Relative Abundance of Three Species of Spider Mites (Acari: Tetranychidae) on Cotton, as Influenced by Pesticides and Time of Establishment

Relative Abundance of Three Species of Spider Mites (Acari: Tetranychidae) on Cotton, as Influenced by Pesticides and Time of Establishment P. J. TRICHILO,' L. T. WILSON,' AND D. GONZALEZ* Department of Entomology, Texas A&M University, College Station, Texas 77843 J. Econ. Entomol. 83(4): 1604-1611 (1990) ABSTRACT Cotton, Cossypium hirsutum L., was planted in experimental plots in the west central San Joaquin Valley of California, during two consecutive seasons. Plants were treated with dicofol, methyl parathion, and/or permethrin, to manipulate densities of spider mites, Tetranychus spp., within the plots. Spider mite densities were also modified by inoculative releases of Tetran ychus turkestani Ugarov & Nikolski. Several times throughout each season, the relative abundance (Le., proportion) of each of three spider mite species, Tetranychus urtlcae Koch, T. pacificus McGregor, and T. turkestant, was determined. In 1982, T. pacificus was the dominant species in all treatments, with a relative abundance of 0.85 in the dicofol plots, and 0.67 in the untreated control. However, T. pactficus was less dominant, and significantly lower in relative abundance in plots treated with methyl parathion (0.53) than in dicofol-treated plots. Conversely, T. turkestani was higher with methyl parathion (0.34) than with dicofol(o.13). In 1983, trends were similar to 1982, although T. urticae was the dominant species in the untreated control, with a relative frequency of 0.49. Permethrin and dicofol had a negative impact on the relative abundance of both T. turkestani and T. urticae. Populations of T. urticae exhibited susceptibility to dicofol, relative to the dicofol resistance of T. pacificus. Early-season inoculative releases of T. turkestani increased the relative abundance of T. turkestani especially in untreated plots. However, late releases were less effective than early releases in promoting the buildup of T. turkestani, relative to the other two species. KEY WORDS Arachnida, Gossypium, spider mites, Tetranychus A VALUABLE AID in the management of cotton, Gossypium hirsutunz L., has been the ongoing development of economic injury levels for spider mites, Tetran ychus spp. (Acari: Tetranychidae), which are serious pests of cotton in the San Joaquin Valley of California (Leigh 1963; Leigh & Burton 1976; Wilson et al. 1983, 1985). Most studies of spider niites have focused on either the twospotted spider mite, Tetranychtrs urttcae Koch, the strawberry spider mite, Tetranychus turkestani Ugarov & Nikolski, or the Pacific spider mite, Tetrunychus pacificus McGregor, or have lumped them into the general category Tetranychus spp. (Canerday & Arant 1964, Furr & Pfrimmer 1968, Mistric 1969, Reynolds et al. 1982). Because the amount and type of damage to cotton varies with each species (Leigh 1963, Brito et al. 1986), it is desirable to segregate such data accordingly. Greater refinement of management programs would be possible with the establishment of thresholds for each of the three spider mite species that inhabit cotton. To date, Few data are available on the relative abundance of these three species Formerly. Department of Entomology, University of California. Davis, Calif. 95616. 2 Division of Biological Control, University of California. Riverside, Calif. 92521. during the cotton growing season, even though evidence suggests that they are not equally abundant in space or time (Leigh 1963, Jackson & Leigh 1967, Dennehy et al. 1987a, Grafton-Cardwell et al. 1987). A precursor to determining species-related injury levels is simply understanding the relative frequency of each spider mite species on cotton, as affected by various biological and environmental factors. Some of these factors include predation by natural enemies, pesticide interference, and migration. To our knowledge, there are no reports that quantify the effects of any of these factors on the relative abundance of spider mites on cotton. Recent investigations have established that the proportion of leaves infested with spider mites, P(I), is a reliable estimate of spider mite population densities on cotton, as well as an excellent criterion for action thresholds (Wilson et a]. 1981, 1983,1985). Evidence suggests that cotton plants can sustain higher P(I) than previously believed, and that 0.5-0.8 can be considered conservative levels (Goodell & Roberts 1985). To determine the validity of a higher action threshold, we treated spider mite populations with dicofol at 0.2 P(1) and at 0.8 As a contrast to dicofol, methyl parathion was also included in the study. This method simultaneously allowed us to study the effect of various acaricides 0022-0493/90/1604-1611$02.00/0 Q 1990 Entomological Society of America

