganisIns ...... 3

III. Methods...... 5

IV. Results...... 9

V. Discussion...... ,...... 12

VI. Literature Cited .. ,...... 21

Appendix A: Figures ...... , ...... 26

Figure 1 ...... 27

Figure 2 ...... 28

Figure 3 ...... ~,., ... t ••• • , ..... •• ,,,,, ." •• ~t ••••• "t' ••••••• ~ .111 •••••• •••• • , ...... ".,29

Figure 4 ...... 30

Figure 5.."...... ".. ., ...... It •• 'f ...... " ...... "'.f •• '.~ ,•• "ftl •• ~ •••••• t' ...... , •• ~ r." 31 Figure 6 ...... 32

Figure 7 ...... 33

Figure 8 ...... 34

Appendix B: Tables ...... 35

Table 1 ...... 36

Table 2 ...... 37

vii INTRODUCTION

A wealth of studies provides compelling evidence that herbivores can have

diverse influences on individual plants, affecting survival, growth, architecture, tissue

chemistry, phenology, and fecundity (see reviews by Harper 1977; Crawley 1983,

1989, 1997; Strong et a1. 1984; Anderson 1987; Howe and Westley 1988;

Abrahamson 1989; Bryant et a1. 1991a; Whitham et a1. 1991; Coley and Barone

1996; Karban and Baldwin 1997; Tollrian and Harvell 1999). The majority of this

research documents that herbivores have negative effects on their host plants,

although a few suggest that plants compensate or even overcompensate in response

to herbivory (Paige and Whitham 1987; Maschinski and Whitham 1989; Whitham

et a1. 1991; Escos et aI. 1996; Alados et al. 1997; Sacchi and Conner 1999; but see

Bergelson and Crawley 1992a, 1992b; Bergelson et al. 1996). The direction and magnitude of plant responses to herbivory are closely linked to individual plant characteristics, such as size and phenology, as well as competitive interactions with other plants and abiotic factors (Maschinsld and Whitham 1989; Whitham et at.

1991).

The susceptibility and responses of plants to herbivory change during their lifetimes. In terms of mortality, seeds and seedlings are generally regarded as the most vulnerable life stages (Crawley 1989; Hanley et al. 1995, 1996; but see

Weltzin et a1. 1998). Although usually less apparent to herbivores than more established plants (Feeny 1976; Rhoades and Cates 1976; Crawley 1983), seedlings are often less well defended against herbivores and also have lower proportions of carbohydrate resources available for recovery from herbivory (Crawley 1983; sensu

Bryant et al. 1995). As plants mature, the impacts of herbivores typically shift from outright mortality to reduction in plant biomass and reproductive capacity (Crawley

1983,1989; Ehrlen 1995). However, to my knowledge, these generalizations have not been well substantiated by field studies that evaluate the effects of a single 2 herbivore species on different life stages of a host plant. Such studies are particularly important for long-lived plant species, in order to address the potentially confounding effects of site variation and multiple herbivores in the process of assessing age-specific responses to herbivory. During tmy given life stage of a host plant, vertebrate and invertebrate herbivores are likeJy to have considerably different effects (Crawley 1989; Strauss 1991; Hulme 1994). These two groups of herbivores differ substantially in size, metabolic rate, feeding pattern, mobility, and dietary preference (Crawley 1989; Lindroth 1989). Thus, the relative importance of these herbivores may shift appreciably from one life stage of a host plant to another. While seedlings and young plants are especially vulnerable to damage by invertebrates or smaller vertebrates (Crawley 1983, 1989; Hulme 1994; Hanley et al. 1995, 19%). mature plants, particularly long-lived perennials, may be more likely to sustain damage from larger vertebrates (Crawley 1983). One option for evaluating the impact of different herbivore species on plant performance is to study a host plant species that is attacked by both vertebrate and invertebrate herbivores. Through their direct effects on a host plant species, one herbivore species can indirectly alter the success of other herbivorous species using the same species (Faeth 1986; Strauss 1991). For example, numerous studies have shown that secondary plant metabolites, produced in response to feeding damage at one point in time, can function as induced defenses that reduce host plant quality for herbivores at a later time (Bryant et aI. 1985; Karban and Baldwin 1997; Agrawal 1999; Tonrian and Harvell 1999). My review of the literature indicates that most studies of plant-mediated indirect effects have focused on interactions between temporally separated insect herbivores. Nevertheless, a handful of studies consider plant­ mediated interactions between more distantly related taxa, such as mammals and invertebrates (Strauss 1991; Roininen et al. 1997; Suominen et al. 1999a, 1999b). 3

These studies have reported widely varying results: for example, Roininen et al. (1997) found that herbivory by moose indirectly benefited two hymenopteran herbivore species, whereas Strauss (1991) found that previous herbivory by two beetle species decreased the likelihood of future browsing by white-tailed deer. Herbivory by vertebrates and invertebrates is likely to have especially pronounced effects on ecological systems if they alter the survival, growth or reproduction of nitrogen-fixing host plants plants (Ritchie and Tilman 1995; Maron and Connors 1996; Ritchie et al. 1998; Maron and Jefferies 1999). This is because these taxa produce nitrogen-rich leaf litter that can increase the availability of this essential nutrient beneath their canopies (Vitousek et al. 1987; Vitousek and Walker 1989).

I designed this project to assess the effects of vertebrate and invertebrate herbivores on seedlings,juveniles, and reproductive-age individuals of a nitrogen­ fixing shrub, Lupinus chamissonis, in a sand dune ecosystem. Using three exclosure experiments, I have addressed the following questions: 1) Do vertebrate and invertebrate herbivores affect the survival, growth, or architecture of individual

Lupinus seedlings/juveniles? 2) Do the effects of vertebrate herbivores on Lupinus individuals vary with the life stage of this species? and 3) Do vertebrate and invertebrate herbivores of Lupinus interact with each other on their shared host plant species?

STUDY SITES AND ORGANISMS I conducted this study in a coastal hind dune system on Bodega Head,

California (380 191 N, 1230 3' W). This region has a Mediterranean-type climate, with 90% of the annual precipitation falling from October through April (Barbour et a1. 1973). The sandy soils of this system are fast-draining, nitrogen-poor, and strongly alkaline (Barbour et al. 1973; J. H. Cushman, unpublished data). Unlike many dune systems in California, this site is not dominated by Ammophila arenaria 4

(Poaceae; European beachgrass), and still supports a relatively diverse native flora. The dominant woody plants are the rapidly growing, nitrogen-fixing shrub Lupinus chamissonis (Fabaceae; silver bush lupine) and the slowly growing shrub Ericameria ericoicles (Asteraceae; dune mock heather), comprising 13% and 20% cover, respectively. Recmitment in these two species differs considerably (J. H. Cushman, P. 1. Warner, personal observation). Seedlings of Lupinus appear in profusion following substantial winter rains and subsequent warming of dune soils, while Ericameria recruitment appears to be more episodic. L. chamissonis responds quickly to seasonal precipitation, with rapid shoot elongation, and typically, more veltical development than horizontal when plants are young (P. 1. Warner, personal observation). Inflorescences are produced on new shoots, starting when plants are a few years old. Lupinus arboreus occurs sporadically in the dunes, but is much more abundant in nearby coastal prairie and bluff communities

(P. J. Warner and J. H. Cushman, personal observation). A considerable diversity of annual and perennial forbs and grasses also grows at this site, with the interstices among shrubs generally dominated by the exotic annual grass, Vulpia bromoides

(poaceae). Bodega Head supports a wide variety of vertebrate herbivores (Barbour et a1.

