Acari: Eriophyidae - BioOne

POPULATION ECOLOGY

Ecology and Aerobiology of Dispersing Citrus Rust Mites (Acari: Eriophyidae) in Central Florida J. C. BERGH1 IFAS, Citrus Research and Education Center, University of Florida, 700 Experiment Station Road, Lake Alfred, FL 33850

Environ. Entomol. 30(2): 318Ð326 (2001)

ABSTRACT Aerial dispersal of citrus rust mite, Phyllocoptruta oleivora (Ashmead), in central Florida showed a diel periodicity peaking between late morning and early afternoon. The abiotic factors that best described the dispersal pattern were solar radiation, time, and leaf wetness; whereas wind speed, humidity, temperature, and rainfall had minimal effect. The longevity of adult mites removed from fruit was inversely related to constant temperatures between 25 and 35⬚C. The longevity of mites removed from fruit at 2-h intervals between 0700 and 1300 hours and exposed to ambient temperature and humidity outdoors was inversely related to the time of removal from the host. There was a linear relationship between the number of mites captured in traps and population density on fruit. Mites left fruit harboring extremely low populations and some fruit supporting dense populations yielded hundreds of dispersing mites per day. Mites were readily carried on air currents between adjacent citrus groves. Nearly all of the mites captured in dispersal traps were adult females, and were found in greater proportions in traps than would be expected from the sex ratio of mites on fruit. Studies using isolated fruit in the laboratory showed that a single virgin or inseminated female could initiate a local population through oedipal mating and sibmating. The data are discussed in relation to the selective forces that may shape the dispersal behavior of citrus rust mite and to the potential impact of aerial dispersal on mite management and the development of acaricide resistance. KEY WORDS Phyllocoptruta oleivora, dispersal, traps, mite management

CITRUS RUST MITE, Phyllocoptruta oleivora (Ashmead), is a key pest of citrus grown in humid climates. Cumulative feeding on the surface of fruit by P. oleivora results in cosmetic damage and rejection of fruit for the fresh market. In Florida citrus, the majority of pesticide applications each year are for management of this mite. Colonization of new citrus hosts by P. oleivora is primarily by aerial dispersal (Bergh and McCoy 1997). Like other eriophyid species (Sabelis and Bruin 1996), adult citrus rust mites exhibit dispersal behaviors, including standing erect on their caudal lobes and leaping, which are thought to elevate them above the boundary layer within which they live (Frost 1997). Mites that release their hold on the host plant are borne away on air currents and can be captured in traps (e.g., Slykhuis 1955, Nault and Styer 1969, Easterbrook 1978, Bergh and McCoy 1997). Most eriophyid mites, including citrus rust mite, are host-speciÞc (Cromroy 1979). Although mites actively initiate dispersal, once aloft, dispersal is a passive process resulting in their random deposition throughout the environment. Mites deposited on a plant can probably choose to stay or leave, depending on its suitability as a host, but it is believed that dispersalrelated mortality is high (Jeppson et al. 1975, Sabelis 1 Current address: Alson H. Smith, Jr., Agricultural Research and Extension Center, Virginia Polytechnic Institute and State University, 595 Laurel Grove Road, Winchester, VA 22602.

and Bruin 1996). In Florida, the citrus rust mite is regularly exposed to extreme climatic conditions that may increase the risks associated with dispersal, relative to those for species of eriophyids living in temperate climates. Daily maximum temperatures from June through September often reach levels that are lethal (Reed et al. 1964) or exceed the limits for reproduction and development (Allen et al. 1995), and frequent late afternoon or evening storms bring heavy precipitation and high wind. Heavy dew that frequently develops during the night often does not dry until midmorning, and the combination of dew and rainfall frequently results in citrus fruit and foliage remaining wet for extended periods. Being only ⬇150 ␮m long (Jeppson et al. 1975), adult citrus rust mites are physically constrained from dispersing from wet fruit and foliage. Although P. oleivora is one of the most economically important pests of citrus, little is known about its population dynamics. Given the fundamental importance of immigration and emigration to the structure and development of arthropod populations, the overriding objective of the study reported here was to improve our understanding of how aerial dispersal may inßuence the population dynamics and management of citrus rust mite. SpeciÞc objectives of laboratory and Þeld experiments spanning 3 yr were to determine the following: (1) the effect of abiotic, environmental factors on the temporal pattern of dispersal and the survival of dispersing mites; (2) the