August 1990 TR~CHILO ET AL.: RELATIVE ABUNDANCE OF SPIDER MITES ON COTTON 1605 upon the relative abundance of each spider mite species. Because cotton has historically been heavily treated with pesticides, spider mites on cotton are coininonly exposed to these chemicals, and often develop resistance that leads to spider mite outbreaks (Bartlett 1968, Reynolds et al. 1982, Iftner & Hall 1984, Penman & Chapman 1988). It is of fundamental importance in studying spider mite population dynamics to have access to, or to induce (e.g., through chemical mediation), a broad range of densities that can be compared. Investigators have found it convenient to test the efficacy of natural enemies by chemically treating cotton plants (Ehler et al. 1973, Trichilo & Leigh 1986), and permethrin often has been used as an experimental device to control natural enemies (Braun et al. 1987a, Penman & Chapman 1988). The use of chemicals such as permethrin to regulate natural enemies simulates the management practice of using these compounds to control key pests, because insecticide applications often result in occasional or moderate pests such as spider mites becoming severe problems (Bartlett 1968, Iftner & Hall 1984, Trichilo & Leigh 1986). Treating spider mite populations with permethrin allowed us to better analyze spider mite species relationships at economically injurious levels. Spider mite populations often peak in middle to late season or have peaks with variable timing from year to year. Thus, spider mites can only be evaluated reliably in terms of their effect on intermediate and late fruit formation, and early fruit maturation, Inoculating cotton plants with a particular spider mite species allows analysis of spider mite infestation on plants of varying age, as well as the effect of establishment time on the relative abundance of that species. Because its feeding damage is the most severe of the three species (Brito et al. 1986), T. tiirkestani was chosen for inoculative release in our study. Materials and Methods Experimental Design. Data on the relative abundance of spider mites on cotton were collected during the summers of 1982 and 1983 at the University of California Westside Field Station. The cultivar Acala SJ-2 was planted 5 April during each season, and thinned to a stand density of 100,000 plants/ha. In each of the 2 yr, individual plots measured 2,800 m2 (56 rows by 50 m). Experimental plots were irrigated and managed using standard practices, although insecticide and acaricide applications were made in accordance with each experimental design. In 1982, a factorial experiment, which totaled 6.7 ha, was developed in the form of a split plot design, in which four 1.7-ha plots were divided into six subplots, each subplot representing an acaricide treatment. Each block consisted of one of four levels of the insecticide permethrin. The acaricide treatments consisted of (1) an untreated control, (2) dicofol at 0.56 kg (AI)/ha (half label rate), applied when designated plots reached 0.2 P(I), (3) dicofol(l.12 kg[ai]/ha) at 0.2 P(I), (4) dicofol(l.12 kg [AI]/ha) at 0.8 P(I), (5) methyl parathion at 0.56 kg (AI)/ha (half label rate) applied on 2 July, and (6) methyl parathion at 1.12 kg (AI)/ha on 2 July, The permethrin blocks consisted of (1) an untreated control, (2) permethrin at 0.112 kg (AI)/ ha (full label rate), applied on 11 June and 17 June, (3) identical to 2, but with a third application of permethrin on 2 July, and (4) identical to 3, but with permethrin at 0.056 kg (AI)/ha (half label rate). The use of permethrin was an attempt to stimulate an increase in spider mite densities. To ensure at least minimal infestations, a small number of spider mites on leaves collected from an infested cotton field with damage resembling that of T. turkestani was released throughout all plots on 4,9, 15, and 22 June. The factorial design, which was based on the assumption of spatial homogeneity, was used to minimize the extent of chemical and arthropod drift betvfeen plots, and thus better ensure that measurements were treatment effects and not artifacts of small plots. Plots were sampled for relative species abundance on 20 July, 4 August, 15 August, and 1 September. Sample dates were chosen to bracket in time the period of peak bloom to peak boll formation. Samples consisted of 40 mainstem leaves per plot, one leaf sampled per plant. The leaf chosen varied with plant growth, and represented the mainstem node most likely to have spider mites (Le., 6-9 nodes from the terminal) (Wilson et al. 1983). Species identification was conducted on 10 male spider mites from each 40-leaf sample. Males were identified on the basis of adeagus morpliology, as described by Tuttle & Baker (1968). Voucher specimens of the three species are present in the University of California-Davis Department of Entomology Museum. Relative abundance