BIOLOGICALCONTROL Habitat Diversification for Improving Biological Control: Abundance of Anagrus epos (Hymenoptera: Mymaridae) in Grape Vineyards
BROOK C. MURPHY, JAY A. ROSENHEIM, ANDJEFFREY GRANETT Department of Entomology, Universityof California, Davis, CA 95616
Environ. Entomol. 25(2): 495-504 (1996) ABSTRACT A frequently cited habitat diversificationtactic is the use of prune tree refuges that support overwintering populations of Anagrus epos (Girault), a mymarid egg parasite of the western grape leafhopper, Erythroneura elegantula Osborn, in vineyards.Here we test the effect of prune trees on early-season abundance of adult A. epos in vineyards. A. epos was found in vineyards downwind of prune trees at more than twice the densities of vineyards lackingpnllle trees, despite significantvariation in A. epos immigration from sources outside the pnme tree-vineyard system. Densities of A. epos overwintering within prune trees ex- plained a significantamount of the variationin A. epos trap capture in vineyards.Furthermore, another factor associated with prune trees was found to influence A. epos abundance in vine- yards: a windbreak effect created by the prune trees concentrated dispersing A. epos on the leeward side of the prune trees, tllereby further enhancing A. epos numbers.
KEY WORnS Anagrus epos, Erythroneura elegantula, habitat diversification,conservation biologicalcontrol, grape
HABITATDIVERSIFICATIONHASlong been promot- Altieri 1992), most studies have been qualitative ed as a tactic to conserve natural enemies and en- rather than quantitative in nature. Furthermore, hance biological control of insect pests and as a there are no current tactics that have been widely integrated method of developing sustainable pest adopted for pest control in production agriculture. control systems in production agriculture. Both ap- Here, we test and examine the ecological mecha- proaches argue that habitat diversification can pro- nisms associated with prune tree habitats planted vide essential resources for natural enemies, such adjacent to grape vineyards on a key natural enemy as feeding sites, alternate hosts or prey, or over- species. wintering sites, that can enhance their abundance Anagrus epos-Erythroneura elegantula System. in the surrounding environment and help prevent A frequently cited example used to illustrate the pest outbreaks (Herzog and Funderburk 1985, Na- effect of habitat diversification on natural enemy tional Research Council 1989, van Emden 1990, effectiveness is the planting of overwintering hab- Altieri 1992). itats for Anagrus epos (Girault) near grape vine- Whether viewed as a biological control tactic or yards to enhance biological control of the western as an integrated approach to sustainable pest con- grape leafhopper, Erythroneura elegantula Os- trol, habitat diversification in or around crop fields born. can have a profound effect on how insect herbivore Erythroneura elegantula is a major pest of and nahlral enemy populations interact. An under- grapes in many regions of the western United standing of the ecological mechanisms involved States. High leafhopper numbers result in eco- will also provide a rich resource for the develop- nomic losses caused by cosmetic damage to grape lIlent of pest management tactics in production ag- berries from leafhopper frass, reduced vine vigor riculture (Herzog and Funderburk 1985). from heavy leaf feeding and leaf loss, and fruit One of the key hypotheses underlying the con- damage from sun exposure. Furthermore, high cept of habitat diversification is that many pest densities of adult leafhoppers can disrupt harvest problems are the result of a loss of habitat critical by flying into the eyes, nose, and mouth of field for supporting natural enemy populations (Letour- laborers. neau 1987, Russell 1989, Andow 1991). Although A key natural enemy of the grape leafhopper is the ecological basis behind habitat diversification A. epos, an important egg parasite of E. elegantula and natural enemy effectiveness has been dis- as well as other leafhopper species (Gordh and cllssed for many years, and many proposed systems Dunbar 1977, Williams 1984). A. epos has been have been evaluated in a preliminary way (re- reported to parasitize a significant proportion of E. viewed by Flint and Roberts 1988, Russell 1989, elegantula eggs within commercial vineyards
0046-225X196/0495-0504$02.00/0 © 1996 Entomolngical Society of America 496 ENVIRONMENTAL ENTOMOLOCY Vol. 25, no. 2
(Doutt and Nakata 1973). A major h1ctor hypoth- A. epos in vineyards or by examining E. elegantula esized to be associated with the abundance and eggs and comparing rates of egg parasitism. Col- effectiveness of A. epos is the presence of nearby onization appeared to be enhanced in vineyards habitat that supports alternate leafhopper hosts downwind from prune orchards or riparian habi- (Doutt and Nakata 1965, 1973). A. epos overwin- tats containing blackberry brambles. They con- ters within leafhopper eggs. E. elegantula overwin- cluded that the greater abundance seen in field ters as an adult, thus, an alternate source of leaf- trials supported the overwintering refuge hypoth- hopper eggs is required to support overwintering esis. However, because these initial experiments A. epos numbers. used unreplicated comparisons, the results, al- Doutt and Nakata (1965) first observed that though consistent with the refuge hypothesis, were vineyards located downwind from riparian habitats preliminary in nature. had higher levels of E. elegantula egg parasitism The interpretation presented by Doutt and Nak- and lower E. elegantula numbers relative to vine- ata (1973) and Kido et a1.