Effects of Predation and Competition on the Population Applied Ecology 1997, 34, Dynamics of Tetvanychus Pacificus on Grapevines 8 78-8 8 8 R

Home , Competition (biology), Metaseiulus

Joirriicrl o/ Effects of predation and competition on the population Applied Ecology 1997, 34, dynamics of Tetvanychus pacificus on grapevines 8 78-8 8 8 R. HANNA, L.T. WILSON*, F.G. ZALOM and D.L. FLAHERTYt Deparlmeii~of Etiromology. Utiiversiry of’ California, Dacis, Californiu, USA

Summary 1. The Pacific spider mite Tetranyhus pacificus and the Willamette spider mite Eot- etmnyclzus rvillamettei are herbivore pests of grapevines in California. The two spider mite species share a common and often effective phytoseiid predator, the Western orchard predatory mite Metaseiitlus occidentalis. It has been suggested that E. willamettei may be beneficial in vineyards because it may have a negative impact on the more damaging T. pacijicus through their shared predator or through some form of interspecific competition. We conducted field and greenhouse experiments to determine the relative effects of these interactions between the two herbivores on the population dynamics of T.pacijicus in ‘Thompson Seedless’ grape vineyards. We also used the field data to generate a functional relationship for the combined impact of E, willamettei and M. occidentalis on T. pacificus. 2. Predation and predator-mediated apparent competition were the only factors affecting T. pacificus densities in the field experiment. The addition of the predatory mite M. occidentalis alone resulted in a significant reduction in T. pacificus densities, while the addition of E. willamettei alone had little impact on T. pacificus densities. The greatest reductions in T. pacijicus densities occurred in plots where both the predatory mite M. occidentalis and E. willamettei were added. The predatory mite occurred earliest and increased at the greatest rate in plots where it was released along with E. willamettei. From April to June, M. occidentalis abundance was positively correlated with E. willanzettei abundance while T. pacijicus abundance in July and August was negatively correlated with rM. occidentalis abundance in the period April to June. 3. Under greenhouse conditions, the addition of E. willamettei to T. paciJicus vines resulted in a marked decrease in T. pacificus population growth rate. In contrast, doubling T.pacificus abundance in the absence E. willametteidid not affect T.pacijicus population growth rate, indicating that interspecific competition is more important than intraspecific competition in determining T. pacijicus abundance. 4. Contrary to the greenhouse data and other studies in ‘Zinfandel’ vineyards, we found no evidence that E. willamettei competed with T. pacificus in the absence of predation in a Thompson Seedless vineyard. We suggest that the outcome of interactions between E. willamettei and T. pacificus may be affected by grape cultivar, growing region, a combination of cultivar and region-specific differences, or size and timing of vine infestations with E. willamettei. 5. In Thompson Seedless vineyards, vine infestation with E. willamettei would be most useful in increasing the effectiveness of the predatory mite M. occidentalis in controlling T. pacijicus, primarily through an elevated predatory mite abundance prior to the development of large T. pacificus populations. Key-words: Acari, Eotetranychus willamettei, Metaseiultu occidentalis, predator- mediated apparent competition, predator-prey interactions, Vitis tlinijera. Journal of Applied Ecology (1997) 34, 878-888

* Present address: Department of Entomology, Texas A&M University, College Station, TX 77543, USA. 1997 British t Present address: University of California Cooperative Extension, County Civic Centre, Agricultural Building, Visalia. CA Ecological Society 93291, USA.