was determined by calculating the fraction of each species, in relation to the total number of male spider mites identified. In 1983, a modified factorial experiment encompassing 7.3 ha was also implemented in the form of a split plot design. In this study, T. turkestani from laboratory-reared colonies was released into the field on five dates (8 June, 22 June, 6 July, 20 July, 3 August) corresponding to five experimental blocks. Each block (about 1.1 ha) was divided into four subplots (split plots). Except for the control, subplots received two applications of permethrin at 0.112 kg (AI)/ha, 4 d apart, with the second application 2 d before the release of T. iurkestani. Each subplot consisted of (1) an untreated control that received no mites, (2) mites re1eased, Permethrin at 0.112 kg (AI)/ha, and dicofol (1.12 kg [AI]/ha) applied at 0.2 P(I), (3) mites, permethrinv and dicofol applied at 0.8 P(I), and (4) mites, Permethrin, but no dicofol. In addition, a sixth block was used to test for

1606 JOURNAL OF ECONOMIC ENTOMOLOGY Vel. 83, no, 4 pe a c Untreeted Control T.prciAcus - T.mllreauni ----- T.urricse 800 1000 1200 1400 1600 1800 PHYSIOLOGICAL TIME ( b12oc) After Plantlng Fig. 1. Fluctuation in relative abundance of three species of spider mites on cotton in untreated control, in relation to time of season (Le., degree-days [ D] > 12 C from planting), in 1982. spatial homogeneity, and accounted for the modified part of the design. This block received no pesticides and was divided into six subplots. Five of these subplots received inoculative releases of T. turkestani, corresponding to the mite release dates of each of the other five blocks. The sixth subplot, a control, received no spider mites. Each of the five mite release treatments within block VI were added to the overall analysis, which thus became a modified 5 x 5 factorial design. The experimental design allowed for the testing of homogeneity between blocks, in which an analysis of variance that encompassed the factorial design was used on the untreated, uninoculated control data from each of the six blocks. Given that each untreated plot originated from a different block, the null hypothesis H,, was that no difference existed between blocks. Plots were sampled on five dates, which included 26 July, 1-2 August, 15-18 August, 29-31 August, and 14-16 September. However, because of low mite population densities, the first two sample dates provided an insufficient number of males and were not included in the analysis. Sampling and male identification methods in 1983 were identical to those in 1982, except that 20 males, rather than 10, were identified per 40-leaf sample. Source of T. trrrkestani. In 1983, colonies of T. turkestani, originally obtained from cotton in the southern San Joaquin Valley, were maintained on cotton cotyledons in an isolated chamber at =3O0C, under constant fluorescent illumination. Spider mite colonies were established to reach peak densities corresponding to dates of inoculative release in the field. On appropriate release dates, mite-infested cotyledons from the colony were systematically placed approximately every 2 m, every other row, throughout the plot on individual cotton plants, between the stem axis and a mainstem leaf axil. The mainstem leaf node chosen represented the I most likely focus for a spider mite infestation (see Wilson et al. 19S3). Statistical Analysis. Analysis of variance (ANOVA) was conducted using SAS statistical packages (SAS Institute 1985). Because the data in this study involved proportions which totaled unity, when species were analyzed simultaneously (as a class variable),.-\nova did not produce valid F values for main effects, other than species. The analysis did, however, allow determination of significant interaction among most factors, with the assumption that the highest order interaction becomes the error term (Zar 1984). Significant speciesmain effect interaction can be interpreted as a significant main effect, because a proportional change in one species must be accompanied by an opposite change in at least one of the other species, which is thus an interaction. The proportional problem also was circum\ented by analyzing the data by species, and thus removing species as a variable from the analysis. Because proportional data are often not normally distributed, all analyses were performed before, and after, an arcsine square root transformation of the data (Steel dr Torrie 1980). Although most of the analyses became slightly more conservative as a result of the data transformation, significant ANOVA effects were unchanged. Results presented in the figures and text are untransformed means of data summed over sample date. Tukey s multiple comparison test (Tukey 1953) was used to distinguish significant differences between treatments at the 5% level, and all reported significant main effects and interactions are at a = 0.05. 