1973), of which black-tailed deer (Oclocoileus hemionus columbianus) and California hares, also known as black-tailed jackrabbits (Lepus califomicus) , are especiaUy common. Black-tailed deer feed on a variety of forbs (herbaceous dicots) and shrubs, including L. chamissonis and Eriogonum latifolium (polygonaceae), whereas hares graze primarily on grasses and forbs (1. H. Cushman and P. J. Warner, personal observation). Both herbivores deposit large amounts of nutrient­ rich feces and urine throughout this coastal dune. Other mammalian consumers in this system include brush rabbits (Sylvilagus bachmani), meadow voles (Microtus calijornica), deer mice (Peromyscus maniculatus), and striped skunks (Mephitis 5

mephitis). The dunes also feature at least two abundant mollusc species: the native snail Helminthoglypta arrosa and the non-native snail Helix aspersa. Both species are primruily active during the rainy season, feeding on a variety of perennial and annual plants.

In addition to mammals and monuscs, L. chamissonis hosts a number of insect herbivores, including root borers, miners of shoot tips, inflorescences and seeds, folivores, and stem- and leaflet-gallers, such as the galling midge Dasineura lupinorum (Diptera: Cecidomyiidae; 1. H. Cushman and P. 1. Warner, personal observation). METHODS Seedling Experiments

In late January 1996, I initiated a seedling experiment designed to assess the impacts of mammalian herbivores on the survival, growth, and architecture of Lupinus chamissonis juveniles. I selected 200 seedlings, each less than one month old, from throughout the largest contiguous population of L. chamissonis in the dunes. Each seedling was less than one month old, with two healthy cotyledons and no more than two true leaves. I randomly assigned them to either exposure to or protection from vertebrate herbivores. Exclosures consisted of cylindrical poultry­ mesh cages, 45 cm high x 20 cm in diameter, anchored into the sand with steel U­ shaped stakes. Periodically, I increased the sizes of cages to allow for increased plant size. These exclosures did not significantly alter ambient light levels, wind speed, air or soil temperatures (P. J. Warner, unpublished data). Every two to four weeks, from 15 February 1996 through 15 January 1998, I collected data on seedling survival, diameters, and heights. I measured the maximum height and three diameters of each plant's canopy to determine a mean radius, and estimated plant volume using the formula for the volume of a hemisphere 6

3 (V =2/3 1t r). Starting with the appearance of branches in late spring, I recorded the numbers of branches on each plant every four weeks.

In April 1997, I randomly selected five compound leaves on each surviving lupine seedling, and measured the lengths of the three longest leaflets of each leaf. I then pooled the leaflet lengths on each plant (15 leaflet lengths) to generate a mean leaflet length per seedling.

In 1997, I established a second seedling experiment to assess the joint effects of mammalian (OdocoUeus) and molluscan (Helminthoglypta and Helix) herbivores on lupines. Established in late January 1997, this experiment consisted of a two-way factorial design, with mammals (present or absent) and molluscs

(present or absent) as the grouping factors. I grouped 160 newly genninated seedlings into 40 blocks offour, with seedlings within blocks matched for physical proximity and microhabitat. I randomly assigned each plant to one of four possible treatments: mammalian herbivores excluded, molluscan herbivores excluded, both herbivores excluded, and both herbivores present. As in the 1996 study, I excluded vertebrate herbivores with cylindrical poultry-mesh cages (snails were not deterred by these cages; P. 1. Warner, unpublished data), To exclude snails, but allow access by all other herbivores, I placed Snail Barr® copper cylinders, 5 cm high x 12 em diameter, around seedlings. When the mucous membranes of molluscs come into contact with the copper, an electrical current is generated that repels the molluscs. Terrarium trials substantiated the effectiveness of this material as a mollusc deterrent (P. 1. Warner, unpublished data). PromPebruary 1997 through June 1997, I recorded survival, diameters, and heights of seedlings every two weeks, and every four weeks thereafter, through 15 January 1998. I estimated plant volumes and counted numbers of branches in the same fashion as in the 1996 seedling experiment. 7

Mature Shrub Experiment In April 1996, I established a third exclosure experiment involving reproductively mature L. chamissonis shrubs. This experiment consisted of a two­ way factorial design, with mammalian herbivores (present or absent) and block as the grouping factors. I grouped the 72 plants into 18 blocks of four, based on size similarity and physical proximity of plants. In each block, I assigned two shrubs to herbivore exclosures, with the other two plants unmanipulated to serve as controls. Exclosures consisted of 1 meter-high steel pOUltry-mesh fencing supported by bamboo stakes. I increased the size of exclosures to allow for shrub growth, as needed. In order to assess the effects of mammals on the growth of individual shoots, I randomly selected branches in each of four canopy quarter-sections of shrubs. These branches had a minimum length of 1 em, and were approximately 10, 60, and 110 cm from the center of a shrub, and at the shrub perimeter. Smaller shrubs had at least eight marked shoots, and the larger shrubs a maximum of 16. Every six months from May 1996 through January 1998, I calculated plant volume, as in the seedling experiments, and measured the length of marked branches. Beginning with the first appearance of flowers in March 1997, I counted the numbers of inflorescences on these plants approximately every two weeks for ten weeks. In June 1997, I collected and dried (at 600 C) up to 20 fruits harvested from each plant, and counted and weighed all seeds, pooling the fruits from each plant to establish mean seed weights. Biomass Assessments While biomass can be an excellent predictor of reproductive success in plants (Crawley 1983), biomass sampling is destructive, and prevents the gathering of further data. Since I was interested in long-term lupine responses to herbivory, I evaluated the possibility that plant volume was a valid and non-destructive predictor of biomass. To determine the 8 relationship between plant volume and aboveground biomass, I selected 25 lupine plants of various sizes from the dune population in June 1998. I first determined plant volumes as outlined previously, then cut the primary stem of each plant at ground level. I removed accumulated sand and detritus from lower branches, dried the plants at 60° C for 72 hours, and weighed them.