0046-225X/01/0318Ð0326$02.00/0 䉷 2001 Entomological Society of America

April 2001

BERGH: ECOLOGY AND AEROBIOLOGY OF DISPERSING CITRUS RUST MITES

relationship between population density and dispersal; (3) the dispersal distance in commercial citrus groves; (4) the sex-ratio of dispersing mites; and (5) the potential for dispersing females to initiate new populations through oedipal and sibmating. Materials and Methods Traps. Three types of traps were used to capture dispersing mites: unidirectional, omnidirectional, and Þlter paper. The unidirectional trap was a 2.5 by 7.6-cm glass microscope slide with silicone grease (Dow Corning, Midland, MI) applied thinly to 5.1 cm along the length of one side. The slide was inserted into a vertical slit cut in a 1.5-cm i.d. (0.6-inch) conduit pipe coupling attached to a 1.5-m section of conduit pipe, and held in place by the second screw on the coupling. The conduit pipe was inserted vertically into the soil so that the trapping surface was at ⬇1.2 m elevation. The unidirectional trap was designed to capture airborne mites coming only from the direction the trap was facing. Captured mites were counted using a dissecting microscope at 16 Ð25⫻. The omnidirectional trap was designed to capture airborne mites regardless of wind direction, and it was placed in row middles within a grove. The trap consisted of a 1.5-m length of 1.5-cm i.d. conduit pipe with a threaded adapter attached to one end. A PVC adapter was screwed to the conduit pipe adapter and a 0.3-m section of 2.2-cm o.d. PVC pipe (seven-eighths in) inserted into the unthreaded end of the PVC adapter. The trapping surface was a thin layer of silicone grease applied in a 1.5 cm wide band along the length of one side of a 7 cm long by 2.5-cm wide strip of clear transparency Þlm (Apollo W0100C, Apollo Presentation Products, Ronkonkoma, NY). The strip was wrapped around the PVC pipe, 8.0 cm from one end, and held in place using double-sided tape attached to the pipe. The conduit pipe was inserted vertically into the ground and the trapping surface was at an elevation of ⬇1.7 m. After trapping, the strip was laid ßat on a 10 by 3.8-cm piece of clear glass (0.25 cm thick) using double-sided tape, and mites were counted as described above. The Þlter paper trap was designed to capture mites dispersing from individual fruit growing singly in the outer canopy and hanging vertically at 0.75Ð1.25 m above the ground. Other fruit near the fruit selected were pruned, but foliage was pruned minimally. The trapping surface was a 7-cm-diameter Þlter paper circle (Whatman No. 50, Whatman International, Maidstone, England) marked with a grid pattern, sitting on wet cotton in a petri dish (19 cm diameter). A screened cage (1.5-mm2 mesh) with an 18 by 18-cm wood bottom was screwed to an adapter on a 60 cm long section of conduit pipe (1.8 cm o.d.; 0.71 inch), which inserted into an 80 cm long section of pipe (1.9 cm i.d.; 0.75 inch). After the thicker section of pipe had been placed in the ground below a fruit, an adjustable collar around the thinner pipe allowed the cage bottom to be positioned ⬇5 cm below a fruit suspended

319

within the cage. Fruit were prevented from swaying within the cage, using string to secure the branch to the tops of opposite corner posts. A sliding wooden sleeve around the bottom of the cage allowed the petri dish to be centered below the fruit and replaced without touching or moving the fruit. After trapping, the Þlter paper circles were partially dried and live mites were counted using backlighting. Because hundreds of mites were often captured by these traps, counts were by made placing a black ink dot on each mite. Temporal Distribution. To examine the temporal distribution of aerial dispersal, four omnidirectional traps were deployed in two ÔValenciaÕ orange groves near Lake Alfred, FL (Citrus Research and Education Center (CREC) and Berry), for seven nonconsecutive days in September and October 1997. Beginning at 0700 hours, the traps were replaced at 2-h intervals until 1900 hours and then again at 0700 hours the following day. Preliminary studies in 1996 had shown that very few mites were trapped between 1900 and 0700 hours. Live mites captured were counted immediately after trap replacement. This test was repeated in September 1999 using Þve Þlter paper and Þve omnidirectional traps in a Valencia orange grove at the CREC for four consecutive days. Omnidirectional traps were replaced at the same intervals as in 1997. Filter paper traps were adversely affected by rain and therefore were not deployed from 1900 to 0700 hours. Data from each site in 1997 and each trap-type in 1999 were analyzed separately, using PROC analysis of variance (ANOVA) of SAS (SAS Institute 1988) and the least signiÞcant difference (LSD) test to compare the number of mites captured among trapping intervals. All statistical comparisons of data from this and subsequent tests were made at the 5% level. Data recorded hourly by the CREC weather station were used in a stepwise regression (SAS Institute 1988) to determine which of the following variables best described the dispersal pattern in 1997 and 1999; time of day, solar radiation (Mjcm-2 in 1997, W/m2 in 1999), temperature (⬚C), relative humidity (%), wind speed (kph), rain (cm), and leaf wetness (% of leaf surface). Different instrumentation used in 1997 and 1999 measured solar radiation differently, precluding an analysis of pooled data. Survival of Mites Off the Host. The survival of mites off the host under constant and ambient conditions was examined. A cohort of adult mites was reared from late-stage nymphs transferred from Þeld-collected to clean fruit late in the afternoon. These fruit were held under humid conditions in a controlled environment chamber set at 25⬚C and a photoperiod of 14:10 (L:D) h. At 0630 hours on the morning of the second day thereafter, 45 adults were placed individually in the depression on “hanging drop” microscope slides and conÞned there by a square piece of Þne mesh, nylon screen glued to the slide using ElmerÕs School glue (ElmerÕs Products, Columbus, OH). At 0700 hours, slides with mites were placed on plastic trays, with the screen facing upward, and held in partially covered, clear plastic boxes lined with wet paper towel in con-

320

ENVIRONMENTAL ENTOMOLOGY

trolled-environment chambers set at 25, 30, and 35⬚C (15 mites per temperature). Beginning at 0800 hours, mortality was assessed at 1-h intervals until all mites had died, using a dissecting microscope at 10 Ð16⫻. Dead mites had a contracted body with legs curled underneath the prodorsal shield. This test was repeated twice in late August 1999. One-way ANOVA followed by the Tukey test were used to compare the longevity of mites among temperatures. To determine the effect of time of removal from the host on the mortality of mites exposed to ambient conditions outdoors, a cohort of adult mites was generated as described previously. At 2-h intervals from 0700 to 1300 hours, groups of 15 mites were removed from fruit and placed individually in depression slide cages. The slides were placed on plastic trays, with the screened side facing up, and the trays were placed under a cardboard canopy on a wooden table exposed to full sunlight. Beginning 1 h after removal from the host, mortality was assessed at 1-h intervals until all mites had died. Ambient temperature and humidity data were provided by the CREC weather station, located ⬇75 m from where the test was performed. This experiment was repeated twice in early September 1999. One-way ANOVA followed by the Tukey test were used to compare the longevity of mites removed from the host at different times. Population Density. Mite population density on 10 fruit was measured in the afternoon in an unsprayed Valencia orange grove at the CREC. Using a 16⫻ hand-lens, the Rogers et al. (1994) modiÞcation of the Horsfall-Barratt system (Horsfall and Barratt 1945) was used to determine the number of mites in four lens Þelds around the circumference of each fruit. This method involves assigning the mites to one of seven size classes (0, 1Ð3, 4 Ð 6, 7Ð12, 13Ð25, 26 Ð50, and ⬎50) without counting the actual number of individuals. After density measurements were taken, the fruit were suspended with the screened cage of the Þlter paper trap. The traps were deployed the following day from 0800 to 1200 hours, and captured mites were counted. Ten new fruit were used each day for Þve nonconsecutive days in August 1999. Least-squares linear regression on pooled data were used to describe the relationship between mite density and the number of mites captured. Sex Ratio. Mites were captured in omnidirectional and Þlter paper traps and sampled from fruit using a small camelÕs-hair brush on Þve nonconsecutive days in September and October 1999. Samples were taken on different days from two groves at the CREC, Lake Alfred. Five heavily infested fruit were selected in the afternoon and suspended within the screened cage of the Þlter paper traps. At 0800 hours the following morning, a trap was placed in each cage and Þve omnidirectional traps were also deployed in the row middles within the grove. Traps were retrieved at 1200 hours. At 1000 hours on each day that traps were deployed, mites from Þve heavily infested fruit from the same grove were brushed onto the greased surface of microscope slides.