1982 Results Variation Over Time. In the untreated control, although there \vas fluctuation, T, pacijcus was consis ten tl y the most dominant species throughout the season in 1982 (Fig. 1). Applications of dicofol that corresponded to 0.2 P(1) occurred on 9 July and 15 July (Table 1). Dicofol treatments at 0.8 P(1) occurred on 16 July and 21 July, for Blocks II-IV. However, Block I did not reach 0.8 P(1). Both methyl parathion treatments were applied on 2 July. The first samples were taken at least 18 d after the application of methyl parathion, 5 d after the application of dicofol at 0.2 P(I), and in between dicofol treatments at 0.8 (I). Effect of Pesticides. Relative frequency of T. pacificus was significantly higher than that of the other two species. Relative abundance of T, pacifficus, 0.85, was significantly higher in the presence of dicofol (half rate) at 0.2 P(I) than when treated with methyl parathion at half rate, 0.53, but was not significantly higher than the untreated control, 0.67 (Fig. 2). Conversely, T. turkestani was significantly lower, 0.13, in the dicofol treatment (half rate) at 0.2 P(I), than in the methyl parathion treat-

August 1990 TRICHILO ET AL.: RELATIVE ABUNDANCE OF SPIDER MITES ON COTTON 1607 Table 1. Dates of experimental mite releases, and insectieide/acaricide applications during the years 1982, 1983 TreatmenP Block Permethrin C D0.2(%) D0.2 D0.8 MP(%) MP 1982 1 - - 15 July 15 July - 2 July 2 July I1 11 June, 17 June 15 July 15 July 21 July 2 July 2 July 111 11 June, 17 June, 2 July - 15 July 15 July 16 July 2 July 2 July 1V 11 June, 17 June, 2 July (2%) 9 July 9 July 16 July 2 July 2 July Block Permethrin Mite release Treatment b C D0.2 D0.8 ND Mi 1983 I 2 June, 6 June 8 June 25 June 5 August I1 16 June, 20 June 22 June 9 July 22 July - 22 July 5 August - - 22 July 12 August V 28 July, 1 August 3 August - 30 June 12 August - - 111 30 June, 4 July 6 July IV 14 July 18 July 20 July C, no pesticides; D0.2, dicofol applied at 0.2 P(1); DO.2(%), D0.2 at half label rate; D0.8, dicofol at 0.8 P(1); MP, methyl parathion; MP(%), MP at half label rate; (dh), permethrin at half label rate. C, no pesticides, no mites released; ND, no dicofol; Mi, mites released, no pesticides. ment at half rate, 0.34 (Fig. 2). Relative abundance of T. turkestani was 19% lower than T. pacificus in plots treated with methyl parathion at half rate, but 72% lower than T. pacificus in plots treated with dicofol at 0.2 P(1) (half rate). However, the effect of dicofol did not significantly change the relative abundance of T. furkestani from that of the untreated control, and the relative frequency of T. urticae was not significantly changed as a result of any of the treatments (Fig. 2). There was no significant difference in the relative abundance of each species between plots treated with dicofol at 0.2 P(1) and those treated at 0.8 P(1). In 1982 there was no significant change in species relationships as a result of permethrin applications, although T. pacificus was significantly dominant (Fig. 3). 1983 Spatial Homogeneity. Analysis of the untreated plots from each block resulted in a significant species effect, but no significant species-block interaction, and no significant block effect. Therefore, we did not reject H, and concluded that for our study, the blocks were spatially homogeneous. Variation Over Time. Based on spider mite infestation rates in 1983, dicofol at 0.2 P(1) was applied on 25 June, 30 June, 9 July, or 22 July, and at 0.8 P(1) on 22 July, 5 August, or 12 August (Table 1). Several trends observed in 1982 were again evident. However, with only three usable sample dates, this variable was of little use in determining trends over time. One trend in 1982 that did not occur in 1983 was the dominance by T. pacijicus in the untreated control. In the 1983 untreated control, T. urticae, 0.49, was significantly higher in relative abundance than T. turkestani, 0.07, and slightly higher than T. pacificus, 0.44 (Fig. 4). Effect of Pesticides. As in 1982, T. pacificus was dominant, 0.70, in the presence of dicofol (half rate) at 0.2 P(I), and registered significantly higher than it did in the untreated control, 0.44, in spite of the releases of T. turkestani in the dicofol treatments (Fig. 4). However, the relative abundance of T. urticae was significantly lower in the same dicofol plots, 0.23, than in the control, 0.49. The relative abundance of T. turkestani was significantly lower in mite release plots treated with dicofol and permethrin, 0.10, than in untreated mite release plots, 0.34 (Fig. 4). Again, as in 1982, there was no significant difference in relative species abundance between dicofol at 0.2 P(1) and 0.8 P(1). There was no significant difference in the relative frequency of T. pacificus between the untreated control and