(1984) also rested on an yards located distant from these' habitats. They untested assumption; namely, that an increase in found that blackberry brambles, Rubus spp., abun- parasite captures or egg parasitism near refuges dant in riparian habitats, support Dikrella eaZifor- was the direct result of A. epos emergence from niea (Lawson), a year-round host of A. epos. They overwintering sites. A 2nd potential mechanism observed that early spring emergence of A. epos that could explain increased parasite captures from blackberry results in earlier vineyard coloni- downwind of prune trees or trees found in riparian zation by parasites, which coincided with oviposi- habitats is a windbreak effect generated by the tion of overwintered E. elegantula. They also spec- physical structure of these habitats. A windbreak ulated that earlier vineyard colonization at higher effect occurs when a structural barrier creates hlr- densities produced a stronger numerical response bulence in the airflow and a sheltered zone on the and thereby improved biological control (Doutt leeward side of the barrier. The turbulent airflow and Nakata 1973). Unfortunately, field trials deter- may act to increase the probability that a dispers- mined blackberry brambles grown away from ri- ing insect will settle into these sheltered zones, parian habitats did not maintain abundant popu- thereby enhancing colonization. Dispersing insects lations of A. epos, and further attempts were from many families, including wasps in the family abandoned (Flal1erty et al. 1985). Mymaridae, have been shown to accumulate be- Kido et at. (1984) demonstrated that prune trees hind windbreaks (Lewis and Stephenson 1966). support the prune leafhopper, Edwardsiana prun- Wilson et a1. (1989) showed that A. epos reach ieola (Edwards), which overwinters in the egg higher densities in pnmes when artificial wind- stage and serves as an overwintering host for A. breaks are positioned on the upwind side of prune epos. Emergence of overwintering A. epos from trees. Corbett and Rosenheim (in press) have also overwintering E. prunieola eggs laid beneath the recently demonstrated a pattern of A. epos cap- bark of prune twigs also was found to coincide with tures in vineyards downwind of pnme trees that is 1st-generation E. elegantula oviposition in vine- consistent with the operation of a windbreak. Thus, yards, suggesting that pnme trees also serve as an enhanced A. epos abundance could be the result overwintering habitat for A. epos. Studies in both of overwintering populations within the prune commercial and experimental plantings of pnille trees, a windbreak effect, or some combination of trees demonstrated that A. epos populations could those 2 factors. be supported year-round. Williams (1984) and In the current shldy, we tested 2 key hypotheses Kido et al. (1984) also showed that A. epos reared with respect to the effect of overwintering sites on from E. prunieola or D. ealifomiea can parasitize A. epos abundance in vineyards as follows: (1) that E. elegant1.l1aeggs successfully, and vice versa. Fi- prune trees adjacent to vineyards enhance early- nally, it has been shown that elevated A. epos pop- season A. epos numbers in vineyards, and (2) that ulations were associated with higher E. elegant1.l1a differences in abundance are the result of differ- egg parasitism in vineyards near pnme trees (Kido ences in the densities of overwintering A. epos et al. 1984, Pickett et aI. 1990). Although these emerging from pnme trees. In addition to those 2 studies established the basic framework of the hypotheses, we also analyzed capture data for ev- grape leafhopper-A. epos interaction, there have idence of a windbreak effect influencing A. epos been no replicated field trials that assess the effi- abundance downwind of pnme trees. cacy of prune tree habitats or that test the ecolog- ical mechanisms underlying the effect. Materials and Methods Anagrus epos Abundance. Doutt and Nakata (1973) and Kido et al. (1984) hypothesized that Experimental Design. A. epos colonizes virtu- overwintering refuges near vineyards increase ear- ally all grape vineyards at some point during the ly-season A. epos vineyard colonization as a result growing season; the time and density of abundance of overwintering A. epos emerging from refuges can, however, be highly variable. E. elegantula and dispersing into nearby vineyards. They tested populations also can be variable among vineyard these predictions in field trials using yellow sticky sites in a given year. Because of the vmiation of traps to measure early-season abundance of adult both parasite and leafhopper numbers, a rigorous April 1996 MURPHY ET AL.: HABITAT DIVERSIFICATION FOR BIOLOGICAL CONTHOL 497
