Economically Beneficial Ground Beetles . The specialized predators Pheropsophus aequinoctialis ( L . ) and Stenaptinus jessoensis ( Morawitz ) : Their laboratory behavior and descriptions of immature stages ( Coleoptera , Carabidae , Brachininae )

Adults of Pheropsophus aequinoctialis (L.) (Coleoptera, Carabidae, Brachininae, Brachinini), are largely nocturnal predators and scavengers on animal and plant materials. Th e daily food consumption of a pair of adults is the equivalent to 1.2-2.3 large larvae of Trichoplusia ni (Hübner) (Lepidoptera, Noctuidae). Larvae developed under laboratory conditions on a diet restricted to mole cricket eggs (Orthoptera, Gryllotalpidae); none survived under any other diet off ered, thus they are specialists. Large numbers of brachinine eggs were laid in the laboratory, even on a paper towel substrate, and in all months of the year albeit with a strong suggestion of an annual peak in oviposition. Many eggs failed to hatch, but those that did so incubated an average 13.5 days. Many neonate larvae failed to feed and died. On average, the larvae that developed took 25.9 days to do so on an average 38.4 mole cricket eggs. Th e pupal period averaged 20.4 days, so the total developmental period was 59.9 days from oviposition to emergence of adult off spring at 26°C. After initial trials, an improved method of handling adults and rearing immature stages was developed, resulting in initiation of feeding by most neonate larvae and control of contaminating organisms (nematodes, mites, and Laboulbeniales). Most neonate larvae need to be in a cell or pit of sand (or earth) resembling a mole cricket egg chamber before they will feed on mole cricket eggs. Th e cause of infertility of many eggs was not resolved because it continued under the improved handling method for adults which permitted weekly mating; the presence of Wolbachia spp. ZooKeys 14: 1-36 (2009) doi: 10.3897/zookeys.14.188 www.pensoftonline.net/zookeys Copyright JH Frank et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Launched to accelerate biodiversity research A peer-reviewed open-access journal

Stenaptinus s. str.has 114 described species (Lorenz 2005a, b).Th ere are descriptions of the fi rst instar S. hispanicus (Dejean) (Emden 1919), and S. africanus (Dejean) (Boldori 1939).Habu andSadanaga (1965, 1969) described and illustrated all three instars and a rearing method for S. jessoensis (Morawitz).Th e fi rst instar is an active triangulin, the second and third instars are hypermetamorphic.Th e larvae develop only in real or simulated mole cricket egg chambers, only on a diet of Gryllotalpa mole cricket eggs.Experimental evidence for those statements was not provided by Habu andSadanaga (1965, 1969).Th e adults are generalist predators, feeding on various insects, including pests, and ovipositing in June and July (Habu and Sadanaga 1965).In China, fi ve artifi cial diets for overwintered female S. jessoensis were compared in terms of longevity of the beetles, egg production, egg fertility and incubation time; some of the artifi cial diets were almost as good as a diet of various insects on which each female produced 42.2 eggs, the last female survived until mid-July, 31.3% of eggs hatched, and mean incubation time was 12.3 days (Li 1988).
Pheropsophus s. str.has seven described species (Erwin 1970); however, a few others are as yet undescribed and numerous synonyms need to be checked; the genus is in need of a modern revision.Th e most widespread and markedly variable species, P. aequinoctialis (L.), has been reported from Argentina (Catamarca, Jujuy), Bolivia, Brazil, Costa Rica, Ecuador, Mexico (Yucatán), Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela (Erwin 2001).Adult P. aequinoctialis have a crepitating behavior like other Brachinini, producing quinones (Zinner et al. 1991) and they are nocturnal, running on sandy trails or riverine beaches, hiding during the day under stones, grass clumps, and drift logs and often in aggregations; they are predatory on other insects and also will eat some plant materials, such as ripe fruits of Astrocaryum sp., a palm (Reichardt 1971).Adult P. aequinoctialis feed on adult Scapteriscus mole crickets in sand-fi lled containers in the laboratory (A. Silveira-Guido, pers. comm.).Adult P. rivieri (Demay) inhabit seasonally-inundated fl oodplains in the Amazon drainage of Brazil, and share the water banks with Scapteriscus mole crickets; dissections of females revealed that the reproductive period is confi ned to the fi rst three months of falling water levels (Zerm and Adis 2003).Immature stages of Pheropsophus have heretofore not been described, and we do that here.
We compared food consumption and diel behavior of adults of S. jessoensis and P. aequinoctialis, their oviposition, fertility of eggs, and development time of immature stages, and contrasted the results of feeding the larvae on various diets.We describe the immature stages of P. aequinoctialis and contrast them with those of S. jessoensis, which we also redescribed in part, here.Notes are also provided about structural attributes of the larvae of Brachinus in contrast to those of Pheropsophus and Stenaptinus.Although Brachinus has no conceived biocontrol importance, recent knowledge about the ecology and behavior of its species (Juliano 1983(Juliano , 1984(Juliano , 1985a(Juliano , b, 1986a, b, c;, b, c;Saska and Honek 2004) is useful for comparative purposes.