those plots that received mites and permethrin, but no dicofol (Fig. 4). However, the relative abundance of T. pacificus was significantly lower in mite release plots that did not receive permethrin, 0.21, than in those that did, 0.43. In the nonpermethrin treatment, the relative abundance of T. pacijicus was significantly lower than that of T. urticae, and marginally lower than that of T. turkestani. Both T. urticae and T. turkestani exhibited (not significantly) lower relative abundance in mite release plots treated with permethrin, 0.33 and 0.23, respectively, than in mite release plots without permethrin, 0.45 and 0.34 (Fig. 4). Effect of Inoculative Release. The proportion of T. turkestani was directly related to time of season, and was significantly higher in those plots represented by the early release dates, 0.34 and 0.27, than the late release dates, 0.04, 0.09, and 0.08 (Fig. 5). There was a significant release datespecies interaction, which suggests that each species responded differently, depending on the particular time during the season T. turkestani entered the system. Relative abundance of T. pacificus and

1608 JOURNAL OF ECONOMIC ENTOMOLOGY w T.NlkCstMi T. URicae z K 1.c 0.8 0.6 B 0.4 T. turkesrani Means for Treatments 1.6 T.unicac - n 0.2 If2 Dicofol Dicofol Dicofol In MP Mp Aca%ie O.ZP(0 0.2P(I) O.SP(1) TREATMENT Fig. 2. Average effects of two acaricides on relative abundance of three species of spider mites on cotton in 1982. Treatments were as follows: no acaricide; dicofol (0.56 kg [AIl/ha = '/2 dicofol) at 0.2 P(1); dicofol (1.12 kg [AI]/ha) at 0.2 P(1); dicofol (1.12 kg [AI]/ha) at 0.8 P(1); methyl parathion at 0.56 kg (AI)/ha (=% MP); methyl parathion at 1.12 kg (AI)/ha. Vertical bars = +I SE. urticae were correspondingly lower in early mite release plots. In Fig. 5, average relative abundance includes data from the untreated plots with no mites released. By observing individual treatments, it becomes apparent that spider mite releases were effective in significantly raising the average relative abundance of 7'.turkestani over the untreated control on 22 June, 20 July, and 3 August, but less so on 8 June (Fig. 6). T. turkestani populations were consistently low in untreated, uninoculated plots, but generally higher (except for 6 July) in plots that received mites but no pesticides. They were, however, moderately high in dicofol plots, and permethrin-no dicofol plots, representing the 8 June (first) release date. The proportional dominance by T. turkestani over the other two species in the 22 June release date plot that did not receive permethrin or dicofol, 0.78, was not observed in samples from later release dates (Fig. 6). The drop in relative abundance of T. turkestanl in plots of the late release dates (Fig. 5) resulted primarily from reduced frequencies in untreated plots compared with the 22 June release date, and a strong expression of dicofol susceptibility in dicofol-treated plots (Fig. 6). Discussion Variation Over Time. Investigators have speculated that abundance of spider mite species is a function of geography and climate (Leigh 1963, Brito 1980). It has also been suggested that the relative abundance of each species varies according to time of season; T. ttirkestanl is an early season species, T. urticae late season, and T. pacificus 0.0 PERMETHRIN LEVEL Fig. 3 Average effects of four permethrin levels on relative abundance of three species of spider mites o,, cotton in 1982. Levels were as follows: no permetilrin; permethrin at 0.112 kg (AI)/ha, applied on 11 June arid 17 June; permethrin applied on 11 June, 17 June, and 2 July; permethrin at 0.056 kg (AI)/ha (=Vi permethrin), applied on 11 June, 17 June, and 2 July. Vertical bars + 1 SE, increases throughout the season (Jackson 81 Leigh 1967, Leigh & Burton 1976, Brito 1980). However, the reason for this seasonality is open to speculation. Dennehy et al. (1987a) found that T. urticue was dominant early in the season (June), and later gave way to T. pacificus (August). The authors suggested that species dominance is tied to acaricide resistance, and that the seasonal change 0.8 z E 0.6 I? 0 0.4 a n 0.2 Means for Blocks I-V T. curkcsmni T.pacilicus o.o Untreated Permcthrin Penncihrin Pemethrin No Pcnnerhrin COnlrOl Dicofol 0.2P(I) Dicofol 0.8P(I) No Dicofol No Dicofol NO Mites Mites Mites Mites Mites TREATMENT Fig. 4. Average effects of dicofol, permethrin, and inoculative releases of T. turkestani (mites) on relative abundance of three species of spider mites on cotton in 1983. Treatments were as follows: no permethrin, no dicofol, and no mites released; permethrin at 0.112 kg (AI)/ha, dicofol (1.12 kg [AI]/ha) at 0.2 P(I), and mites released; permethrin, dicofol at 0.8 P(I), and mites released; permethrin, no dicofol, and mites released; no permethrin, no dicofol, and mites released. Vertical bars = +1 SE.

