* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download early primary succession on mount st. helens: impact of insect
Survey
Document related concepts
Mission blue butterfly habitat conservation wikipedia , lookup
Molecular ecology wikipedia , lookup
Weed control wikipedia , lookup
Banksia brownii wikipedia , lookup
Storage effect wikipedia , lookup
Ecology of Banksia wikipedia , lookup
Plant breeding wikipedia , lookup
Gartons Agricultural Plant Breeders wikipedia , lookup
Ecological succession wikipedia , lookup
Theoretical ecology wikipedia , lookup
Plant defense against herbivory wikipedia , lookup
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Transcript
Ecology, 83(1), 2002, pp. 191–202 q 2002 by the Ecological Society of America EARLY PRIMARY SUCCESSION ON MOUNT ST. HELENS: IMPACT OF INSECT HERBIVORES ON COLONIZING LUPINES JOHN G. BISHOP1 Department of Botany, University of Washington, Seattle, Washington 98195-5325 USA Abstract. Lupinus lepidus var. lobbii, the earliest plant colonist of primary successional habitats at Mount St. Helens, can dramatically influence successional rates and ecosystem development through N fixation and other facilitative effects. However, 15 yr after the eruption, lupine effects remained localized because high rates of population growth in newly founded patches (l 5 11.2, 1981–1985) were short lived ( l 5 1.51, 1991–1995), despite widespread habitat availability. To investigate this paradox, I examined 60 colonizing lupine patches, ranging from low-density patches at the edge of the expanding lupine population, to high-density patches in the population core, including 41 patches created in 1992. Survival, reproduction, and herbivory (for a subset of plants and years) were measured on ;12 000 plants for up to 5 yr (1991–1995). Stem-boring, leaf-mining, and seed-eating lepidopterans and anthomyiid flies, feeding in edge patches and in low-density margins of core patches, strongly affected edge patch demography, but not that of central core areas. For example, in 1994–1995, 77% of edge plants were afflicted by tortricid stem borers vs. 24% in the core, and from 1993–1995 gelechiid leafminers infested 68% of plants in the youngest edge patches, vs. only 8% at the core. Associated adult mortality was 88%, compared to ;30% when absent. Seed predators consumed 36% (range 4–92%, 1991–1995) of seeds in both core and edge patches, with seed loss negatively correlated with seedling number the following year. In contrast to edge patches, resource-dependent seedling mortality appears to govern short-term dynamics in high-density core areas. In conclusion, edge region herbivore effects can account for the observed 7.5-fold difference in l between patches founded 10 yr apart. Locally, increased lupine mortality creates access to high-quality sites, facilitating succession, but at larger spatial scales diminished population growth is likely to retard facilitation and succession. Key words: colonization; demography; facilitation; herbivory; invasion; legume; Lepidoptera; Lupinus; Mount St. Helens (Washington, USA); primary succession; seed predation. INTRODUCTION For many colonizing plants, establishment in primary successional habitats depends upon amelioration of physical conditions, either abiotically or through the facilitative effect of other species. For example, studies of soil development conclude that nitrogen is in short supply early in primary succession, a deficiency that appears especially striking on soils deriving from volcanic ash and lava flows, where N may be undetectable initially (Vitousek et al. 1987, del Moral and Bliss 1993). In such systems, the presence of nitrogen-fixing species is capable of radically altering the pace and pattern of ecosystem and community development (Clements 1916, Connell and Slatyer 1977, Marrs et al. 1983, Vitousek et al. 1987, Vitousek and Walker 1989). Under the facilitative model of succession, the rates of population increase and spatial spread of nitrogen-fixing species critically affect the rate of sucManuscript received 1 November 1999; revised 27 May 2000; accepted 30 June 2000; final version received 15 January 2001. 1 Present address: School of Biological Sciences, Washington State University, 14204 NE Salmon Creek Avenue, Vancouver, Washington 98686 USA. E-mail: [email protected] cession, and it is therefore of interest to understand the factors that control the population dynamics of these species. Consumers of nitrogen-fixing species, e.g., herbivores, may positively impact rates of succession by liberating resources, or they may decrease rates by demographically preventing facilitation. Most studies examining consumer effects on primary succession involve marine systems, where inhibition and tolerance models of succession may be more important than facilitative models (Lubchenco and Gaines 1981, Farrell 1991, Sousa and Connell 1992, Hixon and Brostoff 1996). These studies provide examples of consumermediated deceleration, acceleration, and deflection of successional rates and trajectories (Hixon and Brostoff 1996). The only study demonstrating herbivore effects on terrestrial primary succession is that of Bach (1994), showing that on lake shore sand dunes, chrysomelid beetles specializing on willow decreased host abundance, thereby changing community trajectories. Currently, herbivores are not considered an influential factor in terrestrial primary succession (e.g., Walker and Chapin 1987, del Moral and Wood 1988, Wood and Anderson 1990, Miles and Walton 1993). This is par- 191 192 JOHN G. BISHOP Ecology, Vol. 83, No. 1 PLATE 1. Lupinus lepidus in low-density edge regions attacked by (a) stem-boring Hysticophora spp. (foreground) and (b, c) leaf-mining Chionodes spp., visible as light-colored webbing and woven gravel retreats. ticularly true for invertebrate herbivores, whose ability to more generally influence plant population and community dynamics is a topic of debate (Brown et al. 1987, 1988, Crawley 1989, Gange et al. 1989, Ehrlén 1995, Louda and Potvin 1995, Strong et al. 1995, Louda and Rodman 1996, Root 1996, Maron 1997). In this paper, I describe patterns of herbivory and the demographic consequences of that herbivory in a population of the legume Lupinus lepidus var. lobbii that is colonizing the primary successional landscape created by Mount St. Helens’ 1980 eruption. In 1981, several individual lupines colonized the otherwise barren north slope of Mount St. Helens from remnant populations elsewhere on the volcano. For the next several years the lupine population spread rapidly outward from this initial invasion focus and lupines were the most successful colonist of upland primary successional habitats (del Moral et al. 1995). By 1990, the lupine population comprised a central core region of extremely high lupine density