Abundance of Leaflet Galls

In April 1997, in order to assess the effects of mammalian herbivores on Dasineura lupinorum, a common leaflet galler of L. cliamissonis, I counted the numbers of galls on each juvenile plant survi ving from the 1996 seedling expeliment. In late May and early June 1996, I censllsed 34 non-experimenta11upine plants of varying sizes throughout the dune site to determine the relationship between shoot length and abundance of Dasineura galls. This was accomplished by randomly selecting five shoots per plant, measuring their lengths, and counting the number of Dasineura galls on those shoots. I then calculated the mean shoot length and mean number of galls for each of the 34 shrubs. Statistical Analyses

I performed all statistical analyses using the JMP 3.1 program (SAS

Institute). In the 1996 seedling experiment, I analyzed data on lupine survival, volume, and branch density using one-way ANOVAs, with the presence or absence of mammalian herbivores as the single grouping factor. I similarly analyzed survival data from the 1997 seedling experiment, except that I used two-way ANOV As, with mammalian herbivores (present or absent) and molluscs (present or absent) as the grouping factors. Seedling mortality, and the resulting loss of degrees of freedom, precluded using block as a grouping factor for these analyses. For all plants in these two experiments, I calculated the area beneath the curves generated by plotting each of the three dependent variables against time. I then used these areas (one per plant) as response variables in the one- or two-way ANOVAs. For seedling volume and branch density data, I log-transformed the 9 response variables to correct for heterogeneity of variances. This analysis was ideal because it allowed me to evaluate how herbivores influence seedling survival, volume, and branch density through time, without violating the assumptions of independence that are central to ANOVA. In the mature shrub experiment. I analyzed data on shrub volume and shoot length using two-way ANOVAs, with mammalian herbivores (present or absent) and block (1-18) as the grouping factors. As with the seedling data, I used areas beneath the curve as response variables for shrub volume and shoot length. I log­ transformed both shrub volume and shoot length data to correct for heterogeneity of variances. I collected data on number of inflorescences and seed weight only once, and thus used the actual values as response variables in the ANaVA analyses rather than the areas beneath curves. In the 1996 seedling experiment, I analyzed data on lupine leaflet lengths and number of Dasineura galls using one-way ANOV As. Because I measured these variables just once each. without a time dimension, I used the actual values as the response variables, rather than the areas beneath the curves, as discussed above. I used linear regression analysis to evaluate the relationship between 1) shrub volume and aboveground dry biomass, and 2) shoot length and insect gall abundance. For both, I log-transfonned data to correct for heterogeneity of variances. RESULTS Seedling Experiments In both 1996 and 1997, seedlings protected from mammals survived in higher proportions than unprotected seedlings (Figure 1). Although the effect was not significant in either year (Table 1A), the difference in survival between treatments was consistent over two seasons. In the 1997 experiment, molluscs did 10 not have a significant effect on seedling survival, nor were any of the interaction terms involving molluscs significant (Table lA). In contrast to the trend towards decreasing seedling survival, mammalian herbivores had a large and significantly negative effect on seedling volume in both the 1996 and 1997 experiments (Figure 2; Table IB). After two years of growth, the 1996 seedlings protected from mammalian herbivores were, on average, over five times larger in volume than those exposed to mammals. In the 1997 seedling experiment, I observed a similar pattel11, with protected seedlings more than foUl' times larger than their uncaged counterparts after one year. Molluscs did not have a significant impact on seedling volume in 1997, nor was the interaction between molluscs and mammals significant (Table IB). Mammalian herbivory significantly increased branch density in both 1996 and 1997 seedling experiments (Figure 3; Table lC). After 20 months of growth, the 1996 seedlings exposed to mammalian browsing had a four-fold greater branch density than protected plants. After 11 months, unprotected seedlings in the 1997 experiment had, on average, more than three times the branching density of protected plants. Thus, mammalian herbivores had the effect of increasing branch density in both seedling experiments. As a result, protected plants generally had a more open architecture than unprotected plants. Unlike the dramatic impact of mammals on the architecture of the 1997 seedlings, molluscs had only a subtle, and not significant, effect on branch density in young Lupinus plants (Figure 3; Table IC). Mammals had a significantly negative effect on mean leaflet size in the 1996 seedlings, as measured in April 1997, when the plants were about 15 months old (Figure 4; Table ID). Protected juvenile plants had leaflets about 20% greater in average length than unprotected plants. 11

Mature Shrub Experiment

Unlike the clearly negative effects of mammalian herbivores on L chamissonis seedlings and juveniles, results from the mature shrub experiment are more ambiguous (Figures 5 & 6; Table 2). Mammals did not have a significant effect on the mature shrub volumes, as measured from May 1996 through

December 1997 (Figure 5A; Table 2A). During this period, shrubs protected from mammalian herbi vores increased an average of 184% in volume, while unprotected plants increased by 152%. While the blocking factor alone was highly significant, the mammals x block interaction term was not.

During the same time period, mammals had a significantly negative effect on the lengths of individual shoots on the same shrubs (Figure 5B; Table 2B). Plants protected from mammalian herbivores had longer shoots than those left exposed to mammals: shoots on protected plants increased in mean length by 442%, while shoots on unprotected shrubs increased by 383%. Again, the block effect was significant, but the mammals x block tenn was not significant (Table 2B).

Mammals had a significant and positive effect on the numbers of inflorescences produced by shrubs in the spring of 1997 (Figure 6; Table 2C).

Plants exposed to herbivores had, on average, 1.5 times the number of inflorescences than protected shrubs. The mammal x block term was also significant (Table 2C).

Mammals did not have a significant effect on mean seed weights (Figure 6;

Table 2D). As with other parameters in the mature shrub experiment, seed weight varied significantly according to block (Table 2D). However, the mammals x block interaction term was not significant (Table 2D).

Biomass Assessment

As shown in Figure 7, linear regression analysis revealed that aboveground dry biomass of L. chamissonis increased significantly with increasing plant volume ,

12

• )= (y =-4.59 + 0.92x; R2;:: 0.95; FI 2 414.06; p < 0.0001). Since plant volume predicted 95% of the vmiation in plant biomass, non-destructive measurement of plant volume was an accurate substitute for destructive sampling of aboveground biomass.

Abundance of Leaflet Galls

As shown in Figure 4 and Table IE, plants protected from mammalian herbivores had significantly more Dasineura galls on leaflets than did unprotected plants. In addition, for non-experimental plants, regression analysis indicated that the abundance of Dasineura galls increased significantly with shoot length (Figure 2 8~ y;:: -1.82 + 1.48x~ R =0.55; FI.32;:: 38.82; p < 0.0001). DISCUSSION

In this study, I have presented experimental evidence indicating that mammalian herbivores have substantial influences on a nitrogen-fixing shrub, Lupinus c/utmissonis, in a nitrogen-poor dune environment. In general, this work has demonstrated that these herbivores significantly reduce volume and change the architecture of younger plants, and influence the use of lupines by an invertebrate herbivore species. In contrast, mammalian herbivores did not significantly affect volume in mature shrubs, but reduced individual shoot lengths and increased inflorescence production. These results suggest that the effects of herbivores on lupines are strongly life stage-specific. Since Lupinus is associated with increased pools of plant-available nitrogen in dune soils (1. H. Cushman. unpublished data), these herbivores may also be important mediators of soil fertility in this system.