August 1990 TRICHILO ET AL.: RELATIVE ABUNDANCE OF SPIDER MITES ON COTTON 1609 0.8 t T.~ltiCa~ 0 0.4 p: n 0.2 T, Means!or Treatments 1-5 T T z l.o 0.8 0 0.6 I- K : 0 0.4 K n. 0.2 T. turkestanl NO Permethrin.No Dicolol.No Miles Pcmcthrin.Dicofol0.2P(I).Mitcs RrmeLrin.Dicofo1 O.BP(I).Mitcs @ Pencthrin.No Dicofol.Mitcs NO PCrmClhrin.No Dicofol.Miter nn v.v 8 June 22Junc 6July 20July 3 August RELEASE DATE Fig. 5. Average effects of T. turkestani inoculative releases on relative abundance of three species of spider mites on cotton in 1983. Vertical bars = $1 SE. 0.0 8 June 22 June 6 July 20July 3August RELEASE DATE Fig. 6. Average effects of pesticides and inoculative release of T. turkestani, versus release date, on relative abundance of T. turkestani (i.e., proportion of total mites) in the, field. See Fig. 4. in species represented an inter- as well as intraspecific shift from susceptible to resistant phenotypes. Another report indicated that T. turkestani was dominant in early season (May and June), but later (August) T. urticae had become dominant (Grafton-Cardwell et al. 1987). These data suggest higher late-season frequencies of T. turkestand than we have observed. Spider mite species have perpetuated within the cotton system, in spite of an onslaught of pesticides, and it is difficult to define the normal state. Thus, an untreated control may no longer represent the normal condition, and consideration of competition among spider mite species should include the effect of chemical mediation. Moreover, there is a great potential for the continued reinvasion of resistant phenotypes into relatively small experimental plots from external cultivated and noncultivated (i.e., weed) hosts. Effect of Acaricide. Resistance to sulfur, dicofol, and propargite on cotton has been well documented for T. pacificus and T. urticae, whereas T. turkestani has remained conspicuously susceptible to all three compounds (Dennehy & Granett 1984; Dennehy et al. 1987a,b; Grafton-Cardwell et al. 1987). Our results support these findings. Dicofol conferred a selective advantage to T. pacificus, and (compared with the effect of methyl parathion) negatively affected the relative abundance of T. turkestani. The nonsignificant reduction in relative abundance of T. turkestani in dicofol plots compared with the control does not contradict the reported susceptibility of this species to dicofol. Relative abundance of T. turkestani was consistently low, even in untreated plots. This frequency was lowered further in dicofol plots, although statistical significance was not achieved. The relative abundance of T. turkestani in dicofol-treated plots can be considered a basal frequency, especially because of inoculations with spider mite-infested leaves ear- lier in the season. Thus we can assume that frequencies of T. turkestani in dicofol-treated plots were the lowest achievable in our experiments, irrespective of statistical significance. At such low frequencies, some spider mites will invariably escape pesticide exposure, and thus, always be present to some degree. Although there are reports of dicofol resistance in some strains of T. urticae (Dennehy et al. 1987a), our results suggest that populations of this species in our study were susceptible to dicofol. Dennehy et al. (1987b) suggest that dicofol resistance is unstable. However, resistance may never have existed in our populations of T. urticae. That the relative abundance of T. turkestani was significantly higher in the presence of methyl parathion than with dicofol suggests that this species was not negatively affected by methyl parathion. McWhorter (1982) reported evidence of organophosphate resistance in T. turkestani, and methyl parathion has also been reported to stimulate increased fecundity in T. urticae (Maggi & Leigh 1983). The decrease in relative abundance of T. pacificus from dicofol to methyl parathion treatments indicates a loss of the selective advantage it gained in the presence of dicofol. Pesticide resistance can contribute in large part to the competitive fitness of spider mites on cotton. Carey & Bradley (1982) showed that there were essentially no differences in reproductive fitness, based on the intrinsic rate of increase, I, between the three spider mite species. Thus, the proportional dominance by T. pacificus must be due to an external selective advantage, such as chemical mediation. The potential reservoir for T. puci$~us is vast (e.g., cotton, alfalfa, grapes, almonds), and experimental control plots are likely to reflect the overall relative species abundance of the environment around them. Untreated experimental Plots are small and are easily inundated by spider mites that dominate surrounding acreages. Immigration

