This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy

Biological Control 55 (2010) 63–71

Contents lists available at ScienceDirect

Biological Control

journal homepage: www.elsevier.com/locate/ybcon

Laboratory evaluation of aggregation, direct mutual interference, and functional response characteristics of Pseudacteon tricuspis Borgmeier (Diptera: Phoridae)

Donald C. Henne *, Seth J. Johnson

Department of Entomology, 404 Life Sciences Building, Louisiana State University Agricultural Center, Baton Rouge, LA 70803, USA article info abstract

Article history: Studies of parasitoid behavior yield information on behavioral strategies, and are crucial tests of theoret- Received 19 November 2009 ical models. The studies reported here provided insights into behavioral and functional responses of the Accepted 2 July 2010 red imported fire ant (Solenopsis invicta Buren) parasitoid, Pseudacteon tricuspis (Borgmeier) that were Available online 8 July 2010 previously unknown. Laboratory studies were performed to quantify aggregative responses of P. tricuspis adults to variable host ant densities, determine effect of direct mutual interference between multiple ovi- Keywords: positing P. tricuspis females confined with host S. invicta, elucidate the effect of confining one or two addi- Parasitoid behavior tional males with premated females on progeny sex ratios, and determine the form of the functional Oviposition response of individual ovipositing P. tricuspis. Consistent with theory and field observations, P. tricuspis Searching efficiency tended to aggregate at host patches containing greater numbers of ants. Some evidence of direct mutual interference was found, as per capita oviposition success declined when more than one female was con- fined and weak, but significant, reductions in searching efficiency were found when more than two P. tric- uspis females are simultaneously ovipositing. The presence of one or two males did not appear to affect ovipositional efficiency of solitary premated females, but normally male–biased sex ratios of progeny trended toward a 1:1 ratio when the number of males was increased from zero to two. None of the linear parameters in the logistic models were significantly different from zero suggesting that P. tricuspis had constant, type 1, attack rates regardless of host density, at least under our laboratory experimental design. Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction spatially heterogeneous (Hassell and May, 1973; Chesson and Murdoch, 1986; Godfray and Pacala, 1992). Small and/or sparsely The study of host–parasitoid interactions has produced a wealth distributed host populations escape parasitism spatially and/or of theory. Ever since the development of simple theoretical models temporally in refugia because they are at low risk to parasitism. by Thompson (1924) and Nicholson (1933), a proliferation of re- Conversely, in a continuous time framework density-dependent search has shown that many interacting factors determine how host mortality theoretically destabilizes the interaction (Murdoch many hosts a female parasitoid can successfully parasitize. These and Stewart-Oaten, 1989). However, other factors are also impor- factors include host density, parasitoid density and the spatial dis- tant when parasitoids aggregate that can stabilize host–parasitoid tribution and density of hosts (Hassell and May, 1973; Beddington, interactions. 1975; Cook and Hubbard, 1977). The study of insect pests and their Hassell and Varley (1969) and Hassell and May (1973) recog- biological control have benefited from these theoretical insights, as nized the importance of behavioral interactions between multiple there is considerable interest in establishing the mechanisms by searching conspecifics that encounter one another, also known as which parasitoids control host densities (Stiling, 1987). direct mutual interference. Multiple simultaneously ovipositing fe- One prediction of optimal foraging theory is that parasitoids males may engage in aggressive interactions with conspecifics, should aggregate in higher density host patches in a density- resulting in delayed searching and more time wasted (Visser and dependent way to achieve maximal oviposition rates (Charnov, Driessen, 1991; Visser et al., 1999; Hassell, 2000), thereby leading 1976; Cook and Hubbard, 1977). This has long been suggested as to declining rates of host parasitism as parasitoid density increases an important stabilizing factor allowing for the persistence of dis- (Free et al., 1977). These interactions present unique problems for crete time host–parasitoid interactions, because parasitism risk is individual parasitoids when faced with optimal foraging decisions (Maynard Smith, 1974), such as maximizing host parasitism rates. The resulting contribution of these interactions, if sufficiently * Corresponding author. Present address: Texas AgriLife Research, 2301 Exper- iment Station Road, Bushland, TX 79012, USA. Fax: +1 806 534 5829. strong, can lead to the long-term stability of host–parasitoid inter- E-mail address: [email protected] (D.C. Henne). actions (Hassell, 2000).

1049-9644/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.biocontrol.2010.07.001 Author's personal copy