marily in vineyards with prune trees planted spe- cifically for pest management purposes, we might have obtained a seriously biased sample of vine- yards for at least the 2 following reasons: (1) these vineyards might be more likely to have experi- enced severe leafhopper problems, motivating growers to look for enhanced means of control, and (2) growers using prune tree plantings might be a relatively "progressive" subset of all growers and might also differ in other aspects of vineyard management. Because 10 of the 12 prune tree sites were commercial pnme orchards and were not planted for pest control, we feel that the lack of true randomization is unlikely to introduce a sys- tematic bias into the analysis. Vineyard A. epos Abundance. Vineyard plots were monitored to estimate densities of immigrat- ing A. epos and E. elegantula during the early sea- son, before parasites or leafllOpper adults began emerging from the 1st generation of reproduction in vineyards. For both prune tree and control vine- yard plots, 2 yellow sticky traps (75 by 125 mm yellow plastic cards, Hilcor Plastics, Los Ange]es, CAl coated with Tanglefoot were attached to each Fil(. 1. Pairedvineyardplot locationsin northernand of 6 wooden poles (12 traps total per site). Poles ct'ntra\ California.The explodedview showsthe orien- were placed in a transect at 10-m intervals along tationof pnme tree and controlvineyardplots. the 3rd vine row from the upwind edge of each vineyard block. Traps were oriented perpendicular to the predominant wind direction and positioned evaluation of prune tree-enhanced biological con- 0.6 and 1.2 m above the vine canopy to minimize trol required a large number of replicates. Surveys any influence of vine canopies on trap capture. were conducted throughout the major wine grape Traps were deployed beginning 1 April (7 paired viticultural regions of central and northern Cali- sites), 15 Apri] (3 additional pairs), and 1 May (2 fornia to locate 26 paired commercia] vineyards, final pairs), and were replaced twice monthly one with adjacent upwind pnme trees and the oth- through 15 June 1992. All traps were taken to the er lacking pnme trees (Fig. 1). Paired vineyards laboratory and examined under a dissection micro- were matched for grape cultivar, trellising, ancl scope to count the number of adult A. epos cap- management practices. Control (nonprune tree) tured. vinl'yards were positioned O.~.O kilometers from Four vineyard plots were monitored weekly be- refuge sites and either parallel to or upwind of ginning 15 April to estimate the beginning of leaf- pnll1e tree vineyard sites to minimize movement hopper oviposition and the 1st-generation (Fl) of ]eafllOppersor parasites between treatment and nymphal emergence. The beginning of oviposition control vineyards. The paired plots were grouped was determined by examining field-collected grape into 2 viticultural growing regions defined by the leaves under a dissection microscope for the pres- cUllmlative degree days occurring between April ence of E. elegantula eggs. The presence of and October (Winkler et al. 1974). Comparisons nymphs was determined by examining grape leaves were made between cooler growing regions in the field. «3,000 DD) and warmer regions (>3,000 DD). Prune Tree Effect. The prune tree sites varied Two of the pnme tree sites used in the study were with respect to tree number, tree size, and orchard planted specifically for leafhopper control; the re- management practices. These differences in turn mainder were commercial or remnant prune or- could influence overwintering A. epos densities chards growing adjacent to the vineyard plots. and cause variation in the treatment effect within Given the nature of the pnme tree habitat being our design. Therefore, to control for variation in tested, we were unable to ensure tnle randomiza- the pnme refuge effect, and to test hypotheses tion within our experimental design. True random- about the contribution of A. epos overwintering in ization would have required us to assign prune tree prunes to vineyard abundance, we sampled pmne plantings randomly to plots and would have re- twigs to estimate overwintering A. epos densities quired a delay of 3-5 yr for the pnme trees to (details below). We also recorded the nlllnber of become established and support large populations trees and orchard management practices and es- of the alternate leafhopper host. As a result, we timated the average rate of tree growth to deter- used vineyard sites located near previously estab- mine if these factors influence densities of over- lished prune trees. Had we chosen to work pri- wintering A. epos (detai]s below). 498 ENVIRONMENTAL ENTOMOLOGY Vol. 25, no. 2
The primary difference in management practic- was analyzed using a split-plot analysis of variance es found among sites was whether or not trees (ANOVA) model. Growing region was the main were irrigated. Seven of the 12 sites used in the factor within the model, and presence or absence study received little or no supplemental water dur- of a prune tree site was the subfactor. Each vine- ing the growing season. We classified prune tree yard pair was treated as a statistical block to control sites that had a permanent irrigation system as ir- for spatial variability in A. epos densities. Because rigated trees. Prune trees that were occasionally the number of vineyard pairs sampled increased as provided supplemental irrigation using portable the season progressed, separate univariate analyses systems, or that were