Shipments of 50,000 commercially reared Aphytis melinus DeBach were obtained from each of the five insectaries that produce and sell this parasitoid to citrus growers in California for control of California red scale, Aonidiella aurantii (Maskell). Shipments were received from each insectary every 2 mo over a period of a year to assess variability in quality through time and between insectaries. As indices of quality, we assessed the percentage of live parasitoids (both sexes) 1, 3, 7, 14, and 28 d after receipt of the shipment, shipment sex ratio, and the size of female wasps. We found a fair amount of variation in the percentage of A. melinus that were alive on different sampling days. Despite the fact that all insectaries rear A. melinus in temperature controlled rooms and all of our studies were done at 22°C, wasp mortality occurred more quickly in the colder months of November, January through February, and March. Similar trends were observed with sex ratios; many of the insectaries had male-biased sex ratios in the colder months, especially January through February. Wasp size varied significantly for each of the insectaries throughout the year, with the summer months of July through August yielding significantly smaller females than other months. Collectively our results have important implications for biological control on citrus in California.
California red scale, Aonidiella aurantii (Maskell), first appeared in California in the 1870s (Comstock 1880) and historically has been considered the number one arthropod pest of California citrus (Ebeling 1959, Clausen 1978a, Flint et al. 1991, Morse et al. 2007). At least 35 attempts were made to import predators and parasitoids for red scale biological control between 1889 and the 1950s (reviewed in Clausen 1978b). In 1961, Harold Compere (1961) concluded that biological control agents probably would never be able to control this pest in citrus. However, with the introduction of Aphytis melinus DeBach (Hymenoptera: Aphelinidae) into California in 1956–1957 from Pakistan and India, by the early 1960s excellent biological of red scale was achieved in southern California. Although Aphytis lingnanensis Compere already was established and controlling red scale in California coastal areas, A. melinus proved to be a much more effective biological control agent in the inland coastal valleys (Luck and Podoler 1985). Insectaries began rearing and releasing A. melinus, adapting methods developed previously for A. lingnanensis (DeBach and White 1960), and by 1962, A. melinus had become established as an effective biological control agent, adding to the biological control exerted by other parasitoids and predators (Luck 2006) to such a degree that chemical control was rarely needed in southern California (Flint et al. 1991, Morse et al. 2007).
Nearly 70% of California's citrus acreage is now located in the San Joaquin Valley (SJV) of California and in contrast to southern California, it was previously believed that biological control of red scale was ineffective in the SJV because of extreme temperatures in the summer and winter (Riehl et al. 1980, Morse et al. 2007). This perspective changed after a biologically-based integrated pest management (IPM) research and demonstration project in Tulare Co. of the SJV run from 1987 to 1991 (Haney et al. 1992, Forster and Luck 1997, Luck et al. 1997). The key new component of this IPM program was augmentative release of 247,100 insectary-reared Aphytis melinus/ha/yr (16 releases, one every 2 wk mid-February to mid-September) to control red scale, coupled with the management of other arthropod pests by using economic treatment thresholds and use of selective pesticides that allowed A. melinus to persist.
After the success of the aforementioned demonstration project, the biologically-based IPM program for red scale became quite popular in the SJV, especially as this insect began to exhibit resistance to available organophosphate and carbamate insecticides (Grafton-Cardwell 1994 ; Grafton-Cardwell et al. 1997, 2001 ; Grafton-Cardwell and Stuart-Leslie 1998). However, in 1998, an emergency use exemption was granted, allowing the use of pyriproxyfen for red scale control. Just before the introduction of pyriproxyfen, the use of A. melinus for red scale control peaked with augmentative releases on ≈30% of SJV groves (Rill et al. 2008). Because of the high efficacy of pyriproxyfen, <10% of SJV groves still use augmentative release of wasp parasitoids (Rill et al. 2008).