64 D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71

Studies of insect predation rates at variable host densities led to (3) elucidate the effect of confining 1 or 2 additional males with al- the derivation of the well-known type I, II and III functional response ready mated females on progeny production and sex ratios, and (4) curves (Holling, 1966). Solomon (1949) defined the functional re- determine the form of the functional response of individual ovipos- sponse as the density-dependent rate of attack of a single natural en- iting P. tricuspis to varying host densities. emy to changes in the number of hosts available. In other words, the functional response describes the relationship between the per capi- 2. Materials and methods ta predation (parasitization) rate of a predator (parasitoid) and prey density (Holling, 1959, 1961, 1966), and is a fundamental basis of all Monogyne S. invicta colonies were collected at the Louisiana Agri- 0 trophic (consumer–victim) interactions (Mills and Lacan, 2004). The cultural Experiment Station in St. Gabriel, Louisiana (30° 16 N, 91° 0 three types of functional responses were derived according to the 05 W). As of 2006, expanding populations of P. tricuspis in Louisiana relative shape of the curve. The type I functional response character- had not yet reached this location. Colonies were separated from soil izes arthropod predators (and parasitoids) that search for hosts ran- in the laboratory by the drip flotation method (Banks et al., 1981). domly in a patch and attack at an increasingly linear rate to a Ants from each colony were then sieved to yield host ants that were maximum level, at which point attack rates become independent within the preferred size class for P. tricuspis females (approximately of increasing prey density (a combination of density-dependent 1 mm head width (see Morrison et al., 1997)). and density-independent responses (Chong and Oetting, 2006; Para- Laboratory trials were performed using a large enclosed Plexi- Ò julee et al., 2006)). The type II functional response, or ‘disk’ equation, glas cage (120 cm 60 cm 60 cm) that was illuminated by an describes the predation rate as a non-linear function of prey density. overhead flicker-free fluorescent lamp (Hi-Lume Electronic Fluo- As host density increases, the number of hosts that can be attacked in rescent Dimmer Ballast, 0.70 amps, serial #4LZ22, by Lutron Corp., a fixed period of time hyperbolically reaches an asymptote, at which Coopersburg, PA) and heated by a 75 W infrared lamp. Plaster point the predator is spending all its time handling prey (Holling, blocks saturated with water were placed on the middle and cor- 1961; Parajulee et al., 2006). However, as host density increases ners of the cage floor to provide humidity. Trials were conducted the proportion of hosts parasitized by a type II parasitoid decreases when temperatures inside the cage were approximately 26–28 °C exponentially (inverse density dependence) (Chong and Oetting, and had 80–90% RH. At least 200–300 newly emerged P. tricuspis 2006; Parajulee et al., 2006). The type III functional response applies were released inside the cage prior to the trials. To minimize var- when the number of prey killed sigmoidally reaches an asymptote, iance in performance of P. tricuspis, only flies less than 1 day-old where prey killed increases in proportion up to an inflection point were used. While many S. invicta colonies were used, to reduce var- and then decreases in proportion (Parajulee et al., 2006). Therefore, iation individual trials used ants from the same colony. Except functional responses are critical to descriptions of predation and where otherwise indicated, all statistical analyses were conducted Ò parasitism (Hassell, 2000), and can also be useful for parasitoid con- using Prism versions 4.03 and 5.02 (GraphPad Software, Inc., San servation (Parajulee et al., 2006). Diego, CA). All statistical analyses (described below) were con- Beginning in the late 1990’s, several species of parasitoids in the ducted at a significance level of a = 0.05. genus Pseudacteon Coquillet (Diptera: Phoridae), collectively re- ferred to as ‘decapitating flies,’ have been introduced to the United 2.1. Aggregative responses of P. tricuspis States as biological control agents of the red imported fire ant, Solen- opsis invicta Buren (Hymenoptera: Formicidae) (Henne et al., 2007). Laboratory experiments were conducted to examine the relation- Parasitic phorid flies strongly mediate interspecific competitive ship between host ant density and numbers of attacking P. tricuspis. interactions among ant species (Feener, 1981; Feener and Brown, Ant densities used were typical of densities found at hot dog bait sta- 1992; Folgarait and Gilbert, 1999; Morrison, 1999, 2000; Orr et al., tions under field conditions (Henne and Johnson, unpublished data). 1995, 2003). Solenopsis spp. workers will reduce or terminate forag- Prior to experiments, ants were weighed and placed in 90 mm ing activity in response to attacks by Pseudacteon flies (Feener and 15 mm Petri dishes. A weighed 0.10 g random sample contained Ò Brown, 1992; Orr et al., 1995; Morrison, 1999). Females of these sol- 108 ants. The inner walls of the Petri dishes were coated with Fluon itary endoparasitoids insert a single egg into host ants when they are to prevent ants from escaping. First, 10 replications were conducted engaged in various activities outside of the nest. The maggot feeds on with three host ant densities (1.0 g (1080 ants), 0.25 g (270 ants), 1 internal head structures and eventually pupariates inside the decap- and 0.06 g (65 ants) – each successive density =4 of the previous high- itated host’s empty head capsule (Porter et al., 1995, 1997). Pseudact- est density). Next, 33 replicates were conducted with four host densi- eon phorid flies are considered an important factor in maintaining ties (0.5 g (540 ants), 0.25 g (270 ants), 0.12 g (135 ants), and 0.06 g lower abundances of S. invicta in South America (Porter et al., (65 ants) – each successive density ½ of the previous highest density). 1992), and may be useful in suppressing S. invicta populations in Finally, 12 replicates were conducted with five host densities (0.5 g, the United States. (540 ants) 0.25 g (270 ants), 0.125 g (135 ants), 0.06 g (65 ants), and Under field conditions, aggregations of >5 (but occasionally >100) 0.03 g (32 ants) – each successive density ½ of the previous highest Pseudacteon tricuspis Borgmeier can be observed at individual dis- density). During each trial, flies were counted at Petri dishes at 1, 5 turbed S. invicta mounds (Henne and Johnson, 2008, 2009). Both and 10 min after lids were removed from the dishes. In all trials, Petri sexes are attracted to host aggregations and mating occurs while fe- dishes were separated by at least 5 cm. A preliminary experiment re- males are actively searching for hosts (Porter et al., 1997; Porter, vealed no differences in attractiveness of hosts attacked for several 1998). Additionally, aggressive interactions between conspecific hours vs. hosts that were never attacked. After each trial all Petri males and females can be commonly observed under both labora- dishes were covered, removed, and a new set of ants randomly se- tory and field conditions (Morrison and Porter, 2005a, Pers. Obs.). lected from a pool of preweighed ants was placed in the cage. To allow Males are promiscuous and will mate with the same female multiple flies to settle down, a minimum of five minutes was permitted to times (Porter et al., 1997, Pers. Obs.). However, nothing is known elapse before a new trial was performed. Trials on the three, four about P. tricuspis aggregation, direct mutual interference and the and five ant density levels were performed on different days within functional response of individual females, necessitating exploratory one week, and up to 10 trials were performed on a single day. research into these areas. The objectives of this study were to: (1) quantify aggregative responses of P. tricuspis adults to variable host 2.1.1. Statistical analysis densities, (2) determine effect of direct mutual interference between Data consisting of counts are frequently Poisson rather than pairs of ovipositing P. tricuspis females confined with host S. invicta, normally distributed, with the result that the mean and variance Author's personal copy

D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 65 will not be independent but will tend to vary together (Sokal and the lids were removed, the flies allowed to escape or aspirated, and