surrounded by numerous, low-density edge patches up to 2 km from the core (Fig. 1; del Moral et al. 1995, Fagan and Bishop 2000). Intervening areas between the edge and core patches, as well as the expanding margins of core patches, contain low-density, patchily distributed lupine. By 1992, population growth rates (l) in recently founded, lowdensity edge region patches had dropped to an average of l 5 1.51 (mean of 13 edge patches studied from 1991 to 1995), far below those observed during the early stages of colonization (l 5 11.2, for the original colonizing patch, 1981–1985; Fagan and Bishop 2000). L. lepidus on Mount St. Helens has been the focus of numerous studies demonstrating its ability to ameliorate physical conditions. For example, L. lepidus can fix more than 7 kg·ha21·yr21 of N, compared to background rates of only 2 kg·ha21·yr21 of N, and dramatically increases organic matter accumulation (Kerle 1985, Halvorson et al. 1991, Halvorson et al. 1992). Live lupines competitively suppress other ruderal species, whereas dead lupines facilitate these same species (Morris and Wood 1989, del Moral and Bliss 1993, Titus and del Moral 1998a). However, L. lepidus’s potential impact on community development has been tempered by its surprisingly slow rate of spread in recent years, associated with decreased l (Halvorson et al. 1992, del Moral and Bliss 1993). A number of hypotheses could explain recent decreases in rates of lupine population increase and spatial expansion. For example, dispersal limitation, caused by lupines’ relatively large seeds (4–8 mg), could curtail spatial spread (Wood and del Moral 1987). However, long-distance dispersal does occur. Founder individuals were .4 km from the nearest surviving patches, in 1991 edge patches existed .1 km from the January 2002 HERBIVORY, DEMOGRAPHY, AND COLONIZATION 193 FIG. 1. Map of study patches on the Pumice Plains and debris avalanche. Locations were obtained by standard survey methods using a TopCon Total Station. Patches with dates denote location and founding dates of core patches. Contour intervals are 14.7 m. Elevations are in feet (3600–4400 feet 5 1100–1340 m). The crater is 1.5 km southeast. Large solid circles 5 core patches; small solid circles 5 young edge, experimentally established; solid squares 5 young edge, natural; solid triangle 5 old edge; diamonds 5 source of transplants. Southern 1984 core patch is drawn to scale. The fourth core patch is ;700 m north-northeast of the 1981 patch. Inset shows location of Mount St. Helens National Volcanic Monument, denoted by open rectangle. Tick marks in inset represent 468159 N latitude and 1228109 W longitude. nearest seed source (personal observation), and seeds are occasionally deposited in traps many meters from lupine populations (del Moral and Wood 1988, Wood and del Moral 1988, Wood and del Moral 2000). Thus, while dispersal limitation may play a role, it is not an adequate explanation for changes in spread rates. Another plausible explanation is that lupines colonized most suitable habitat shortly after the eruption, and fare poorly in currently available habitats. This explanation is also inadequate. L. lepidus is characteristic of highaltitude pumice communities, and resource-poor gla- cial moraines (Franklin and Dyrness 1973, Kruckeberg 1987). Following the eruption, L. lepidus colonized substrates 200–600 m below pre-eruption elevations, .1000 m below its elevation elsewhere in the region. At these lower elevations, decreased drought stress and longer growing seasons are known to increase lupine growth rates, size, and fecundity relative to plants at pre-eruption elevations (Braatne and Bliss 1999). Another plausible hypothesis is that decreased l is the demographic consequence of insect herbivory. High levels of seed predation and other insect herbivore JOHN G. BISHOP 194 damage have been documented in L. lepidus at Mount St. Helens (Halvorson et al. 1992, Bishop and Schemske 1998, Fagan and Bishop 2000; see Plate 1). In addition, short-term removal experiments demonstrate that these herbivores can reduce population growth and rates of spread (Fagan and Bishop 2000). Here I present the results of a large-scale (60 patches), long-term (5 yr) demographic survey of core and edge patches that allows extrapolation from the experimental results of Fagan and Bishop. Specifically, I describe patterns of herbivory that differ between core and edge patches and that are closely related to host density. I also use extensive demographic data to ask whether the damage by different herbivore guilds likely affects lupine demography (survivorship, fecundity, and population growth), and whether these effects vary among core and edge portions of the population. METHODS Species and study site The study site was formed on 18 May 1980 by a massive debris avalanche, followed by pyroclastic flows (incandescent gas and pumice), lahars, and tephra deposition. The lupine patches in this study lie in a 60km2 primary successional landscape, atop 4–80 m of newly emplaced rock, and 300–1500 m from the nearest secondary successional habitats. Thus, there was no pre-eruption seed bank, herbivores, or other biotic legacy surviving in these sites. The study patches are scattered across 6 km2, with an elevation range of 1150–1350 m (3700 ft–4400 ft in Fig. 1). Environmental conditions on the Pumice Plains are stressful; the pumice and rock substrates initially had little waterholding capacity, few nutrients, and nitrogen was undetectable (Nuhn 1987, Halvorson et al. 1991, 1992, del Moral and Bliss 1993). Although yearly precipitation is high (2.4 m), there is pronounced summer drought (rainfall in July and August is frequently ,10 mm/mo), and surface temperatures reach 508C (Reynolds and Bliss 1986). Lupinus lepidus possesses several features, e.g., the ability of seedling roots to quickly reach moister subsurface soil, that allow it to avoid drought stress (Braatne and Bliss 1999). Mature plants of L. lepidus var. lobbii are ,15 cm high, with prostrate herbaceous stems spreading up to a 45-cm radius from a fibrous caudex. L. lepidus is self-compatible, and pollen addition experiments in large patches show that fruit set is not pollinator limited (J. G. Bishop, unpublished data). Seeds are explosively dehisced, but have no obvious adaptations for secondary dispersal, moving primarily by wind and during snowmelt. The flowers, flower stalks, and fruits are frequently attacked by lupine aphids (Macrosiphum albifrons), pentatomid bugs (Chlorochroa ligata), larvae of an anthomyiid fly (Crinurina sp.), and by two lepidopteran species specializing on lupines, the microlepidopteran Schinea sueta and the lycaenid Plebejus Ecology, Vol. 83, No. 1 icarioides montis (Bishop and Schemske 1998). Larvae of several lepidopteran species also mine the leaves (Gelechiidae: Chionodes spp.) and bore into the caudex (Tortricidae: Hystricophora sp.); these are also lupine specialists and rarely attack adjacent L. latifolius. Damage by leafminers is easily recognized as a woven mass of whitened or yellowed leaves, and damage by caudex borers as a gray pallor accompanied by wilting (see Plate 1); both types of damage are readily quantified. Damage to fruits and seeds by chewing insects is more subtle and must be assessed through dissection. Field methods Twelve patches (3 core, 9 edge) were selected for demographic study in 1991, and 7 additional patches (1 core, 6 edge) were selected in 1992. Core patches, founded 1981–1984, include the original colonizing patch and several high-density satellites. Edge patches are widely spaced secondary or tertiary foci, generally located uphill of the core, and can be divided into two age groups: those founded 1987–1988 (‘‘old edge’’) and those founded 1990–1991 (‘‘young edge’’). These 19 patches represent a substantial proportion of the core and older edge patches that existed by 1991. Extensive surveys in 1994 identified only three additional large patches in the region illustrated in Fig. 1, and about 40 additional patches among edge patches, mostly comparable to young edge patches. To increase the sample size of newly-founded patches and to examine the probability of patch establishment, I established 41 additional edge patches prior to the start of the 1992 growing season. Each experimental patch consisted of eight 1-yr-old individuals arranged in a circle 1 m in diameter. Founders were from one or two nearby patches (Fig. 1), representing likely sources of natural colonists. Sites were chosen without regard to microsite characteristics, except that they were not placed within gullies and were located .30 m from existing patches so as to avoid external demographic influence. These patches are similar in density, percent cover, and size distribution to natural edge patches, so they are grouped with ‘‘young edge’’ patches. To provide an indication of successional status, percent cover of lupines and percent cover of all other plants were estimated in 1993 for each patch by estimating the number of 10 3 10 cm squares occupied by each species. Differences among patch types (core, old edge, young edge) were examined using Tukey tests for multiple comparisons (Zar 1984). All edge patches were censused by establishing a permanent grid around the patch. In core patches, because of their much higher density, five or nine 25 3 25 cm quadrats were placed randomly along transects running from the periphery to the center of the patch. An additional four or seven 1 3 1 m quadrats were located at patch margins in order to sample the lowdensity periphery of the core. The number of quadrats January 2002 HERBIVORY, DEMOGRAPHY, AND COLONIZATION varied because additional quadrats were placed in larger patches. When increased patch size made sampling entire patches and quadrats infeasible, sections of grids or quadrats were subsampled and some patches were dropped from the study (2 in 1994, and 13 in 1995). All patches were censused in late August (15–28 August), but plants were usually located and tagged in June and July. All plants were marked, and their location determined to the nearest centimeter, to allow relocation. Plant diameter, which correlates strongly with biomass (Pearson’s r 5 0.95, df 5 49, P , 0.0001), and herbivore damage were measured at each census. Herbivory.—Foliage damage by leaf-mining Chionodes caterpillars, which is conspicuous, was rare until 1993. It was quantified first as present or absent (1993), then, in 1994 –1995, on a 0 –5 scale of leaf destruction: 0 5 0 –5%, 1 5 6–25%, 2 5 26– 50%, 3 5 51–75%, 4 5 76–95%, 5 5 95–100% destruction). Caudex-boring Hystricophora sp. was not apparent from 1990 to 1992, but in 1993–1994 it caused nearly 100% mortality in several patches not under study. Mortality caused by Hystricophora between August 1994 and July 1995 was quantified by surveying live plants and plants that had died since August 1994 for Hystricophora presence. Because dead lupines take . 1 yr to decay, and because Hystricophora forms extensive galleries in the root and caudex, stem borer presence was easily identified in summer, 1995. Its presence in live plants is detected by wilting, gray pallor, and the accumulation of frass surrounding holes in the caudex. Reproduction.—I censused each plant’s infructescences and the number of fruits on a sample of seven infructescences per plant, if available, at roughly 2-wk intervals. In addition, a sample of infructescences was collected to estimate seeds per fruit and damage by seed predators. Sample sizes fluctuated because (a) in some years (i.e., 1993) weather conditions resulted in little or no reproduction, and (b) to minimize demographic effects, fruits were not collected from patches with low fruit set. Overall for the years 1991–1996, seed production and damage was measured for 2957 fruits from 347 plants from core patches, and 2594 fruits from 369 plants from edge patches. Fruits were checked for (a) the presence of damage, indicated by an entry or exit hole produced by a seed predator; (b) the total number of mature or nearly mature seeds, not counting aborted seeds; and (c) the number of these seeds damaged by chewing. Not counting aborted seed results in underestimation of insect damage, because many are caused by inflorescence-attacking hemipterans and homopterans. Seed production before predation was estimated for each plant by multiplying total fruits (known for every plant) by average seeds per fruit (from that patch or a nearby one of the same patch type). Using average seeds per fruit to calculate seeds per plant is reasonable, as most of the 195 interplant variation within patches in seed production is due to fruit number rather than seeds per fruit (Bishop and Schemske 1998). The number of seeds matured (i.e., after predispersal predation) was estimated by discounting each plant’s seed production by estimates of proportion of seeds damaged by chewing insects. Analyses.