Vol. 30, no. 2

The mites captured in each omnidirectional trap were concentrated in small spots of grease on a glass slide for storage. A subsample of ⬇30 live, adult mites from each Þlter paper trap and ⬇50 adults from each fruit was similarly collected and stored. The species and gender was determined for all of the mites captured in omnidirectional traps and for groups of 25 from the subsamples taken from Þlter paper traps and fruit. Mites were Þrst degreased and Þxed in Hemo-De (Fisher, Pittsburgh, PA), then cleared in NesbittÕs solution and slide-mounted in groups in a drop of NesbittÕs solution. A phase contrast microscope was used for identiÞcation of species and gender. Gender determinations were based on examination of the genitalia (Lindquist 1996). The proportion of female P. oleivora in each sample was calculated and transformed by arcsine square-root before comparisons among samples from each day using ANOVA and the Tukey test. Dispersal Distance. The aerial movement of mites between adjacent citrus groves was examined at three locations near Davenport, FL, in September and October 1997. This experiment used the east/southeast wind that prevails in Florida during the summer and the fact that unidirectional traps capture only mites being carried from one direction. At each location, two commercial citrus groves with north-south oriented rows were separated by a Þeld or a secondary road. The east grove at each site supported a heavy population of P. oleivora and was considered the source of dispersing mites. The adjacent grove, west of each source grove, was considered the sink for dispersing mites. At the Highway 547 (H547) and Minute Maid Road (MMR) sites, a Þeld separated ÔHamlinÕ orange groves by 98 and 103 m, respectively. Four unidirectional traps were deployed ⬇3 m from the edge of the source grove, spaced ⬇12 m apart, with trapping surfaces facing east. Lines of four east-facing traps were also deployed approximately half-way between the two groves and at ⬇2 m from the edge of the sink groves. At the Orchid Drive site, a road separated Valencia orange groves by 13 m. Only two lines of four traps (12 m apart) were deployed at Orchard Drive, one at ⬇2.4 m from the source grove and the other at ⬇1.2 m from the sink grove. Traps were deployed for three consecutive 7-d intervals at H547 and MMR, and for two consecutive 7-d periods at Orchard Drive. Traps were replaced after each 7-d interval and the number of mites captured in each trap was recorded. Comparisons of trap-catches among distances from the source grove at each site used one-way ANOVA followed by the Tukey test (SAS Institute 1988). Linear regression with a repeated measures function was used to predict the log number of mites trapped as a function of distance. Nonrandom Mating. The potential for the development of local populations based on oedipal mating and sibmating was examined by generating cohorts of inseminated and virgin females. Unsprayed Valencia oranges were collected from the Þeld in early June, washed in distilled water, dried, and dipped in liquid

April 2001 Table 1. and 1999


321

Mean ⴞ SD number of citrus rust mites captured in dispersal traps deployed in citrus groves near Lake Alfred, FL, in 1997

Grove

Trapping interval (time of day) 7Ð9

9 Ð11

CREC (omnidirectional) Berry (omnidirectional)

0.71 ⫾ 1.1ab 1.0 ⫾ 1.0a

4.0 ⫾ 2.5c 15.7 ⫾ 8.8c

CREC (omnidirectional) CREC (Þlter paper)

1.0 ⫾ 0.8a 27.8 ⫾ 5.2ab

6.0 ⫾ 4.7b 266.3 ⫾ 153.0c

11Ð13

13Ð15

1997 2.1 ⫾ 1.1b 1.9 ⫾ 1.3ab 14.4 ⫾ 5.3bc 12.7 ⫾ 7.1bc 4.0 ⫾ 1.6bc 124.5 ⫾ 51.8b

1999 1.3 ⫾ 1.0ac 58.8 ⫾ 32.5ab

15Ð17

17Ð19

19 Ð7

1.3 ⫾ 0.9ab 9.0 ⫾ 6.5b

0.6 ⫾ 0.9a 2.7 ⫾ 1.8a

0.6 ⫾ 0.8a 1.4 ⫾ 1.7a

1.0 ⫾ 1.2a 40.3 ⫾ 40.6ab

0.8 ⫾ 1.0a 11.0 ⫾ 11.0a

0.5 ⫾ 0.6a NA

Four omnidirectional traps per site for 7 d in September and October 1997 at the CREC and Berry groves, respectively. Five omnidirectional and Þlter paper traps at the CREC for 4 d in September and October 1999. Means within rows followed by the same letter are not signiÞcantly different at the 5% probability level by ANOVA and the LSD test.