Determining the identity of mammalian herbivores that feed on L chamissonis is critical in order to evaluate the results generated from my exclosure experiments. In this study system, the candidates are black-tailed deer (Odocoileus), California hares (Lepus), brush rabbits (Sylvilagus), skunks (Mephitis), voles (Microtus), and deer mice (Peromyscus). The small size of both Microtus and 13

Peromyscus eliminates them from consideration, given that they could easily pass through the exclosure fencing used in this study (J. H. Cushman, unpublished data). However, the exclosure cages were capable of excluding the four remaining herbivore species. Brush rabbits are not particularly abundant in the Bodega Head dune system (1. H. Cushman, personal observation), and skunks are omnivores that feed pIimm'ily on invertebrates (Jameson and Peeters 1988). Based on extensive observations and physical evidence, I hypothesize that deer, not hares, are the herbivores primarily responsible for the observed impacts on Lupinus. Four lines of evidence SUppOlt this view. First, scat produced by hares and brush rabbits is almost entirely comprised of graminoid fragments, whereas that of deer is not (J. H. Cushman, personal observation). Second, much of the browsing damage sustained by lupine shrubs was on shoots well out reach of hares and rabbits. Third, on multiple occasions, I have observed deer feeding on lupines and often found their hoofprints around recently browsed plants. Lastly, in a separate field study, I found that lupine shrubs initially protected from herbivory for 15 months were severely damaged by both leaf-stripping and the removal of entire shoots, once the cages were removed (P. 1. Warner, unpublished data). These types of herbivory are characteristic of the feeding damage inflicted by large ungulate browers in other systems (Danell et a1. 1994; Mulder and Harmsen 1995). In summary, these observations strongly suggest that black-tailed deer are the primary mammalian herbivores of L. chamissonis in this dune system. Deer had a dramatically negative impact on volume in juvenile lupines (Figure 2). Because volume and aboveground biomass are so strongly correlated (Figure 7), the decreases in plant volume shown experimentally can be readily translated into reductions in aboveground biomass. Deer also had a significantly negative effect on leaflet size (Figure 4), a result that should further contribute to overall biomass loss. Although not addressed in these experiments with juvenile 14

lupines, reduction in biomass has often been shown to influence reproduction negatively, through delayed or decreased flowering (Crawley 1983; Mulder and Harmsen 1995), decreased vegetative reproduction (Bonser and Reader 1995), and reduced fruit and seed production (Crawley 1983; Hendrix 1988; Bonser and Reader 1995; Ehrlen 1995; Louda and Potvin 1995). Herbivore damage to shoots can also delay or alter sexual expression (Whitham and Mopper 1985; Juenger and Bergelson 1997). If such effects occur in this dune system, then seedling recruitment could be substantially delayed or reduced, with subsequent impacts on population size and dynamics. In addition to their effect on plant volume, deer substantially altered the architecture of juvenile shrubs (Figure 3). The increased branch density of unprotected lupines was a joint consequence of reduced plant volume (Figure 2) and an increase in the number of lateral branches (P. J. Watner, unpublished data). In most plant species, the removal of apical meristems releases lateral meristems from apical dominance and results in the growth of lateral branches (Crawley 1983; Whitham and Mapper 1985). Thus, browsed plants can respond by growing from lateral or apical meristems that are undamaged or inaccessible to herbivores (Crawley 1983; Marquis 1996). Some studies have found that an increase in the density of woody branches may deter further browsing in the short term (Belovsky and Schmitz 1994; Mulder and Harmsen 1995). However, others have found that herbivores commonly return to the same plants repeatedly (Bryant et al. 1985; Danell et a1. 1994). These results suggest that herbivores can induce architectural changes in plants that influence the timing of future browsing, which is contingent upon the recovery of palatable tissues in the host plants. An unexpected result from this study is that two distantly related taxa - deer and a cecidomyid fly - interact with each other on their shared host plant. As shown in Figure 4B, I found greater numbers of Dasineura galls on juvenile plants 15 protected from deer than on unprotected plants. In addition, I determined that regardless of browsing history, the number of cecidomyid galls on shrubs was positively correlated with lupine shoot length (Figure 8). Although I lack data on this subject, three hypotheses may explain, individually or collectively, the negative effects of deer browsing on gall abundance. First, by feeding on branch tips, deer may inadveltently consume insect galls, thus reducing gall abundance on unprotected plants. Second, given that protected plants had larger leaflets (Figure 4A), the gall-forming flies may either choose to oviposit on larger leaflets or their offspring have a greater chance of surviving on them (see Price 1997). Third, deer herbivory during the winter months may induce defensive responses in lupines, such as increased alkaloid content in leaves or an increased number of spicules on leaves, which may in turn reduce plant quality for the cecidomyid (see Harrison and Karban 1986; Johnson and Bentley 1988, 1991; Harrison 1994). I suspect that all three mechanisms operate in this lupine system, and that preferences for long shoots and induced defenses are especially important. The seedling stage is a vulnerable period for most plants, and herbivores commonly reduce seedling survival (Harper 1977; Crawley 1983; Hulme 1994; Ehrlen 1995; Hanley et a1. 1995, 1996; Crawley 1997). However, neither of my seedling experiments (1996 or 1997) demonstrated that protected lupines survived in significantly greater numbers (Figure 1). Nevertheless, the subtle negative trend towards increased mortality in unprotected juvenile plants, considered in concert with the more pronounced effect on juvenile plant biomass, suggests that mammalian herbivores may exhibit a plant size-dependent response in their choices of forage. In both experiments, when plants were approximately five months old (late spring to early summer), the survival rate of unprotected plants continued to fall while the rate among protected plants stabilized. This pattern may be related to a seasonal shortage of alternative forage for deer, given that lupines are one of the few 16

food resources available in the dune environment dUling the dry summer and early autumn months. Alternatively, the rapid growth of young lupines may render them

more apparent to mammalian herbivores. Under this scenario, plant size could be a crucial determinant for the timing of initial mammalian browsing on lupine juveniles,

as demonstrated by Hulme (1994) in English grasslands.

