Oecologia (Berl) (1982) 55:185-191 Oecologia Springer-Verlag 1982

Inflorescence : A Cost/Benefit Analysis for the Host Plant, Haplopappus venetus Blake (Asteraceae)

Svata M. Louda* Biology Department, San Diego State University, San Diego, CA 92182, USA

Summary. Predators on flower visitors, such as spiders, hand, it is known that spiders show a numerical (Green- could influence plant reproduction by determining the bal- stone 1978) or reproductive (Wise 1975, 1979) response to ance between pollination and seed predation by insects. increased prey density in specific cases, and insects on plant This study examines the net effect of predation by the in- inflorescences do provide temporary increases in prey con- florescence , Peucet& viridans (Hentz), for seed pro- centration. Several families of spiders characteristically duction by a native plant species on which it hunts. Both hunt on flowers (Comstock 1940; Marden 1963; Gertsch pollination and seed set of Haplopappus venetus (Astera- 1979; Morse 1979, 1980), and these predators may be "part ceae) were reduced on branches with spiders; however, the of a plant's battery of defenses against herbivores" (Price release of viable, undamaged seed was higher on inflores- et al. 1980). However, such predator defense against inflo- cence branches with spiders than on those without. Occur- rescence herbivores carries an implicit cost, the potential rence of P. viridans was associated with the flat-topped in- reduction of pollination by insects; predation by spiders florescence branch structure characteristic of H. venetus may be analogous to interference by ants with pollinators rather than with the vertical structure of its congener, H. or with plant parasites in ant-plant mutualisms (Carroll squarrosus. Thus, the interaction should be a reinforcing and Janzen 1973; Bentley 1976; Messina 1981 ; Skinner and selective pressure on inflorescence branch morphology of Whittaker 1981). H. venetus over time. Two factors providing constraints If the balance between pollination and predation by on the degree and rate of coevolution of the plant-spider insects is positive for the plant when a spider is present, interaction are suggested by the results: (1) the critical role one would predict selective reinforcement of traits, such of phenological synchrony and (2) the opposing require- as morphological adaptations, which attract spiders and ments of interacting species and of subsequent life history which reinforce the interaction and facilitate the mutualistic stages within a species. relationship between the plant and its defenders (Janzen 1967). Few relevant studies exist to test this hypothesis. Furthermore, data on such higher order interactions may provide insight into constraints on the degree and the rate Introduction of coevolution among interacting species. Interaction with insects can be an important aspect of plant The purpose of my study was to examine the effect biology (Harris 1973). Not only are many plants dependent of predation by the inflorescence spider, viridans on insect visitation for pollination (e.g. Faegri and van der (Hentz), on seed production by a plant, Haplopappus Pijl 1971 : Richards 1978) but, in addition, most plants are venetus Blake (Asteraceae), on which it hunts. The central subject to insect herbivory (Salisbury 1942; Whittaker question was: what is the net effect of predation by P. 1979), The visual and spatial concentration of floral re- viridans on inflorescence insects for the reproductive output sources that are necessary to attract pollinating insects of H. venetus? This involved two subsidiary questions: (1) (Baker and Hurd 1968) also make floral and developing is pollination decreased significantly in the presence of an seed tissues conspicuous to flower- and seed-feeding insects. inflorescence spider? and (2) is destruction of floral tissues Consumption of floral tissue and unreleased seed can have and seeds by insect seed predators reduced significantly by an important effect on plant fecundity (Salisbury 1942; spider foraging? In addition, I asked whether there was Janzen 1971; Bohart and Koerber 1972; Harper 1977; an association between spider occurrence and inflorescence Louda 1978; Lamb 1980; Zimmerman 1980b) and on plant branch structure. I hypothesized that: (1) localized foraging establishment (Louda 1978, 1982a, b, 1983). by P. viridans was consistent with optimal foraging theory Predators on flower visitors could influence the balance and contributed significantly to observed variation in seed between the opposing processes of insect pollination and set and seed release by H. venetus (Louda 1978, 1983); and insect-caused predispersal seed predation. The role of high- (2) the morphological form of the inflorescence branch of er order interactions in plant reproduction is not well H. venetus was attractive to spiders, enhancing continued known (Cates et al. 1977; Price et al. 1980). On the other interaction and net positive outcome for the plant. Conse- quently, data were collected on spider occurrence on H. * Present address: Duke University, Pivers Island, Beaufort, venetus, with a flat-topped inflorescence branch (Fig. 1 A), NC 28516, USA and on a closely related co-occurring species, H. squarrosus

0029-8549/82/0055/0185)$01.40 186

1979). Consequently, female spider site tenacity is excep- tionally high, from bud initiation through seed release. The newly hatched first instars remain in the egg sac 12-14 days before they molt and leave the sac (Whitcomb et al. 1966). The spiderlings thus emerge at the end of September or beginning of October (Lowrie 1963; Louda, personal obser- vation). The newly emerged spiderlings feed and eventually disperse, overwintering as second or third instar spiderlings. Seven to 9 instars (286-301 days) are required to reach ma- turity (Whitcomb et al. 1966).