878 led to speculations that the presence of E. willamettei, 879 Introduction R. Hunnu et al. especially during spring, may prevent outbreaks of T. Direct effects of predation and interspecific competition pacificus, possibly by leading to an early increase in have long been implicated as important factors affect- populations of M. occidmtalis (Flaherty & Huffaker ing the population dynamics of arthropods in natural 1970; Flaherty & Hoy 1971). If this hypothesis is true, systems (e.g. Schoener 1983; Connell 1983; Sih et a/. then predator-mediated apparent competition 1985). More recently, it has been suggested that indirect between T. pacificus and E. willanwttei may be an effects produced by shared predation (i.e. one or more important factor in the dynamics of this acarine pred- predator species sharing more than one prey species) ator-prey system. More recently, results from green- may also cause profound effects on prey population house and field experiments have suggested that nega- dynamics (e.g. Sih et al. 1985; Kerfoot & Sih 1987). tive effects of E. willameftei on T. pac(fictrs may be One class of indirect effects of shared predation results caused by interspecific competition or ‘induced resist- in population and community patterns similar to those ance’ (English-Loeb & Karban 1988; Karban & Engl- produced by interspecificcompetition (Holt 1977,1984; ish-Loeb 1990). However, these alternative responses Holt & Koller 1987; Holt & Lawton 1994; Holt, Grover were cultivar-specific. While they were apparent on & Tilman 1994). Holt (1977) termed this interaction ‘Zinfandel’ and ‘Chardonnay’ cultivars, they were predator-mediated apparent competition, which arises weak or lacking on ‘Thompson Seedless’ grapevines. from strong predator numerical responses to prey den- Also lacking were the relative effects of interspecific sities. Despite the theoretical appeal of apparent com- and intraspecific competition on spider mite densities. petition and the existence of numerous examples of Clearly, several biological processes may be impor- predator-prey interactions that are in line with its pre- tant in determining population abundance and pest- dictions (Holt 1984), there have been few experimental status of spider mites in vineyards. In this paper we demonstrations of its existence (Collyer 1964; Schniitt report the results of a field experiment designed to 1987; Settle & Wilson 1990; Huang & Sih 1990; assess the relative significance of predation, inter- Karban, Hougen-Eitzman & English-Loeb 1994). specific competition, and predator-mediated apparent Although apparent competition theory was developed competition on the short- and long-term dynamics of for prey that do not compete directly (Holt 1977), coni- 7”. pac@ws populations in Thompson Seedless vine- petition for resources and predator-mediated apparent yards in the San Joaquin Valley of California. We competition are not mutually exclusive hypotheses and also report on the results of a greenhouse experiment are likely to be complimentary forces determining sizes designed to compare the relative significance of inter- of animal populations (Holt 1984; Holt et al. 1994; Sih specific competition (concurrent feeding by E. wil- et al. 1985). lumettei) and intraspecific competition on the population In this paper, we explore the relative effects of pre- growth of T. puc$icus. In conducting this study, we dation, interspecific competition, and predator- were also motivated by the practical desire (i) to deter- mediated apparent competition on the population mine the viability of vine infestation with the less dynamics of the Pacific spider mite Tetranychus damaging E. willarnertei to increase the effectiveness pacificus McGregor (Acari: Tetranychidae), which is of M. occidentalis in reducing densities of the more a serious pest of grapevines Vitis uinfera L. in the San damaging T.pacificu.~, and (ii) to generate a functional Joaquin Valley of California. Large populations of relationship for predicting the impact of M. occi- this spider mite species are capable of inflicting con- dentalis densities on T. pac$cu.s populations in the siderable damage to vine growth and grape yield and presence of different levels of E. wil/mne/fei. quality (Flaherty et NI. 1992). The Willamette spider mite Eofetranychus willamettei (McGregor) (Acari: Methods Tetranychidae) is also commonly found on grapevines, but generally causes considerably less damage LlFE HlSTORY OF THE MlTES than comparable densities of T. pacificus, particularly in vineyards of the Southern San Joaquin Valley, Cal- Both T. pacificus and E. cvillumettri are endemic to ifornia (Flaherty 1969). Both T. pacificus and E. the far western United States. They were first reported willamettei are attacked by the predatory mite on grapes in California vineyards by Lamiman (1 935). Meta&idu.s ( = Typhlodromus = Galendromus) Both species are found throughout California, but T. occidentulis (Nesbitt) (Acari: Phytoseiidae), which is ppc~ficicusis most commonly found in the warm interior their principal biological mortality agent in vineyards TetrFnychus pyc@:us, like (Flaherty et al. 1985). arine’genus Tetr~/?ychus,has That T. pactficus and E. willamettei Co-occur on a wide host range, infesting a variety of perennial and grapevines and share the same predator has led to annual plant species, and is a pest of major crops in speculations that the two species might compete the western United States (Pritchard & Baker 1955). $> 1997 British EcologicalSociety, directly (interspecific Competition) or through the In contrast, E. willumettei infests only a few perennial .iouma/qf’A,lp/ipc/ shared predator (predator-mediated apparent com- plant species including apple Ma/u.s spp., antelope Ecology. 34, 878-888 petition), Observations from several vineyards have brush Purshia tridenratu, box elder Acer negundo cali- 880 ,/ornictin~,service berry Amelunchier sp., Quercus gar- predation, interspecific competition, and predator- Predution und ryanu, and both wild and cultivated Vitis spp. (Frazier mediated apparent competition on the population competilion in an & Smith 1946; Pritchard & Baker 1955). While the dynamics of T. puc us. We conducted this experi- acarine communi/)) biology of T.pac$cus is well known on several hosts, ment in a section of a 35-year-old Thompson Seedless little is known about the biology of E. willamettei on vineyard (located near Madera, California) where T. hosts other than grapes. pac1ficu.s reached outbreak levels during the previous Both T.pucificus and E. willumettei are residents of year, while E. willumettei and M.occi~kntolis were rare vineyards; they overwinter as mated adult females (R. Hanna, unpublished data). We used a randomized under loose bark (Flaherty & Huffaker 1970; Kinn & complete block design with four treatments and three Doutt 1972). In late winter and early spring, adult replicate blocks. Each subplot was four vine rows females of both species become active and colonize (14.6 m wide) by 10 vines (24.4 m long), and was sep- newly expanded leaves soon after bud break (Flaherty arated from other subplots within each block by 15 & Hufhker 1970). Generally, E. willurnettei appears buffer vines (36.6 in). Four vine rows (14.6 m) served to be more active and able to reproduce faster than as buffers between blocks. This experimental design T. pucijcus during the cool spring temperatures (R. represented a trade-off between large plot size that Hanna, personal observation). It is not known, minimized between-plot interference, and small plot however, if the differences in early season activity size that minimized between-plot variability. We chose between the