parafÞn wax, leaving about one-third of the surface unwaxed. The circular, unwaxed area was divided into four cells using Tangletrap (Tanglefoot, Grand Rapids, MI), and a single nymph, about to enter the imagochrysalis stage (Sternlicht and Goldenberg 1971), was transferred from Þeld-collected fruit to each cell. The fruit were held for 2 d under humid conditions in a controlled-environment chamber at 25⬚C. Citrus rust mite reproduction is arrhenotokous (Swirski and Amitai 1959), and virgin females that had developed from isolated nymphs were identiÞed by their deposition of male eggs within the cell. Inseminated females were generated by placing virgins on the unwaxed area of fruit on which numerous males (generated previously by virgin females) had deposited spermatophores (OldÞeld et al. 1970) during the previous 24 h. To maximize the probability of insemination, virgin females were transferred to areas with many spermatophores and left on the fruit for 4 h (OldÞeld and Newell 1973). Forty unsprayed Valencia fruit were washed and waxed as described above, leaving a 3.5-cm-diameter circular area unwaxed. A dissecting microscope was used to check for extraneous citrus rust mites, their eggs, or other arthropods in the unwaxed area. Using a Þne-tipped, indelible ink marker and a template, four 0.5-cm2 observation areas were marked at approximately equidistant locations around the perimeter of the unwaxed area. A single inseminated or virgin female mite was placed in the center of the unwaxed area on fruit (n ⫽ 20 per group). The fruit were placed in covered, plastic boxes in a controlled environment chamber set at 25⬚C and a photoperiod of 14:10 (L:D) h. High humidity was maintained in the boxes by lining the bottom with wet paper towel. After 7 d, the entire unwaxed area on each fruit was examined for the presence of the female mite and any of her progeny. During this examination, several adults were found in the unwaxed area on Þve fruit (two with inseminated and three with virgin females). The extraneous adults were assumed to have developed from citrus rust mite eggs overlooked during the inspection on the Þrst day of the test, and those fruit were discarded. At 7, 21, and 28 d after the onset of the test, the number of eggs, juveniles, and adult mites inside the four observation areas on the remaining fruit was recorded.

The chi-square test with Yates correction for continuity was used to compare the number of inseminated and virgin females that had initiated a population on fruit by day 28. The criterion used for population initiation was the presence of more than Þve individuals from all life stages counted within the four, 0.5-cm2 observation areas on day 28. One-way ANOVA followed by the Tukey test were used to compare the mean number of eggs, juveniles, and adults found in the observation areas after 28 d. Results Temporal Distribution. The capture of dispersing mites by omnidirectional traps in the Berry and CREC groves in 1997 showed a diel periodicity. The onset of dispersal was pronounced, beginning after 0900 hours (Table 1). Mite captures were statistically largest between 0900 and 1100 hours at the CREC, and numerically, though not statistically, largest during the same interval at the Berry grove. In 1997, 75.8 and 71.8% of the total catch occurred between 0900 and 1500 hours at the Berry and CREC groves, respectively, and very few mites were captured after 1700 hours. The same diel dispersal patterns were recorded at the CREC grove in 1999 and were independent of the type of trap or sample size (n ⫽ 2,116 and 59 mites in Þlter paper and omnidirectional traps, respectively) (Table 1). The dispersal peak between 0900 and 1100 hours was signiÞcantly different from other trapping intervals in the Þlter paper traps, and numerically largest in the omnidirectional traps. The majority of mites were captured during the 6-h period from 0900 to 1500 hours (84.9 and 76.3% in Þlter paper and omnidirectional traps, respectively). The effect of abiotic variables on the capture of dispersing mites was generally consistent among the four data sets (Table 2). Solar radiation was the best one-variable model in all cases. The relationship was improved considerably by a two-variable model incorporating time and leaf wetness for three data sets, and solar radiation and humidity for one data set. None of the other variables made important improvements to the models. Survival of Mites Off the Host. The longevity of a cohort of adult mites off the host plant was inversely related to the constant temperature to which they

322


Vol. 30, no. 2

Table 2. Results of stepwise regression describing the effect of abiotic factors on the capture of dispersing citrus rust mites in traps replaced at intervals throughout the day

Grove/trap

Month/ year

CREC/omni Berry/omni

One-variable model Parameter estimate

(SE)

9/97

Intercept Solar radiation

10/97

CREC/omni CREC/Þlter paper

Two-variable model

Model r2

F

0.64 (0.36) 0.66 (0.20)

0.186

10.74


1.40 (1.40) 4.28 (0.71)

0.437

36.52

9Ð10/99


0.26 (0.66) 0.006 (0.002)

0.327

12.65

9Ð10/99


0.350

11.85

⫺2.11 (31.92) 0.24 (0.07)

P

Parameter estimate

(SE)

Intercept 5.43 (0.82) Time ⫺0.25 (0.06) Leaf wetness ⫺2.65 (0.63) ⬍0.001 Intercept ⫺11.40 (5.29) Solar radiation 5.73 (0.89) Humidity 0.16 (0.06) ⬍0.01 Intercept 12.92 (2.63) Time ⫺0.68 (0.17) Leaf wetness ⫺0.75 (0.17) ⬍0.01 Intercept 586.79 (99.11) Time ⫺32.08 (6.40) Leaf wetness ⫺28.39 (6.39) ⬍0.01

Model r2

F

P

0.364

13.18

⬍0.001

0.505

23.43

⬍0.001

0.441

9.86

⬍0.001

0.559

13.33

⬍0.001

No three-variable model signiÞcantly improved the correlation coefÞcient, according to the maximum r2 improvement procedure.

were exposed (Table 3). Adult mites removed from the host at 2-h intervals between 0700 and 1300 hours and exposed to ambient temperature and humidity showed an inverse relationship between longevity and time of removal (Table 4). Population Density. There was a signiÞcant, linear relationship between population density and the number of mites captured (Fig. 1). Mites were captured even when populations on fruit averaged less than one per lens Þeld and ⬎300 mites were captured from some fruit during the 4-h trapping interval. Sex Ratio. With exceptions in two traps, the mites sampled from Þlter paper and omnidirectional traps were adult females (Table 5). The percentage of female mites in samples from Þlter paper and omnidirectional traps was signiÞcantly greater than from samples of mites brushed from fruit in all but one comparison. The percentage of adult male mites in samples brushed from individual fruit ranged from 0 to 52% (mean ⫾ SD ⫽ 16.8 ⫾ 13.5%). Dispersal Distance. At H547 and MMR, more mites were captured next to the upwind, source grove than in traps halfway between the groves or next to the downwind grove; whereas at Orchard Drive, the number of mites trapped next to the source and sink groves was not different (Fig. 2). The estimated percentages and 95% conÞdence intervals of mites that left the source grove and arrived at the sink grove was 22% (1, Table 3. Mean ⴞ SD longevity (hours) of a cohort of adult citrus rust mites removed from fruit and exposed to constant temperatures Temp, ⬚C