Rohlf, 1995). Therefore, P. tricuspis counts were log10 x + 1-trans- the containers removed from the cage. formed before analysis to achieve normality and stabilize vari- After all flies were removed, a small (1 cm3) plaster block that ances. A profile ANOVA was performed on treatment effect (i.e. was saturated with distilled water was placed in the containers host density) and time (1, 5 and 10 min) (PROC GLM, SAS Institute, to provide humidity. A 1 ml drop each of water and 10% sugar 2002). A profile ANOVA is a multivariate test (similar to repeated water was also deposited on the bottom of the containers and measures analysis; Simms and Burdick, 1988), but allows for the the lids closed. Numerous small pinholes were made in the lids sample trials to be non-independent in time. As no significant ef- to provide ventilation. Containers were placed in an environmental fects of time or time treatment effects on fly abundances were chamber (Percival Intellus 136 VL) with temperature set at a con- found in any of the experiments (P > 0.05), the # flies present at stant 28 °C and a 14:10 photo/scotoperiod. Every two days, the each dish was averaged over the three temporal evaluations and containers were cleaned of middens, and fresh water and sugar analyzed using a one-way ANOVA followed by Tukey’s test. water provided. Containers were randomly rearranged in the chamber daily. Parasitized ants began to die approximately 10 days after exposure to female P. tricuspis. For two subsequent 2.2. Direct mutual interference weeks, decapitated heads from individual replicates were carefully removed daily with soft forceps and placed onto moistened filter Two laboratory trials were performed to determine the effect of paper inside individual 90 mm 10 mm Petri dishes bearing the increasing the density of P. tricuspis females when confined with a same information as the source containers. The dishes were sealed constant density of S. invicta workers. In both trials 1, 2 or 3 female with paraffin laboratory film to maintain humidity and held in an P. tricuspis were confined with 0.5 g of S. invicta workers (ca. 540 environmental chamber under the same conditions as described ants). In both trials, each parasitoid density was replicated four above until adult emergence. times Ants were weighed and placed into individually labeled plas- Ò tic containers (Ziploc 236 ml snap lid containers) lined with Flu- 2.2.1. Male interference Ò on to prevent ants from escaping. After newly emerged flies An experiment was performed to determine if individual female were released inside the cage, several hours were allowed to elapse parasitism rates were affected by confining males with day-old al- before experiments were initiated. We found that newly emerged ready mated females (i.e. females that were confined previously P. tricuspis are either not, or are weakly, responsive to host S. invicta with other males and females). Mated solitary females were cap- pheromones for 1–2 h (Pers. Obs). Male P. tricuspis also appear to tured from an aggregation of flies that were attacking S. invicta in- undergo an obligate dispersal phase after emergence, as they side the cage. Females were then confined with zero, one and two would actively fly along the top of the cage near the lights, while males. Each treatment was replicated eight times. The procedure females would sit on the walls of the cage. was similar to that described above, except that the combination After this post emergence phase, a single container with of males and females that entered the containers was manipulated approximately 1.0 g of S. invicta workers (ca. 1000 ants) was placed with an aspirator. The post experiment handling of S. invicta and in the center of the cage to prime the flies to begin attacking hosts, puparia was the same as described above. and to allow the two sexes to mate. Adults of P. tricuspis mate while females are attacking hosts (Porter, 1998). As phorid parasitoids lo- 2.2.2. Statistical analysis cate their hosts by detecting ant semiochemicals (Porter, 1998; Numbers of hosts parasitized were ln – transformed prior to Morrison and King, 2004), ants were lightly probed to elicit alarm analyses, as above. Replicates with zero hosts parasitized were behavior and production of alarm pheromones, which attracted omitted from analyses, as female P. tricuspis in these replicates both male and female P. tricuspis. either failed to successfully parasitize at least one host, were cap- After 15 min, the primer container was removed and experi- tured and killed by S. invicta inside the container, or were other- mental containers were placed into the cage with their lids on. wise defective. A one-way ANOVA and Tukey’s HSD tests were Individual containers were randomly chosen among the replicates performed on numbers of hosts parasitized by 1, 2 or 3 females. and the lid was opened, allowing flies access to host S. invicta. Dis- The proportion of total hosts encountered by parasitoids per tinguishing between the two sexes was straightforward as they unit time (per capita searching efficiency) can be quantified in display different hovering behaviors; females are generally larger terms of the rate of decline in searching efficiency as parasitoid and hover a few mm from their hosts, while males are usually density increases (Hassell, 2000). Changes in per capita searching smaller and tend to maintain a larger distance between themselves efficiency (s)ofP. tricuspis in relation to parasitoid density were and S. invicta, and can also be distinguished from females by their estimated from the following equation (Visser and Driessen, 1991): searching behavior. When searching for females, males tend to ho-  ver in place and turn from side to side at approximately 45–90° 1 N s ¼ ln t ð1Þ from center (see also Porter, 1998). They also tend to spend more Pt Nt Na time searching among hosts that are more densely aggregated (Pers. Obs.). When males descended into experimental containers, where Pt is density of P. tricuspis, Nt is the number of hosts and Na is an aspirator was used to remove them. the number of hosts killed. Searching efficiency was regressed When the predetermined numbers of female P. tricuspis des- against ln host density using least squares regression. cended into individual containers and were confirmed to be attack- Finally, the effect of confining additional males with solitary fe- ing hosts, the lid was snapped into place and set aside in the cage. The males on progeny sex ratios was tested with a Pearson chi-square time that the lid was closed was written on a label on the lid and the test. The null hypothesis was that progeny sex ratios were not next container was opened. This process was continued until all rep- different among treatments. Progeny sex ratios were also tested licates had the desired number of females. However, at the same against a hypothesized 1:1 ratio with a Pearson chi-square test. time that flies were being confined with S. invicta, previously com- pleted replicates were tapped lightly to induce ant alarm behavior 2.3. Functional response and maintain fly activity. When under attack by P. tricuspis, S. invicta tended to cluster and required occasional disturbance to disperse Four laboratory trials were conducted to determine the shape of them. Flies were confined with hosts for two hours, after which the functional response when confining a single female P. tricuspis Author's personal copy