two spider mite species is due to differ- three replicates of larger plots, as opposed to a greater ences in critical temperature thresholds or the shape number of replicates and smaller plots, to minimize of the development-temperature curves. Both species between-plot interference. We reduced sampling error produce several generations during the grape growing by taking a large number of sample units (Le. grape season and can reach high densities during the summer leaves) from each plot. months. Treatments were randomly assigned to plots within Metuseiulus occidentalis is the dominant natural each of three blocks. In one treatment, we added E. enemy of spider mites in California vineyards (Flah- willamettei and M. occidmtalis. In two other treat- erty & Huffaker 1970; Kinn & Doutt 1972). Although ments we added either E. willunwttei or M. occi- indigenous to western North America, M. oecidentalis dentulis. We did not manipulate predator and prey is now found in many areas of the world (McMurtry abundance in the fourth treatment, which served as 1982). It is associated with many mite species on many the control. [Another experiment was conducted to annual and perennial host-plants (Schuster & Prit- determine the impact of M. occic/ent~ili.sand T. chard 1963). Metuseiulus occidentulis can be found in puc~ficus on E. willamcttei Hanna (1992)j. Eot- vineyards throughout the year, and is capable of long- etranychus willumettei needed for our experiment was and short-range dispersal between crop hosts (Hoy collected from a nearby vineyard where this mite was 1982). In vineyards, M. occidentalis overwinters under abundant. We treated source vines with methomyl bud scales as diapausing, non-gravid and mated adult (Dupont Co., Richmond, VA) 1 week prior to col- females (Hoy & Flaherty 1970). In the spring, over- lecting E. willumettei to eliminate naturally occurring wintering females colonize young leaves where they individuals of M. occidrntulis. Methomyl is toxic to feed on E. willainrttei, T.pacificu.~ and the tydeid mites M. occidentulis (with a residual toxicity of at least 5 Homeopronemutus (Pronematus) anconai (Baker) and days), and relatively harmless to E. willumettei (Hoy Pronematus ubiquitous McGregor. et a/. 1979). Vines in the experimental area were not Soon after the initiation of vegetative growth, pred- treated with methomyl at any time during the season. ators and prey colonize the newly emerging leaves, On 10 May 1989, we added 156 +_ 45 (+ 1SE) imma- and remain on the lower leaves of the growing shoots ture and adult E. willumettei individuals to each vine for much of the early spring. During mid-to-late within each of the E. willunwttei release plots. We spring and in the summer, predators and prey move also added 67 & 19 (I1 SE) mobile individuals of M. outwards along the elongating shoots (Hanna et ul. occidentalis (Biotactics Inc., Riverside, CA) per vine 1996). Tetrunychuspuc~ficusand E. willumettei overlap in predator release plots on 21 May 1989. considerably in their distribution on the vines but We sampled all plots at approximately 3-week inter- during the summer, T. pacificus occurs in greater vals between 24 April and 2 June 1989, and at approxi- abundance on sun-exposed areas of the vines, and mately 2-week intervals between 2 June and 14 August E. willumettri occurs throughout the vines but with 1989. On each sampling date, we collected 20 leaves slightly greater abundance on shaded leaves (Hanna from the middle six vines of the two centre vine rows et ul. 1996). These overlapping distributions on vines within each plot. On the first three sampling dates, we may increase the likelihood of direct or indirect inter- collected leaves from the basal region of vine shoots. actions between T. pacificus and E. willumettei. During the remainder of the season, leaves were selected from the middle portion of shoots. On all sampling (<> 1Y97 British FIELD EXPERIMENT dates, shoot selection was divided equally between Ecological Society. Journal II/ Applied We conducted a field experiment during the 1989 four zones of the vine canopy: south, north, top and E~(~lo~~.34. 878-488 growing season to determine the relative effects of interior. Although sampling was stratified, shoots 88 1 within zones and leaves from each mid-shoot region as the ratio of final mite and initial mite density) were R. Hunna et al. were selected randomly. All life stages of spider mites analysed with a one-way analysis of variance. We used and M. oc*cidentalison leaves were counted under a planned contrasts to test for the effects of interspecific dissecting microscope in the laboratory. and intraspecific competition on changes in T. We used univariate repeated measures analysis of puc$cus population size. variance of the two main effects (M. occidentalis addition and E. willanwtiei addition) and their inter- Results action to determine the impact of M. occidentalis and E. willamettei on the abundance of T. pacc@cus, E. F I ELD EX P E R 1 M E NT willamettei and M. occidentalis. We also quantified the effects of sampling date and the interaction of We conducted this field experiment to test the relative sampling date with the main effects. All analyses were effects of predation, interspecific competition, and conducted with SAS PROC GLM (SAS Institute predator-mediated apparent Competition on T. 1989). Due to heterogeneity of variance inherent in pacificus abundance. We assumed that rates of mor- census data, all analyses were performed on log-trans- tality due to predation by M. occidentalis, and the formed values of the dependent variables. effects of interspecific competition, can be inferred from between-treatment differences in T. puc$'cu.s abundance. Seasonal trends of T. pucificus, E. 1~~;l- GREENHOUSE EXPERIMENT lamettei and M. occidentulis are presented in Figs I, 2 We conducted a greenhouse experiment on 2-month- and 3. The results of univariate repeated measures old grapevines to measure the effect of interspecific analysis of variance on each species are presented in and intraspecific competition on changes in T. Table 1. puc~ficuspopulation size. We established two solitary We succeeded in establishing large differences in E. and one mixed species treatments. One solitary treat- willnmetiei densities in our experimental plots. Our ment received 15 adult T. pac$cu.s females, and the efforts were especially aided by low background levels other solitary treatment received 30 adult T.puctfictrs of E. willamettei in the control plots, where E. id- females. The mixed-species treatment received 15 lumettei abundance remained very low throughout the adult T. pac$cus females and 15 adult E. willumettei experiment (Fig. 1). The addition of E. willamc.rtPi females. There were five replicate vines for each treat- explained the greatest proportion of variation in num- ment. All vines were uniform in size and age, with a bers of this species. In plots where E. willamcttc~iwas single shoot and a total of seven leaves for each shoot. added, its seasonal density reached a significant 2.7- Vine and mite source and culture were as described fold higher level than in non-addition plots. The by Hanna & Wilson (1991). Eight days after the addition of M. occidenialis appeared to reduce E. id- addition of spider mites to the vines, all leaves were lamettei population size but this effect was not sig- harvested and chilled to 3 "C, and all stages of spider nificant in our experiment (Fig. 1 and Table I). Fur- mites were counted in the laboratory. thermore, the effect of E. willumeitei augmentation on Changes in T. pucijicus population size (expressed its seasonal density was independent of our efforts to