4. Flower Head Insects Three groups are attracted to the flower heads: phytopha- gous species, parasitoid/hyperparasitoid species, and pollen or nectar foraging species. At least eleven phytophagous species forage on the developing flowers and seeds (Louda 1978, 1983). The most conspicuous are three tephritid flies: Urophora formosa Coquillet, Trupanea femoralis Thomp- son, and Paroxyna murina Doane. The most destructive phytophagous species are microlepidopterans in three fami- Fig. 1. Inflorescence branch structure of Haplopappus. A=H. lies: Pterophoridae, Tortricidae, and Gelechiidae (Soph- venetus Blake, B = H. squarrosus H.&A. ronia sp.). In addition, the developing ovules are fed on by larvae of two pteromalid wasps and of a curculionid H.&A., with a vertical inflorescence branch (Fig. 1 B) and weevil, Anthonomus ochreopilosus Dietz. Additionally, the with higher pollination and predation rates in the same flowers attract phytophagous thrips: Frankliniella occiden- climatic area (Louda 1978, 1982a, b). talis (Pergando), F. minuta (Moulton), and Thrips tabaci Lindeman. The phytophagous insects attract parasitoids System Studied and hyperparasitoids. A eurytomid (Eurytoma sp.), a eulopid (Tetrasticus sp.), and a parasitoid species of ptero- I. Host Plant malid attack the tephritid flies. An ichneumonid parasitoid Haplopappus venetus Blake is a small shrub, 50-150 cm tall, attacks the larvae. Finally, the pollen and nectar for- that is characteristic of the coastal sage scrub vegetation agers attracted to the flower heads include honey bees (Apis from central California to central Baja California, Mexico mellifera), a chrysidid wasp, and a halictid bee. (Munz and Keck 1959; Mooney 1977). H. venetus occurs primarily in disturbed microhabitats, such as alluvial fans, arroyos, and overgrazed horse pastures. Vegetative growth Methods occurs in winter and spring. In July, flower heads are ini- tiated and flowers occur in August-September; seeds are Plant Phenology released in October-November (Fig. 2). Development of flower head buds, flower presentation and seed maturation were recorded for all heads on three inflo- 2. Study Area rescence branches on each of three plants (N= 9 branches). The main study site was a coastal, disturbed plot at the The censuses included growth and number of flower heads junction of Carmel Valley Road and Interstate Highway in five developmental stages and were done biweekly from 5 in Del Mar, California, 22 km north of the City of San 22 July to 27 December 1976. Developmental stages of Diego. This site was adjacent to the back of Penasquitos flowers and seeds were defined as follows: (1) small buds Lagoon, an area with some of the largest stands of H. were heads less than 4.0 mm total length; (2) large buds venetus observed in San Diego County. Site characteristics were unopened heads from 4.0 mm up to presentation of and vegetation description are presented elsewhere (Louda floral buds and less than three opened flowers; (3)flowering 1978). heads were those with at least three florets with open, bright yellow floral tubes; (4) maturing heads were those with 3. Inflorescence Spider fewer than three fresh flowers but with no more than two Peucetia (Oxyopes) viridans (Hentz), the Green Lynx seeds missing and released; and (5) releasing heads were Spider, is the most conspicuous and common member of those with more than two seeds dispersed but at least two the Oxyopidae, occurring in the southern United States, seeds remaining in the head. Following the latter stage, Mexico, and Central America (Gertsch 1979). The adults the heads were considered empty. (female= 14-16 mm body length, male= 11-13 mm) are diurnal, visual hunters that forage on plants (Gertsch 1979). Phenology Western specimens of P. viridans are often associated with Insect seed predator occurrence was sampled biweekly by wild buckwheat, Eriogonum fasciculatum (Brady 1964; collecting ten heads in each developmental stage (N= 50/ M.H. Greenstone, personal communication). Maturation date). These were collected from plants intercepted along and mating occur in August (Comstock 1940). Egg sacs a random transect. Up to five heads, a maximum of two are constructed in September-October (Whitcomb et al. in any developmental stage, were taken from an individual 1966; Louda, personal observation) and are defended pers- plant. These heads were returned to the laboratory and istently by the female until spiderling emergence (Gertsch dissected within 24 h. In addition, marked plants (N= 15) 187