66 D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 with variable densities of S. invicta. Ants were placed in plastic con- were invariant over the 10 min time interval, but the overall total tainers (ZiplocÒ 236 ml snap lid containers) that were lined with Flu- number of flies that recruited slightly increased over time. In all onÒ to prevent ants from escaping. The procedure of confining treatments, significant differences in numbers of aggregating flies female P. tricuspis with host S. invicta and maintenance of ants was were found among the different host densities, with higher num- the same as described above for the interference trials. In trials 1– bers of flies aggregating at the highest host densities: (three host 3, individual female P. tricuspis were confined with 135, 270, 540, densities (df = 2, 27; F = 14.53; P < 0.0001; Fig. 1A), four host densi- 810 and 1080 ants, with each host density replicated four times. In ties (df = 3, 128; F = 16.08; P < 0.0001; Fig. 1B), five host densities trial 4, individual female P. tricuspis were confined with 25, 50, 100 (df = 4, 55; F = 19.3; P < 0.0001; Fig. 1C). and 200 ants, with each host density replicated eight times. Post experiment maintenance of exposed ants was the same as described 3.2. Direct mutual interference above for the interference experiment. Sex ratios of progeny adults were determined every second day when emergence began. 3.2.1. Total hosts parasitized The total number of S. invicta hosts successfully parasitized by 2.3.1. Statistical analysis the limited range of female P. tricuspis densities evaluated was sig- To distinguish among the three types of host dependence in the nificantly different in trial 1 (P = 0.02, Fig. 2A), but not in trial 2 functional response, a two-step approach recommended by Juliano (2001) was followed. First, the shape of the functional response curve on the percentage of ant hosts successfully parasitized by 1.5 A P. tricuspis as a function of ant density was determined by logistic maximum likelihood regression (PROC CATMOD, SAS Institute, b 2002). The logistic model is as follows: 1.0 2 3 Na expðP0 þ P1N0 þ P2N0 þ P3N0Þ a ¼ 2 3 ð2Þ N P. tricuspis a 0 1 þ expðP0 þ P1N0 þ P2N0 þ P3N0Þ

# 0.5 where the parameter N0 is the host density, Na is the number of 10 hosts parasitized, and P0, P1, P2, and P3 are the logistic regression log parameters associated with the slope of the curve. The null hypoth- esis is that the linear parameters are not significantly different from 0.0 zero. A type I functional response is indicated by linear terms not 65 270 significantly different from zero (i.e. zero slope), a type II functional 1080 response by a significant negative value of P0, and a type III func- Host ant density tional response by a positive P0 parameter and a negative P1 (qua- dratic) parameter. If the linear parameter computed from the 0.8 B logistic regression is not significantly different from zero, it indi- c cates no effect of increased host density on the proportion of hosts 0.6 parasitized and the type I functional response is fitted to the data by b,c the following linear equation (Parajulee et al., 2006): a,b Na ¼ a þ bN0 ð3Þ 0.4 P. tricuspis a where Na is the number of hosts parasitized, N0 is the host density, # and a and b are the intercept and slope of the attack rate prediction 10 0.2 line, respectively. log Second, if the appropriate functional response form is deter- mined to be type II or III, parameter estimation of a (attack con- 0.0 stant) and b (functional response asymptote) are achieved by 65 135 270 540 fitting the numbers of ants parasitized at variable host densities Host ant density to the appropriate functional response selected by the logistic pro- cedure using a non-linear least squares procedure (PROC NLIN, SAS 1.0 c Institute, 2002). Equations for types II and III functional responses C are given in Juliano (2001). Since the results indicated P. tricuspis c 0.8 females attack according to a type I functional response (see Sec- tion 3), the slopes and intercepts of the mean number of hosts par- b 0.6 asitized in relation to host density, and the proportion parasitized a,b

in relation to host density for trials 1–3 were compared with an P. tricuspis 0.4

ANCOVA (Sokal and Rohlf, 1995). Replicates where no puparia #

were produced (i.e. no successful attacks occurred) were excluded 10 a from the analysis, as these females either were defective or other- 0.2 log wise were captured and killed by S. invicta. 0.0

3. Results 32 65 135 270 540 Host ant density 3.1. Aggregative responses of P. tricuspis

Fig. 1. Aggregation responses of P. tricuspis (log10-transformed mean ± SE) to (A) Flies quickly recruited to Petri dishes containing host S. invicta. three levels of host density, (B) four levels of host density, and (C) five levels of host In general, proportional fly abundances among treatment levels density. Author's personal copy

D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 67

(P = 0.87, Fig. 2B). The total numbers of hosts parasitized remained -1.0 generally static over the range of females evaluated. A

3.2.2. Searching efficiency -1.5 Although there was a declining trend in per capita searching efficiency, no correlation was found between the log searching effi- ciency and the log number of P. tricuspis in trial 1 (R2 = 0.23; df = 1, 10; F = 2.941; P = 0.12) (Fig. 3A). However, the correlation -2.0 was marginally significant in trial 2 (R2 = 0.31; df = 1, 11; F = 4.85; P = 0.0499) (Fig. 3). Therefore, interference among several oviposit- ing P. tricuspis females may be important at densities of more than Log search efficiency (s) three simultaneously ovipositing females. -2.5 0.0 0.2 0.4 0.6

3.2.3. Male interference Log Pt The presence of additional males confined with mated solitary -1.5 females did not have a significant effect on the total number of suc- B cessfully parasitized hosts (P > 0.05). The chi-square analysis showed no significant effect on progeny sex ratios from having -2.0 additional males confined with already mated females (df = 2, 132; v2 = 0.3; P = 0.86). However, male to female sex ratios shifted over the range of treatments, 3.6:1 (0 males), 3:1 (1 male), -2.5 2.8:1 (2 males). Overall, the progeny sex ratios deviated signifi- cantly from a hypothetical 1:1 ratio (no males: df =1, v2 = 14.7, -3.0 P = 0.0001; 1 male: df =1, v2 = 8.0, P = 0.005; 2 males: df =1, v2 = 12.8, P = 0.0003). Log search efficiency (s) -3.5 3.3. Functional response 0.0 0.2 0.4 0.6

Log Pt None of the linear parameters in the logistic models were significantly different from zero (P > 0.05), suggesting that P. tricu- Fig. 3. Results of direct mutual interference experiments, showing per capita searching efficiency (s of Eq. (1))ofP. tricuspis in relation to female density: (A) trial 1, (B) trial 2.