Table 1. Repeated measures iinalysis of variance* on mite counts where E. ii,illumettei and M. occitlen/uli.s were simultaneously nianipulated to test their impact on the population dynamics of T.pacificus

T. pucificus E. willanzc~irri M. occirliwtulis - - Source of variation d.f. F P F P F P

- -_____ -I Between treatments? M. occirlenkdis addition 1 67.50 C0.005 0.85 0.391 32.16 0.001 E. willanwtiei addition I 6.59 0.043 107.61 <0,005 4.54 0,077 Interaction I 3.81 0.098 1.08 0.338 5.36 0,060

Within treatmcntst Date 7 2 18.30 < 0.005 97.97 <0.005 32.70 <0,005 Date x M. occidmlalis addition 7 28.70 <0.005 1.84 0.104 21.58 <0.005 Date x E. willumetieiaddition 7 2.33 0.042 11.69 <0 005 2.12 0.062 Date x M.occidenlnlis addition x E. ivillnme/tei addition 7 2.43 0.035 1.44 0.217 2.24 0,049

*Analysis performed on log-transformed values of the dependent variables. 8') I997 British ?All block effects had a P-value greater than 0.05. Error degrees of freedom for belween-treatlllerlt efrects = 6. Error Ecological Society, degrees of freedom for within-treatment effects were adjusted where necessary with the Huynh-Feldt Epsilon (c = 1.066 Jotirncrl o/ Applied for T. pucifictis, E = 1.188 for E. willrimet/ei and t: = 973 for M. occ~idontulis).Degrees of freedom were not adjusted for Ec,olog~,,34, 878-888 e z I, Unadjusted residual d.f. = 42. 882 Predation and --e E. willamenel and competition in an ucarine community - E. willametlei added

April June July August Date of sampling Fig. 1. Abundance of E. iuillame/tc.ion grapevines from 24 April to 14 August 1989. Data are mean density of E. ivillamelrei per leaf with standard errors.

manipulate predator densities (Fig. 1 and Table 1). average seasonal density of T. pacificus was greater Although M. occidentalis is generally capable of con- on vines where E. willumettei was not added. There trolling E. willamettei densities, we cannot make any were also indications that E. willamettei releases might conclusions on the impact of M. occidentalis on E. have increased the success of M. occidenlalis in reduc- willumettei (when present alone) in this study because ing the size of T. pac(ficus populations (P = 0,098). A T. pacificus was present in all plots. This issue is con- lack of significant interaction effect was probably due sidered in another study that examined the impact of to large subplot variability and the cumulative nature T. pacificus and predator addition on the population of errors associated with the interaction effect. We dynamics of E. willamettei (Hanna 1992). cannot interpret the E. willumettei effect as evidence Fluctuations in E. willamettei abundance through for some form of interspecific competition because the the season was indicated by a significant overall date effect of E. willanwttei addition was averaged over effect (Table l), with these fluctuations being affected predator plots and therefore contained some indirect most by the addition of E. willamettei. Densities of effects. Post hoc comparisons of pairs of cell means this mite increased earlier and faster where it was (or simple effects) were necessary for partitioning the added (linear time contrast, P = 0.007). Rate of direct and indirect effects of E. willamettei on T. change in E. willamettei densities was not affected by pucificu.~densities. Also, the interaction between sam- predator release (P = 0.246), and the temporal pling dates and treatments provided greater sensitivity dynamics of E. willametfei were independent of pred- for testing the null hypothesis of treatment effects, and ator release (P = 0.591). allowed subplots to act as their own control to a Tetrunychus pacificus densities differed con- greater extent than in only main effect analysis (Dr siderably between treatments (Fig. 2). The average Neil Willits, Statistical Laboratory, University of seasonal density of T.pacificus in non-predator release California, Davis, personal communication). plots was 2.8-fold greater than in predator release Since conservatism was desirable for post hoc com- plots. Tetranychus pac$cus densities were also nega- parisons, we used Tukey’s technique (Neter, Wasser- tively affected by the presence of E. willamettei. The man & Kutner 1985) to test each treatment mean

200 r E. willamenei and

- E. willamettei added - Control

April June July August Q 1997 British Ecological Society, Date of sampling Journal of Applied Fig. 2. Abundance of T. pacijicus on grapevines from 24 April to 14 August 1989. Data are mean density of T. pacIficus per Ecology. 34, 878-888 leaf with standard errors. 883 --e E. willamellei and R. Hannu et al. M. occidenlnlis added - M. occidenlnlis added T

April May June July August Date of sampling Fig. 3. Abundance of M. occidentalis on grapevines from 24 April to 14 August 1989. Data are mean density of M. occicicv~tulis per leaf with standard errors.