were examined biweekly and spider occurrence was re- Table 1. Occurrence of an adult spider in relation to host plant corded. Estimates of the relative frequency of Peucetia viri- size: 150 Haplopappus venetus plants were examined at the Carmel dans on both H. venetus and the closely related H. squarro- Valley Road site 25-28 October 1978 sus were made (25-28 October 1978). Plants were measured and presence or absence of a spider was noted for all plants Parameter Without spider With spider (N= 101) (N= 49) taller than 50 cm (N= 150 individuals/species). )? SE .g SE p a Seed Production in Relation to Spider Presence Every H. venetus individual over 50 cm tall within a 2 m x Branches per plant 48.3 3.19 99.4 4.37 * 25 m transect was examined on 1 2 November 1977 to eval- Tallest flowering 50.6 3.57 97.1 3.78 * uate seed production and damage in relation to spider pres- branch height (cm) ence. The tallest central flowering branch for twenty, equal- sized plants, ten with and ten without P. viridans, were a Mann-Whitney U test, * =P<0.05 collected and dissected in the laboratory. I recorded the total number of heads, their developmental stage, and dam- age. All flowers initiated and ovules present were scored Table 2. Flower head production and damage in relation to the for size, stage, and condition (Louda 1978). Identical proce- occurrence of adult Peucetia viridans on the tallest flowering branch dures were used to evaluate seed destruction on H. squarro- of equal-sized individuals of Haplopappus venetus a at the Carmel sus (Louda 1978, 1982a, b). Valley Road site on 1 November 1977

Results Number produced/Branch Branch p c Spider Occurrence (N= 10/treatment) Without With Spiders were associated with larger individuals of H. venetus spider spider rather than with earlier plants or with larger inflorescence branches among plants. H. venetus size, measured as maxi- 2 SE 2 SE mum height and as total branches per plant, was greater For all flower heads initiated b for those plants with adult spider than for those without Total number 78.9 9.86 81.8 8.12 ns spiders (Table 1). Flower phenology was similar between Number damaged 67.9 10.17 45.8 6.51 * the two groups, and production of flower heads for sampled plants was the same (Table 2). The flowering phenology By head developmental category was similar between branches with and without spiders (Ta- Small flower heads ble 2). In addition, the proportion of heads which had dis- Total number 14.8 0.94 15.8 1.70 ns persed their seeds by the sampling date was not significantly Number damaged 11.3 1.20 13.1 1.63 ns greater, i.e. earlier, on those branches with spiders than Maturing and releasing for those without (Table 2). Consequently, spiders occurred heads on larger plants rather than on plants with larger or earlier Total number 53.7 8.88 53.9 5.41 ns terminal inflorescence branches. Number damaged 51.5 8.93 27.7 4.97 ** Within an individual plant, however, the occurrence of Empty, released heads P. viridans was related to the flowering phenology of Total number 10.4 1.97 12.9 1.61 ns Number damaged 5.1 0.85 4.3 0.47 ns branches. First, 47 of the 49 adult spiders (95%) on the seed production transect were on the plant's tallest flower- a Mean size (height) of sampled plants: with spider=95.6 cm ing branch. Second, for plants observed in 1976 (Louda (SE=2.87) and nearest, equal-sized without spider=89.1 cm 1978), the tallest flowering branch was also the first to flow- (SE=2.56) er (93.3% of the time, N= 15 plants/site, 3 sites). So, spiders b Partitioned in subsequent section by size and stage of develop- were associated with the inflorescence branch of a plant ment at the end of the season: (1) small heads were __<0.4 cm that had the earliest vegetative and floral development. involucre, (2) large heads were > 0.4 with maturing and/or re- Flower development and spider appearance on the leasing developed seeds, and (3) empty heads were large heads tallest branch of a plant coincided. Flower anthesis ranged that had released all of the matured seeds. c Mann-Whitney U Test, * =P<0.05 and ** =P<0.01 from late August to the end of October (Fig. 2). The pre- dominant flowering period, when the highest proportion of heads were in the flowering stage (38%, N=941), was between 21 September and 7 October (Fig. 2). At the same iiiinJllt Small Buds I ...nn I nlllllll I time, the cumulative frequency of spiders observed on ex- Large Buds nnnnl perimental plants reached over 50% (54%, N=26) by 1 Flowering I Iiiiiiiii i IIiiiiii October and 65% by 7 October (Fig. 3). Spiders increasing- Maturing : iiilUl I I I IIIIInl I ly utilized inflorescences as flower development accelerated; Releasing iiiiiiii : III,~III relative abundance of adult spiders was correlated with the Empty I relative frequency of the flowering stage (Fig. 3: Spearman A S 0 N D Rank Correlation Coefficient = 1.0, P_-<0.05). 1976 Fig. 2. Phenology of flower head, flower, and seed development The adult spiders occurred on both species of Haplopap- for Haplopappus venetus at the Carmel Valley Road site (1976): pus; however, they were significantly more frequent on H. range of occurrence of each stage and two week period of predomi- venetus, which has flat-topped inflorescence branches nance of each stage, when the highest numbers of the stage were (Fig. 1 A), than on H. squarrosus, with vertical inflorescence observed. See text for stage definitions 188

9--Q Spiders flowering branch of the plant. In contrast, only three of H-roll Flowers the 150 H. squarrosus plants examined at the same time 0-...0 insects 1.00 - (2%) had an adult spider. Two of the three were females m --7...~.~ ~ tending an egg sac.