—For each of the three types of herbivore damage, leaf mining, caudex boring, and seed predation, there are two classes of analyses. I first asked whether each type of herbivory differs between patch type (core and edge), and whether there is an additional effect of plant density. Second, I analyzed the effects of patch type, herbivory, and density, on mortality, seed and seedling production, and population growth. Regression analyses included year as a covariate, when appropriate. In edge patches, density was calculated by defining area as the rectangle bounding the outermost plants. Variables were transformed to improve model fit, but not to de-correlate residuals. Instead, significance levels for multiple and stepwise regressions were determined by randomization tests performed as follows. For each regression analysis, dependent variable data were randomly permuted among cases 4999 times, while holding the independent variable data constant, to create 4999 new data sets. The regressions were repeated for each permuted data set, and the P value was calculated as [rank of the actual regression coefficients among 4999 random ones 4 5000] (Manly 1991). Table 3 lists dependent and independent variables. Proportion of seeds damaged in each patch was analyzed with ANOVA using sequential sums of squares, with year and patch type as factors. Significance levels were determined using randomization tests as for the regression analysis, with P values calculated as the [rank of the actual F statistic 4 5000] (Manly 1991). All analyses were done using S-Plus 3.3 (Statsci 1994). Logistic regressions were used to determine (1) the effect of patch type (core vs. edge), density, and plant diameter on Hystricophora infestion (1994–1995 only, for four edge and four core patches) and (2) whether leaf damage or Hystricophora infestation affected the probability of surviving until the next census, with patch type, year, and plant diameter as covariates. For years in which degree of leaf damage was measured, logistic regression was used to determine whether plants with the most damage (.74%) had an increased probability of death. Survivorship distributions (probability of surviving until a given age) were estimated for 1991–1995 for all patch type and cohort combinations using Kaplan-Meier estimates for 5246 plants (Kalbfleisch and Prentice 1980). Effects of patch type and cohort on survivorship distributions were analyzed with an accelerated life model, assuming a Weibull distribution. Details of this analysis are available in Appendix A. Appendix B contains photographs showing the impact of insect herbivores. JOHN G. BISHOP 196 Ecology, Vol. 83, No. 1 TABLE 1. Site and sample characteristics by patch type. Statistics are reported as median (range). Site Patches surveyed Year founded Area sampled (m2) Edge, young Edge, old Core 50 6 4 1990–1991 1987–1988 1981–1985 90 (48–600) 225 (144–300) 4 (1–10) Initial population size Density (plants/m2) 3 (1–24) 131 (35–160) .40,000 0.13 (0.001–1.70)a 1.01 (0.03–5.06)b 73 (0.25–1622)c Notes: Number of plants sampled does not include dead plants. Values of density and percent cover with different superscript letters are significantly different, P , 0.01. RESULTS density and herbivory results from herbivory concentrated at the margins of the patch. Similarly, though quantified only in 1994–1995, damage by caudex-boring Hystricophora increased through the study period, and was much higher in edge patches (Table 4), and Hystricophora presence was positively density dependent in edge patches but negatively density dependent in core patches. Logistic regression indicated that core plants at 1.16-m spacing were 153 more likely to be infested by Hystricophora than plants at 0.16 m, whereas the odds of infestation for widely spaced plants at the edge is 0 relative to plants at 0.16-m spacing (odds 5 probability of infestation 4 probability of no infestation; Table 4). Seed production and predation.—Seed production per plant (post predation) decreased dramatically through the study period (compare 1991–1992 with 1993–1995 in Table 2). The presence of Chionodes leafminers, which increased over time, was associated with 32 fewer seeds per plant (Table 3, line 4), a 97% decrease compared to 33 6 71 seeds per plant (mean 6 1 SD of all populations, after seed predation). Seed loss to predispersal seed predators, primarily lycaenid caterpillars (Plebejus icarioides) and anthomyiid fly larvae (Crinurina spp.), averaged 36% across all populations and years. ANOVA and post hoc comparisons indicated that mean seed loss was greater in core patches and decreased over time (effect of patch type: F 5 49, df 5 1, P , 0.0001; effect of year: F 5 11, df 5 4, P 5 0.001; patch type 3 year interaction: F 5 6, df Percent cover of lupines was .11003 higher in the core than at the edge; percent cover of other species was 2653 higher at the core than at the edge, and median patch densities were 5613 greater in the core than in young edge patches (Table 1). These results provide independent support for lumping patches into classes representing different stages of colonization. Note that the densities in edge patches appear too low to invoke either intraspecific or interspecific resource competition as an explanation for the 7.5-fold decrease in population growth in these regions (Table 2). All 41 transplant patches survived the first growing season, with a median of 3 survivors per patch, and 38 of 41 survived for 6 yr. Natural and transplant young edge patches were similar in percent cover (both 0.4%) and lupine density (mean 6 SD; 0.03 6 0.1 plants/m2 vs. 0.08 6 0.05 plants/m2 in 1992). Foliage and caudex damage.—Foliage damage by leaf-mining Chionodes caterpillars increased dramatically through the study period. While leaf damage was not apparent in 1991–1992, by 1995 a median of 68% of plants were damaged in young edge patches, and the median damage score averaged across patches was 1.85 (2 5 26–50%, Table 2). Edge patches suffered much greater damage than core patches (Table 2, Fig. 2), and increasing density of L. lepidus resulted in decreased damage at both the core and the edge, even with year and patch type entered first as covariates (Table 3, lines 1 and 2). At the core, this negative relationship between TABLE 2. Mean of population and quadrat means (6 1 SD ). Edge Measure Years Younger Older Core 1993–1995 0.68 6 0.4 a 0.48 6 0.3b 0.08 6 0.1c Proportion plants damaged Average area damage 1994–1995 1.85 6 1.1a 1.43 6 1.5a 0.18 6 0.3b Seed number 1991–1992 1993–1995 231 6 425 51 6 120a 174 6 358 7 6 33b b 161 6 394b 35 6 60ab l ( 5 e r) 1981–1985 1991–1995 NA 1.51 6 0.75 NA 1.46 6 0.93 11.3 6 2.7 0.91 6 0.19 a Notes: Plant damage refers to leaf mining by Chionodes spp. Average area damaged is the mean leafminer damage class on a scale of 0 (0–5%) to 5 (95–100%). Values in the same row with different superscript letters are significantly different (P , 0.05) after applying variable-wide sequential Bonferroni correction. Population growth rate (l) is the median of the geometric mean Nt11/Nt for each patch taken across years. Core data from 1981–1985 are courtesy of C. M. Crisafulli. NA 5 not applicable. January 2002 HERBIVORY, DEMOGRAPHY, AND COLONIZATION 197 TABLE 1. Extended. Percent cover lupine 0.03 (0.01–0.1)a 0.06 (0.02–0.15)a 66.3 (4–92)b Percent cover other plants 0.12 (0.00–1.9)a 0.12 (0.06–0.5)a 31.8 (0.00–108.8)c Number of live plants sampled 1991 1992 1993 1994 1995 41 399 291 213 603 757 1300 1715 1894 1433 1828 1496 1148 98 298 5 4, P 5 0.018; P-values estimated by randomization test; total df 5 215). The strong patch type 3 year interaction corresponds to the observation that over 7 yr of observations, percent fruit damage in core patches undergoes a damped oscillation between 4% and 82%, whereas damage in edge patches declines after 1991 (Fig. 3). Regression analysis reveals that the effect of seed predation carries over into the next year’s seedling abundance. Seedlings per square centimeter of leaf area were significantly negatively affected by the proportion of fruits damaged, but not affected by the total number of seeds (before predation) in the previous year (Table 3, line 5). Survivorship and annual mortality.