against the control (which did not receive M. occi- considerably between treatments (Fig. 3). This pred- dentalis or E. willamettei). Of interest were three com- ator apparently was able to colonize vines where it parisons used to test the simple effects on T. pacificus was not released or was initially absent. Average sea- by (i) adding only M. occidentalis, (ii) adding only E. sonal M. occidentalis abundance was l ,5-fold higher willamettei, and (iii) adding both E. willamettei and on release vines compared to non-release vines despite M. occidentalis. The first and second comparisons, the presence of greater prey abundance 011 non-release respectively, measured the separate direct effects of vines. The larger average seasonal abundance of M. M. occidentulis and E. willamettei on T. pacificus occidentalis in release plots was the probable cause of abundance. The third comparison measured the lower T. pacificus abundance in these plots compared indirect components in the effects of M. occidentalis with those where M. occidentalis was not added. There and E. willamettei on T.pacificus abundance. was also some indication that the addition of E. wil- The addition of M. occidentulis in the absence of E. lamettei to the system resulted in an increase in the willanzettei releases resulted in a twofold reduction in average seasonal abundance of M.occidenta1i.r (Table T. pac$cus abundance (P < 0.05), whereas the I, P = 0.077), and this effect was dependent on addition of E. willamettei in the absence of M. occi- whether or not M. occidentalis was released (main dentalis releases resulted in negligible reductions in T. effects interaction, P = 0.060). Again, due to the mar- pacificus abundance, thus supporting the hypothesis ginal non-significance of E. willanzettei addition and that E. willumettei has little direct effect on T. the interaction between E. willamettei and M. occi- pacificus. The greatest reduction in the size of T. dentalis addition, we performed post hoc comparisons puc$cus populations occurred in the presence of both of treatment means to clarify the nature of the main M. occidentalis and E. willaniettei. Addition of these effects and their interaction. two species resulted in a 4.7-fold reduction in T. The addition of E. willamettei alone had little effect pacificus densities. on the average seasonal density of this predator In addition to differences in the average seasonal (P > 0.05). This outcome was expected as predators abundance of T. pac$icus, there were also differences were almost absent from April to the end of June on in the temporal dynamics of this species resulting from non-release vines. The average seasonal abundance of the manipulation of M. occidentulis and E. willunzettei M. occidentalis was greatest where this predator was (Fig. 2 and Table I). Although predator releases released on vines that also received E. willamettei explained the greatest proportion of the temporal vari- (P < 0.05), and was also significantly greater in M. ation in T. prrc$icus abundance, this effect also occidentalis release plots without E. wilbmettei depended on E. willamettei addition. The addition of addition compared with control plots (P < 0.05). E. willamettei alone had little effect on the rate of Analysis of temporal changes in M. occidenialis increase of T.pacificus compared to the control (linear abundance indicated that there was a highly sig- time contrast, P > 0.05), whereas predator addition nificant sampling date by treatment effect on this spe- significantly reduced the rate of increase of T.pacijcus cies' abundance, with the greatest effect being due to regardless of E. willamettei addition (P < 0.05). How- predator release. There was also an indication that E. ever, T. pacificus abundance appeared to have willumettei addition afyected the temporal dynamics increased faster in the presence of M. occidentalis of M. occidentalis as indicated by the date and E. 0 1997 British EcologicalSociety, alone than in the presence of M. occidentalis and E. willanzettei interaction (P = 0.062), and by the three- Jourltal qfAp[i/jed wil/amettei together (Fig. 2). way interaction between date, E. willurnettei addition Ecologv, 34, 878-888 Average seasonal densities of M. occidentalis varied and M. occidentulis release (P = 0.049). It is therefore r = 0.864 P c 0.001

I 0.2 0.4 0.6 0.8 1.0

Mean relative density of E. willamellei per leaf Fig. 4. Relationship between relative E. ~1,illarneIIeidensity(expressed as a proportion of the combined densities of E. ii~il/an(,//ei and T. pucIficus) and loy-transformed M. occidenictlis density during the period 24 April to 29 June 1989. Each data point is the mean density of mites per leaf from an experimental unit (n = 12).

necessary to interpret changes in predator abundance occidentalis was more abundant in release plots com- in terms of this three-way interaction. This was best pared with non-release plots (two-factor ANOVA with examined with the analysis of predator abundance blocking, P = 0.002). Eotetranpclius willumettei during the period 24 April to 29 June, which preceded release also resulted in a net increase in M. occidentulis any appreciable predator increase in plots where they (P = 0.01). with this effect being greatest where M. were not released (Fig. 3). Linear time contrasts over occidentulis was released together with E. ivillanwltei the entire experiment were not appropriate because (P = 0.022), indicating that the addition of E. cd- M. occidentulis colonized non-release vines and con- lunzettei as alternative prey for M. occidentulis sumed all prey by the end of the experiment. The increased the numbers of this predator during April, predator’s numerical response (although delayed) in May and June. This is also indicated by a highly non-release plots masked the effects of early season E. significant correlation between relative E. willaniettei ivillumettei densities on predator release. abundance and M. occidentalis abundance during the Meta.reiulu.7 occidentalis was almost completely period from 24 April to 29 June (Fig.4; r = 0.864; absent from all plots on 24 April and 11 May. Pred- P < 0.001). Furthermore, there was a highly sig- ator densities began to increase after the 2 June sam- nificant negative correlation between M. occiclentrrlis pling in the release plots, but remained very low in abundance in April, May and June and T. pacificus non-predator release plots (predators released on 21 abundance in July and August (Fig. 5; r = 0.81 1; May). There were significant main effects and inter- P = 0.001). The predator’s numerical response to the action effects on M. occidentulis abundance during the addition of E. willuniettei and the resulting increase in period between 24 April and 29 June. Metaseiulus the predator’s impact on T.pacificus densities are both

r = 0.811 P=O.OOI

11t 0 0 .I -2 -1 0 1 2

Mean density of M. nccideralalis per leaf (log transformed) $I> 1997 British Fig. 5. Relationship between April-June M. occideniulis density and July-August T. pucrJcus density. Each data point is the Ecological Society, log-transformed mean number of T.pac/fieus or M. occidenialis from an experimental unit (n = 12). The x-axis represents the .Journnl of’ Applied mean density of M. oceidcnrulis per leaf for the period of 24 April to 29 June, and the )!-axis represents the mean density of T. go log^., 34, 878-888 pnc{/ieus per leaf for the period of 29 June to 14 August. 885 R. Hnnna et al.

T. pacipcus (15) T. pacifcus (30) T. pacifcus (15) + E. wiiiamettei (15) Initial mite density Fig. 6. Relative change in 7'. pac(ficus density (expressed as the ratio of final and initial 7'.pacificus densities) as affected by interspecific and intraspecific competition in the greenhouse experiment. Vertical bars are standard errors of the means.