Despite the recent success of pyriproxyfen, relying solely on chemical control of red scale has long-term negative ramifications. Pesticide resistance is inevitable in chemically-based control programs that involve limited classes of chemistry and frequent applications, requiring growers to increase the number of chemical applications and consequently increasing the cost of chemical control. However, biological control is more sustainable and less costly in the long run in addition to being safer for consumers and the environment (Haney et al. 1992, Van Lenteren 2008).
Until the mid-1980s, quality control of natural enemies had not even been considered. Today there are quality control guidelines for some natural enemies, but testing is rare and there are no criteria specifically for A. melinus (Van Lenteren 2008). The methods commercial insectaries employ to produce A. melinus were developed in the late 1950s for the related A. lingnanensis (DeBach and White 1960), have changed little since then, and there has been little research or evaluation conducted with A. melinus, largely because the present production system appears to be working well. It is well known that providing large host scales is critical to producing a female-biased sex ratio as well as large and fit Aphytis (Luck et al. 1982, Luck and Podoler 1985, Opp and Luck 1986, Walde et al. 1989, Murdoch et al. 1992). Parasitoid size and longevity can by improved by manipulating host size and diet regimes (Collier 1995, Heimpel and Rosenheim 1995, Heimpel et al. 1997, Luck and Forster 2003).
In the long-term, we believe citrus growers in California would be wise to reduce broad-spectrum pesticide use and maximize use of biologically-based IPM that emphasizes augmentative releases of A. melinus for red scale control (Morse et al. 2007). The research reported herein was intended as a preliminary survey for California insectaries producing A. melinus to assist in determining to what degree wasp quality varies between insectary and over the year. We estimated the quality (and presumably the ability of the wasp to control scale populations) of A. melinus by measuring the sex ratio, size, and percentage of individuals still alive on a schedule of sampling days after receipt of the shipment. Because females are the effective sex in biological control, sex ratio differences can greatly influence the efficacy of biological control efforts (Heimpel and Lundgren 2000). Size also was used as a fitness proxy because it is often positively correlated with the field performance of adult female parasitoids (Kazmer and Luck 1995, Bennett and Hoffmann 1998, Roitberg et al. 2001, Sagarra et al. 2001).
Materials and Methods
Aphytis melinus were obtained from all five of the California insectaries that were in production at the time the study was run, although a number of other “secondary” suppliers buy and resell this wasp by using stock from these primary producers (Associates Insectary, Santa Paula, CA; Foothill Agricultural Research, Corona, CA; Good Bugs Insectary, Exter, CA; Mulholland Citrus, Orange Cove, CA; and Sespe Creek Insectary, Lindsay, CA; insectaries were randomly coded A-E to maintain anonymity). The rationale for the study was explained in advance to each insectary and results were discussed in a group meeting once the study was done. Each insectary overnight shipped ≈50,000 adult wasps every 2 mo for 1 yr (six shipments per insectary). Aphytis were sent to us in a paper cup with a plastic lid that had pin holes for ventilation with honey applied to the interior surface of the lid as a food source. The entire cup was wrapped in wet newspaper and shipped chilled on ice packets. Shipment methods were exactly the same as those used to ship to growers. Upon arrival of each shipment at the University of California, Riverside (UC Riverside), a hole was punched in the cup lid and A. melinus were anesthetized with CO2 for 30 s (see the next section on CO2 anesthesia). While anesthetized, ≈150 A. melinus were spooned into each of fifteen 50-mm-diameter plastic petri dishes (Fisherbrand, Pittsburgh, PA). Petri dishes then were placed gently into individual 0.47-liter jars (wide mouth pint size glass Kerr Mason jars, Jarden Corporation, Rye, NY). Jars were covered with a fine mesh cloth that was streaked with organic U.S. Grade A honey (Natural Directions Foods, Inc., Tigard, OR). After transfer, live wasps were allowed to move out of the cups and into the jars for 20 min. Individuals that did not move from the dish to the jar were presumed dead and were discarded. This procedure ensured that we started our study only with live individuals at the time the shipment arrived at UC Riverside. Food was provided to each jar every 3 d by applying thin streaks of honey through the mesh cloth cover of the jar. All studies were carried out in a laboratory kept at a temperature of 22 ± 3°C.