SE) 40 A ± b spis parasitism rates follow a type I functional response, at least under laboratory conditions (Table 1 and Fig. 4). Therefore, attack 30 rates appear to be host density-independent. The results of the AN- COVA for comparing slopes and intercepts of mean number of a 20 hosts parasitized in relation to host density, and the mean propor- a tion parasitized in relation to host density for trials 1–3 indicated no significant differences between either slopes or intercepts 10 [mean hosts parasitized (slopes: dfn =2, dfd =9; F = 0.49; P = 0.63) (intercepts: dfn =2,dfd = 11; F = 0.30; P = 0.74); calculated

untransformed total hosts pooled slope for trials 1–3 is 0.002 and the pooled intercept is 0 2.45], [mean proportion hosts parasitized (slopes: dfn =2,dfd =9; parasitized/container (mean 1 2 3 F = 0.64; P = 0.55) (intercepts: dfn =2,dfd = 11; F = 0.12; P = 0.89); # females/container calculated pooled slope for trials 1–3 is 0.0000118 and the pooled intercept is 0.015]. B

SE) 10 ± a 8 4. Discussion a a 6 4.1. Aggregative responses

In this study, aggregative responses of P. tricuspis to host densi- 4 ties were density-dependent under laboratory conditions. Simi- larly, under field conditions, Morrison and King (2004) found that 2 increasing the number of non-nestmate S. invicta workers at baits

untransformed total hosts already occupied by S. invicta led to enhanced numbers of P. tricu- 0 spis, presumably because increased alarm pheromone production parasitized/container (mean 1 2 3 by fighting non-nestmates attracted more flies. Furthermore, # females/container Morrison and Porter (2005b) established a positive correlation be- tween P. tricuspis abundance and S. invicta density in north-central Fig. 2. Results of laboratory trials evaluating total host S. invicta parasitized/female: Florida. In the laboratory experiments described in our paper, P. (A) trial 1, (B) trial 2 (untransformed mean ± SE shown) at three levels of female P. tricuspis density. Bars with the same letters are not significantly different at tricuspis continued to aggregate at the higher host densities, even a = 0.05. when Petri dishes were covered and the dishes rearranged. Perhaps Author's personal copy

68 D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71

Table 1 Results of the maximum likelihood estimates by PROC CATMOD for the functional response of P. tricuspis to varying host densities.

Trial Logistic regression parameters Estimate ± SE {2 df P-value

1 Intercept 3.71 ± 1.10 11.38 1 0.0007

P0 0.005 ± 0.007 0.60 1 0.44 2 8.9 106 ± 1.2 105 0.58 1 0.44 P0 3 5.2 109 ± 5.9 109 0.77 1 0.38 P0 Likelihood ratio 24.00 13 0.03 2 Intercept 2.49 ± 0.96 6.80 1 0.009

P0 0.01 ± 0.007 2.31 1 0.13 2 1.4 105 ± 1.3 105 1.08 1 0.30 P0 3 6.04 109 ± 7.1 109 0.72 1 0.40 P0 Likelihood ratio 7.37 11 0.77 3 Intercept 4.89 ± 1.06 21.23 1 <0.0001

P0 0.005 ± 0.007 0.56 1 0.45 2 1.0 105 ± 1.2 105 0.95 1 0.33 P0 3 6.3 109 ± 6.6 109 0.91 1 0.34 P0 Likelihood ratio 16.37 13 0.23 1–3 Intercept 3.60 ± 0.57 39.73 1 <0.0001

P0 0.004 ± 0.004 1.15 1 0.28 2 4.5 106 ± 6.8 106 0.43 1 0.51 P0 3 2.2 109 ± 3.6 109 0.39 1 0.53 P0 Likelihood ratio 55.90 48 0.15 4 Intercept 0.17 ± 0.67 0.06 1 0.80

P0 0.05 ± 0.03 2.77 1 0.10 2 4.1 104 ± 3.3 104 1.53 1 0.22 P0 3 1.1 106 ± 1.0 106 1.19 1 0.28 P0 Likelihood ratio 24 <0.0001

P. tricuspis are able to learn and visually distinguish between rela- of ultimate mating success when females also appear at high den- tive patch sizes in space, and it is thought that parasitoids can prof- sity host patches. it most when they aggregate and spend most of their time in patches where host densities are highest (Free et al., 1977). 4.2. Direct mutual interference A lot of frenzied activity occurs when adults of P. tricuspis aggre- gate at patches of host S. invicta. Not only are females competing Some evidence of direct mutual interference was found when with one another for access to hosts, males also aggressively com- two or three female P. tricuspis were confined in small laboratory pete with other males for access to females (Pers. Obs.). A similar containers, as per capita oviposition success (measured as number pattern of activity occurs with Scatophaga stercoraria L. (Diptera: of hosts killed) declined when more than two females were con- Scatophagidae), where intra-male competition for females at dung fined. This study did not demonstrate any reductions in estimates pats is strong, as males outnumber females by 4:1 (Parker, 1974). of searching efficiency of at least 2 or 3 simultaneously ovipositing Male P. tricuspis are variable in size, with larger males often as P. tricuspis females. However, Visser and Driessen (1991) warn that large as some females (Pers. Obs.). Presumably these larger males it is important to consider population and generation level effects live longer and have an advantage in competing with smaller of mutual interference on estimates of searching efficiency, as dis- males for mates (Morrison et al., 1999). persal from patches containing high densities of conspecifics can P. tricuspis females are probably pro-ovigenic and egg-limited lead to enhanced searching efficiency if hosts are uniformly dis- parasitoids, but no information is available on fecundity of P. tric- tributed. However, this study did not evaluate nor allow dispersal uspis. However, fecundity of the related Pseudacteon wasmanni Sch- of females between host patches. mitz ranges from 30 to nearly 300 eggs (Zacaro and Porter, 2003). Field studies of P. tricuspis populations in Louisiana revealed Egg-limited parasitoids characteristically have short handling that approximately 50% of P. tricuspis aggregations at disturbed times (Getz and Mills, 1996; Mills and Lacan, 2004). Indeed, han- S. invicta mounds include 1–3 females (Henne and Johnson, dling time in P. tricuspis is very short (<1 s) in relation to overall 2009). It was extremely difficult to consistently confine more than time spent searching. A small ratio of handling to search time in three P. tricuspis females together in small containers in the labora- parasitoids that are confined to a single patch for a longer time tory experiments described here because some females tended to than by choice could result in a linear functional response (Hassell, leave the container when too many females were present. There- 2000, but see Mills and Lacan, 2004). Therefore, the problem for fore, direct mutual interference may become more important when parasitoid females, such as P. tricuspis, is to parasitize as many higher densities of ovipositing P. tricuspis are simultaneously pres- hosts as possible in its short lifespan (Wajnberg, 2006). Similarly, ent than those evaluated in this study. The intensity of interactions males should maximize their fitness by mating with as many fe- usually increases at higher parasitoid densities, leading to greater males as possible. Natural selection will enhance strategies that mutual interference and overall suppressed searching efficiency optimize not only female reproductive success, but mating success of the parasitoid population (Visser and Driessen, 1991). In con- as well (Cook and Hubbard, 1977; Parker, 1978). In the laboratory, trast, indirect mutual interference is a reduction in searching effi- males were often observed to appear at higher density host ciency at the population level due to superparasitism (Visser and patches before females, a strategy that would increase probability Driessen, 1991). Superparasitism of individual S. invicta workers Author's personal copy