evidence of the occurrence of predator-mediated Numerous studies have shown that M. occidentalis apparent competition between E. willamettei and T. strongly suppresses spider mite densities on several pacificus. crop plants (McMurtry 1983). On grapevines, this predation effect on T. pacificu.~was caused by strong predator numerical and functional responses that INTERSPECIFIC AND INTRASPECIFIC were probably enhanced by the aggregated spatial COMPETITION IN THE GREENHOUSE distribution of T. pacificus and area restricted search The effects of interspecific and intraspecific com- by M. occidentalis (Hanna & Wilson 1991). petition on the population growth of 7'. pac(ficus in This study has also shown that the impact of pre- the greenhouse are summarized in Fig. 6. The presence dation on T. pacificus by M. occidentalis can be of E. willamettei on vines with T.pacificus led to a 37% enhanced by the presence of E. willamettei as alternate reduction in the change of T.pacificus population size prey, particularly during the early growth phase of T. (F,,,2= 5.04, P = 0,044) compared with vines without pac(ficus populations. The observed impact of shared E. willamettei, indicating that some form of inter- predation on T.pacificus abundance was hypothesized specific competition occurred between the two species. by Flaherty (1967), who observed a negative cor- Furthermore, under the conditions of this experiment relation between the abundance of E. willunwttei and T. pacificu.r was not affected by conspecifics T. pacificus. We demonstrated that the addition of (F,,lz= 1.77, P = 0.208), indicating that intraspecific both E. willamettei and M. occidentulis caused the competition is not as important as interspecific com- greatest reductions in 7.pacificus abundance (Fig. 2), petition in determining T.pacificus abundance. These which was apparently caused by an enhanced numeri- results must be interpreted with caution, however, cal response by M.occidentulis (Fig. 4). since the ultimate test of competition should be con- Our study is one of only few studies (Collyer 1964; ducted in the field (Schoener 1983). The greenhouse Schmitt 1987; Huang & Sih 1990; Settle & Wilson experiment was intended only to help us understand 1990; Karban et al. 1994) that have experimentally the relative effects of interspecific and intraspecific demonstrated the existence of predator-mediated competition on changes in T.pacificus population size apparent competition. We have shown that predator- under conditions which would have been difficult to mediated apparent competition is a significant eco- achieve in the field. logical mechanism affecting the dynamics of spider mites on Thompson Seedless grapevines, and that apparent competition was caused by a strong M. occi- Discussion dentalis numerical response to the addition of E. wil- This study has shown that the direct effect of pre- lamettei. The potential for apparent competition dation and the indirect effect of predator-mediated would have been weakened, however, if the numerical apparent competition strongly affected the population response by M. oceidentalis had been dampened by dynamics of T. pacificus in the field. Tetranychus intraguild predation (Holt 1977; Holt et (11. 1994; Holt pac$cus densities were significantly greater in control & Lawton 1994). c) 1997 British Ecological Society, plots (initially free from predators) compared with Although several insect and mite predators can be journal?f &,>lied M. occidentalis release plots. That predation strongly found on grapevines in California, the six-spotted Ecology, 34,878-888 affected T. pacificus abundance is not surprising. thrip Scolothrips sexniaculatus (Pergande) is the only 886 other spider mite predator found in large numbers in Loeb (1990) and English-Loeb et ul. (1993). The Predution mu' vineyards and is primarily associated with high spider strength of the negative interactions between the mites conipetition in an mite densities (Flaherty et u1. 1992). Scolothrbs sex- (in the absence of predation) may be affected (i) by ucurine community niaculatus prefers spider mites but it can feed on M. cultivar-specific differences (Zinfandel vs. Thompson occidentalis when spider mites are scarce (R. Hanna, Seedless), (ii) by climatic variables associated with unpublished data). Therefore, intraguild predation is the two geographical regions where the studies were most likely to occur during the declining phase of the conducted (Lodi is generally cooler than Madera), seasonal spider mite cycle, when both M. occidentulis (iii) by a combination of cultivar and region-specific and S. .sexn~~iculatushave consumed most of the spider differences, or (iv) by differences in the timing and size mite prey. Intraguild predation did not play a role in of E. willunzettei releases [negative effects are more our experiment because S. .sexmuculutus was absent pronounced when E. willarnettei is present shortly during much of the duration of the experiment. after the initiation of vegetative growth (Hougen-Eitz- In the greenhouse study, the change in T. pacificus man & Karban 1995)l. These differences in exper- population size was significantly lower in the presence imental conditions may have affected the outcome of of E. willumettei, indicating that this species also com- the experiments and our conclusions regarding the petes with T. pucificus in the absence of predators. ecological mechanisms affecting the dynamics of this This finding is consistent with earlier studies (English- acarine community. Loeb & Karban 1988; English-Loeb, Karban & The experiments reported in this paper were con- Hougen-Eitzman 1993) where previous or concurrent ducted in part to test the relative significance of the feeding by E. willunzettei resulted in reductions in T. ecological mechanisms affecting spider mite dynamics pacificus abundance; however, the mechanisms of in vineyards, and to obtain practical information that these interactions between T. pacificus and E. wil- can be used by pest management practitioners to reduce luniettei