Three jars from each shipment were selected randomly to be destructively sampled 1, 3, 7, 14, and 28 d after the arrival of each shipment. For destructive sampling, A. melinus were anesthetized with CO2 and then killed by saturation with 70% ethanol. These animals then were collected on 70-mm-diameter filter paper (Whatman, Piscataway, NJ) by using a 186-ml Buchner funnel (Fischerbrand, Pittsburgh, PA) and immediately were counted and sexed. When processing jars on sampling days, individuals that were desiccated were counted as having died sometime between the setup and destruction of the jar and those that were not desiccated were counted as alive. The eyes and abdomen of A. melinus begin to desiccate within 4 h of being killed with ethanol, allowing for accurate determination of whether an individual was living or dead at the time of destructive sampling (C.J.V., unpublished data). This method was used to assess the percent alive and dead (those dying between when the shipment arrived and when the destructive sampling was done) as well as sex ratio measurements. Five live females and five dead females (when available) then were selected randomly from each of these samples to measure adult size. In total, 150 size measurements (75 live, 75 dead) per insectary were performed every 2 mo. Size was calculated in 0.01-mm increments with an ocular micrometer by measuring the hind tibia length (HTL). HTL is used here as a proxy for overall adult body size because it can be linearly measured and is resistant to desiccation (Nicol and Mackauer 1999).
There is conflicting evidence on the effects of CO2 anesthesia on insects (Nicolas and Sillans 1989). Some studies report adverse effects ranging from a decrease in male fertility (Champion de Crespigny and Wedell 2008), reduced longevity (Perron et al. 1972), and fecundity (Dawson 1995), whereas others report no effect at all (Parkman and Pienkowski 1991, Perrotti and Maroli 1993). We therefore performed our own experiment to test the effect of CO2 exposure on A. melinus longevity before collecting data for this study. We exposed adults to a constant pressure of 100% CO2 for 0 s (control), 1.5, 3, 5, 10, and 30 min. We found no significant difference in the percentage of A. melinus alive 3 d after exposure (mean alive for the control was 45.4% and the means for treatments ranged from 38.4% to 50.5%; F=0.5904, df=5, P > 0.706) and therefore assumed an anesthesia period of 30 s would not adversely affect these insects.
All statistical analyses were performed using SAS 9.2 software for Windows (SAS Institute 2008). To determine the effect of the three key factors, 1) shipment date, 2) insectary, and 3) sample day (day of destructive sampling post receipt of the shipment) on size data, we performed a three-way analysis of variance (ANOVA). A logistic regression model (Hosmer and Lemeshow 2000) was used to determine the effects of the explanatory variables (shipment date, insectary, sample day) and their interactions on the percentage measurements that required arc-sine square-root transformation (sex ratio and percentage alive). Data were converted to a binary response (i.e., females were coded as one and males as zero, live individuals were coded as one and dead as zero) and tested with a Wald χ2 test. Where significant treatment effects were detected, the REGWQ option within PROC GLM was used to make posthoc pairwise comparisons of means.
Both shipment date (time of the year) (for females, Wald χ2=0.0262, df=5, P=1.0; for males, Wald χ2=0.0088, df=5, P=1.0) and insectary source (for females, Wald χ2=0.0137, df=4, P=1.0; for males, Wald χ2=0.0048, df=4, P=1.0) failed to significantly affect the percentage of either sex that remained alive over the five destructive sampling days. As one might expect, an increase in sample day (period of time after receipt of the shipment) resulted in a significant reduction in the percentage of live females (Wald χ2=912.3, df=4, P < 0.0001). All of the interactions between these three main effects had a significant effect on the percentage of living females (shipment*insectary, Wald χ2=231.5, df=20, P < 0.0001; shipment*sample day, Wald χ2=578.8, df=20, P < 0.0001; insectary*sample day, Wald χ2=216.3, df=16, P < 0.0001; and shipment*insectary*sample day, Wald χ2=913.9, df=80, P < 0.0001). Female wasps from insectaries B and D showed higher survival rates than at least one other insectary on three of five sample days (Table 1). Females from insectary E had a lower survival rate than one insectary on days 1 and 28 and a lower rate than all four insectaries on days 3 and 14. Figure 1 shows the variability in female survival rate across the five insectaries and six shipments.