never irrigated, were classi- were conducted for each sampling period. The fied as nonirrigated. analyses were conducted to test the null hypothesis Overwintering A. epos Densities. Anagrus epos that A. epos abundance rates in vineyards are in- emerge from overwintering prune leafhopper eggs dependent of the presence or absence of prune oviposited beneath the bark of 1-6-yr-old branch trees. wood (Mulla 1957). Because 2-6-yr-old branch Prune Tree Effect. To test if differences in wood can accumulate A. epos emergence holes vineyard parasite abundance can be explained by over several years, we sampled only 1-yr-old the emergence of overwintering A. epos from branch wood to distinguish the current year's pnme trees, multiple regression analyses were per- emergence holes from those of previous years. formed using A. epos emergence hole densities Prune twigs were collected in late June and early from prunes mId A. epos capture in control vine- July, after the completion of A. epos emergence. A yards as independent variables and A. epos capture minimum of 4 pnme twigs (l-yr-old) sampled from in prune tree vineyards as the response variable. 10 randomly selected prune trees at each site was We interpret A. epos trap caphlfe in a control vine- examined (see twig sampling details described be- yard as an estimate of the local abundance of par- low). The length and diameter of twigs were mea- asites moving into vineyards in the absence of sured to calculate the surface area of 1-yr-old prune trees (area-wide A. epos numbers). We test- growth (area = 7T * d * I, where d is 1-yr-old twig ed the null hypothesis that differences in A. epos diameter, I is twig length). Prune twigs were then abundance are independent of variation in over- examined under dissection microscopes for the wintering A. epos densities within prune trees. presence of A. epos emergence holes. These data Multiplicative Effects. We evaluated the poten- were then used to estimate the average number of tial influence of a \vindbreak effect on A. epos emergence holes per square centimeter of twig abundance in vineyards using model II regression surface area for each refuge site. analyses (Sokal and Rohlf 1981) to examine the PrWle Tree Growth. Average tree growth var- relationship between A. epos captures in the con- ied substantially among prune tree sites. To eval- trol vineyards (the independent variable) and A. uate the possible influence of this factor on A. epos epos caphlfes in prune tree vineyards (the depen- densities, we recorded the following 3 parameters dent variable) (statistical model: prune tree vine- as measures of tree growth: average length, diam- yard captures = 13* control vineyard captures + c eter, and number per tree of 1-yr-old twigs. The + error). If overwintering A. epos within prune average length and diameter of twigs were deter- trees were the only factor affecting A. epos trap mined by randomly selecting 10 trees within each capture, we would expect the slope of the regres- pnme tree site for sampling. A main branch was sion equation (13) to approximate 1.0, with an in- removed from each compass direction of each tree crease in the intercept (c) reflecting an additive for a total of 40 branches per site. Branches were increase in the number of A. epos contributed by taken to the laboratory, and the length and diam- pmne trees. If, on the other hand, the slope pa- eter of each 1-yr-old twig were recorded. rameter is significantly greater than 1.0, a multi- The total number of 1-yr-old twigs per tree was plicative effect in pnme tree vineyard trap capture estimated by randomly selecting 4 trees from each is present, indicating the presence of another fac- prune tree site for sampling. Each tree was divided tor, such as a windbreak effect, increasing A. epos in half vertically and horizontally to establish 4 capture independent of any contribution from quadrants. One quadrant position was randomly overwintering A. epos emerging from pnme trees. chosen in each of the 4 trees to count hvigs. The This analysis was used becanse model I regres- 4 hvig counts, 1 example from each quadrant po- sion produces a biased estimate of the slope for sition, were pooled to estimate the mean number low values of the correlation coefficient (Sokal and of hvigs per tree. Rohlf 1981, Pagel and Harvey 1988). Model II Prwle Tree Site Traps. YellowstiCh)'traps also analysis is also the most appropriate method for were used to monitor A. epos and E. prunicola analyzing the functional relationship between 2 adult numbers within pnme trees. For each pnme variables when both variables are subject to exper- site, 1 trap was placed in 3 randomly selected imental error (Sokal and Rohlf 1981). A principal prune trees 2.0-2.5 m from the ground. Traps axis (major axis) model II regression estimates the were replaced twice monthly as described above. slope parameter by estimating the principal axis Statistical Procedures. The average number of between 2 normally distributed random variables A. epos caught on traps within the vineyard plots (Pagel and Harvey 1988). We used this analysis to April 1996 MURPHY ET AL.: HABITAT DIVERSIFICATION FOR BIOLOGICAL CONTHOL 499