1610 JOURNAL OF ECONOMIC ENTOMOLOGY Vel. 83, no. 4 from other areas where acaricides are used favors the increase in resistant species such as 7'. pacijcus over susceptible T. turkestani. Effect of Permethrin. Failure of permethrin to alter relative species abundance in 1982 suggests that the relationship among the three spider mite species is relatively stable, and to a degree, resists perturbation. In 1983, T. pacificus showed a slightly positive response to permethrin, again indicative of the minimal effect of this chemical on relative species stability. Braun et al. (1987a,b) reported that permethrin was not toxic to spider mites, and that sublethal levels had no effect on reproductive rate. The authors concluded that permethrin reduced the levels of natural enemies, thereby stimulating spider mite outbreaks. Iftner & Hall (1983, 1984), reported that permethrin caused feeding disruption and dispersal in T. urticae, and in another case, higher fecundity and faster development. Spider mites generally have higher reproductive rates at low densities when dispersed, than at high densities when clumped (Wrensch & Young 1978). Penman & Chapman (1988) suggested that this difference is because at lower densities, mites are less subject to density-dependent competition. We observed that permethrin had a slightly (not significant) negative effect on the relative abundance of T. turkestani and T. urticae that was significantly amplified by the application of dicofol. It is unlikely that predation alone was species selective for spider mites on cotton, and therefore predation would not account for fluctuation in relative spider mite abundance. Dominance by T. pacificus suggests that because this species was the most abundant (due to dicofol resistance), it was most readily able to benefit from a chemical stimulus or reduced predation pressure, which are potential outcomes of permethrin applications. Effects of Inoculative Release. As an experimental treatment, inoculative release in 1983 moderately increased the relative abundance of T. turkestani in the field, and the timing of the releases proved to be critical, with only the early releases giving rise to an increase in establishment of T. turkestani. The importance of establishment time is further accentuated in Fig. 6, which indicates that T. turkestani was sufficiently well established before the first release date and could maintain moderately high relative abundance even in the presence of dicofol. Late releases resulted in lower levels of T. turkestani in untreated plots than did early releases, which suggests that natural populations of T. turkestani were not sufficiently present later in the season to bolster the late releases. Therefore, dicofol susceptibility was more readily expressed in late releases than in early releases. Moreover, late releases allowed progressively more time for the other species to increase (through reproduction and immigration from surrounding fields), relative to the constant numbers of T. turkestani that were released. Yield loss is directly proportional to date of infestation, and in general, earlier infestations give rise to greater losses (Furr & Pfrimmer 1968, Miss tric 1969, Wilson 1986). Because degree of damage is species dependent, time of spider mite establishment is of even greater consequence when relative species abundance is considered. With the avail. ability of a species component, management models should be able to generate greater specifkity as to the degree of expected damage, based on date of spider mite establishment within the cotton system, Acknowledgment This research was supported in part by a grant to [,, T. Wilson from the U.S. Environmental Protection Age,,. cy and USDA funded National Consortium for Intcgrated Pest Management, and by a grant to D. Gonza& and L. T. Wilson from the University of California systemwide Integrated Pest Management Program, References Cited Bartlett, B. R. 1968. Outbreaks of two-spotted spidermites and cotton aphids following pesticide treatment. I. Pest stimulation vs. natural enemy destruction as the cause of outbreaks. J. Econ. Entomol. 61: 297-303. Braun, A. R., J. M. Guerrero, A. C. Bellotti & L. T. Wilson. 1987a. Relative toxicity of permethriri to Monon ychellus progressivus Doreste and Tetranythus urttcae Koch (Acari: Tetranychidae) and their predators Amblyselus limonicus Garman & Mc- Cregor (Acari: Phytoseiidae) and Oltgota minuta Cameron (Coleoptera: Staph ylinidae): bioassays and field validation. Environ. Entomol. 16: 545-550, 1987b. Evaluation of possible nonlethal side effects of permethrin used in predator exclusion experiments to evaluate Amblyseius limonicus (Acari: Phytoseiidae) in biological control of cassava mites (Acari: Tetranychidae). Eviron. Entomol. 16: 1012-1018. Brito, R. M. 1980. Factors affecting the distribution and abundance of three species of Tetranychus spider mites on cotton and the effect of their damage on transpiration and photosynthesis, Ph.D. dissertation, University of California, Riverside. Brito, R. M., V. M. Stern & F. V. Sances. 1986. Physiological response of cotton plants to feeding of three Tetranychus spider mite species (Acari: Tetranychidae). J. Econ. Entomol. 79: 1217-1220. Canerday, T. D. & F. S. Arant. 1964. The effect of late season infestations of the strawberry spider mite, Tetran ychus atlnnticus, on cotton production. J. Econ. Entomol. 57: 931-933. Carey, J. R. & J. W. Bradley. 1982. Developmental rates, vital schedules, sex ratios, and life tables for Tetranychus urticae, T. turkestani, and T. pacificus (Acarina: Tetranychidae) on cotton. Acarologia 23: 333-315. Dennehy, T. J. & J. Granett. 1984. Monitoring dicofol-resistant spider mites (Acari: Tetranychidae) in California cotton. J. Econ. Entomol. 77: 1386-1392. Dennehy, T. J., J. Granett, T. F. Leigh & A. Colvin. 1987a. Laboratory and field investigations of spider mite (Acari: Tetranychidae) resistance to the selective acaricide propargite. J. Econ. Entomol. 80: 565-574. Dennehy, T. J., E. E. Grafton-Cardwell, J. Granett