—The odds of dying increased through the study period and were twice as great in core patches as in edge patches, corresponding to a decrease in survivorship from 0.65 to 0.32 (logistic regression: effect of each additional year on odds 5 4.413, P , 0.0001; effect of edge vs. core 5 1.933, P , 0.0001; N 5 8867; effect indicates increase in odds of death). Logistic regression did not detect significant effects of Chionodes presence on mortality from 1993 to 1995, an effect that may have been masked by Hystricophora, but did indicate that a leaf damage score .4 (.75% leaf loss, N 5 109) raised the overall probability of death in 1994–1995 from 0.80 to 0.90 (P 5 0.03, df 5 2261). Graphical inspection of age-specific survivorship curves suggests that patch type, cohort, and year have strong effects on survivorship (Fig. 4). These conclusions are supported by the accelerated life model analysis for patch type and cohort, which shows decreasing survivorship in younger patches and later cohorts (P , 0.0001 for each level of each factor, df 5 5239, r2 5 0.41; see Appendix A for details). Of particular note is that seedling mortality (as indicated by the transition from the first to the second growing season) was 40% higher in core patches than in edge patches (excluding core 1991 and edge 1994). In contrast, postseedling mortality was comparable between patch types. The pattern at the core is attributable to high lupine establishment and rapid growth of other species in 1991, followed by low establishment thereafter. In 1994 survivorship was low in all patch types for all cohorts (Fig. 4) demonstrating that high mortality late in the study is not attributable to a deterministic 4 or 5 yr lifespan. In edge patches this appears to be due to Hystricophora-caused mortality, whereas in core patches Hystricophora was less common (Table 4), and survivorship is comparable to previous years. Population growth (l 5 Nt11/Nt) was higher on average in edge patches than in core patches, and highest in the young edge patches (Table 2), but even in edge patches l averaged seven times less than in the original patch 10 yr earlier. l fluctuated greatly between patches and years, with r (5 loge l) ranging between 21.2 and .5 in edge patches in 1992–1993 alone (data not shown, see Fig. 5 for range over all years). In addition, population growth appears to be strongly affected by density (Fig. 5), even when patch type and year are entered first as covariates (Table 3, line 3). In other words, negative density dependence is as strong across the range of densities found in young edge patches (median density 5 0.13 plants/m2) as it is in core patches (median density 5 73 plants/m2). DISCUSSION FIG. 2. Magnitude of leaf damage caused by Chionodes spp. (patch averages), vs. density. The change of damageclass scale on the bottom graph reflects a difference in scoring method between 1993 (present/absent) and later years [scale of 0 (no damage) to 5 (100% damage)]. Note logarithmic plant density scale. Lines are local regression (LOESS) estimates. Stem-boring, leaf-mining, and seed-eating lepidopterans and anthomyiid flies, feeding in edge patches and in the low-density margins of core patches, markedly increased mortality and decreased propagule production. This conclusion, in combination with the experimental results of Fagan and Bishop (2000), con- JOHN G. BISHOP 198 Ecology, Vol. 83, No. 1 TABLE 3. Results of regression analyses of leafminer damage and demographic rates. Independent variables† Year Dependent variable Coeff. A) Leafminer damage Plants damaged by Chion- 20.03 odes, 1993–1995 (%) Average Chionodes dam0.08 age, 1994–1995 r in patch or quadrat 20.33 [5loge(l)] B) Lupine seed production Seed production per reproductive plant§\ C) Seedling density Seedlingst11/cm2 of lupine\ Patch type P Coeff. P Density (Nt /m2)‡ Coeff. P Patch type 3 Density Coeff. P r2 N 0.16 0.34 0.0002 20.09 ,0.0001 20.06 0.008 0.56 236 0.61 0.93 0.0001 20.33 ,0.0001 20.30 0.004 0.51 143 1.97 ,0.0001 20.21 ,0.0001 20.22 0.008 0.33 249 ,0.0001 1.4 0.0006 0.01 0.64 224.6 214.5 20.08 0.0002 0.005 0.006 232.2 20.12 0.0002 0.02 3.2 0.0002 0.27 2023 20.001 0.87 0.16 109 Notes: Results are from stepwise regression with variables entered in order from left to right; patch type is a dichotomous variable with core 5 0, edge 5 1, except as noted. Listed under each independent variable are regression coefficients (Coeff.) and P values. † All significance levels determined by randomization tests are based on regressions of 4999 permuted data sets. ‡ Variable was loge-transformed to improve model fit. § Top coefficient is for old edge relative to young edge, bottom values are for core relative to young edge. \ Multiple regression. tradicts the view that insect herbivores do not influence colonization by native species or of primary successional habitats, suggesting instead that specialist insect herbivores restrict population growth and spatial spread of lupines colonizing Mount St. Helens. For example, in 1994, 77% of edge plants were infested with the Hystricophora stem borer, and infested plants had 97% mortality, compared to 59% for uninfested plants and 25–40% for edge plants in the previous 3 yr (Fig. 4). TABLE 4. (A) Hystricophora infestation and associated percentage mortality in edge vs. core patches; (B) results of logistic regression analysis for the effect of patch type and nearest-neighbor distance (NN) on Hystricophora presence. Mortality in 1995 not caused by Hystricophora was partly attributed to the Chionodes leafminer, which infests the majority of edge plants and removes .25% of leaf area on average (Fig. 2, Table 2). Herbivores also curtailed edge plant seed production. In 4 of 6 yr, pre-dispersal seed predators removed 37–69% of maturing seeds (Fig. 3) and directly affected seedling numbers the next season. Moreover, leaf damage by Chionodes prevented many plants from producing any seeds. Because established seedlings often have high survivorship in edge patches (Fig. 4), and L. lepidus A) Hystricophora infestation Category Infested Uninfested Edge (N 5 703) Core (N 5 345) Plants (%) Mortality (%) Plants (%) Mortality (%) 77 23 97 59 24 76 95 93 B) Regression analysis (R2 5 0.78, N 5 1048) Variable Intercept Edge relative to core NN at core (m2) NN at edge (m2) Coefficient Odds P 0.29 0.47 2.71 210.04 1.60 15.03 0.00 0.22 0.000 0.000 0.000 Notes: (A) The ‘‘Plants’’ columns report the percentages of infested and uninfested plants. (B) Coefficients indicate the increment in log odds. ‘‘Odds’’ refers to the odds of infestation relative to core or edge plants with 0.16-m spacing. For example, edge plants are 1.63 more likely to be infested than are core plants. FIG. 3. Time series of percentage fruit damaged with pointwise 95% binomial confidence intervals. Estimates are means of plant means. Values for 1990 are anecdotal, but based on a large number of fruit dissections (Bishop and Schemske 1998); values for 1996 are from Fagan and Bishop (2000). January 2002 HERBIVORY, DEMOGRAPHY, AND COLONIZATION 199 Edge vs. core Herbivory in core patches was lower compared to edge patches and had relatively little demographic impact. Hystricophora and Chionodes caterpillars were scarce in core patches. When present, they were concentrated in the lower density margins of the core patches, where their presence might easily be overlooked, yet where models predict the greatest effects of decreased propagule production on spatial spread. Fagan and Bishop (2000) found no effect of herbivore removal on l in core patches, but their experiment was not designed to distinguish between central and marginal zones of the core. Greater herbivory in edge patches and core margins, compared to the denser center of the core, runs counter to expectations based on resource concentration, wherein the densest host populations sustain the highest herbivore density. While negative relationships between host density and herbivory are not uncommon, the underlying mechanisms remain obscure (Kunin 1999). It seems unlikely here that greater herbivore load at low-density is attributable to the larger basin of attraction of each plant, because the herbivores are FIG. 4. Survivorship curves: comparison of patch type and cohort (year of germination); N 5 8867 plants. Age 1 is the end of the first growing season for each cohort. A small fraction of plants survive past 5 yr. has a short-lived seed bank (Bishop 1996), the loss of seeds may directly decrease recruitment to adult stages. Removal of the Chionodes leafminer and several less abundant herbivores from two edge patches in 1995 produced a temporary fivefold increase in population growth rate (l), mediated primarily by increased plant growth and seed production (Fagan and Bishop 2000). Our combined studies show that such leafminer effects occur throughout edge region patches and have persisted from 1993 to 1999. Fagan and Bishop did not find any significant effect of herbivore removal on mortality, nor of predispersal seed predators in edge patches. The current study suggests that the fivefold effect of removal actually underestimates the impact of insect herbivores because both pre-dispersal seed predation and Hystricophora-caused mortality are frequently greater than they were in 1995–1996. Thus, herbivore impacts likely account for the observed 7.5-fold decrease in l compared to initial rates of l 5 11.3 (Table 2). Models of spatial spread predict that diminished l at the edge of an expanding population will greatly affect spread rates, suggesting that herbivory in edge regions may explain L. lepidus’ failure to rapidly exploit available habitat (Skellam 1951, Moody and Mack 1988, Andow et al. 1990, Fagan and Bishop 2000). FIG. 5. Annual population growth (r 5 loge l) for patches and quadrats as a function of density, by patch type, for 1991– 1995. Line slopes are from multiple regression; horizontal lines denote r 5 0. For the abscissa label, the original units of density were Nt /m2. 200 JOHN G. BISHOP nearly absent from the central core areas, and because this pattern has persisted for 7 yr. Alternatively, high loads on edge and margin plants may reflect resource quality for herbivores, which may in turn be a function of plant stress, plant vigor, or nutrient stoichiometry (Price 1991, Elser and Urabe 1999). Margin and edge plants do appear more vigorous than their central counterparts. A third hypothesis for core–edge differences is that faunal buildup is more advanced in core areas, and supports higher trophic levels that suppress herbivores. In fact, spatial variation in tritrophic interactions is likely at Mount St. Helens. The abundance of generalist predators and parasitoids of lepidopteran larvae is much higher in the core than at the edge, as are vertebrate predators, especially birds (Fagan and Bishop 2000; J. Edwards, unpublished data). Subtle differences in faunal succession might also explain the damped oscillation in fruit damage on core plants over a 7-yr period (1990–1996), and the contrasting monotonic decrease at the edge (Fig. 3). Density-dependent mortality and population growth Several observations imply that high mortality in core patches (Fig. 4, 1992–1994 cohorts) is attributable to resource competition. Intraspecific density reaches 1600 plants/m2 in some quadrats, median intraspecific neighbor distance was .7 cm only in 1991, and cover of all species averages 54% (Table 1). Seedling mortality was much higher than that for established plants, except in 1991 (Fig. 4), a hallmark of competition (Silvertown and Lovett Doust 1993). Finally, population growth in core patches appears negatively density dependent (Fig. 5), even when year is entered as a covariate (Table 3), and despite the fact that herbivory is concentrated in low-density margins. This suggests that insect-caused mortality and propagule loss may be unimportant to core population dynamics. This suggestion is confirmed by the effect of insect removal, which increased seed production in core patches without affecting r (Fagan and Bishop 2000). Whether plant populations are often limited by propagule loss to herbivores is a continuing controversy (Crawley 1989, 1990, 1992, Louda 1989). In lupines at Mount St. Helens, the demographic significance of propagule loss varies spatially. High levels of seed predation in core patches may have little effect on l because competition predominates, but propagule production in edge patches is closely linked to l and rate of spread. However, core patches may depend on seed production, since recruitment following disturbances caused by erosion, lahars, and large mammals depends on the newly formed, but short-lived, seed bank (Bishop 1996). Long-lived seed banks are thought to dampen population-level effects of seed predation in other lupine species, but these systems too are frequently disturbed and seed banks could become depleted by seed predation (Breedlove and Ehrlich 1972, Dollinger et