45 tected in 18 of 20 vineyard plots. The increasing ~.. 40 A. r3 Control vineyards 3 35 o Treatment vineyards A. epos capture over time indicated that parasites •.. began moving into vineyards in April, several is 30 •.. weeks after overwintered E. elegantula adults be- ~ 2S gin moving into vineyards. Overwintered adults .,•• 20 ~ IS typically begin moving into vineyards from their ...: 10 overwintering habitats during March (Doutt and ..•i Nakata 1965, Williams 1984). E. elegantula eggs -< were detected in vineyards beginning in late April (Fig. 2C), demonstrating that the appearance of A. e 2S B•.. B. epos coincided with the beginning of the oviposi- !l 20 tion period of overwintering leafllOpper adults. •.. The large increases in A. epos capture during the ~ IS 4th sample period in late May coincided with the oS 1st appearance of A. epos emergence holes in 10 grape leaves, marking the emergence of adults of ~t '" the Fl parasite generation (Fig. 2D). Also during ~ 5 the 4th sample period, Fl leafhopper nymphs be- ..•"3- gan appearing on grape leaves in most vineyards -< 4/15-./18 4/24·511 S/8-S/1S 5/22-5/29 613·6/10 (Fig.2E). Date Regression analyses were done to assess the re-
F. Lealltopper eggs lationship between the final density of A. cpos c. 11I1111I11I1111I11I11I emergence holes in 1-yr-old prune twigs and trap 11I _ FlA. epos emergence capture in pnme trees to estimate the time of A. D. epos emergence from pnme twigs. During the 1st E. sample period (15-18 April), only a single prune tree site had an A. epos capture. Regression anal- Fill:. 2. Seasonal patterns of A. epos and E. elegantula yses for the 2nd (24 April-l May) and 3rd (8-15 in vineyard plots. (A) A. epos adult trap captures (mean May) periods revealed that A. emergence ± SE) in control and treatment plots. (B) E. elegantula epos adult trap captures (mcan ± SE) in control and treatment holes explained a significant amount of the varia- plots. (e) Period of E. elegantrtla oviposition in vineyard tion in trap captures (R2 = 0.88, n = 9, P = 0.002 plots. (D) Pel;od of A. epos F I emergence. (E) Period of and R2 = 0.42, n = 12, P = 0.022, respectively). FI nymphal emergence. Hatched lines indicate early ap- No relationship was found between area-wide trap pearance in some vineyard plots. Solid lines indicate ap- captures in control vineyards and trap capture pem'anl'e in all plots. within pnme trees (R2 = 0.05, n = 9, P = 0.56 for the 2nd sample period; R2 = 0.17, n = 12, P = test the null hypothesis that there is no multipli- 0.18 for the 3rd sample period). The conclllsion cative increase in A. epos trap capture in prune from these analyses is that sticky traps in prune tree vineyards relative to control vineyard captures. trees reflect the A. epos emergence pattern from For all statistical analyses except the model II prune trees and are not influenced significantly by regressions, trap captures were subjected to a log(x area-wide A. epos moving from outside sources. + 1) transformation, and A. epos emergence hole The high degree of correlation during the 2nd data were square root-transformed to normalize sample period suggests that peak overwintering the distribution of means. Mean comparisons tests emergence of A. epos from prune twigs occllrred were accomplished using single degree of freedom between 18 April and 1 May. tests (orthogonal contrasts), and all statistical anal- Vineyard A. epos Abundance. We confined our yses except the model II regressions were per- analyses of A. epos abundance to the 1st 3 sample formed using the JMP statistical program for the dates when A. epos captured on traps were most Macintosh (SAS institute 1989). A Bonferroni cor- likely to have originated from sources outside the rection for multiple comparisons was used to main- vineyard. Traps in prune tree vineyards captured tain the total experimentwise a error rate at 0.0.5 >2 times the number of parasites as did traps in for the ANOVA and regression analyses (Sokal and control vineyards during the 2nd and 3rd sample Rohlf 1981). dates (Fig. 2A; Table 1). No significant differences in capture were detected between the 2 growing regions, but variation in A. epos captures was sub- Results stantial across the paired vineyards as indicated by Seasonal Patterns. Examination of the early- the significant block effect. season pattern of parasite and leafhopper trap cap- Prune Tree Effect. Modell regression analyses hires revealed that during the 1st sample period in revealed that captures of A. epos in pnme tree early to mid-April, 1 or more A. epos were caught vineyards were closely correlated with captures in in 7 of the 14 vineyard plots monitored (Fig. 2). control vineyards for the 2nd and 3rd sample dates During the 2nd sample period, A. epos was de- (R2 = 0.53, n = 10, P = 0.016, R2 = 0.65, n = 12, 500 ENVIRONMENTAL ENTOMOLOGY Vol. 25, no. 2