August 1990 TRICHILO ET AL.: RELATIVE ABUNDANCE OF SPIDER MITES ON COTTON 1611 K. Barbour. 1987b. Practitioner-assessable bioassay for detection of dicofol resistance in spider mites (Acari: Tetranychidae). J. Econ. Entomol. 80: 998-1003. Ehler, L. E., K. G. Eveleens & R. van den Bosch. 1973. An evaluation of some natural enemies of cabbage looper on cotton in California. Environ. Entomol. 2: 1009-1015. Furr, R. E. & T. R. Pfrimmer. 1968. Effects of early-, mid-, and late-season infestations of two-spotted spider mites on yield of cotton. J. Econ. Entomol. 61: 13-lG-1447. Goodell, P. B. & B. Roberts. 1985. Implementation of a presence-absence sampling method for spider mites on cotton, pp. 170-171. In Proceedings, Beltwide Cotton Producers Research Conference. National Cotton Council of America, Memphis, Tenn. Grafton-Cardwell, E. E., J. Cranett & T. F. Leigh. 1987. Spider mite species (Acari: Tetranychidae) response to propargite: basis for an acaricide resistance management program. J. Econ. Entomol. 80: 579-587. Iftner, D. C. & F. R. Hall. 1983. Effects of fenvalerate and permethrin on Tetranychus urticae Koch (Acari: Tetranychidae) dispersal behavior. Environ. Entomol. 12: 1782-1786. 1984. The effects of fenvalerate and permethrin residues on Tetranychus urticae Koch fecundity and rate of development. J. Agric. Entomol. 1: 191-200. Jackson, C. & T. F. Leigh. 1967. Sulfur for suppression or control of tetranychid mites on cotton. J, Econ. Entomol. 60: 30-33. Leigh, T. F. 1963. Considerations of distribution, abundance, and control of acarine pests of cotton. Adv. Acarol. 1: 14-20. Leigh, T. F. & V. E. Burton. 1976. Spider mite pests of cotton, Leaflet 2888, Division of Agricultural Sciences, University of California, Berkeley. Maggi, V. L. & T. F. Leigh. 1983. Fecundity response of the twospotted spider mite to cotton treated with methyl parathion or phosphoric acid. J. Econ. Entomol. 76: 20-25. McWhorter, C. M. 1982. Spider mites: a status report, pp. 189-191. In Proceedings, Beltwide Cotton Producers Research Conference. National Cotton Council of America, Memphis, Tenn. Mistric, W. J. 1969. Damage by the strawberry spider mite on cotton when infestations commenced at the beginning, middle, and end of the flowering period. J. Econ. Entomol. 62: 192-195. Penman, D. R. & R. B. Chapman. 1988. Pesticideinduced mite outbreaks: pyrethroids and spider mites. Exp. Appl. Acarol. 4: 265-276. Reynolds, H. T., P. L. Adkisson, R. F. Smith & R. E. Frisbie. 1982. Cotton insect pest management, pp. 375-441. In R. L. Metcalf & W. H. Luckman [eds.], Introduction to insect pest management. Wiley, New York. SAS Institute. 1985. SAS user's guide: statistics, version 5 ed. SAS Institute, Cary, N.C. Steel, R. G. & J. H. Torrie. 1980. Principles and procedures of statistics. A biometrical approach. McGraw-Hill, New York. Trichilo, P. J. & T. F. Leigh. 1986. The impact of cotton plant resistance on spider mites and their natural enemies. Hilgardia 54: 1-20, Tukey, J. W. 1953. The problem of multiple comparisons. Ditto, Princeton University, Princeton, N. J. Tuttle, D. M. & E. W. Baker. 1968. Spider mites of the southwestern United States and a revision of the family Tetranychidae. University of Arizona Press, Tucson. Wilson, L. T. 1986. Developing economic thresholds in cotton, pp. 308-344. In R. E. Frisbie & P. L. Adkisson [eds.], Integrated pest management on major agricultural systems. Texas A&M University, College Station. Wilson, L. T., T. F. Leigh & V. Maggi. 1981. Presence-absence sampling of spider mite densities on cotton. Calif. Agric. 35 (7 & 8): 10. Wilson, L. T., D. Conzalez, T. F. Leigh, V. Maggi, C. Foristiere & P. Coodell. 1983. Within-plant distribution of spider mites (Acari: Tetranychidae) on cotton: a developing implementable monitoring program. Environ. Entomol. 12: 128-134. Wilson, L. T., D. Gonzalez & R. Plant. 1985. Predicting sampling frequency and economic status of spider mites on cotton, pp. 168-170. In Proceedings Beltwide Cotton Producers Research Conference. National Cotton Council of America, Memphis, Tenn. Wrensch, D. L. & S. S. Y. Young. 1978. Effects of density and host quality on rate of development, survivorship, and sex ratio in the carmine spider mite. Environ. Entomol. 7: 499-501. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, N.J. Recetued for publication 12 June 1989; accepted 3 October 1989.