Snails and slugs have decidedly negative impacts on seedling survival in

many plant species (Hulme 1994; Ehrl6n 1995; Hanley et a1. 1995,1996; Crawley

1997), and snails abound in my dune study system during moist, cool peliods of the year (generally November - March). However, seedling mortality was not significantly affected by mollusc herbivory in my experiment. The late winter and

entire spring of 1997 was a relatively dry peliod, with only about 18.5 cm of rain

falling from February 1 through June 30, compared with 50.8 cm for the same

period in 1996. Snails were much less abundant in the dunes during spdng 1997

compared with the previous year (p. J. Warner, personal observation), and I suspect that drought was a critical factor in reducing mollusc activity. Due to the low

precipitation levels in 1997, an evaluation of the importance of snail herbivory in this lupine system is difficult. They may be extremely impoltant in wet years, but were

not influential in a dry one. Mature lupines exhibited markedly different responses to mammalian herbivores than juvenile plants. First, deer did not significantly reduce the volume of mature shrubs (Figure SA). While I can offer a number of plausible explanations for this result, none alone is sufficiently sound. Volume differences between browsed and unbrowsed plants may have been undetected because 1) deer did not browse any of these larger individuals, 2) my methods were too coarse for detecting volume change in larger plants, 3) deer browsed mature shrubs less intensively than juvenile plants, 4) browsed plants compensated for tissue losses, or 5) the length of time between volume measurements obscured actual short-term tissue losses. I can 17

refute the first hypothesis, having demonstrated that protected plants had significantly longer shoots than unprotected individuals (Figure 5B). The second hypothesis is also not supported by my data: although tissue losses in larger shrubs are likely to be disproportionately small compared with their volumes, my methods readily detected volume changes in all plants, regardless of plant size or treatment. The remaining hypotheses can be considered collectively in order to demonstrate the complexity of plant responses to herbivory (see Maschinski and Whitham 1989; Whitham et a1. 1991; Danell et al. 1994; Bonser and Reader 1995; Jaremo et a1. 1996). Browsing intensity and timing are directly related to the feeding behaviors of herbivores, as well as to plant location, size and growth, and indirectly, to many environmental factors. Larger mammalian browsers tend to be browsing generalists (Crawley 1989; Danell et a1. 1994), and larger woody plants are less likely to sustain proportionately large amounts of tissue loss and can also physiologically compensate. The capability of plants to compensate for tissue losses and the rate of compensation are also influenced by environmental factors (Whitham et aI. 1991),

as well as by the continued risk of browsing (Jaremo et a1. 1996). Thus, if browsing caused only minor changes in volume in larger shrubs, they may have been able to compensate with substantial regrowth that quickly replaced lost tissues. Figure 5B shows that lupine shoots elongate rapidly regardless of treatment, and several studies have demonstrated that plants exposed to herbivores can compensate for individual shoot reductions (paige and Whitham 1987; Paige 1992; Danell et al. 1994; Marquis 1996). Therefore, the ability of plants to compensate for tissue losses can ameliorate short-tenn effects of browsing, unless the effects of repeated bouts of browsing are measured over extended periods of time (see Sacchi and

Connor 1999). Although this study was not designed specifically to address the degree of plant compensation in response to herbivory, one of its most striking results is that 18 unprotected mature shrubs produced significantly greater numbers of inflorescences than protected lupines (Figure 6). Since herbivores did not have a negative effect on seed weight (Figure 6) or on numbers of seeds per fruit (P. J. Warner, unpublished data), an increase in the numbers of inflorescences could be interpreted to SUppOlt the overcompensation hypothesis. According to this controversial hypothesis (see Bergelson and Crawley 1992a, 1992b; Bergelson et a1. 1996), plants achieve greater reproductive success through herbivore~induced changes that result in increased flower and fruit production (paige and Whithrun 1987; Maschinski and Whitham

1989; Whitham et a1. 1991; Paige 1992; Escos et a1. 1996). Since L. chamissonis produces inflorescences on cUlTent-year shoot tissue (p. J. Warner, personal observation), browsing can stimulate floral production through the removal of apical meristems and the growth of multiple new shoots from lateral medstems. Although my data on compensation are suggestive, such short-tenn effects do not necessarily result in greater long-term reproductive success. This is a particularly important consideration in longer-lived perennials for which the impacts of herbivory on reproductive fitness may take several years to ascertain (Sacchi and Connor 1999). For instance, over several years, browsed plants could produce fewer flowers than those protected from herbivores. In addition, browsing could stimulate the production of inflorescences with fewer flowers than those of unbrowsed branches or plants. In light of these unresolved issues, while my data are consistent with the overcompensation hypothesis, they do not provide irrefutable support. By influencing the survival, growth or reproduction of dominant nitrogen-fixing plants, herbivores can indirectly alter the fertility of soils in areas they frequent (Ritchie and

Tilman 1995; Maron and Connors 1996~ Ritchie et al. 1998; Maron and Jefferies 1999; Sirotnak and Huntly 2000). Nitrogen-fixing taxa produce nitrogen-rich leaf litter that increases the availability of soil nutrients beneath their canopies (Vitousek et a1. 1987; Hobbie 1992). Herbivore-mediated reductions in lupine growth or abundance should lead 19 to reduced deposition of nitrogen-rich litter and thus, reduced supplies of plant-available nitrogen in the soiL Such effects should be especially important in sand dune environments, which are known for their fast-draining, nutrient-poor soils (Barbour et a1. 1973; Kachi and

Hirose 1983; J. H. Cushman, unpublished data), Although not quantified in this study, I hypothesize that black-tailed deer indirectly reduce nitrogen availability in dune soils by altering growth and litter deposition by Lupinus chamissonis. Three lines of evidence support this view. First, Cushman et a1. (unpublished manuscript) have presented non­ experimental evidence from both 1997 and 2000 indicating that the soil underneath lupine shrubs is associated with significantly higher levels of ammonium and nitrate than adjacent shrub-free patches. Second, I have presented experimental data revealing that deer significantly reduce shrub volume and biomass in juvenile plants (Figures 2 and 7). Third, McNeil and Cushman (unpublished dala) found that lupine shrubs exposed to deer herbivory had significantly less leaf litter as well as reduced amounts of plant-available nitrogen in the soil underneath their canopies than plants protected from these herbivores. The findings summarized in this paper demonstrate the inherently variable impacts of vertebrate and invertebrate herbivores on mUltiple life history stages of a nitrogen-fixing plant species at a single study site. Deer consistently, if subtly, increased the rate of seedling mortality: while statistically insignificant in each of two successive years, this influence on lupine survival is neither negligible nor inconsequential if considered from a cumulative perspective. As young plants developed, the effects of deer browsing became more pronounced, resulting in significant reductions in aboveground biomass and changes in plant architecture. In contrast, molluscs appeared to have little impact on seedling survival or juvenile plant growth, although a reliable assessment of molluscan impacts was confounded by dry weather patterns. Deer browsing had less clear-cut effects on the growth of mature shrubs, significantly decreasing shoot lengths but not affecting shrub biomass. However, feeding by deer increased inflorescence production, thereby providing evidence consistent with overcompensation, at least as a short-term plant response. Deer 20 browsing on lupines also appeared to influence the use of shoots and leaves by a galling cecidomyid fly, an indication that larger mammalian herbivores, through direct impacts on their host species, may also have indirect impacts on other herbivores in this system. While not the focus of this particular study, the effects of deer browsing on lupine growth and fecundity may have impoltant ramifications for the longer-tetm reproductive capabilities of individual plants, and consequently, for the size and distribution of the lupine population. Finally, because Lupinus chamissonis is known to increase nitrogen pools in this dune environment, deer may be important indirect mediators of soil fertility. 21

LITERATURE CITED Abrahamson, W. G. 1989. Plant-Animal Interactions. McGraw-Hill Book Company, New York,NY.