27 Aug 1999 30 Aug 1999

25 (n ⫽ 12)

30 (n ⫽ 15)

35 (n ⫽ 15)

14.7 ⫾ 1.2a 14.5 ⫾ 1.3a

9.2 ⫾ 1.4b 10.3 ⫾ 2.8b

4.7 ⫾ 0.9c 5.7 ⫾ 1.0c

Means within rows followed by the same letter are not signiÞcantly different at the 5% probability level by ANOVA and the Tukey test. Sample sizes vary as a result of a few mites being discarded because of being stuck in the glue or escaping from the cage.

35), 13% (0.3, 20), and 150% (73, 307) for the H547, MMR, and Orchard Drive sites, respectively. Nonrandom Mating. Based on the criterion used, equal numbers of inseminated (n ⫽ 13) and virgin (n ⫽ 12) females initiated populations (␹20.05, 1 ⫽ 1.261, P ⬎ 0.25). The mean (⫾SD) number of eggs, juveniles, and adults counted in the observation areas after 28 d was the same whether they were produced by virgin or inseminated females: eggsÐinseminated ⫽ 29.3 ⫾ 44.9; virgin ⫽ 12.3 ⫾ 7.5; F ⫽ 1.68; df ⫽ 1, 23; P ⫽ 0.2078; juvenilesÐinseminated ⫽ 10.6 ⫾ 16.1; virgin ⫽ 5.2 ⫾ 3.3; F ⫽ 1.31; df ⫽ 1, 23; P ⫽ 0.2633; adultsÐinseminated ⫽ 10.1 ⫾ 14.0; virgin ⫽ 3.3 ⫾ 2.2; F ⫽ 2.70, df ⫽ 1, 23; P ⫽ 0.1140. Discussion The dispersal behavior of the citrus rust mite appears to be shaped by a combination of abiotic factors. The onset of dispersal each day is probably triggered by the drying of dew. Mites submerged in rain or dew are physically prevented from dispersing by the waterÕs surface tension. This effect may account for the consistent contributions of solar radiation and time and leaf wetness to the one- and two-variable models describing the relationship between abiotic factors and the dispersal pattern, and may partially explain the lack of dispersal at night. Peak dispersal activity occurs from late morning until early afternoon, declining thereafter to relatively low levels during the remaining daylight hours. In contrast to the results reported here, the aerial dispersal of other eriophyid species was correlated with wind velocity alone (Staples and Allington 1956) or a combination of temperature and wind velocity (Nault and Styer 1969). These variables had negligible impact on the dispersal model developed for citrus rust mite, probably due to the different species of mites studied and to the different sampling regimes. Whereas the current study used 2-h sample intervals throughout most of the day and 12-h intervals overnight, Nault and Styer (1969) and Staples and Alling-

April 2001 Table 4.


323

Longevity of a cohort of adult citrus rust mites removed from the host at intervals and exposed to ambient conditions Time of removal from fruit

2 Sept 1999 Mean ⫾ SD longevity (hours) Temp, ⬚C % RH 3 Sept 1999 Mean ⫾ SD longevity (hours) Temp, ⬚C % RH

0700 hours

0900 hours

1100 hours

1300 hours

(n ⫽ 15) 7.6 ⫾ 0.5a 22.8 99.1 (n ⫽ 15) 7.3 ⫾ 0.5a 21.4 100

(n ⫽ 13) 5.8 ⫾ 0.4b 27.4 80.4 (n ⫽ 15) 5.1 ⫾ 0.5b 27.7 63.3

(n ⫽ 15) 4.1 ⫾ 0.3c 30.7 65.5 (n ⫽ 14) 3.7 ⫾ 0.5c 30.8 53.3

(n ⫽ 14) 3.6 ⫾ 0.9c 31.6 56 (n ⫽ 14) 2.6 ⫾ 0.5d 33.6 38.9

Means within rows followed by the same letter are not signiÞcantly different at the 5% probability level by ANOVA and the Tukey test. Sample sizes vary as a result of a few mites being discarded because of being stuck in the glue or escaping from the cage.

ton (1956) related environmental data to 24-h captures of Aculodes dubius and Aceria tulipae (tosichella), respectively. A 24-h sampling regime may or may not reveal the key environmental factors that regulate the dispersal of eriophyid mites, especially if the dispersal of a given species shows a diel periodicity. Frost (1997) discussed the conditions that favor the successful arrival of an aerially dispersing eriophyid mite at a new host, including the spatial distribution of hosts and the duration of survival during dispersal, as inßuenced by the miteÕs buoyancy and ability to avoid desiccation. Adult grain rust mites, Abaracus hystrix (Nalepa), increased their production of wax Þlaments in response to high temperatures (Frost 1997). Compared with unwaxed individuals, waxed mites lived longer at low humidity and showed reduced terminal

Fig. 1. Least squares linear regression of the relationship between the population density of citrus rust mites on fruit and the number of dispersing mites captured in traps. Parameter estimate ⫾ SE; intercept: 5.30 ⫾ 2.37, trap 11.73 ⫾ 1.56, model r2 ⫽ 0.546; F ⫽ 56.58; df ⫽ 1, 47; P ⬍ 0.0001.