are not well known. Eotetrunjtchus willamettei the occurrence of damaging T.pac~ficus outbreaks. We can compete with T.paciJicus (i) for resources (exploit- have determined that infestation of Thompson Seedless ative competition), (ii) by direct interference, or (iii) vines with E. willumettei is a viable management by an induced response such as shown by English- approach. However, we were unable to show (with the Loeb & Karban (1988) and English-Loeb et ul. (1993). densities and timing of E. willumettei infestation) that If exploitative competition was more important than this mite alone would reduce T. pacificus densities. other mechanisms of competition between E. wil- Under the conditions of our experiments, vine infes- lumettei and T. pacificus, population growth of the tation with E. willamettei was most useful in inc latter should have been negatively affected by intra- M. occidentalis densities, which increased the effec- specific Competition. Both species of spider mites have tiveness of this predator in controlling T. pczcificu.~ similar physiological effects on grapevines under field populations. Furthermore, as indicated in Figs 4 and and greenhouse conditions (Welter et ul. 1989), yet T. 5, this study provides useful information for predicting pacificus population growth was lower in the presence predator effectiveness in controlling T. pac;ficus popu- of E. willumettei than in the presence of equal densities lations under varying levels of E. willumettei. of conspecifics (Fig. 6). If these similar physiological Although we have shown that vine infestation with effects are related to depletion of resources, then E. willamettei was useful in reducing T.puci~5cus infes - exploitative competition between E. willamettei and tations, caution should be exercised in using this T. pacificus is not as important as other mechanisms tactic, because the pest-status of E. wi//uniettei of competition. depends on the grape cultivar and its geographical Although we were able to document in the green- location. Vine infestation with E. willmiettei is house that E. willamettei competed with T. pucificus unlikely to cause significant problems on Thompson in the absence of predation, we did not obtain any Seedless grapes in the Southern San Joaquin Valley, evidence for the presence of competition (other than where E. willumettei rarely causes significant losses in predator-mediated apparent competition) between grape yield and quality. Eotetranychus cvilluiizettei is the two species in the field, contrary to findings by more likely to be a pest on wine grape cultivars (e.g. Karban & English-Loeb (1990) and English-Loeb et Zinfandel, Chardonnay and Pinot noir) in dry-land a/. (1993) (but see Karban et ul. 1994). Working in vineyards in the foothills of the Sierra Nevada moun- Zinfandel vineyards in the Northern San Joaquin Val- tains (Welter et d. 1989), the Northern San Joaquin ley (Lodi, California), these authors found that E. Valley (R. Haniia & F. Zalom, unpublished data), willumettei negatively affected T. pucijicus abundance and the coastal valleys (Flaherty e/ al. 1992). If vine independent of the effects of predation. In contrast, we infestation with E. willamettei is used, augmentative did not find any evidence that E. willumetteicompeted releases and/or conservation of M. occirlentulis should with T. pucificus (in the absence of predation) in lower the risk of the development of damaging E. Thompson Seedless vineyards in the Southern Sail willumettei populations on grapes grown in vulnerable 0 1997 British Ecological Society, Joaquin Valley (Madera, California). Several expla- regions, while increasing the viability of vine infes- Journal of Applied nations are offered for the lack of concordance tation with E. wil1umettc.i to control T. pucificu.~out - Ecology, 34,878-888 between our study and that of Karban & English- breaks. I 887 Acknowledgements of predalion and conzpc.ti/ion.PhD thesis, University of I R. Hanna et al. California, Davis. The authors wish to thank J. Carey, A. Corbett, R. Hanna, R. & Wilson, L.T. (1991) Prey preference by Mti- Holt, J. Rosenheim, T. Schoener, D. Spiller and S. useiulus occidentalis (Acari: Phytoseiidae) and ihe role of Welter for their helpful comments on earlier versions prey aggregation. Biological Control, I, 51-58, Hanna, R., Wilson, L.T., Zaloin, F.G., Flaherty, D.L. & of this manuscript. We especially thank M. Sabelis for Leavitt, G.M. (1996) Spatial and temporal dynamics of his review and his numerous suggestions that sub- spider mites in ‘Thompson Seedless’ vineyards. Eni)iron- stantially improved the quality of the manuscript. We meniul Entonio/ogj,,25, 370-382. also thank N. Willits for statistical advice, G. Nelson Holt, R.D. (1977) Predation, apparent competition, and the for allowing us to use his vineyard to conduct our structure of prey communities. Tlworeticul fopultiiio~i Biology, 11, 197-229. research, and G. Leavitt for logistical support. The Holt, R.D. (1984) Spatial heterogeneity, indirect inter- manuscript was prepared while the senior author was actions, and the coexistence of prey species. Amcvican supported by grants from the USDA Sustainable Nuturulist, 124, 377-406. Agriculture Research and Education Program, the Holt, R.D. & Kotler, B.P. (1987) Short-term apparent coni- USDA National Research Initiative Grants Program, petition. American Naturulisi, 130, 41 2-430. Holt, R.D. & Lawton, J.H. (1994) The ecological conse- the California Raisin Advisory Board, and the Cali- quences of shared natural enemies. Annuul Reoicw qf Ecol- fornia Table Grape Commission. ogy and Systeniafics, 25, 495-520. Holt, R.D., Grover, J. & Tilman, D. (1994) Simple rules of interspecific dominance in systems with exploitative and apparent competition. Tho American Naturalist, 144, 741- References 777. Hougen-Eitzman, D. & Karban, R. (1995) Mechanisms of Collyer, E. (1964) The effect of an alternative food supply interspecific competition that result in successful control on the relationship between two Typhlodromus species and mites following inoculations of Willamette mite Panorzychus ulmi (Koch) (Acarina). Eniomologia E.wper- on grapevines. Oecologia, 103, 157-1 61, inztntalis ei Applicatu, 7, 120-124. Hoy, M.A. (1982) Aerial dispersal and field efficacy of a Connell, J.H. (1983) On the prevalence and relative import- genetically improved strain of the spider mite predator, ance of interspecific competition: evidence from field M. occidentalis. Enloniologiu E.wi~~~ri/ii~,titali.