Agrawal, A. A. 1999. Induced responses to herbivory in wild radish: effects on several herbivores and plant fitness. Ecology 80: 1713-1723.

Alados, C. L., F. G. Barroso, A. Aguirre, and J. Escos. 1996. Effects of early season defoliation of Anthyllis cytisoides (a Mediterranean browse species) on further herbivore attack. Jou111al of Arid Environments 34:455-463. Anderson, D. C. 1987. Below-ground herbivory in natural communities: a review emphasizing fossorial animals. Quarterly Review of Biology 62:261-286.

Barbour, M. J., R. B. Craig, F. R. Drysdale, and M. T. Ghiselin. 1973. Coastal Ecology. University of California Press, Berkeley. CA, USA. Belovsky, G. E. and O. J. Schmitz. 1994. Plant defenses and optimal foraging by mammalian herbivores. Jou111al of Mammalogy 75:816-832. Bergelson, J. M., and M. J. Crawley. 1992a. Overcompensation in response to mammalian herbivory: the disadvantage of being eaten. American Naturalist 139:870-882. _. 1992b. The effects of grazers on the petformance of individuals and populations of scarlet gilia, /pomopsis aggregata. Oecologia 90: 435"444. Bergelson, J. M., T. Juenger, and M. J. Crawley. 1996. Regrowth following herbivory in Ipomopsis aggregata: compensation but not overcompensation. American Naturalist 148: 744-755. Bonser, S. P. and R. J. Reader. 1995. Plant competition and herbivory in relation to vegetation biomass. Ecology 76:2176-2183. Bryant, I. P., G. D. Wieland, T. Clausen, and P. Kuropat. 1985. Interactions of snowshoe hare (Lepus americanus) and feltleaf willow (Salix alaxensis) in Alaska. Ecology 66:1564-1573.

Bryant, I. P., F. D. Provenza, J. Pastor, P. B. Reichardt, T. P. Clausen, and J. T. du Toit. 1991 a. Interactions between woody plants and browsing mammals mediated by secondary metabolites. The Annual Review of Ecology and Systematics 22:431446. Bryant, I. P., I. Heitkonig, P., Kuropat, N., and N. Owen-Smith. 1991b. Effects of severe Defoliation on the long-term resistance fo insect attack and on leaf chemistry in six woody species of the southern Mrican savanna. American Naturalist 137:50-63. Bryant, J. P., and R. Iulkunen-THtto. 1995. Ontogenic development of chemical defense by seedling resin birch: energy cost of defense production. Ecology 21 :883-896.

Coley, P. D., and J. A. Barone. 1996. Herbivory and plant defenses in tropical forests. Annual Review of Ecology and Systematics 27:305-335. Crawley, M. J. 1983. Herbivory: Dynamics of Plant-Animal Interactions. University of California Press, Berkeley. 22

__. 19~9. The rel~tiv~ importance of vert~brate and invertebrate herbivores in plant population dynamlcs. /,n E. A. Bemays, editor. Insect-Plant Interactions, Vol. 1, pp. 45­ 71. eRC Press, Boca Raton, FL. __. 1997. Plant Ecology. Blackwell Science Ltd., Oxford, UK. Danell, K, R. Bergstrom, and L. Edenius. 1994. Effects of large mammalian browsers on architecture, biomass, and nutrients of woody plants. Journal of Mammalogy 75:833­ 844.

Escos, 1., F. O. Ban'oso, C. L. Alados, and L. Garcia. 1996. Effects of simulated herbivory on reproduction of a Mediterranean semi-arid shrub (Anthyllis cytisoides L.). Acta Oecologica 17:139-149.

Ehrien,1. 1995. Demography of the perennial herb Lathynts vernus. 1. Herbivory and individual peiforrnance. Journal of Ecology 83:287-295. Faeth, S. H. 1986. Indirect interactions between temporally separated herbivores mediated by the host plant. Ecology 67:479-494. Feeny, P. 1976. Plant apparency and chemical defense. Recent Advances in Phytochemistry 10: 1-40. Hanley, M. E., M. Fenner, and P. J. Edwards. 1995. The effect of seedling age on the likelihood of herbivory by the slug Deroceras reticulatum. Functional Ecology 9:754­ 759. __. 1996. The effect of mollusc grazing on seedling recruitment in artificially created grassland gaps. Oecologia 106:240-246. Harper, J. L. 1977. Population biology of plants. Academic Press, London, UK. Harrison S. 1994. Resources and dispersal as factors limiting a population of the tussock moth (Orgyia vetusta), a flightless defoliator. Oecologia 69:354-359.

Harrison, S., and R. Karban 1986. Effects of an early-season folivorous moth on the success of a later-season species, mediated by a change in the quality of a shared host Lupinus arboreus. Oecologia 69:354-359. Hendrix, S. D. 1988. Herbivory and its impact on plant reproduction. Pages 246-263 in 1. L. Doust and L. L. Doust, editors. Plant Reproductive Ecology. Oxford University Press, Oxford, England. Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336-339. Howe, H. F., and L. C. Westley. 1988. Ecological Relationships of Plants and Animals. Oxford University Press, New York. NY. Hulme, P. E. 1994. Seedling herbivory in grassland: relative impact of vertebrate and invertebrate herbivores. Journal of Ecology 82:873-880. 23

Jameson, E. W., and H. 1. Peeters. 1988. CalifomiaMammals. University of California Press; Berkeley.

Jaremo, 1., P. Nilsson, and 1. Tuomi. 1996. Plant compensatory growth: herbivory or competition? Oikos 77:238-247.

Johnson, N. D., and B. L. Bentley. 1988. The effects of dietary protein and lupine alkaloids on the growth and survivorship of Spodoptera eridania. Journal of Chemical Ecology 14:1391-1403.