Means followed by the same letter within a row (i.e. for a specific sampling day) are not significantly different (as determined using posthoc pairwise comparisons of means using the REGWG option within PROC GLM).
Wasp Sex Ratio.
Sampling day was not included as a main effect in the logistic regression model for sex ratio because the jar replicates were assigned randomly to a sampling day and therefore should not differ in their sex ratio. Both of the main effects, shipment date (Wald χ2=3191.9, df=5, P < 0.0001) and insectary (Wald χ2=1246.5, df=4, P < 0.0001), and their interaction (Wald χ2=4643.4, df=20, P < 0.0001) showed significant impacts on sex ratio. Table 2 shows the mean proportion females and the differences in sex ratios between shipments and insectaries can be seen in Fig. 2 . All of the insectaries had male-biased sex ratios during shipment 4 (January through February). Although insectary B shipped a higher percentage of females than at least one insectary on four of six samples, differences between insectaries were not significant when averaged across all shipments.
Means followed by the same letter within a row are not significantly different (as determined using posthoc pairwise comparisons of means using the REGWG option within PROC GLM). Mean sex ratios are the average from all six destructive sampling periods following receipt of each shipment.
Female Wasp Size.
All of the main effects significantly affected the size of both live (shipment date, F=12.54, df=5, P < 0.0001; insectary, F=36.32, df=4, P < 0.0001; and sample day, F=7.59, df=4, P < 0.0001) and dead females (shipment date, F=15.02, df=5, P < 0.0001; insectary, F=45.27, df=4, P < 0.0001; and sample day, F=11.88, df=4, P < 0.0001) as did the interaction between shipment date and insectary for live (F=3.99, df=20, P < 0.0001) and dead females (F=7.34, df=20, P < 0.0001). Table 3 shows the mean live female sizes after combining data from the first two sampling days. These data were chosen because they are the most biologically relevant to growers and the pool of live A. melinus individuals was large enough from which to choose truly random samples. Females from insectaries B and E were significantly larger than females from at least one insectary during six and five of the sampled shipments, respectively. When averaged over all shipments, samples from insectary B had the largest females (0.257 mm), whereas those from insectary C had the smallest (0.232 mm).
Means followed by the same letter within a row are not significantly different (as determined using posthoc pairwise comparisons of means using the REGWG option within PROC GLM).
Our results suggest that all five of the insectaries display reduced quality and fitness parameters of A. melinus during certain times of the year. Decreased longevity and male-biased sex ratios were evident in the cooler months of November–March. Smaller females were produced in the warmer months of July–August. Variation in longevity, sex ratio, and size persisted despite a relatively constant production environment of ≈60% RH and 25°C at all of the insectaries.
One caveat of our study is that we did not evaluate the percentage of A. melinus that were alive upon arrival of the shipments, nor the number of wasps that actually were sent. We did not do this because the insectaries knew where they were sending the shipments and thus, we were concerned these measurements might not be productive. Because we began our experiments with only live wasps, it is unknown what percentage of the remaining wasps were alive or dead. Another limitation to this study is the lack of field performance measures, which may or may not correlate well with laboratory results. Flight tests might be relevant because dispersal ability can deteriorate under mass rearing conditions as well as the preparation and shipment of these relatively fragile wasps. Identifying an easily measured parameter in the laboratory, such as fluctuating asymmetry, that predicts field success would be ideal for producers and growers (Leary and Allendorf 1989, Parsons 1992, Van Lenteren 2008).