• 0.75- | Interaction with Insects // An increase in spider frequency (Fig. 3) accompanied an 0.50- .2 increased probability of flower visitation by both pollina- Y tors and seed predators. Cumulative frequencies of adult 0.25- spiders and of flower- and seed-feeding insects in all stages of development were identical between mid-September and ....o,';~ mid-October (Fig. 3). The frequency of immatures (all I I l i I I I stages) of the seed predators was highest in this period l A S 0 N 1976 (Fig. 4). The increase initially reflected oviposition by teph- Fig. 3. Cumulative occurrence over the flowering season of adult ritid flies between 15 September and I October and addi- Peucetia viridans individuals (e), open flowers of Haplopappus tionally by other insects, especially between 1-15 October venetus (u), and immature insect seed predators (all stages) in flow- (Fig. 4). Also, peak flowering, and therefore pollinator ac- er heads of H. venetus (o) at the Carmel Valley Road site, 1976 tivity, occurred during the period between 21 September and 7 October (Fig. 2)...... Tephritid Flies 0--4~ Other Insects When adult spiders were present, insects caused less 0.7- damage to flower heads. The number of maturing heads and the proportion of all flower heads that were damaged Ii by insects were significantly lower on branches with an 0.5- I k adult spider than on those without (Table 2). Spiders were /?-, observed capturing potential seed predator adults and inter- / \ fering with oviposition (Louda, personal observation). In- ."~ //i \. terestingly, the number of small buds that were damaged was not decreased on branches with spiders (Table 2). Small buds appear on terminal inflorescence branches early in the season (Fig. 2), prior to the median of the frequency J A S 0 N distribution of spiders on these branches (Fig. 3). 1976 Spiders apparently interfered with insect pollinators. Fig. 4. Relative frequency of tephritid flies (e = eggs, larvae, prepu- Lower levels of seed set were associated with the presence pae and pupae) and other insect seed predators (. = all stages) of adult spiders. The number of achenes, single-seeded fruit, in flower heads of Haplopappus venetus at the Carmel Valley Road set per flower head on branches without spiders was 5.4, site, 1976 25-28% of the total florets initiated (Louda 1978). When spiders were present, the number of seeds set per head was branches (Fig. 1B). Of the 150 H. venetus examined at the between 3.6 and 3.9, 17-/8% of florets initiated (Table 3). coast in late October 1978, 32.7% had at least one adult In addition, spiders were observed both capturing flower Peucetia viridans spider. Most of these (98.0%) were territo- visitors and interrupting visitation (Louda, personal obser- rial, persistent females with egg sacs. Among the 49 adult vation). Thus, it is likely that spiders were responsible for P. viridans observed on H. venetus, 95% were on the tallest the reduction in the proportion of flowers which were pol-

Table 3. Flowers produced and seed set on the tallest flowering branch of Haplopappus venetus in relation to the presence of the spider, Peucetia viridans, at Carmel Valley Road on 1 November 1977 Number/Flower head Tallest flowering branch p" Without spider With spider

N 5( SE N R SE

Total flowers initiated/Head b In heads with no insect damage 110 19.1 0.87 375 19.6 0.29 ns In heads with insect damage c 679 22.0 0.69 451 21.8 0.88 ns Total pollinated flowers/Head d In heads with no insect damage 22 5.4 0.68 262 3.6 0.49 * In heads with insect damage c 515 5.4 0.49 277 3.9 0.38 * Undamaged matured seeds/Headd In heads with no insect damage 22 5.4 0.68 262 3.5 0.49 * In heads with insect damage c 515 2.9 0.42 277 2.8 0.94 ns " Mann-Whitney U Test, z approximation, *=P=<0.05 b For all flower heads, independent of stage of development Flower heads with any evidence of insect feeding or oviposition on phyllaries, flower head receptacle, internal florets, or developing ovules and seeds a For maturing and releasing flower heads only 189 linated in a head, from around 26% to about 17% (Ta- of each group in the interaction, is critical to the net ble 3), an average reduction in branch fecundity of about outcome for the plant. one-third. The outcome of this specific set of interactions was posi- The net effect of spiders on development and release tive for the host plant in this case, but secondary interac- of viable seed by H. venetus, however, was positive. On tions need not be so. Even in this case at least two con- branches with spiders the lower proportion of flowers pol- straints were evident. First, the net outcome for the plant linated per head (Table 3) was offset by a highly significant depends on the timing of occurrence of the spiders in rela- increase in number and proportion of maturing flower tion to the timing of flowering. Even a slight variation in heads that escaped damage by insects (Table 2). Thus, aver- the phenology of the interaction would lead to one of two age production of viable seed on flowering branches with different results. If spider colonization had been 1-2 weeks spiders was higher than on those without: J?