velocity in a column of still air, which implies increased buoyancy and potentially greater time aloft. Citrus rust mites do not show any apparent morphological adaptations that could enhance survival during dispersal; however, they may maximize their potential Þtness through adaptive migratory behavior. Measurements of the mortality of mites removed from the host suggest that dispersal during a 2- to 3-h period following the drying of dew should increase survival by reducing the duration of exposure to the higher temperatures and lower humidity which characterize the afternoon hours (Tables 3, 4, and 6). Based on the environmental conditions characteristic of the period of peak dispersal in the morning, the late afternoon and early evening periods would also appear favorable for the dispersal of mites (Table 6). The lack of dispersal from late afternoon onward may be an adaptive response to the brief, but intense rainfall that frequently occurs in the interior of central Florida during the summer months. These storms rarely occur before ⬇1400 hours, but are common in the late afternoon and evening. Citrus rust mites can attach themselves Þrmly to the host plant using their caudal lobes, and are probably not washed off by heavy rain. Whether this attachment behavior occurs in response to changing barometric pressure or humidity associated with the onset of these storms is unknown. J.C.B. (unpublished data) has found that citrus rust mites transferred to a wet surface cease moving immediately and do not struggle or attempt to walk. This behavioral response to wetting may be an adaptation that conserves energy during regular periods of immersion in rain and dew. The aerial dispersal of citrus rust mites from fruit was independent of population density; mites were captured from 49 of 50 fruit sampled, even though the density of mites on nine fruit averaged less than two per lens Þeld. Bergh and McCoy (1997) captured dispersing mites very early in the season, when populations on fruit were nearly undetectable. These results concur with those of previous studies showing that aerial dispersal of eriophyid mites occurs throughout the season and seems to be independent of population density or host plant quality (Sabelis and Bruin 1996). Lawson et al. (1996) reported similar observations for the European red mite, Panonychus ulmi

324 Table 5.


Vol. 30, no. 2

Mean ⴞ SD percentage of female citrus rust mites in samples of mites brushed from fruit and captured while dispersing Brushed from fruit

Filter paper traps

Omnidirectional traps

Date (1999)

Na

% female

n

% female

n

% female

17/9 28/9 30/9 7/10 12/10

125 125 125 125 124

67.2 ⫾ 18.0a 81.6 ⫾ 12.8a 95.2 ⫾ 3.3a 88.0 ⫾ 6.3a 83.9 ⫾ 4.8a

125 125 125 111 115

100 ⫾ 0b 100 ⫾ 0b 100 ⫾ 0b 100 ⫾ 0b 98.2 ⫾ 2.5b

5 7 3 26 20

100 ⫾ 0b 100 ⫾ 0b 100 ⫾ 0b 88.6 ⫾ 18.6ab 100 ⫾ 0b

Means within rows followed by the same letter are not signiÞcantly different at the 5% probability level based on ANOVA and Tukey test analyses on arcsine transformed percentages. a Total number of Phyllocoptruta oleivora in subsamples of 25 mites taken from Þve fruit and Þve Þlter paper traps per day. Sample size from omnidirectional traps represent the total number of P. oleivora captured by Þve traps each day.

(Koch), and speculated that a density-independent proportion of mites is committed to disperse. Sabelis and Bruin (1996) made the same tentative interpretation from data on vagrant species of eriophyids. The dispersal behavior of citrus rust mite is gender-speciÞc, concurring with KrantzÕs (1973) observations on the rust mite Aculus cornutus (Nalepa). The underlying mechanisms that trigger dispersal in some proportion of females in every population remain unknown. Gender-speciÞc dispersal may have contributed to some of the variability not explained by the model describing the relationship between population density on fruit and the number of mites captured in traps. Because the Þlter paper traps capture mites leaving a single fruit, the sex ratio of mites on a fruit would inßuence the proportion of the population available for trapping. On each of 2 d, a single fruit with a high mite population yielded very few dispersers. The percentage of males in samples brushed from individual fruit with high populations varied from 0 to 52%, and Swirski and Amitai (1960) reported that the percentage of male citrus rust mite on leaves and fruit in late summer in Israel could exceed 80%. It is impossible to distinguish male and female citrus rust mites without examining the genitalia, and this lack of information on the sex ratio on fruit from which dispersing mites were trapped may have introduced some variability into the model. The Þtness of dispersing female citrus rust mites depends ultimately on their arrival at suitable host plants and their contribution to or initiation of a population. Relative to eriophyid species colonizing hosts that are patchy in distribution, citrus rust mites in Florida may have a higher probability of arriving at a new host because of citrus monoculture over large and contiguous areas. However, dispersal-related mortality is probably high and may be partially affected by within-tree rust mite distributions. Allen and McCoy (1979) found high rust mite populations in the northand south-bottom quadrants of trees, where temperatures were favorable for development, and the lowest mite densities in the south-top quadrant, where lethal temperatures were recorded. Given this distribution pattern, research on the aerial dissemination of fungal spores from a plant canopy may help our understanding of the dispersal success of citrus rust mite. Aylor (1990) reviewed the dissemination of fungal pathogens by wind. The majority of

fungal spores are often produced in the lower portions of the plant canopy because of favorable conditions there. Lower wind-speed and less turbulence in the lower canopy limit the escape of those spores except during a period around midday, when wind-speed and turbulence near the ground are usually highest. Gradients of disease severity typically decrease rapidly with increasing distance from the source of spores. As suggested by the distribution of citrus rust mites within a tree, the majority of dispersal is from lower portions of the canopy. A potential consequence of this is a shorter distance traveled, relative to mites that leave from higher in the canopy, and a large percentage of dispersing mites settling out of the air onto nonhost plants. This settling of dispersing mites should be examined using ßat traps positioned horizontally at ground level with the trapping surface facing upward. One aspect of the aerial dispersal of citrus rust mite that has received only cursory attention is the longdistance transport of mites. In previous studies (J.C.B., unpublished data), sticky traps were deployed at an elevation of ⬇25 m by attaching them to the frame of two Þre towers in central Florida for two 2- to 3-wk periods in July and August 1996. Citrus rust mites were captured in all eight traps on each tower during each sample period. However, eriophyid mites carried upward from the canopy and transported long distances would likely suffer extremely high mortality as a function of prolonged exposure to the elements, and a greatly reduced probability of being deposited on a suitable host plant. Regardless of the mortality that occurs during dispersal, these data show that some mites were carried from a grove to an adjacent grove downwind and that the percentage of mites arriving at the downwind location depended on the distance between the groves. For groves separated only by a secondary road, which occurs commonly in central Florida, the regression model predicted a greater percentage of mites arriving at the sink grove than had left the source grove. Although this prediction is nonsensical and probably a function of the sampling method, it does indicate the magnitude of movement possible between adjacent groves. Bergh and McCoy (1997) discussed the potential implications of such movement on mite management programs in central Florida, where the majority of citrus is grown in adjacent 2- to 9-ha blocks and where pest management practices