D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 69

B SE 0.0175

7.5 ±

SE A ± 0.0150 0.0125 5.0 0.0100 0.0075 2.5 0.0050 0.0025 0.0 0.0000 Mean hosts parasitized 0 250 500 750 1000 1250 0 250 500 750 1000 1250 Host density Mean proportion parasitized Host density

C SE 6.5 ± 0.035 D SE 6.0 ± 5.5 0.030 5.0 4.5 0.025 4.0 3.5 0.020 3.0 0.015 2.5 2.0 0.010 1.5 1.0 0.005 0.5 0.0 0.000 Mean hosts parasitized 0 250 500 750 1000 1250 0 250 500 750 1000 1250 Mean proportion parasitized Host density Host density SE

6.5 E ± 0.025 F SE 6.0 ± 5.5 0.020 5.0 4.5 4.0 0.015 3.5 3.0 2.5 0.010 2.0 1.5 0.005 1.0 0.5 0.0 0.000 Mean hosts parasitized 0 250 500 750 1000 1250 0 250 500 750 1000 1250 Mean proportion parasitized Host density Host density

7.5 SE

G ± H

SE 0.020 ±

0.015 5.0

0.010 2.5 0.005

0.0

Mean hosts parasitized 0.000 0 250 500 750 1000 1250 0 250 500 750 1000 1250 Host density Mean proportion parasitized Host density SE

25 I ± 0.35 J SE

± 0.30 20 0.25 15 0.20 0.15 10 0.10 5 0.05

0 0.00 Mean hosts parasitized hosts Mean 0 50 100 150 200 250 0 50 100 150 200 250 Mean proportion parasitized Host density Host density

Fig. 4. Results of functional response trials: A, C, E, G, and I are mean hosts parasitized (mean ± SE) at varying levels of host density; B, D, F, H, and J are mean proportion of hosts parasitized at varying levels of host density: (A and B) trial 1; (C and D) trial 2; (E and F) trial 3; (G and H) trials 1–3 pooled; (I and J) trial 4.

by P. tricuspis has been observed in the laboratory on numerous criminate between parasitized and unparasitized hosts. A labora- occasions, suggesting that P. tricuspis females are unable to dis- tory experiment comparing attractiveness of hosts exposed to P. Author's personal copy