=286 seeds/ earlier or flower presentation or development had been branch (SE= 11.77) and J?=243 seeds/branch (SE=9.96) delayed, pollination would have been prevented. If, instead, respectively (Mann-Whitney U Test, P<0.05). The de- spider colonization of the inflorescence branch had been crease in seed set in the presence of spiders (from 395 to 1-2 weeks later or flower presentation had been early, seed 328/branch; Mann-Whitney U Test, P < 0.02) was counter- predation would not have been reduced. The timing, which acted by an increase in seed matured (from 243 to is critical to the net effect on the host plant's seed produc- 286/branch), a net increase of 17.7% in seed production tion, reflects a variety of factors to which plant, and spider, per branch (Mann-Whitney U Test, P< 0.05). and floral insects must respond independently. For example, since flowering phenology is influenced by both environmental and genetic factors (Sorenson 1941 ; Jackson Discussion 1966; Louda, personal observation), exact flowering time The essential feature of the interaction between spiders and will shift between years. Independent response to environ- insects for the host plant is the net outcome. The relative mental conditions has the potential of changing the relation intensities of the spiders' interactions with flower visitors of flower presentation and spider occurrence and, thus, pro- versus with insect consumers determines the influence of vides a potential constraint on the development of the inter- each group on successful seed release by the host plant. action. Second, the linkage is additionally constrained by Pollination success was lower on branches with spiders, but the negative effect that complete insect exclusion from the insect damage to seeds was also reduced on those branches. flower heads would have on spider reproduction through The net result was an increase in the number of viable lowering the potential food supply for the spiderlings. The seeds matured and released where spiders were present. The key point, however, is that the results suggest that there impact of spiders on overall fecundity of individuals of Hap- are significant potential constraints on the degree of interde- lopappus venetus will be determined by: plant size pendence between the plant and the spiders. (branches/plant), number of spiders per plant, and number The occurrence and timing of the adult spiders, sit-and- of branches utilized by each spider. However, since seedling wait predators, were consistent with foraging and consumer recruitment by H. venetus was directly proportional to the theory (Roughgarden 1976; Pyke et al. 1977). An inflores- number of undamaged seeds released under all environmen- cence that attracts large numbers of insects provides a local- tal conditions examined (Louda 1978, 1983), predation by ized patch of increased prey density. For a female P. viri- Peucetia viridans has potentially significant implications for dans, location on the inflorescence branch should maximize H. venetus' reproductive success. A consistent increase in the probability of encountering prey, particularly within plant fecundity as the result of the higher order interaction the constraints of constructing and defending an egg sac. - among spiders, pollinators, and plant predators - could The appearance of female spiders on inflorescence branches lead to significant cumulative, long-term effects for the pop- from the foliage was correlated with increased insect occur- ulation dynamics of this native plant. The data further rence there, suggesting a response by female spiders to in- suggest that there is a temporal" window" for flower preda- creased resource availability. Generally, spiders' habitat tors, one of the spiders' prey groups, that allows persistence preferences are influenced as much by prey availability as and significant plant impact by predispersal predators even by physical conditions (Hallander 1967, 1970; Turnbull in the face of high densities of spiders. 1965, 1972; Riechert and Tracy 1975; Morse 1979; Olive Peucetia viridans is not alone in its potential for this 1980). Response to prey density is logical since food-limited type of indirect effect on its plant host. Parallel examples reproductive success occurs and has been demonstrated for: may include salticid and thomicid spider predation on the orb spiders (Wise 1975, 1979), a lycosid (Kessler 1971), tephritid, Orellia occidentalis, developing in the flower and for the agelenids Agelenopsis potteri and A. aperta (Rie- heads of the Platte thistle, Circium canescens (Lamp 1980), chert and Tracy 1975). My results, thus, are consistent with and also other interactions of the Asteraceae with flower- predicted behavior of female spiders in relation to variation and seed-feeding insects (Louda, unpublished data). In ad- in prey concentrations. dition, predatory ant/plant interactions on inflorescences The observations of female occurrence and foraging also should have a similar cost associated with their anti-herbi- support the hypothesis of