April 2001


325

Table 6. Representative temperature, humidity and wind speed data recorded at Lake Alfred, FL, between the 14th and 16th day of the 4 mo that span the period of peak citrus rust mite population growth in Florida Month

Fig. 2. Number of dispersing citrus rust mites captured in traps deployed between two adjacent citrus groves at three locations. For each of three 7-d trapping intervals at Minute Maid Road and Highway 547, bars from left to right are data from traps next to the source grove, half-way between the source and sink groves, and next to the sink grove, respectively. For the two trapping intervals at Orchid Drive, left and right bars are data from traps next to the source and sink groves, respectively.

may differ widely between groves. Whether the aerial movement of mites adversely affects the success and cost of management programs in some situations is not known, but is worthy of further investigation. Laboratory tests have shown that a combination of oedipal mating and sibmating can result in the founding of a local population by a single virgin or inseminated female citrus rust mite. It is probable that most female eriophyids are inseminated before dispersal (OldÞeld and Newell 1973), in which case sibmating among F1 progeny would be most relevant to the initiation of a local population. A dispersing female mite might also arrive in the midst of an established population on a new host. Alternatively, when populations are low early in the season or following a pesticide application, there could be few or no conspeciÞcs in her vicinity, and these divergent scenarios would have very different outcomes with respect to the genetic structure of the ensuing population. The

0900 Ð1200 hours

1300 Ð1600 hours

1700 Ð2000 hours

June July Aug Sept

Mean temp (⬚C) and range 30.3 (28.1Ð32.8) 32.0 (26.4Ð35.4) 27.1 (23.8Ð33.7) 30.8 (28.8Ð33.5) 32.0 (26.9Ð34.9) 26.6 (24.2Ð29.1) 31.6 (29.4Ð33.6) 34.2 (31.6Ð35.8) 27.9 (22.4Ð32.4) 28.8 (25.1Ð32.2) 30.6 (28.2Ð33.6) 27.5 (24.5Ð32.9)

June July Aug Sept

Mean relative humidity (%) and range 70.5 (57.1Ð86.5) 59.1 (42.5Ð86.6) 81.2 (45.7Ð100) 65.7 (50.4Ð81.2) 57.8 (45.7Ð76.8) 85.1 (72.3Ð98.5) 71.6 (60.2Ð84.5) 56.5 (49.4Ð69.8) 81.0 (59.4Ð100) 76.9 (59.6Ð90.8) 65.4 (43.7Ð89.3) 76.1 (44.6Ð100)

June July Aug Sept

Mean wind speed (kph) and range 5.9 (4.3Ð7.2) 7.1 (2.6Ð18.5) 5.8 (1.1Ð14.8) 6.8 (3.9Ð10.6) 9.3 (2.9Ð14.0) 7.9 (3.9Ð10.9) 5.8 (3.2Ð9.0) 6.3 (3.9Ð9.9) 9.7 (4.7Ð20.1) 15.6 (4.3Ð25.4) 15.8 (9.0Ð23.0) 10.8 (6.0Ð19.0)

founding of local populations through nonrandom, sibmating could have important implications for the development of acaricide resistance in citrus rust mite populations. Population structure is inßuenced by the size of breeding units, the number of founding females, the sex ratio, immigration and emigration, and nonrandom mating (Roush and Daly 1990). Resistance to dicofol (Omoto et al. 1994) and shifts in susceptibility to abamectin (Bergh et al. 1999) have been reported in Þeld populations of citrus rust mite, and the genetic consequences of these reproductive strategies for the evolution of resistance in citrus rust mite populations merit investigation. The traps used in these studies have proven useful for addressing questions about the aerial dispersal of citrus rust mite. Bergh and McCoy (1997) used unidirectional traps to monitor the dispersal of citrus rust mite through time and reported strong correlations between trap-catch and population density on fruit. They discussed the possibility of using dispersal traps to monitor mite populations for management purposes. This objective was pursued over several years by examining the effects of grove location and age, fruit variety, trap placement, sample interval, and trap type (i.e., unidirectional and omnidirectional) on the relationship between trap-catch and population density. Although some data sets yielded strong correlation coefÞcients, the variability among locations and years was great and was not consistently improved by any approach (J.C.B., unpublished data). Although the traps are useful research tools, they are not suitable for use in a monitoring program upon which pest management decisions are based. Further research on the biology of dispersing citrus rust mites could use the advantages associated with the Þlter paper trap. Whereas unidirectional and omnidirectional traps kill the mites they capture and yield relatively small samples, the Þlter paper trap can provide large numbers of live mites that could be used to address questions about population dynamics and

326


management. For example, comparing the fecundity and longevity of dispersing females with females of known age would establish the age at which females disperse and their potential reproductive capacity. Bergh and McCoy (1997) also captured dispersing citrus rust mites infected with the entomopathogenic fungus, Hirsutella thompsonii. Filter paper traps could be deployed under fruit on which a fungal epizootic was occurring; and by culturing captured mites on media, the effect of a fungal epizootic on the dispersal behavior of populations and individuals could be measured. Acknowledgments Thanks to J. Waldow, T. Wilkinson, S. Blackburn, and S. Sloan for technical assistance; J. Harrison and J. Fan for statistical help; R. Kerr and Holly Hill Groves for their generous cooperation; and the Florida Citrus Production Research Advisory Council for partial Þnancial support. The general assistance of the faculty and staff at the Lake Alfred Citrus Research and Education Center over the past 5 yr is greatly appreciated. Florida Agricultural Experiment Station Journal Series No. R07577.