Although all of the five evaluated insectaries adapted rearing methods from the same protocol (DeBach and White 1960, i.e., using oleander scale, Aspidiotus nerii Bouché, grown on squash plants), they differ in some ways that appear to affect production and fitness of A. melinus. For example, some of the facilities maintain A. melinus adults in a cup with honey for up to 3 d before shipping them out to growers. This practice certainly could affect wasp longevity. Also, some insectaries will sell females that have been used previously to parasitize oleander scale to produce the next generation of A. melinus. This practice undoubtedly affects the fecundity and quite possibly the life expectancy of these synovigenic females that growers purchase.
It is often suggested that wild genetic stock be infused into cultures of biological control agents to avoid loss of genetic variation under the benign conditions of domestication (Hopper et al. 1993, Hufbauer and Roderick 2005). Each of the insectaries evaluated varies in their implementation of this practice with some insectaries not attempting to bring in new genetic material at all, some integrating field collections into the insectary populations, and some completely replacing their populations each fall with new collections. Each of these methods brings with it potential benefits and detriments. Completely replacing the insectary population on an annual basis could reduce “lab” adaptation problems associated with inbreeding; such effects have been documented to negatively affect the fitness of Drosophila melanogaster L. (Spielman and Frankham 1992, Woodworth et al. 2002). However, it is important to know what one is replacing the lab population with and whether it is actually more “fit”. Adaptation to captivity actually increases with higher initial genetic diversity and can occur quite rapidly (Woodworth et al. 2002, Nunney 2003). Similarly, mixing new genetic material with an established colony could increase genetic diversity, potentially leading to hybrid vigor (a.k.a. outbreeding enhancement or heterosis) and greater efficacy of biological control (Hopper et al. 1993 ; Margan et al. 1998 ; Nunney 2002, 2003, 2006). Conversely, if the populations are genetically distinct, the combination could result in outbreeding depression in which the offspring actually have lower fitness.
Introducing new genetic stock into an established colony generally is recommended because of the ramifications of inbreeding and lab adaptation; however, not doing so obviates the potential problems listed above that are associated with using field populations. In the post study meeting, several of the insectaries suggested that studies might be run to examine how often to recollect from the field and what might be done to maintain and produce the best A. melinus for field release. Several studies have shown that biological control agents suffer from bottlenecks in population size based on neutral genetic variation (Baker et al. 2003, Hufbauer et al. 2004, Lloyd et al. 2005). Unfortunately, we can only speculate on these impacts at this time, as it is not known how genetic diversity correlates with the field success of insect biological control agents. We hypothesize that the wide variation in insectary methods for introducing new genetic stock into established captive populations could have fitness consequences for the A. melinus that growers receive and release into the field.
Even though our study clearly shows that certain times of the year are of concern to insectaries producing A. melinus, much research needs to be done to tease apart the reasons for these production problems. One possibility, also of interest to insectary producers, would be to measure the impact scale size on A. melinus size and fitness in the winter versus summer months. It is known that increasing temperature reduces the size of second- and third-instar red scale (Yu and Luck 1988) and that A. melinus will lay more males than females in small scales. In addition, females emerging from small scales are themselves smaller and less fit (Luck et al. 1982). Another consideration is the quality of the squash used as the host plant. Determining which squash qualities (age, color, volatiles, etc.) produce the most and largest oleander scales and using those plants could result in more and larger A. melinus females.
We thank the five insectaries providing A. melinus to us for their cooperation and participation in this study, Lisa Forster and Jan Hare from the laboratory of Robert F. Luck for sharing their expertise, and Sakshi Aggarwal and Corinne Stouthamer for assisting in data collection. Karen Xu, Associate Director of the UCR Statistical Collaboratory, provided statistical assistance. We also thank Drs. Richard Stouthamer and Richard Redak for reviewing earlier drafts of this manuscript. Funding was provided in part by grants from the University of California Statewide Integrated Pest Management Program and the California Citrus Research Board. This research fulfilled part of the dissertation requirements for C.J.V.