differential feeding strategies be- vore benefit. Predatory ants, responding to extra floral or tween male and female spiders (Haynes and Sisojevic 1966; floral nectar production (Carroll and Janzen 1973; Bentley Givens 1978), with female spiders responding primarily to 1976, 1977; Tilman 1978 ; Inouye and Taylor 1979; Skinner prey availability. Effective female foraging is constrained and Whittaker 1981) or to other plant resources such as by reproductive requirements, such as the site tenacity nec- hollow thorns for nesting (Janzen 1966, 1967), can interfere essary in the defense of the egg sac and the establishment with pollinators as well as with insect enemies. These studies of the sac in a location which increases the probability of lead to the suggestion that the phenology of the community food for the young (Givens 1978). The position of female of plants and , i.e., the timing of the occurrence spiders, thus, must be a compromise between exposure to 190 predators such as wasps (Muma and Jeffers 1945; Kurc- as well as between host plant species. This suggests that zewski and Kurczewski 1968; Doris 1970; Olive 1980), pre- there are other factors impinging and limiting the degree to dictability of prey, and reproductive needs. which sit-and-wait predators can exploit even relatively pre- Location on an inflorescence branch should have an dictable, high concentrations of insect prey on inflorescences. added advantage. Food for the spider's young is high on Second order interactions have been postulated as a step the inflorescence branch when they emerge in October, since toward coevolution and mutualism (Janzen 1967; McKaye immature flower- and seed-feeding insects developing in the 1977, 1979; Buss and Jackson 1979; Price et al. 1980). The inflorescences emerge in October-November. Besides being usual assumption is that once an interaction is established, abundant, the emerging microlepidopteran and tephritid it is just a matter of time before the system becomes "fine- seed predators are about the same size as the spiderlings tuned". Some systems are more closely coevolved than and are especially vulnerable from eclosion to wing harden- others (Janzen 1980); we need further analysis to know ing. Thus, foraging and ovipositing insects provide a con- why some interactions become very closely linked and centrated resource for the adult female spider, while emerg- others do not. In this case some constraints on tight linkage ing insects provide a resource for the young spiderlings. appear clear: (1) the pivotal role of phenological synchrony Further work is warranted to test whether prey availability in in the net outcome for the host plant; and (2) the conflicting the form of emerging, newly eclosed insects increases the sur- requirements of plant versus spider and of adult spider vivorship and recruitment of spiderlings as predicted here. versus spiderling. The work on this higher order interaction, Spider occurrence was associated with a specific inflo- thus, suggests two conditions which constrain the degree rescence branch morphology between the two related or the rate of coevolution of primary interactions. The species of Haplopappus examined. The efficiency of exploi- short-term interaction of predator/insect/plant described tation may be related to the area over which the search here led to differences in seed set and seed loss. Spider must be conducted. Haplopappus venetus, the plant used occurrence was patchy. Such patchiness contributed addi- more frequently, has a flat-topped, horizontal flowering tional variation in seed production. Other data on the con- branch (Fig. 1 A); those of Haplopappus squarrosus, an in- tribution of higher order interactions to variance in seed frequently used but available plant, are predominantly of set and loss among plants or frequency of their occurrence vertical orientation (Fig. 1 B). While variation in inflores- are rare. However, the findings of this study suggest that an cence branch form occurs within both species, the interspe- understanding of such interactions is crucial to our ability to cific difference is distinct and characteristic (Fig. 1A, B). explain variation in successful reproduction among plants. It is interesting to note that Eriogonum fasciculatum, another southern California plant on which Peucetia viri- Acknowledgements. I thank M.H. Greenstone and P.R. Atsatt for the encouragement provided by their interest in my initial observa- dans hunts (Brady 1964; M.H. Greenstone, personal com- tions. I appreciate the help contributed by my doctoral committee, munication), also has a distinctly flat-topped inflorescence friends, and family to my work on Haplopappus, especially the branch which is similar in form to that of H. venetus many hours cheerfully spent on data collection by G.A. Baker (Fig. 1A). The horizontal form increases the spatial concen- and G.B. Harvey. Several entomologists generously identified the tration of flowers and, thus, the probability of encountering insects : G. Marsh (Curculionidae), J.A. Powell (Microlepidoptera), flower visitors. The spatial