Ecology, Vol. 83, No. 1 al. 1973, Harrison and Maron 1995, Maron and Simms 1997). In lupines at Mount St. Helens, competition and herbivory govern dynamics in different portions of the population and their relative importance changes over time. Negative density-dependent population growth also occurred in edge patches, where the median density is only 0.13 plants/m2 and total cover averaged only 0.4%, even when year is included first as a covariate (Table 3). Resource competition is an unlikely explanation for negative density dependence at the edge, but positive density-dependent herbivory in edge areas by Hystricophora may be partially responsible. Community and ecosystem implications Farrell (1991) presents a model of consumer-driven succession, wherein the effect of consumers depends on the model of succession (e.g., tolerance, inhibition, or facilitation). According to this view, consumers of pioneer species will increase rates of succession when pioneers inhibit later successional species, but consumers will decrease rates or alter trajectories of succession when early species facilitate later ones. Vertebrate herbivores, for example, can severely delay succession in N-limited grasslands by feeding on a legume that conrols the size of N pools (Ritchie et al. 1998). In contrast, effects of herbivory on dune primary succession (Bach 1994) and marine primary succession (e.g., Hixon and Brostoff 1996) are mediated by interactions that fall under tolerance or inhibition models of succession. At Mount St. Helens, episodic lupine mortality may increase local rates of succession, because invading ruderals are inhibited by live lupines but facilitated by dead lupines (Morris and Wood 1989). This effect is similar to that of L. arboreus in coastal grasslands, where mortality caused by hepialid moth larvae mediates access by exotic plants to lupine-enriched N pools (Maron and Jefferies 1999). However, at the landscape scale, delay in L. lepidus arrival caused by herbivores is likely to slow primary succession through decreased rates of N mineralization, accumulation of organic matter, and other facilitative effects (Halvorson et al. 1991, Halvorson et al. 1992, del Moral 1993, Titus and del Moral 1998b). Thus, at Mount St. Helens insect herbivores may simultaneously prevent buildup of N pools at large spatial scales and mediate access to those pools at smaller scales. ACKNOWLEDGMENTS I thank D. Schemske and members of his lab for guidance, discussion, and comments at all stages of this study, and D. McCrumb, D. Kline, M. Jackson for extensive field assistance. L. Anderson, E. Anderson, P. Chilsen, B. Cook, I. Dews, B. Fagan, A. Holzapfl, K. Koerner, B. Nakamura, I. Parker, D. Parks, J. Ramsey, D. Schemske, J. Titus, K. Warner, and R. Williamson, also provided invaluable field assistance. C. Crisafulli kindly provided unpublished data. Comments by K. Clay, B. Fagan, J. Kingsolver, J. Maron, J. Titus, and three anonymous reviewers improved the manuscript. This work was supported by the UW Plant Molecular Integration and Function Committee, NSF dissertation improvement grant DEB-9213143 and NSF training grant BIR-9256532. January 2002 HERBIVORY, DEMOGRAPHY, AND COLONIZATION LITERATURE CITED Andow, D. A., P. K. Kareiva, S. A. Levin, and A. Okubo. 1990. Spread of invading organisms. Landscape Ecology 4:177–188. Bach, C. E. 1994. Effects of a specialist herbivore (Altica subplicata) on Salix cordata and sand dune succession. Ecological Monographs 64:423–445. Bishop, J. G. 1996. Demographic and genetic variation during colonization by the herb Lupinus lepidus on Mount St. Helens. Dissertation. Department of Botany, University of Washington, Seattle, Washington, USA. Bishop, J. G., and D. W. Schemske. 1998. Variation in flowering phenology and its consequences for lupines colonizing Mount St. Helens. Ecology 79:534–546. Braatne, J. H., and L. C. Bliss. 1999. Comparative physiological ecology of lupines colonizing early successional habitats on Mount St. Helens. Ecology 80:891–907. Breedlove, D. E., and P. R. Ehrlich. 1972. Coevolution: patterns of legume predation by a lycaenid butterfly. Oecologia 10:99–104. Brown, V. K., A. C. Gagne, I. M. Evans, and A. L. Storr. 1987. The effect of insect herbivory on the growth and reproduction of two annual Vicia species at different stages in plant succession. Journal of Ecology 75:1173–1189. Brown, V. K., M. Jepsen, and C. W. D. Gibson. 1988. Insect herbivory: effects on early old field succession demonstrated by chemical exclusion methods. Oikos 52:293–302. Clements, F. E. 1916. Plant succession: an analysis of the development of vegetation. Carnegie Institution of Washington Publication, Washington, D.C., USA. Connell, J. H., and R. O. Slatyer. 1977. Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist 111:1119– 1144. Crawley, M. J. 1989. The relative importance of vertebrate and invertebrate herbivores in plant population dynamics. Pages 45–71 in E. A. Bernays, editor. Insect–plant interactions. CRC Press, Boca Raton, Florida, USA. Crawley, M. J. 1990. The population biology of plants. Philosphical Transactions of the Royal Society of London B330:125–140. Crawley, M. J. 1992. Seed predators and plant population dynamics. Pages 157–191 in M. Fenner, editor. Seeds: the ecology of regeneration in plant communities. C.A.B. International, London, UK. del Moral, R. 1993. Mechanisms of primary succession on volcanoes: a view from Mount St. Helens. Pages 79–100 in J. Miles and D. Walton, editors. Primary succession on land. Blackwell Scientific, London, UK. del Moral, R., and L. C. Bliss. 1993. Mechanisms of primary succession: insights resulting from the eruption of Mount St. Helens. Advances in Ecological Research 24:1–66. del Moral, R., J. H. Titus, and A. M. Cook. 1995. Early primary succession on Mount St. Helens, Washington, USA. Journal of Vegetation Science 6:107–120. del Moral, R., and D. M. Wood. 1988. Dynamics of herbaceous vegetation recovery on Mount St. Helens, Washington, USA, after a volcanic eruption. Vegetatio 74:11–27. Dollinger, P. M., P. R. Ehrlich, W. L. Fitch, and D. E. Breedlove. 1973. Alkaloid and predation patterns in Colorado lupine populations. Oecologia 13:191–204. Ehrlén, J. 1995. Demography of the perennial herb Lathyrus vernus. II. Herbivory and population dynamics. Journal of Ecology 83:297–308. Elser, J. J., and J. Urabe. 1999. The stoichiometry of consumer-driven nutrient recycling: theory, observations, and consequences. Ecology 80:735–751. Fagan, W. F., and J. G. Bishop. 2000. Trophic interactions during primary succession: herbivores slow a plant rein- 201 vasion at Mount St. Helens. American Naturalist 155:238– 251. Farrell, F. M. 1991. Models and mechanisms of succession: an example from a rocky intertidal community. Ecological Monographs 61:95–113. Franklin, J. F., and C. T. Dyrness. 