Table 1. Nested ANOVA of the number of A. epos ation was generated by variation in irrigation prac- (1011:10 trw1sformed) captured on traps in .incyards with tices (Table 4). Irrigated trees were associated with resl)ect to prune refuges, growing regions, and paired vineyard plots (block effcct) greater tree growth (average surface area of I-yr- old twigs) and lower twig numbers than nonirri- Source of Date df F 1''' gated trees. Irrigation was also associated with variation higher A. epos overwintering densities as well as 15-18 Apr Treahnent 1 3.93 0.094 higher trap captures of E. prunicola and A. epos Blocks (region) 5 6.17 0.023* adults in tree canopies. ANCOVA revealed that Region 1 0.13 0.735 both irrigation (F 23.06, df 1, 12, P 0.001) 24 Apr-l May Treatment 1 6.47 0.032* = = = Blocks (region) 8 5.14 0.012" and prune tree number (F = 11.21, df = 1, 12, P Region 1 4.60 0.061 = 0.009) had significant influences on the density 8-15 May Treahnent 1 11.32 0.006" of A. epos emergence holes in prune trees. There Blocks (region) 10 5.72 0.004" Begion 1 3.56 0.086 was a positive linear relationship between the number of prune trees and the density of A. epos "One-tailed tests for hypothesis testing; *, l' < clitical value emergence holes in trees (intercept = -0.0496, after Bonferroni correction (1' < 0.033). slope = 0.0773). Nevertheless, neither pnme tree number nor irrigation had detectable effects on parasite catch in associated vineyards during the 3 P = 0.001, respectively), but not for the 1st sample sample periods (P > 0.047). date (R2 0.48, n 7, P 0.09). The significant = = = We also estimated the average number of A. regressions of A. epos capture between the treat- epos emergence holes per tree and the total num- ment and control plots underlines the importance ber of emergence holes per prune tree site to test of blocking by vineyard pair, a key feature of our if these parameters could explain captures in ad- experimental design. This result also confirms that jacent vineyards. The number of emergence holes area-wide A. epos numbers were a significant source of parasites in vineyards whether pnme per tree was nonsignificant during the 1st sample trees were present or not. period (F = 8.9; df = 1, 6; P = 0.041), significant The multiple regression analyses indicated that during the 2nd period (F = 7.21; df = 1, 9; P = A. epos emergence from prunes was a factor ex- 0.031), and non-significant during the 3rd period plaining A. epos trap captures in prune tree vine- (F = 0.26, df = 1, 11, P = 0.625). The number yards, after controlling for area-wide abundance per prune tree site was, however, nonsignificant from control plot trap captures, during the 2nd during all 3 sample periods (P ~ 0.096). sample period (Table 2). This coincided with the Multiplicative Effects. The principal axis anal- estimated time of A. epos emergence from prune yses (model II regression) of A. epos trap captures twigs as determined by sticky traps in pnme trees. in control and prune tree vineyards suggest that a The partial regression coefficients indicated that multiplicative effect was present during the 2nd emergence from pnmes explained a substantial and 3rd sample periods, when both slope param- amount of the variation in A. epos capture during eters were significantly >1.0 (Fig. 3 B and C). the first 2 sample periods in April, but by the 3rd During the 1st sample period, the estimated slope sample period in early May, area-wide A. epos cap- was similar to the other sample periods, but not tures explained most variation in trap captures. significantly different from 1.0 (Fig. 3A). These re- Variation in Overwintering A. epos Densities. sults suggest that independent of any contribution Prune tree sites varied in terms of the number of from A. epos overwintering in pmne trees, prune trees per site, the rate of tree growth, the number trees were associated with a multiplicative effect of I-yr-old twigs per tree, and the density of A. augmenting A. epos capture on the downwind side epos emergence holes (Table 3). Much of this vari- of prune trees.
Table 2. Multiple regression analysis of the relationship bctwcen the dependent variable, 10fl;lO (A. epos trap •.atch) ill prune trec plots, and the oulcpendent variable, s'luarc root (A. epos emergence density) from prlUlC sites and 10gIO (A. epos trap catch) 01 control plots
Sample date Source of variation df Slope (:1: SE) F R2" pb 15--18 Apr. Whole model test 2, 7 11.9 0.86 0.026 Emergence density 1,7 1.06 (~ 0.325) 10.6 0.75 0.031 Control captu res 1,7 0.67 (:1:0.237) 8.0 0.69 0.047 24 Apr-l May Whole model test 2,10 16.8 0.83 0.002* Emergence density 1,10 2.37(:1: 0.717) 10.9 0.73 0.013* Control captures 1,10 0.84 (:1:0.242) 11.9 0.75 O.oI.l " 8-]5 May Whole model test 2, 12 10.15 0.69 0.005* E mergence density 1, 19- 0.91 (:1:0.825) 1.21 0.35 0.299 Control captures 1.12 1.48 (~ 0.363) 16.74 0.81 0.003*
" H2 for individual factors are partial correlation coefficients (r2). b ", l' < clitical value after Bonferroni correction (1' < 0.017). April 1996 MURPHY ET AL.: HABITAT DIVERSIFICATION FOR BIOLOGICAL CONTROL 501
Table 3. Descrilltive l.armnelers of pnnle lrees mnong experimenlall.lols
Variable Mean SE Range No. pnllle trees per site 422.5 111.9 27-1,216 No. 1-yr-old twigs per tree 1,121.7 177.4 43,5-2,238 l-yr-old twig surface area, cm2 20.3 4.0 3.7-122.5 A. (--pOS t'lllf>rgencp density/cm2 0.021 0.0056 0.0023-0.0640