70 D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 tricuspis parasitism for four days vs. non-parasitized hosts showed 4.5. Conclusions no apparent differences in attractiveness, as equal numbers of flies were attracted (Henne and Johnson, unpublished data). Superpar- The studies conducted in this paper have provided important asitism is probably rare under natural conditions, given that natu- insights into P. tricuspis behavioral and functional responses that ral parasitism rates of S. invicta by P. tricuspis are very low (see were unknown until now. The density-dependent aggregations of Morrison and Porter, 2005a). P. tricuspis observed in the laboratory are consistent with theory and field observations. Mutual interference of conspecific male and females at low densities does not appear to be significant, 4.3. Male interference but may be important at higher densities. The type I functional re- sponse found was unexpected on the grounds that most parasit- This study did not reveal any significant effect of having addi- oids appear to have a type II functional response. It is hoped that tional males confined with solitary premated females. However, the results obtained in this study will stimulate further research the sex ratios shifted when the number of males confined with sol- into Pseudacteon population ecology and test host-parasitoid itary females was increased from zero to two. It is unclear what theory. mechanism(s) is (are) responsible for shifts in sex ratio allocation in Pseudacteon spp. Consistent with host size-dependent-sex allo- cation theory (Charnov et al., 1981), sex ratios of Pseudacteon Acknowledgments spp. have been linked to host size, with more females arising from larger hosts (Morrison et al., 1999). Therefore, secondary sex ratios Thanks are extended to J. Cronin, G. Henderson and T. Schowalt- may simply be an artifact of the size range of available host ants er for helpful comments and criticisms. Approved for publication (Morrison and Porter, 2005a). by the Director, Louisiana Agricultural Experiment Station as Man- Sex ratio shifts have been documented in other parasitoid spe- uscript number 2008-234-2330. cies. For example, Wylie (1965) found that increasing the ratio of Nasonia vitripennis (Walk.) (Hymenoptera: Pteromalidae) females References to host Musca domestica L. (Diptera: Muscidae) resulted in a reduc- tion in the proportion of progeny females. Perhaps female P. tricu- Arditi, R., Ginzburg, L.R., 1989. Coupling in predator-prey dynamics: ratio- spis can adjust sex ratio allocation of progeny by differential dependence. Journal of Theoretical Biology 139, 311–326. selection of host sizes in response to interference by conspecifics. Banks, W.A., Lofgren, C.S., Jouvenaz, D.P., Stringer, C.E., Bishop, P.M., Williams, D.F., Wojcik, D.P., Glancey, B.M., 1981. Techniques for collecting, rearing, and handling The theory of local mate competition (Hamilton, 1967) has been imported fire ants. US Department of Agriculture, Science and Education proposed as an explanation for sex ratio shifts among Hymenop- Administration, Advances in Agricultural Technology, AATS-S-21, p. 9. teran parasitoids, where an increase in the number of female con- Beddington, J.R., 1975. Mutual interference between parasites or predators and its effect on searching efficiency. Journal of Animal Ecology 44, 331–340. specifics at a host patch results in an increase in the proportion of Charnov, E.L., 1976. Optimal foraging, the marginal value theorem. Theoretical male progeny produced. Unfortunately, a similar evaluation of sex Population Biology 9, 129–136. ratio shifts when multiple females were confined was not con- Charnov, E.L., Los-den-Hartogh, R.L., Jones, W.T., 1981. Sex ratio evolution in a variable environment. Nature 289, 27–33. ducted in this study, but this would be an interesting research Chesson, P.L., Murdoch, W.W., 1986. Aggregation of risk: relationships among host– direction to pursue. Nevertheless, a P. tricuspis 3:1 sex ratio parasitoid models. American Naturalist 127, 696–715. (males–females) has been consistently found under field condi- Chong, J.H., Oetting, R.D., 2006. Functional response and progeny production of the Madeira mealybug parasitoid, Anagyrus sp. nov. nr. sinope: the effects of host tions (Henne and Johnson, 2009), and was reproduced under labo- and parasitoid densities. Biological Control 39, 320–328. ratory conditions. Cook, R.M., Hubbard, S.F., 1977. Adaptive searching strategies in insect parasites. Journal of Animal Ecology 46, 115–125. Feener Jr., D.H., 1981. Competition between ant species: outcome controlled by 4.4. Functional response parasitic flies. Science 214, 815–817. Feener Jr., D.H., Brown, B.V., 1992. Reduced foraging of Solenopsis geminata (Hymenoptera: Formicidae) in the presence of parasitic Pseudacteon spp. None of the linear parameters in the logistic models were (Diptera: Phoridae). Annals of the Entomological Society of America 85, 80–84. significantly different from zero suggesting that P. tricuspis had Folgarait, P.J., Gilbert, L.E., 1999. Phorid parasitoids affect foraging activity of constant attack rates regardless of host density under the labora- Solenopsis richteri under different availability of food in Argentina. Ecological Entomology 24, 163–173. tory experimental design. It is possible that host density levels Free, C.A., Beddington, J.R., Lawton, J.H., 1977. On the inadequacy of simple models evaluated in this study were too high, and therefore P. tricuspis at- of mutual interference for parasitism and predation. Journal of Animal Ecology tack rates were operating at an asymptotic level. However, the pro- 46, 543–554. Getz, W.M., Mills, N.J., 1996. Host–parasitoid coexistence and egg-limited encounter portion parasitized clearly support a type I functional response. It rates. American Naturalist 148, 333–347. should also be pointed out that these females were confined with Godfray, H.C.J., Pacala, S.W., 1992. Aggregation and the population dynamics of their hosts and not allowed to disperse freely between host parasitoids and predators. American Naturalist 140, 30–40. Hamilton, W.D., 1967. Extraordinary sex ratios. Science 156, 477–488. patches. A different functional response curve may be relevant un- Hassell, M.P., 2000. The spatial and temporal dynamics of host–parasitoid der more natural situations or laboratory settings where females interactions. Oxford University Press, New York. Viii+200pp. are allowed unrestricted movement (Chong and Oetting, 2006). Hassell, M.P., May, R.M., 1973. Stability in insect host–parasite models. Journal of Animal Ecology 42, 693–726. Type I functional responses among insect parasitoids are rare Hassell, M.P., Varley, G.C., 1969. New inductive population model for insect (see Mills and Lacan, 2004 for examples). However, Turchin parasites and its bearing on biological control. Nature 223, 1133–1137. (2003) argues that differences between type I and type II functional Henne, D.C., Johnson, S.J., 2008. Spatio-temporal population structure of a fire ant responses are minor. It is important to mention that the functional parasitoid. Entomologia Experimentalis et Applicata 129, 132–141. Henne, D.C., Johnson, S.J., 2009. Sampling and dynamics of Pseudacteon tricuspis response is not independent of host density alone and should ac- (Diptera: Phoridae) in Louisiana. Environmental Entomology 38, 539–550. count for the reality that parasitoids rarely exist as single individ- Henne, D.C., Johnson, S.J., Porter, S.D., 2007. Status of the fire ant decapitating fly uals and more likely interact with conspecifics, necessitating the Pseudacteon tricuspis Borgmeier (Diptera: Phoridae) in Louisiana. Florida Entomologist 90, 565–569. need for ratio-dependent functional response studies (Arditi and Holling, C.S., 1959. Some characteristics of simple types of predation and parasitism. Ginzburg, 1989; Mills and Lacan, 2004; Chong and Oetting, Canadian Entomologist 91, 385–398. 2006). Furthermore, effects of temperature on functional response Holling, C.S., 1961. Principles of insect predation. Annual Review of Entomology 6, 163–182. can be important (see Parajulee et al., 2006; Zamini et al., 2006) but Holling, C.S., 1966. Functional response of invertebrate predators to prey density. were not evaluated in this study. Memoirs of the Entomological Society of Canada 48, 1–60. Author's personal copy