1973. Natural vegetation of Oregon and Washington. Oregon State University Press, Corvallis, Oregon, USA. Gange, A. C., V. K. Brown, I. M. Evans, and A. L. Storr. 1989. Variation in the impact of insect herbivory on Trifolium pratense through early plant succession. Journal of Ecology 77:537–551. Halvorson, J. J., E. H. Franz, J. L. Smith, and R. A. Black. 1992. Nitrogenase activity, nitrogen fixation, and nitrogen inputs by lupines at Mount St. Helens. Ecology 73:87–98. Halvorson, J. J., J. L. Smith, and E. H. Franz. 1991. Lupine influence on soil carbon, nitrogen and microbial activity in developing ecosystems at Mount St. Helens. Oecologia 87: 162–170. Harrison, S., and J. L. Maron. 1995. Impacts of defoliation by tussock moths (Orgyia vetusta) on the growth and reproduction of bush lupine (Lupinus arboreus). Ecological Entomology 20:223–229. Hixon, M. A., and W. N. Brostoff. 1996. Succession and herbivory: effects of differential fish grazing on Hawaiian coral-reef algae. Ecological Monographs 66:67–90. Kalbfleisch, J. D., and R. L. Prentice. 1980. The statistical analysis of failure time data. Wiley, New York, New York, USA. Kerle, E. A. 1985. The ecology of lupines in Crater lake National Park. Thesis. Oregon State University, Corvallis, Oregon, USA. Kruckeberg, A. R. 1987. Plant life on Mount St. Helens before 1980. Pages 3–17 in D. Bilderback, editor. Mount St. Helens 1980. University of California Press, Berkeley, California, USA. Kunin, W. E. 1999. Patterns of herbivore incidence on experimental arrays and field populations of ragwort, Senecio jacobea. Oikos 84:515–525. Louda, S. M. 1989. Predation in the dynamics of seed regeneration. Pages 25–51 in M. A. Leck, V. T. Parker, and R. L. Simpson, editors. Ecology of soil seed banks. Academic Press, New York, New York, USA. Louda, S. M., and M. Potvin. 1995. Effect of inflorescencefeeding insects on the demography and lifetime fitness of a native plant. Ecology 76:229–245. Louda, S. M., and J. E. Rodman. 1996. Insect herbivory as a major factor in the shade distribution of a native crucifer (Cardamine cordifolia A. Gray, bittercress). Journal of Ecology 84:228–237. Lubchenco, J., and S. D. Gaines. 1981. A unified approach to marine plant–herbivore interactions. I. Populations and communities. Annual Review of Ecology and Systematics 12:405–437. Manly, B. F. J. 1991. Randomization and Monte Carlo methods in biology. Chapman and Hall, London, UK. Maron, J. L. 1997. Interspecific competition and insect herbivory reduce bush lupine (Lupinus arboreus) seedling survival. Oecologia 110:284–290. Maron, J. L., and R. L. Jefferies. 1999. Bush lupine mortality, altered resource availability and alternative vegetation states. Ecology 80:443–454. Maron, J. L., and E. L. Simms. 1997. Effect of seed predation on seed bank size and seedling recruitment of bush lupine (Lupinus arboreus). Oecologia 111:76–83. Marrs, R. H., R. D. Roberts, R. A. Skeffington, and A. D. Bradshaw. 1983. Nitrogen and the development of ecosystems. Pages 113–136 in J. A. Lee, S. McNeill, and I. H. Rorison, editors. Nitrogen as an ecological factor. The 202 JOHN G. BISHOP 22nd Symposium of the British Ecological Society. Blackwell Scientific, Oxford, UK. Miles, J., and D. W. H. Walton. 1993. Primary succession on land. Blackwell Scientific, London, UK. Moody, M. E., and R. N. Mack. 1988. Controlling the spread of plant invasions: the importance of nascent foci. Journal of Applied Ecology 25:1009–1021. Morris, W. F., and D. M. Wood. 1989. The role of Lupinus lepidus in succession on Mount St. Helens: facilitation or inhibition? Ecology 70:697–703. Nuhn, W. W. 1987. Soil genesis on the 1980 pyroclastic flows of Mount St. Helens. Thesis. University of Washington, Seattle, Washington, USA. Price, P. W. 1991. The plant vigor hypothesis and herbivore attack. Oikos 62:244–251. Reynolds, G. D., and L. C. Bliss. 1986. Microenvironmental investigations of tephra covered surfaces at Mount St. Helens. Pages 147–152 in S. Keller, editor. Mount St. Helens: five years later. Eastern Washington Press, Cheney, Washington, USA. Ritchie, M. E., D. Tilman, and J. M. H. Knops. 1998. Herbivore effects on plant and nitrogen dynamics in oak savanna. Ecology 79:165–177. Root, R. B. 1996. Herbivore pressure on goldenrods (Solidago altissima): its variation and cumulative effects. Ecology 77:1074–1087. Silvertown, J. W., and J. Lovett Doust. 1993. Introduction to plant population biology. Blackwell Scientific, Oxford, UK. Skellam, J. G. 1951. Random dispersal in theoretical populations. Biometrika 38:196–218. Sousa, W. P., and J. H. Connell. 1992. Grazing and succession in marine algae. Pages 425–441 in D. M. John, S. H. Hawkens, and J. H. Price, editors. Plant–animal interactions in the marine benthos. Clarendon, Oxford, UK. Statsci. 1994. S-Plus 3.3. Statsci, Seattle, Washington, USA. Ecology, Vol. 83, No. 1 Strong, D. R., J. L. Maron, P. G. Connors, A. Whipple, S. Harrison, and R. L. Jefferies. 1995. High mortality, fluctuation in numbers, and heavy subterranean insect herbivory in bush lupine, Lupinus arboreus. Oecologia 104:85– 92. Titus, J. H., and R. del Moral. 1998a. The role of mycorrhizal fungi and microsites in primary succession on Mount St. Helens. American Journal of Botany 85:370–375. Titus, J. H., and R. del Moral. 1998b. Seedling establishment in different microsites on Mount St. Helens, Washington, USA. Plant Ecology 134:13–26. Vitousek, P. M., and L. R. Walker. 1989. Biological invasion by Myrica faya in Hawaii: plant demography, nitrogen fixation, ecosystem effects. Ecological Monographs 59:247– 266. Vitousek, P. M., L. R. Walker, L. D. Whiteaker, D. MuellerDombois, and P. A. Matson. 1987. Biological invasion by Myrica faya alters ecosystem development in Hawaii. Science 238:802–804. Walker, L. R., and F. S. I. Chapin. 1987. Interactions among processes controlling successional change. Oikos 50:131– 135. Wood, D. M., and M. C. Anderson. 1990. The effect of predispersal seed predators on colonization of Aster ledophyllus on Mount St. Helens. American Midland Naturalist 123:193–201. Wood, D. M., and R. del Moral. 1987. Mechanisms of early primary succession in subalpine habitats on Mount St. Helens. Ecology 68:780–790. Wood, D. M., and R. del Moral. 1988. Colonizing plants on the Pumice Plains, Mount St. Helens, Washington. American Journal of Botany 75:1228–1237. Wood D. M., and R. del Moral. 2000. Seed rain during primary succession on Mount St. Helens, Washington. Madroño 47:1–9. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey, USA. APPENDIX A Failure time analysis of lupine survival is available in ESA’s Electronic Data Archive: Ecological Archives E083-004-A1. APPENDIX B Photographs showing the impact of insect herbivores are available in ESA’s Electronic Data Archive: Ecological Archives E083-004-A2.