Johnson, N. D., and B. L. Bentley. 1991. Symbiotic N2- fixation and the elements of plant resistance to herbivores: lupine alkaloids and tolerance to defoliation. Pages 45-63 in Barbosa P., Ktischik. V., Jones, C., editors. Microbial Mediation of Plant-Herbivore Interactions. John Wiley and Sons, Inc., New York, NY. Juenger, T .• and 1. Bergelson. 1997. Pollen and resource limitation of compensation to herbivory in scarlet gilia, Jpomopsis aggregata. Ecology 78:1684-1695. __. 2000. Does early season browsing influence the effect of self-pollination in scarlet gilia? Ecology 81:41-48. Kachi, N., and T. Hirose. 1983. Limiting nutrients for plant growth in coastal sand dune soils. Journal of Ecology 71:937-944. Karban, R. and L T. Baldwin 1997. Induced Responses to Herbivory. The University of Chicago Press, Chicago, IL.

Lindroth, R. L. 1989. Mammalian herbivore-plant interactions. Pages 163-206 in W. B. Abrahamson, editor. Plant-Animal Interactions. McGraw-Hill, Inc., New York, NY.

Louda, S. M., and M. A. Potvin. 1995. Effect of inflorescence-feeding insects on the demography and lifetime fitness of a native plant. Ecology 76: 229-245.

Maron, J. L., and P. O. Connors. 1996. A native nitrogen-fixing shrub facilitates weed invasion. Oecologia 22:302-312. Maron, J. L., and R. L. Jefferies. 1999. Bush lupine mortality, altered resource availability, and alternative vegetation states. Ecology 80:443-454. Marquis, R. J. 1996. Plant architecture, sectoriality and plant tolerance to herbivores. Vegetatio 127: 85-97. Maschinski, J., and T. O. Whitham 1989. The continuum of plant responses to herbivory: the influence of plant association, nutrient availability, and timing. The American Naturalist 134:1-19. Mulder, C. P. H, and Harmsen, R. 1995. The effect of muskox herbivory on growth and reproduction in an Arctic legume. Arctic and Alpine Research 27: 44-53.

Paige, K. N. 1992. Overcompensation in response to mammalian herbivory: from mutualistic to antagonistic interactions. Ecology 73:2076-2085. Paige, K. N., and T. O. Whitham..1987. Overcompensation in. response to mammalian herbivory: the advantage of bemg eaten. Amencan NaturalIst 129:407-416. 24

Price, P. W. 1997. Insect Ecology. John Wiley and Sons! New York, NY, USA.

Rhoades, D. F., and R O. Cates 1976. Toward a general theory of plant antiherbivore chemistry. Recent Advances in Phytochemistry 19: 168-213. Ritchie, M. E. and D. Tilman. 1995. Responses of legumes to herbivores and nutrients during succession on a nitrogen-poor soil. Ecology 76:2648-2655.

Ritchie, M. E., D. Tilman, and 1. M. H. Knops. 1998. Herbivore effects on plant and nitrogen dynamics in oak savanna. Ecology 79: 165-177. Roininen H., P. W. Price, and 1. P. Bryant. 1997. Response of galling insects to natural browsing by mammals in Alaska. Oikos 80:481-486. Sacchi, D. F., and E. F. Connor. 1999. Changes in reproduction and architecture in flowering dogwood, Comus florida, after attack by the dogwood club gall, ResseliellCl clavula. Oikos 86:138-146. Sirotnak, 1. M. and N. 1. Huntly. 2000. Direct and indirect effects of herbivores on nitrogen dynamics: voles in riparian areas. Ecology 81:78-87. Strauss, S. Y. 1991. Direct, indirect, and cumulative effects of three native herbivores on a shared host plant. Ecology 72:543-558.

Strong, D. R, 1. H. Lawton, and R. Southwood. 1984. Insects on Plants. Harvard University Press, Cambridge, MA.

Suominen, 0., K. Danell, and R. Bergstrom. 1999a. Moose, trees, and ground-living invertebrates: indirect interactions in Swedish pine forests. Oikos 84:215-226.

Suominen, 0., K. Danell, and J. P. Bryant. 1999b. Indirect effects of mammalian browsers on vegetation and ground-dwelling insects in an Alaskan floodplain. Ecoscience 6:505­ 510.

Tollrian, R. and C. D. Harvell 1999. The Ecology and Evolution of Inducible Defenses. Princeton University Press, Princeton, NJ.

Vitousek, P. M., L. R Walker, L. D. Whiteaker, D. Mueller-Dombois, and P. A. Matson. 1987. Biological invasion by Myricafaya alters ecosystem development in Hawaii. Science 238:802-804.

Vitousek, P. M. and L. R Walker. 1989. Biological invasion by Myricafaya in Hawai'i: plant demography, nitrogen fixation, ecosystem effects, Ecological Monographs 59:247-265.

Weltzin, J. F., S. R. Archer, and R K. Heitschmidt. 1998. Defoliation and woody plant (Prosopis glandulosa) seedling regeneration: potential vs. realized herbivory tolerance. Plant Ecology 138:127-135. Whitham, T. G., and Mopper, S. 1985. Chronic herbivory: impacts on architecture and sex expression of pinyon pine. Science 228: 1089-1091. 25

Whitham, T. G., J. Maschinski, K. C. Larson, and K. N. Paige. 1991. Plant responses to herbivory: the continuum from negative to positive and underlying physiological mechanisms. Pages 227-256 in P. W. Price, T. W. Lewinsohn, O. W. Fernandes, and W. W. Benson, editors. Plant-Animal Interactions: Evolutionary Ecology in Tropical and Temperate Regions. John Wiley and Sons, Inc., New York, NY.

26

APPENDIX A: Figures 27

. 1996 1.0 --0- - Mammals --e-- + Mammals

0.8

0.6

0.4

0.2 ......

B. 1997 1.0 --0­ - Mammals, - Molluscs • - Mammals, + Molluscs 0.8 - -0 -- + Mammals, - Molluscs - - • ­ - + Mammals, + Molluscs

0.6 . 0.4 ":'0' ~ , ~ 'C--o.~~~--- ...... -...... -" ...... 0.2 ---..... -­ ...... -""'...... , ...

O+-~~--~~~--r-~~~--~~~~~T-~~--~T-~~ o 100 200 300 400 500 600 700 800 900 1000 Days Elapsed (Day 1 =1 February) Figure 1. The proportion of Lupinus chamissonis seedlings surviving through time as a function of A) the presence or absence of mammalian herbivores and B) the presence or absence of both mammalian and molluscan herbivores.

28

10~~~~------1 A. 1996 J.~-i--f-H 8 ---0­ - Mammals --.-- + Mammals ....1----1'1- .. t -t~f "r-i i/! 6 ----t.... .r ...