Discllssion wide A. epos captures and A. epos emergence from prune trees coinciding with the onset of oviposi- The goal of this research was to assess the influ- tion by overwintered E. elegantula in vineyards. ence of pnme trees on early-season A. epos abun- dance in vineyards and to test the hypothesized Thus, parasite abundance appears to be synchro- nized with the appearance of host eggs in grape ecological mechanisms underlying this system. Pomp trees were associated with significant in- vineyards. Vineyard A. epos Abundance. The replicated creases in the density of A. epos captured in nearby grape vineyards relative to vineyards lacking prune comparisons of A. epos abundance in vineyards trees. Pome tree vineyards were found, on aver- showed that the presence of overwintering habitats agp, to receive twice the number of early-season produced a substantial early-season increase in the parasite captures. Overwintering A. epos emer- density of A. epos trapped in grape vineyards, a gpnce from prune trees was a significant predictor result consistent with the observations reported by of A. cpos trap capture in nearby vineyards during Doutt and Nakata (1973) and Kido et a!. (1984). the 2nd sample period, which coincides with our The elevated capture rates were seen despite dif- estimated period of A. epos emergence from prune ferences in geographical regions and variation in twigs. Examination of the functional relationship the number of area-wide A. epos captures, which between control and pnme tree vineyard parasite indicates that the prune tree effect operated under captures also revealed a multiplicative response in a wide variety of vineyard management and envi- trap capture on the downwind side of pnme tree ronmental conditions. Based on these results, we vineyard plots. This suggests that another factor concluded that the presence of prune trees was associated with pnme trees, other than overwin- associated with elevated A. epos numbers, and as tering A. epos populations, enhanced parasite a result, parasites may colonize these vineyards at numht'rs in refuge vineyards. higher rates than vineyards lacking nearby pnme Seasonal Patterns. Based on limited field data trees. Doutt and Nakata (1973) and Kido et al. (1984) Ecological Mechanisms. Doutt and Nakata speculated that the effectiveness of overwintering (1973) and Kido et a!. (1984) assumed that any refuges near vineyards was, at least in part, caused increase in parasite abundance in vineyards (in- by the time of A. epos emergence from overwin- ferred using trap data or egg parasitism rates) was tering hosts. They based this hypothesis on the ob- the direct result of overwintering emergence from sen'ation that parasite emergence from prune trees nearby habitats. In the current study,we tested this or blackberries occurs during mid-April, at the ap- assumption by comparing the density of overwin- proximate time overwintered E. elegantula begin tering A. epos emergence holes with A. epos trap oviposition in vim·yards. They further speculated capture in pnme tree vineyards after controlling that enhancing early-season A. epos colonizers, for area-wide colonizers. The significant regression coupled with the high numerical response of the during the 2nd sample period indicated prune parasite (A. epos completes 3 generations for each trees were a significant source of parasite captures E. d('~antllia generation), improves biological con- during the period of A. epos emergence from over- trol of leafhoppers in vineyards. The seasonal pat- wintered E. prunicola eggs. These results reveal 2 tern of leafllOppers and parasites seen in the cur- important points regarding the influence of pnme rent study concurs with these previous trees on parasite abundance in vineyards. The 1st observations. Across the replicated vineyard pairs, is that overwintering emergence is a significant our data show a consistent pattern of both area- ecological factor determining early-season parasite
Table 4. Effeet of irrigalion on prIme lree growlh mld populalions of E. prunicola mId A. epos.
Growth rate, No, twigs A. epos E'l'nmicola A, epos tmppecI" cm2n (:!:SE) emergence holes/cm2 trapped" Irrigated 30.4 :!: 4.98* 642:!: 176** 0,0325 :!: 0.0063* 9.50 :!: 2.09** 22.16:+: ,'5.411** Nonirrigated 13.1 :!: 4.20 1,475:!: 163 Om05 :!: 0.058 1.10:!: 1.93 0.59:t 5.07
a Surface area of 1-yr-old pnmc twigs (mean :t SE). " Numbers of adults caught on traps within prune tree canopies. * Signifkantly different. 0.01 < P < 0.05. ** SigniRcantl)' different, P < 0.01. 502 ENVIRONMENTAL ENTOMOLOGY Vol. 25, no. 2
:l Q Q" -~ ~ A. B. c. .•.. 0.8 5.0 8.0 ""~ • 4.0 E= 0.6 • • 6.0 Q" 3.0 ....= 0.4 4.0 ~tIJ 2.0 0.2 2.0 .a"" 1.0 Q" to;! ~ 0.0 0.0 ~"':l 0.6 0.8 0.0 1.0 2.0 3.0 4.0 5.0 0.0 2.0 4.0 6.0 8.0 ~ ~ 0 No. A. epos captures in control plots Z Fig. 3. Scatterplots of the functional relationship between A. epos capture in pnme tree refuge vineyardsagainst A. epos capture in control vineyards.