References Cited Allen, J. C., and C. W. McCoy. 1979. The thermal environment of the citrus rust mite. Agric. Meteorol. 20: 411Ð425. Allen, J. C., Y. Yang, and J. L. Knapp. 1995. Temperature effects on development and fecundity of the citrus rust mite (Acari: Eriophyidae). Environ. Entomol. 24: 996Ð 1004. Aylor, D. E. 1990. The role of intermittent wind in the dispersal of fungal pathogens. Annu. Rev. Phytopathol. 28: 73Ð92. Bergh, J. C., and C. W. McCoy. 1997. Aerial dipsersal of citrus rust mite (Acari: Eriophyidae) from Florida citrus groves. Environ. Entomol. 26: 256Ð264. Bergh, J. C., D. Rugg, R. K. Jansson, C. W. McCoy, and J. L. Robertson. 1999. Monitoring the susceptibility of citrus rust mite (Acari: Eriophyidae) populations to abamectin. J. Econ. Entomol. 92: 781Ð787. Cromroy, H. L. 1979. Eriophyoidea in biological control of weeds, pp. 473Ð475. In J. G. Rodriguez [ed.], Recent advances in acarology, vol. 1 Academic, New York. Davis, R. 1964. Autecological studies of Rhynacus breitlowi Davis (Acarina: Eriophyidae). Fla. Entomol. 47: 113Ð121. Easterbrook, M. A. 1978. The life-history and bionomic of Epitrimerus pyri (Acari: Eriophyidae) on pear. Ann. Appl. Biol. 88: 13Ð22. Frost, W. E. 1997. Polyphenic wax production in Abaracus hystrix (Acari: Eriophyidae), and implications for migratory Þtness. Physiol. Entomol. 22: 37Ð46. Horsfall, J. G., and R. W. Barratt. 1945. An improved grading system for measuring plant disease. Phytopathology 35: 655. Jeppson, L. R., H. H. Keifer, and E. W. Baker. 1975. Mites injurious to economic plants. University of California Press, Berkeley. Krantz, G. W. 1973. Observations on the morphology and behavior of the Þlbert rust mite, Aculus cornutus (Prostig-

Vol. 30, no. 2

mata: Eriophyoidea) in Oregon. Ann. Entomol. Soc. Am. 66: 709 Ð717. Lawson, D. S., J. P. Nyrop, and T. J. Dennehy. 1996. Aerial dispersal of European red mites (Acari: Tetranychidae) in commercial apple orchards. Exp. Appl. Acarol. 20: 193Ð202. Lindquist, E. E. 1996. External anatomy and notation of structures, pp. 3Ð31. In E. E. Lindquist, M.W. Sabelis, and J. Bruin [eds.], World crop pests: eriophyoid mitesÑtheir biology, natural enemies and control. Elsevier, Amsterdam. Nault, L. R., and W. E. Styer. 1969. The dispersal of Aceria tulipae and three other grass-infesting eriophyid mites in Ohio. Ann. Entomol. Soc. Am. 62: 1446 Ð1455. Oldfield, G. N., and I. M. Newell. 1973. The role of the spermatophore in the reproductive biology of Aculus cornutus (Acarina: Eriophyidae). Ann. Entomol. Soc. Am. 66: 160 Ð163. Oldfield, G. N., R. F. Hobza, and N. S. Wilson. 1970. Discovery and characterization of spermatophores in the Eriophyoidea (Acari). Ann. Entomol. Soc. Am. 62: 269 Ð 277. Omoto, C., T. J. Dennehy, C. W. McCoy, S. E. Crane, and J. W. Long. 1994. Detection and characterization of the inter-population variation of citrus rust mite (Acari: Eriophyidae) resistance to dicofol in Florida citrus. J. Econ. Entomol. 87: 566 Ð572. Reed, D. K., A. K. Burgitt, Jr., and C. R. Crittenden. 1964. Laboratory methods for rearing rust mites (Phyllocoptruta oleivora and Aculus pelekassi) on citrus. J. Econ. Entomol. 75: 130 Ð133. Rogers, J. S., C. W. McCoy, and M. M. Manners. 1994. Standardized visual comparison keys for rapid estimations of citrus rust mite (Acari: Eriophyidae) populations. J. Econ. Entomol. 87: 1507Ð1512. Roush, R. T., and J. C. Daly. 1990. The role of population genetics in resistance research and management, pp. 97Ð 152. In R. T. Roush and B. E. Tabashnik [eds.], Pesticide resistance in arthropods. Chapman & Hall, New York. Sabelis, M. W., and J. Bruin. 1996. Evolutionary ecology: life history patterns, food plant choice and dispersal, pp. 329 Ð 366. In E. E. Lindquist, M. W. Sabelis, and J. Bruin. [eds.], World crop pests: eriophyoid mitesÑtheir biology, natural enemies and control. Elsevier, Amsterdam. SAS Institute. 1988. SAS/STAT userÕs guide, release 6.03 ed. SAS Institute, Cary, NC. Slykhuis, J. T. 1955. Aceria tulipae Keifer (Acarina: Eriophyidae) in relation to the spread of wheat streak mosaic. Phytopathology 45: 116 Ð128. Somsen, H. W. 1966. Development of migratory form of wheat curl mite. J. Econ. Entomol. 59: 1283Ð1284. Staples, R., and W. B. Allington. 1956. Streak mosaic of wheat in Nebraska and its control. Univ. Nebr. Agric. Exp. Stn. Res. Bull. 178. Swirski, E., and S. Amitai. 1959. Contribution to the biology of the citrus rust mite (Phyllocoptruta oleivora Ashm.) C. Oviposition and longevity of males and females. Ktavim 9: 281Ð285. Swirski, E., and S. Amitai. 1960. Sex ratio of the citrus rust mite (Phyllocoptruta oleivora ASHM.) in the citrus grove. Ktavim 10: 225Ð226. Received for publication 28 April 2000; accepted 16 November 2000.