B. 1997 --0- - Mammals, - Molluscs 7 • - Mammals, + Molluscs - - c - - + Mammals, - Molluscs --.-- + Mammals, +Molluscs 6

5

4

3+---~~--~--~--~--~--~~--~--~--~--~~~~--~ o 100 200 300 400 500 600 700 Days Elapsed (Day 1 = 1 February) Figure 2. The volume of young Lupinus charnissonis plants, through time, as a function of A) the presence or absence of mammalian herbivores, and B) the presence or absence of both mammalian and molluscan herbivores. Data were log-normalized to correct for heterogeneity of variances. Vertical lines correspond to +/- one standard error. 29

-2.0 A. 1996 --0- - Mammals --e-- + Manunals !:" " " " " -3.0 '" '"

-4.0

-S.O

-6.0

-7.0

B. 1997 --0- - Mammals, - Molluscs -2.S • - Mammals, + Molluscs - - c - - + Mammals, - Molluscs - -.- - + Mammals, + Molluscs ... ------­ -3.S ,---~--- ~,~ ~~~ c---- ..... -..... f .".." ..... _- ."..'

-4.S

-S.S+---~--~--~--~--~--~--~--~--~--~--~--~~ o 100 200 300 400 SOO 600 Days Elapsed (Day 1 =1 February)

Figure 3. Branching density in young plants of Lupinus chamissonis, through time, as a function of A) the absence or presence of mammalian herbivores and B) the absence or presence of both mammalian and molluscan herbivores. Data were log-normalized to correct for heterogeneity of variances. Vertical lines correspond to +/- one standard error. 30

30 ,...... , A. II + Mammals B. 25 II - Mammals Z -...... -­~ ~ 15 S ..s c:r' OJ) 20 $:l ~ (!.) 0 I-t; .....:l 15 10 ~ e.0 ~ 1--1en (!.) 10 .....:l ~ 5 1--1 § (!.) 5 ~ :E 0

Figure 4. Mean length of leaflets for young Lupinus chamissonis shrubs (A), and the number of Dasineura lupinorum galls on leaflets of young Lupinus shrubs (B), in April 1997, as a function of the presence or absence of mammalian herbivores. Vertical lines correspond to one standard error. 31 14.0 A. - Manunals ? 13.5 + Manunals § .,,; -- ...... _- ...... _- .... ­ 1"-'1 I=-=- .­ // ~------~-~-~ / ~ / rt') / / 13.0 / uS ",'" / '" '" El / '" \....-..I 12.5 '" S ------­ ---­ ...... ~ >0 12.0

11.5+---~--~--~--~--~--~--~--~--~--~--~~

4~------. B.

1-_------1 /

",,;/ '" /

/ /'" / ,;'" / '" '" /'" '" ___ ------*':I:

1+---~--~--~--~---~---r--~---.---~--~-----~~ o 100 200 300 400 500 600 Days Elapsed (Day 1 = 1 May 1996) Figure 5. Plant volume (A) and mean shoot length (B) through time in matue Lupinus chamissonis shrubs as a function of the presence or absence of mammalian herbivores. Data were log-normalized to correct for heterogeneity of variances. Vertical lines correspond to +1- one standard en·or. 32

Figure 6. Fecundity of mature Lupinus chamissonis shrubs in spring 1997 as a function of the presence or absence of mammalian herbivores. A) The mean number of inflorescences per plant and B) the mean dry weight of seeds. Vertical lines correspond to one standard error. 33

r;:;' § 10 ~ ~ b[J d 8 0 C/'J C/'J cIj 6 S ...... 0 ~ 4 C Q '"d 2 d ::;j 0 Sb 0 d) :> 0 ..0 -2 -< 4 6 8 10 12 14 16 Shrub Volume [In cm3 /plantJ Figure 7. Relationship between aboveground dry biomass and shrub volume for Lupinus chamissonis. Both variables were log-normalized to correct for heterogeneity of variances. 34

~ 3.0 ~ ... $ 0 0 ~ 2.0 ~CI.) ~ c;j tJ § (t) 1.0 ::E p ~

O.O+------~------~------~ 1.5 2.0 2.5 3.0 In Mean Shoot Length (cm)IPIant Figure 8. Relationship between the mean shoot length of Lupinus chamissonis shrubs and the mean number of Dasineura lupinorum galls per shoot per plant. Both variables were log-normalized to correct for heterogeneity of variances. 35

APPENDIX B: Tables 36

Table 1. ANOV A tables for an experimental study evaluating the influence of mammalian and molluscan herbivores on the survival, volume, and branch density of Lupinus chamissonis seedlings/juveniles in two experiments (1996 and 1997). Results are also presented for the influence of mammalian herbivores on leaflet lengths and leaflets galled by the midge Dasineura lupinorum in 1996.

Variable Source DF MS F P

A. Seedling/Juvenile Survival

1996 Mammals 1 167715.8 1.21 0.273 Error 192 138486.6

1997 Mammals 1 119929.9 2.27 0.134 Molluscs 1 3605.3 0.07 0.795 Mammals x Molluscs 1 22195.9 0.42 0.518 Error 156 52931.8

B. Seedling/Juvenile Volume

1996 Mammals 1 4974288.6 8.77 0.005 Error 43 567176.0

1997 Mammals 1 2659556.6 16.73 <0.001 Molluscs 1 3878.8 0.02 0.877 Mammals x Molluscs 1 151125.0 0.95 0.334 Error 48 158970.4

C. Branch DensH)!

1996 Mammals 1 1626605.2 10.57 0.002 Error 43 153928.6

1997 Mammals 1 2065967.0 15.32 <0.001 Molluscs 1 370562.6 2.75 0.104 Mammals x Molluscs 1 30713.1 0.23 0.635 Error 49 134821.0

D. Leaflet Lengths

1996 Seedlings Mammals 1 95.5 6.36 0.015 Error 56 15.0

E. Dasineura Galls

1996 Seedlings Mammals 1 1695.8 3.98 0.051 Error 56 426.5 .- 37

Table 2. ANOVA tables for an experimental study evaluating the effect of mammalian herbivores on the volume, shoot lengths, inflorescence number, and seed weights of mature Lupinus chamissonis shrubs.

Variable Source DF MS F P

A. Shrub Volume

Mammals 1 32875.0 0.38 0.540 Block 17 2196749.9 25.61 <0.001 Mammals x Block 17 124484.1 1.45 0.170 EtTOr 36 85788.4

B. Shoot Lengths

Mammals 1 123785021.0 14.01 <0.001 Block 17 1954548.1 2.21 0.027 Mammals x Block 17 1093616.4 1.24 0.294 Error 31 8832948.0

C. Number of Inflorescences

Mammals 1 72517.0 8.64 <0.006 Block 17 111333.4 13.27 < 0.001 Mammals x Block 17 28004.0 3.34 <0.001 Error 36 8392.0

D. Seed Weights

Mammals 1 0.034 0.59 0.447 Block 17 0.221 4.59 <0.001 Mammals x Block 17 0.047 0.84 0.636 Error 36 0.059