(A) 15-18 April (B = 1.705; 97.5% CL, L = 0.58, U = nonestimatable). (8) 24 April-l May (B = 2.38; 97.5% CL, L = 2.05, U = 2.79). (C) 8--15 May (B = 3.10; 97.5% CL, L = 2.2.1, U = 4.85). Confidence limits were established for 97.5% as a correction for multiple comparisons. numbers. The 2nd is that the effect of prune trees be consistent for vineyards with a windbreak pres- on A. epos abundance was dependent not only on ent. Furthermore, this result identifies an impor- the presence of prune trees themselves, but on the tant additional mechanism that may be affecting density of overwintering parasites residing within A. epos vineyard colonization in this system. the trees. This latter point may have important im- These results underscore the importance of in- plications for using prune trees as a commercial vestigating the ecological mechanisms underlying pest control tactic (discussed below). systems of habitat diversification in addition to Lewis and Stephenson (1966) showed that flying tests of overall effectiveness of diversification, Un- insects dispersing in the surrounding environment derstanding the mechanisms involved will enhance accumulate in sheltered regions near physical bar- our ability to manipulate these systems to maxi- riers (natural or artificial windbreaks). They also mize their effectiveness and provide a useful found that insects aggregate behind windbreaks as framework to begin analyses of new systems, a multiplicative function of the number of insects Variation in the Prunc Tree Effect. Because dispersing in the surrounding environment. This overwintering A. epos densities varied among the effect was particularly pronounced for mymarid prune tree sites, we were interested in how the wasps. Using elemental labeling techniques, Cor- management of pnme tree's affects overwintering bett and Rosenheim (in press) found a pattern of A. epos populations and their subsequent abun- A. epos capture suggesting that area-wide A. epos dance in vineyards. Irrigation practices were found numbers accumulate at a rate 4 times greater im- to be a significant factor c:\plaining much of the mediately downwind of prune trees than found up- variation in overwintering A. epos densities ob- wind of trees. Based on these results, they hypoth- served in pnme trees (Table 4) and somc of the esized that a windbreak effect may also be a variation in trap catches in pnme tree vineyards. mechanism enhancing abundance within the prune The number of prune trees within a site also af- tree-vineyard system. We tested this hypothesis fected overwintering A. epos density. How tree across our replicated study by examining the func- number exerts an influence on the density of over- tional relationship between control vineyard trap wintering A. epos (per square centimeter of pnme captures (area-wide colonizers in the surrounding bark) is not known, but we speculate that greater environment) and prune tree vineyard captures numbers of trees may create a more favorable mi- (number of colonizers behind windbreaks), Our re- crohabitat, increasing the number of A. epos that sults revealed that during each of the 3 sample successfully emerge from their overwintering periods, the slope of the relationship was 2::1.7, hosts. Thus, a multiplicative increase in capture was as- Although tree number had a strong positive in- sociated with the presence of prune trees but in- fluence on the density of A. epos emergence holes dependent of any contribution from overwintering per square centimeter of prune tree bark, it was A. epos densities. This result supported the wind- not a predictor of A. epos captures in pnme tree break hypothesis and demonstrated the effect may vineyards. This is an enigmatic result. At the outset April IHH6 MUHl'lIY ET /\1..: HABITAT DIVEHSIFICATlON FOH BIOLOGICAL CONTHOL .'503 of till' study, we expected that pnllle tree sites con- ta\ity, and E. elcgalltula population densities within taining morl' trel'S would producl' a larger pool of grape vineyards. migrants for thl' downwind vineyards. The in- en'ased density (pcr squan> centimeter of bark) of ovcrwintcring A. epos in prune trce sites with larg- Acknowledgments cr tn'e 1lI11l11wrsonly strengthened onr expcctation We extend special thanks to Nerissa Fret'man, Maddy that trcc numbpr would lw a major determinant of Samant, John Tlnnter, Jason May, and Paul Higgs (De- cady-season abundance in prune trep vineyards. partment of Entomology, University of Califol1lia. Davis) Indcl'
National Research Council. 1989. Alternative agri- SAS Institute. 1989. JMP statistical program. SAS In- culture. National Academy, Washington, DC. stitute, Cary, NC. PaA'el, M. D., and P. H. Harvey. 1988. Recent de- van Emden, H. F. 1990. Plant diversity and natural velopments in the analysis of comparative data. Q. enemy efficiency in agroecosystems, pp. 63-80.In M. Rev. BioI. 63: 418-440. Mackauer, L. E. Ehler, and J. Roland [eds.], Critical issues in biological control. Intercept, Andover, UK. Pickett, C. H., L. T. Wilson, and D. L. Flaherty. Williams, D. W. 1984. Ecology of the blackberry-leaf- 1990. The role of refuges in crop protection, with hopper-parasite system and its relevance to California reference to plantings of French prune trees in a grape agroecosystems. Hilgardia 52: 1--33. grape agroecosystem, pp. 151-165. In N. J. Bostanian, Wilson, L. T., C. H. Pickett, and D. L. Flaherty. L. T. Wilson, and T. J. Dennehy [eds.], Monitoring 1989. French prune trees: refuge for grape leafhop- and integrated management of arthropod pests of per parasite. Calif. Agric. 43(2): 7-8. small fruit crops. Intercept, Andover. Winkler, A. J., J. A. Cook, W. M. Kliewer, and L. A. RusseD, E. P. 1989. Enemies hypothesis: A review of Lider. 1974. Climate and soils, pp. 58-75. In Gen- the effect of vegetational diversity on predatory in- eral viticulture. University of California Press, Berke- sects and parasitoids. Environ. Entomol. 18: 590-599. ley. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd Received for publication 5 December 1994; accepted 6 ed. Freeman, San Francisco. November 1995.