concentration of insects is J. Hall (Hemiptera, Diptera), W.H. Evert (Thysanoptera), and G. greater on a flat, horizontal inflorescence branch since the Gordh (Hymenoptera). Discussions with L.P. Buss and K.R. surface area of that configuration was less than that of McKaye were highly instrumental in the development of my think- a cylinder composed of the same number of flower heads ing about higher order interactions. Additional suggestions by M.H. Greenstone. J.L. Hayes, N. Huntly, D.W. Inouye, D.H. in a vertical arrangement. Thus, plant morphology influ- Morse, M.V. Price, M. Stanton, N.M. Waser, and an anonymous ences prey availability and provides a characteristic to reviewer helped improve the manuscript. Support was generously which the spider can respond. provided by the Joint Doctoral Program in Ecology of San Diego The main alternative hypothesis to explain greater State University and the University of California, Riverside, and spider occurrence on H. venetus is that insect visitation and by an N.S.F. Doctoral Dissertation Improvement Grant. prey abundance are higher on H. venetus than on H. squar- rosus. This hypothesis appears unlikely for two reasons. References First, the level of seed predation in coastal areas is higher on H. squarrosus than on H. venetus (Louda 1978, 1982a, Baker HG, Hurd PD Jr (1968) Intrafloral ecology. Ann Rev b); this suggests that both the number and the predictability Entomol 13:385-414 of flower head consumers and their parasitoids is higher Bentley BL (1976) Plants bearing extrafloral nectaries and the asso- ciated ant community: interhabitat differences in the reduction on the less frequently used form, the vertical inflorescence of herbivore damage. Ecology 57 : 815 820 branch of H. squarrosus. Second, flower heads are larger Bentley BL (1977) Extrafloral nectaries and protection by pugna- and successful seed set is higher on H. squarrosus than on cious bodyguards. Ann Rev Ecol Syst 8:407-427 H. venetus in the coastal zone (Louda 1978, 1982a, b). Since Bohart GE, Koerber TW (1972) Insects and seed production. Pp. there is usually a strong relationship between seed set and 1-50 in TT Kozlowski (ed), Seed Biology, Vol 3. Academic pollinator visitation rates (Waser 1978; Pleasants 1980; Press, New York Zimmerman 1980a; N.M. Waser and M.V. Price, personal Brady AR (1964) The lynx spiders of North America north of communication), higher rates of seed set on Haplopappus Mexico (Araneae: Oxyopidae). Bull Mus Comp Zool Harvard squarrosus than on H. venetus (Louda 1978) suggest higher Univ 131(13):429 518 Buss LW, Jackson JBC (1979) Competitive networks: nontransitive insect pollinator visitation on H. squarrosus. Consequently, competitive relationships in cryptic coral reef environments. spider use of Haplopappus venetus over H. squarrosus must Amer Nat 113 : 223-234 reflect something other than absolute abundance of insects Carroll CR, Janzen DH (1973) Ecology of foraging by ants. Ann on the two species. Spider occurrence, interestingly, was Rev Ecol Syst 4:231-258 only very localized and differed among H. venetus plants Cates RG, Orians GH, Rhoades DF, Schultz JC, Tomoff CS (1977) 191

Resource utilization systems: the plant-foliage eater-predator McKaye KR (1979) Defense of a predator's young revisited. Amer system. In: GH Orians, OT Solbrig (eds) Convergent evolution Nat 114:595 601 in warm deserts, pp 166-196. Dowden, Hutchinson, and Ross, Messina FJ (1981) Plant protection as a consequence of an ant- Stroudsburg, PA membracid mutualism: interactions on goldenrod (Solidago sp). Comstock JH (1940) The spider book. Doubleday, Doran and Ecology 62:1433-1440 Company, New York Mooney HA (1977) Southern coastal scrub. In: M Barbour and Doris PR (1970) Spiders collected from mud dauber nests in Mis- J Major (eds), Terrestrial vegetation of California pp 471-489. sissippi. J Kansas Ent Soc 43:10-1/ Wiley Interscience, New York Faegri K, van der Pijl L (1971) The principles of pollination ecolo- Morse DH (1979) Prey capture by the crab spider Misumena caly- gy (Second ed). Pergamon Press, Oxford cina (Araneae: Thomisidae). Oecologia (Berl) 39:309-319 Gertsch WJ (1979) American spiders. Van Nostrand, Reinhold Morse DH (1980) Interactions among syrphid flies and bumblebees Pub, New York on flowers. Ecology 62:81-88 Givens RP (1978) Dimorphic foraging strategies of a salticid spider Muma MH, Jeffers WF (1945) Studies of the spider prey of several (Phidippus audax). Ecology 59 : 309-321 mud-dauber wasps. Ann Ent Soc Amer 38:245-255 Greenstone MH (1978) The numerical response to prey availability Munz P, Keck DD (1959) A California flora. University of Califor- of Pardosa ramulosa (McCook) (Araneae: Lycosidae) and its nia Press, Berkeley relationship to the role of spiders in the balance of nature. Olive CW (1980) Foraging specializations in orb-weaving spiders. Symp Zool Soc Lond 42:183-193 Ecology 61 : 1133-1144 Hallander H (1967) Range and movements of the wolf spiders Pleasants JM (1980) Competition for bumblebee pollinators in Pardosa chelata (OF Muller) and P. pullata (Clerck). Oikos Rocky Mountain plant