D.C. Henne, S.J. Johnson / Biological Control 55 (2010) 63–71 71

Juliano, S.A., 2001. Nonlinear curve fitting: predation and functional response Parker, G.A., 1978. Evolution of competitive mate searching. Annual Review of curves. In: Scheiner, S.M., Gurevitch, J. (Eds.), Design and Analysis of Ecological Entomology 23, 173–196. Experiments, second ed. Chapman & Hall, New York, pp. 178–196. Porter, S.D., 1998. Biology and behavior of Pseudacteon decapitating flies (Diptera: Maynard Smith, J., 1974. The theory of games and the evolution of animal conflicts. Phoridae) that parasitize Solenopsis fire ants (Hymenoptera: Formicidae). Journal of Theoretical Biology 47, 209–221. Florida Entomologist 81, 292–309. Mills, N.J., Lacan, I., 2004. Ratio dependence in the functional response of insect Porter, S.D., Fowler, H.G., MacKay, W.P., 1992. Fire ant mound densities in the parasitoids: evidence from Trichogramma minutum foraging for eggs in small United States and Brazil (Hymenoptera: Formicidae). Journal of Economic host patches. Ecological Entomology 29, 208–216. Entomology 85, 1154–1161. Morrison, L.W., 1999. Indirect effects of phorid fly parasitoids on the mechanisms of Porter, S.D., Pesquero, M.A., Campiolo, S., Fowler, H.G., 1995. Growth and interspecific competition among ants. Oecologia 121, 113–122. development of Pseudacteon phorid fly maggots (Diptera: Phoridae) in the Morrison, L.W., 2000. Mechanisms of Pseudacteon parasitoid (Diptera: Phoridae) heads of Solenopsis fire ant workers (Hymenoptera: Formicidae). Environmental effects on exploitative and interference competition in host Solenopsis ants Entomology 24, 475–479. (Hymenoptera: Formicidae). Annals of the Entomological Society of America 93, Porter, S.D., Williams, D.F., Patterson, R.S., 1997. Rearing the decapitating fly 841–849. Pseudacteon tricuspis (Diptera: Phoridae) in imported fire ants (Hymenoptera: Morrison, L.W., King, J.R., 2004. Host location behavior in a parasitoid of imported Formicidae) from the United States. Journal of Economic Entomology 90, 135– fire ants. Journal of Insect Behavior 17, 367–383. 138. Morrison, L.W., Porter, S.D., 2005a. Phenology and parasitism rates in introduced SAS Institute, 2002. SAS/STAT user’s guide. SAS Institute, Cary, NC. populations of Pseudacteon tricuspis, a parasitoid of Solenopsis invicta. Biological Simms, E.L., Burdick, D.S., 1988. Profile analysis of variance as a tool for analyzing Control 50, 127–141. correlated responses in experimental ecology. Biometry Journal 30, 229–242. Morrison, L.W., Porter, S.D., 2005b. Testing for population-level impacts of Sokal, R.R., Rohlf, F.J., 1995. Biometry the principles and practice of statistics in introduced Pseudacteon tricuspis flies, phorid parasitoids of Solenopsis invicta biological research, third ed. W.H. Freeman, New York. fire ants. Biological Control 33, 9–19. Solomon, M.E., 1949. The natural control of animal populations. Journal of Animal Morrison, L.W., Dall’Aglio-Holvorcem, C.G., Gilbert, L.E., 1997. Oviposition behavior Ecology 18, 1–35. and development of Pseudacteon flies (Diptera: Phoridae), parasitoids of Stiling, P.D., 1987. The frequency of density dependence in insect host–parasitoid Solenopsis fire ants (Hymenoptera: Formicidae). Environmental Entomology systems. Ecology 68, 844–856. 26, 716–724. Thompson, W.R., 1924. La théorie Mathématique de l’action des parasites Morrison, L.W., Porter, S.D., Gilbert, L.E., 1999. Sex ratio variation as a function of entomophages et le facteur du hazard. Annales de la Faculté des Sciences de host size in Pseudacteon flies (Diptera: Phoridae), parasitoids of Solenopsis fire Marseille 2, 69–89. ants (Hymenoptera: Formicidae). Biological Journal of the Linnean Society 66, Turchin, P., 2003. Complex population dynamics: a theoretical/empirical synthesis. 257–267. Princeton University Press, Princeton, New Jersey. xvii+450pp. Murdoch, W.W., Stewart-Oaten, A., 1989. Aggregation by parasitoids and predators: Visser, M.E., Driessen, G., 1991. Indirect mutual interference in parasitoids. effects on equilibrium and stability. American Naturalist 134, 288–310. Netherlands Journal of Zoology 41, 214–227. Nicholson, A.J., 1933. The balance of animal populations. Journal of Animal Ecology Visser, M.E., Jones, T.H., Driessen, G., 1999. Interference among insect parasitoids: a 2, 131–178. multi patch experiment. Journal of Animal Ecology 68, 108–120. Orr, M.R., Selke, S.H., Benson, W.W., Gilbert, L.E., 1995. Flies suppress fire ants. Wajnberg, E., 2006. Time allocation strategies in insect parasitoids: from ultimate Nature (London, U.K.) 373, 292. predictions to proximate behavioral mechanisms. Behavioral Ecology and Orr, M.R., Dahlsten, D.L., Benson, W.W., 2003. Ecological interactions among ants in Sociobiology 60, 589–611. the genus Linepithema, their phorid parasitoids, and ant competitors. Ecological Wylie, H.G., 1965. Some factors that reduce the reproductive rate of Nasonia Entomology 28, 203–210. vitripennis (Walk.) at high adult population densities. Canadian Entomologist Parajulee, M.N., Shrestha, R.B., Leser, J.F., Webster, D.B., Blanco, C.A., 2006. 97, 970–977. Evaluation of the functional response of selected arthropod predators on Zacaro, A.A., Porter, S.D., 2003. Female reproductive system of the decapitating fly bollworm eggs in the laboratory and effect of temperature on their predation Pseudacteon wasmanni Schmitz (Diptera: Phoridae). Arthropod structure and efficiency. Environmental Entomology 35, 379–386. development 31, 329–337. Parker, G.A., 1974. The reproductive behavior and the nature of sexual selection in Zamini, A.A., Talebi, A.A., Fathipour, Y., Baniameri, V., 2006. Temperature-dependent Scatophaga stercoraria L. (Diptera: Scatophagidae). IX. Spatial distribution of functional response of two aphid parasitoids, Aphidius colemani and Aphidius fertilization rates and evolution of male search strategy within the reproductive matricariae (Hymenoptera: Aphididae), on the cotton aphid. Journal of Pest area. Evolution 28, 93–108. Science 79, 183–188.