~et Applicatu, experiments. American Nattrralisi, 122, 66 1-696. 32, 205--2 12. English-Loeb, G.M. & Karban, R. (1988) Negative inter- Hoy, M.A. & Flaherty, D.L. (1970) Photoperiodic induction actions between Willamette mites and Pacific mites: poss- of diapause in a predaceous mite, Merasc4ulus occirlentali,~. ible management strategies for grapes. Entoniologiu Exper- Annals of tho Entoniologicul Socir,ty of America, 63, 960--- inzentalis e/ Applicata, 48, 269-274. 963. English-Loeb, G.M., Karban, R. & Hougen-Eitzman, D. Hoy, M.A., Flaherty, D.L.. Peacock, W. &Culver, D. (1979) (1993) Direct and indirect competition between spider Vineyard and laboratory evaluations of methomyl, dime- mites feeding on grapes. Ecological Entomology, 3, 699- thoate, and permethrin for a grape pest management pro- 707. gram in the San Joaquin Valley of‘ California. Journul of Flaherty, D.L. (1967) The ecology und importance of spider Economic Entomology, 72, 250-255. niiies on grapevines in the Southern Sun Joaquin Vulley with Huang, C. & Sih. A. (1990) Experimental studies on behavi- emphasis on the role of Metaseiulus occidentalis (Ncshitt) orally mediated, indirect interactions through a shared in /heir naturcil cotztrol. PhD thesis, University of Cali- predator. Ecolog.v, 71, 1515-1522. fornia. Berkeley. Karban, R. & English-Loeb, G.M. (1990) A ‘vaccination’ of Flaherty. D.L. (1969) Vineyard trophic complexity and Wil- Willamette spider mites (Acari: Tetranychidae) to prevent lamette mite. Eoreiranychus willamtJtfei Ewing (Acarina: large populations of Pacific spider mites on grapevines. Tetranychidae) densities. E(,ology,SO, 9 I 1-9 16. Journal of Economic, Eniomology, 83, 2252-2257. Flaherty, D.L. & Hoy, M.A. (1971) Biological control of Karban, R., Hougen-Eitzman, D. & English-Loeb, G.M. Pacific mites and Willaiiiette mites in San Joaquin Valley (1994) Predator-mediated apparent competition between vineyards. 111. Role of tydeid mites. Researches on Popu- two herbivores that feed on grapevines. Oc‘cologia, 97, la/ion Ecology, 13, 80-96. 508-5 1 1. Flaherty, D.L. & Hufaker, C.B. (1970) Biological control of Kerfoot. W.C. & Sih, A. (1987) Predaiion: DirectandIndirec/ Pacific mite and Willamette mite in San Joaquin Valley Impacts on Aqumtic Conimuniiies. University Press of New vineyards. I. Role of Metaseiu1u.r occidenfulis. 11. Influence England, Hanover, New Hampshire. of dispersion patterns of Meraseiulus occidenralis. Hilgar- Kinn, D.N. &Doutt, R.L. (1972)Initialsurveyofarthropods diu, 40, 267-330. found in North Coast vineyards of California. Environ- Flaherty, D.L., Wilson, L.T., Stern, V.M. & Kido, H. (1985) mental E/itOnzo/ogy, 1, 508-513. Biological control in San Joaquin Valley vineyards. Bio- Lamiman, J.F. (1 935) Pacific mite, Tetranychus pac$cus logical Conlrol in Agricultural IPM systems (eds M. A. McGregor in California. Journal of Econoniic Entomology, Hoy & D. C. Herzog), pp. 501-520. Academic Press Inc., 28, 900-903. New York. McMurtry, J.A. (1982) The use of phytoseiids in biological Flaherty, D.L., Wilson, L.T., Welter, S.C., Lynn, C.D. & control: progress and future prospects. Recent Ad[x”.s Hanna, R. (1992) Spider mites. Grape Pest Management in KnowledgcJ uf the Phytoseiitlue (ed. M. A. Hoy). pp. (eds D. L. Flaherty, L. P. Christensen, W. T. Lanini, J. J. 23-48. University of California, Division of Agricultural Marois, P. A. Phillips & L. T. Wilson) , 2nd edn, pp. 191- Sciences, Berkeley. 192. University of California, Division of Agriculture and McMurtry, J.A. (1983) Phytoseiid predators in orchard sys- CJ 1997 British Natural Resources, Oakland, California. tenis: a classical biological control success story. Biological Ecological Society, Frazier, N.W. & Smith, L.M. (1946)The Willamette mite on Confrol of Pesfs hy Mires (eds M. A. Hoy, G. L. Cun- .Journal of Applied grapes. Nilgurdia. 17, 191-195. ningham & L. Knutson), pp. 21-46. University of Cali- Ecologl., 34, 878-888 Hanna, R. (1992) Ecology of’ spider mires in vineyards: role fornia, Division of Agricultural Sciences, Berkeley. 888 Neter, J., Wasserman, W. & Kutner, M.H. (1985) App1ic.d.. Settle, W.H. & Wilson, L.T. (1990) Invasion by the vari- Predation and Lineor Siutis/ical Models, 2nd edn. Richard D. Irwin Ilic., egated leafhopper and biotic interaction: parasitism, com- Homewood, Illanois. coinpetition in un petition, and apparent competition. Ecolog,~.71, 146L Pritchard, A.E. & Baker. E.W. (19.55) A euide to the wider . I- 1 ~~ 1470. (uvine c()”uflit,V mites of deciduous fruit trees. Hilgurdiu, 21, 253-287. Sih, A,, Crowley, P., McPeek, M., Petranka, J. & Stohmeier, SAS Institute Iiic. (1989) PROC CLM. SAS/STAT Uxr’,s K. (I 985) Predation, competition, and prey communities: Guide. Version 6,41h edn, Vol. 2. SAS Institute Inc., Cary, a review of field experiments. Annuol REO~(Wof’ Ecolog!. North Carolina. und Sysrcniurics, 16, 269-3 I I, Schmitt, R.J. (1987) Indirect interactions between prey: Welter, S.C., Farnham, D.S., McNally, P.S. & Freeman, K. apparent competition, predator aggregation, and habitat (1989) Effect of Willamette mite and Pacific spider mite segregation. Ecology, 68, 1887-1897. (Acari: Tetranychidae) on grape photosynthesis and Schoener, T.W. (1983) Field experiinents on interspecific stomatal conductancc. Eni;ironmc,nful Enromolog~,18, competition. American Naruralisi, 122, 240-285. 953-957. Schuster, R.O. & Pritchard, A.E. (1963) Phytoseiid mites of California. Hilgurdiu, 34, 191-285. Received 2 Septeinber. 1995: rwision wceriwl9 Augusi 1996

c) 1997 British Ecological Society, Journal of’ Applied Ecolog,~.,34, 878--888