communities. Ecology 61:1446-1459 18 : 360-364 Price PW, Bouton CE, Gross P, McPheron BA, Thompson JN, Hallander H (1970) Prey, cannibalism and microhabitat selection Weis AE (1980) Interactions among three trophic levels: influ- in the wolf spiders Pardosa chelata (OF Muller) and P. pullata ence of plants on interactions between insect herbivores and (Clerck). Oikos 21:337-340 natural enemies. Ann Rev Ecol Syst 11:41-66 Harper JL (/977) The population biology of plants. Academic Pyke GH, Pulliam HR, Charnov EL (1977) Optimal foraging: a Press, New York selective review of theory and tests. Q Rev Biol 52:137-154 Harris P (1973) Insects in the population dynamics of plants. In: Richards AJ (1978) The pollination of flowers by insects. Academic HF van Emden (ed) Insect/plant relationships, pp 201-210. Press, New York Blackwell, Oxford Riechert SE, Tracy CR (1975) Thermal balance and prey availabili- Haynes DL, Sisojevic P (1966) Predatory behavior of Philodromus ty: basis for a model relating web-site characteristics to spider rufus (Araneae: Thomisidae). Can Entomol 98:113-133 reproductive success. Ecology 56:265-284 Inouye D, Taylor OR (1979) A temperate region plant-ant-seed Roughgarden J (1976) Resource partitioning among competing predator system: consequences of extrafloral nectar secretion species - a coevolutionary approach. Theor Pop Biol 9 : 388 424 by Helianthella quinquenervis. Ecology 60:1 7 Salisbury EJ (1942) The reproductive capacity of plants. Bell Ltd, Jackson MT (1966) Effects of microclimate on spring flowering London phenology. Ecology 47 : 407-415 Skinner GJ, Whittaker JB (1981) An experimental investigation Janzen DH (1966) Coevolution of mutualism between ants and of inter-relationships between the wood-ant (Formica rufa) and acadias in Central America. Evolution 20:249-275 some tree-canopy herbivores. J Anim Ecol 50:313-326 Janzen DH (1967) Fire, vegetation structure, and the ant x acacia Sorenson T (1941) Temperature relations and phenology of the interaction in Central America. Ecology 48:26-35 northeast Greenland flowering plants. Medd Gronland Janzen DH (1971) Seed predation by animals. Ann Rev Ecol Syst 125:1-305 2: 465-492 Tilman D (1978) Cherries, ants and tent caterpillars: timing of Janzen DH (1980) When is it coevolution? Evolution 34:611-612 nectar production in relation to susceptibility of caterpillars Kessler A (1971) Relation between egg production and food con- to ant predation. Ecology 59:686-692 sumption in species of the genus Pardosa (Lycosidae, Araneae) Turnbull AL (1965) Effects of prey abundance on the development under experimental conditions of food abundance and food of the spider Agelenopsis potteri (Blackwell) (Araneae: Ageleni- shortage. Oecologia (Berl) 8:93-109 dae). Can Entomol 97:141-147 Kurczewski FE, Kurczewski EJ (1968) Host records for some Turnbull AL (1972) Ecology of the true spiders (). North American Pompilidae (Hymenoptera) with discussion of Ann Rev Entomol 18:305-348 factors in prey selection. J Kansas Entomol Soc 41 : 1-33 Waser NM (1978) Competition for hummingbird pollination and Lamb WO (1980) Predispersal seed predation of the Platte thistle sequential flowering in two Colorado wildflowers. Ecology and its effect on seed production. PhD thesis, Univ of Ne- 59 : 934-944 braska, 173 pp Whitcomb WH, Hite M, Eason R (1966) Life history of the green Louda SM (1978) A test of predispersal seed predation in the popu- , Peucetia viridans (Araneida: Oxyopidae). J Kansas lation dynamics of HapIopappus (Asteraceae). PhD thesis, Univ Entomol Soc 39:259-267 of California, Riverside and San Diego State University, 185 pp Whittaker JB (1979) Invertebrate grazing, competition and plant Louda SM (1982a) Limitation of the recruitment of the shrub dynamics. In: RM Anderson, BD Turner, LR Taylor (eds) Pop- Haplopappus squarrosus (Asteraceae) by flower- and seed- ulation dynamics, pp 207-222. Blackwell, Oxford feeding insects. J Ecol 70(1):43-53 Wise DH (1975) Food limitation of the spider Linyphia marginata: Louda SM (1982 b) Distribution ecology: variation in plant recruit- experimental field studies. Ecology 56:637-646 ment over a gradient in relation to insect seed predation. Ecol Wise DH (1979) Effects of an experimental increase in prey abun- Monogr 52(1) : 25-41 dance upon the reproductive rates of two orb-weaving spider Louda SM (1983) Seed predation and seedling mortality in the species (Araneae: Araneidae). Oecologia (Berl) 41:289-300 recruitment of a shrub, Haplopappus ventus Blake (Asteraceae), Zimmerman M (1980 a) Reproduction in Polemonium: competition along a climatic gradient. Ecology: in press for pollinators. Ecology 61:492501 Lowrie DC (1963) Effects of grazing and intensive collecting on Zimmerman M (1980 b) Reproduction in Polemonium: pre-dispers- a population of the green lynx spider. Ecology 44:777 781 al seed predation. Ecology 61 : 502-506 Marden L (1963) The man who talks to hummingbirds. Nat Geo- graphic Mag 123:80-99 McKaye KR (1977) Defense of a predator's young by an herbivo- rous fish: an unusual strategy. Amer Nat 111 : 301-315 Received April 21, 1982