Materials and methods
A culture of the pest mole cricket Scapteriscus abbreviatus Scudder has been maintained by the University of Florida/Institute of Food and Agricultural Sciences' Mole Cricket Research Program since the 1980s.Th e stock was initially collected by pitfall traps in Broward County, Florida.Rearing methods are to be described by S.A. Wineriter, now with USDA-ARS, Gainesville, FL, who did much to develop them.Th is is an ideal mole cricket to rear because it is multivoltine, thus enabling production of eggs year-around.It may be reared without restriction in Florida because, although it is non-native, populations are established.Its shipment to other parts of the USA would need USDA-APHIS permit because it is a "plant pest" which is subject to restriction of interstate shipping.Maintenance is labor-intensive, but survival is high.As necessary for the work below, Sc. borellii Giglio-Tos, Sc. vicinus Scudder, and Neocurtilla hexadactyla (Perty) were captured in Alachua County, FL and reared by the same methods to produce eggs.Th ose species are all univoltine in northern Florida, so eggs are avail-able only for a few weeks of each year.Th eir survival in culture was poorer or, for N. hexadactyla, much poorer than for Sc.abbreviatus.
We initiated cultures of the house cricket Acheta domesticus (L.) (Orthoptera: Gryllidae) and the mealworm Tenebrio molitor (L.) (Coleoptera: Tenebrionidae).We obtained eggs of Gryllus sp.(Orthoptera: Gryllidae) from T. J. Walker, and eggs and larvae of Trichoplusia ni (Hübner) (Lepidoptera: Noctuidae) from a USDA-CMAVE culture.Th ese, and cucumber slices were off ered to the beetle larvae as alternative diets.Raisins and oatmeal, off ered to adult beetles, and also cucumber, were from a grocery store in Gainesville, FL.

Laboratory behavior of adults
A pair of wild-caught adult P. aequinoctialis was held in each of 10 plastic boxes, 31 cm L × 23 cm W × 10 cm H, fi lled to × 5 cm with moistened sand, initially sterile, from July through October 1986.Additional moisture was provided by deionized water-soaked cotton in a small Petri dish embedded in the sand surface.A triple thickness of non-sterile moist paper towel, 8 × 23.5 cm, was provided on the surface as a shelter.Five large Trichoplusia ni larvae were provided as food in each box.At daily check for 4 months, these larvae were counted; missing or dismembered larvae were noted and the number was increased again to fi ve living larvae, and any that had begun to spin cocoons were replaced.Dismembered larvae were removed.Daily for 10 days, the location of each adult beetle was noted as in the open on the surface, under a piece of paper towel, or in a self-constructed burrow.Pairs of S. jessoensis were treated simultaneously and identically.
First instars (planidia) of both species were sometimes noticed on the sand surface, so oviposition was occurring.Eggs could not be seen on the sand surface, so would have to be extracted from the sand by a fl otation method, or the beetles would have to be induced to oviposit on a more artifi cial substrate, before they could be documented.Remains of T. ni larvae attracted phorid fl ies.Phoretic organisms that had arrived with the adult beetles were not being suppressed.For these reasons we developed a more artifi cial and sterile handling method.
We developed a rearing method in which adults were housed in small groups on crumpled, moist, brown paper towel in 237 ml (8 fl .oz.) plastic "deli" cups with press-on lids.Some females were housed solitarily in 150 ml cups for some recording needs.Paper towel served as oviposition substrate.Once eggs had been removed, the paper towel was autoclaved together with any contaminating organisms.Survival of adults was good and eggs were readily found.Our routine removed these "egg papers" weekly (but daily for some recording needs) and transferred the beetles into transparent plastic boxes, 31 cm L × 23 cm W × 10 cm H, with fresh paper towel and food.As food, we provided T. molitor pupae, oatmeal, and raisins, all of which were observed to be fed upon.After a 2-day exposure to this food, beetles were placed once again in plastic "deli" cups, and the remaining contents of the feeding containers were autoclaved.Feeding containers and plastic cups were washed and dipped in ~ 5% bleach.Although the diet we developed was not perfected by trials, it was adequate because the adults survived well and normally produced many eggs each week.We wore eye protection when handling adult beetles because they are well able to aim their defensive spray toward human eyes.Our fi ngers became stained by their defensive secretions unless we wore gloves.

Oviposition
Eggs produced by field-caught females confined solitarily in 150 ml plastic cups with crumpled, moist paper towel were harvested daily from June 1986 through May 1987 for S. jessoensis.One female (no male) was in each cup.Each was removed from its cup once weekly for one day to another cup where it was confined with a large T. ni larva.At first the beetles were in a room with natural window light supplemented by overhead fluorescent lights only when people were working there.In January 1987, we were required to move them to another room with little illumination because of space shortage, but there we operated overhead fluorescent lights for 9 h/d.Temperature in the building was constant at 26°C.The frequency (eggs/female/day) was recorded during February 1987, for a total of 145 eggs observed.
Similar records for P. aequinoctialis likewise produced a frequency distribution, but the period of observation was continued until early April until 145 eggs had likewise been tabulated.Eggs were transferred by fi ne artist's paintbrush from "egg papers" to discs of brown paper towel, two layers, in small Petri dishes, ~ 5 cm diam.× 1 cm H. Th ese were examined daily and moistened with a fi ne spray of water from a wash bottle.Neonate larvae were transferred by fi ne artist's paintbrush to individual containers.We were surprised by the large number of eggs produced and by the large number of infertile eggs, which eventually molded or collapsed.

Egg fertility
For 10 days from 8 July 1986, the viability of the fi rst 10 eggs from each of 10 fi eldcollected S. jessoensis females was recorded.
We observed low fertility of eggs in S. jessoensis and P. aequinoctialis.Because the bacterial genus Wolbachia may cause cytoplasmic incompatibility in many insects (Werren 1997), we asked A. Jeyaprakash (see Jeyaprakash and Hoy 2000) to test our P. aequinoctialis for the presence of Wolbachia.On confi rmation of its presence, we tried to eliminate it from our laboratory culture in hope this would lead to increased fertility of eggs.In an eff ort to kill the Wolbachia, part of the culture was housed at 35°C for 24 hours, whereas the remaining part was left untreated.

Initiation of larval feeding
At fi rst, for both beetle species, we placed neonate larvae into small plastic Petri dishes (≈ 5 cm diameter × 1 cm height) stocked with mole cricket eggs on moist paper towel, in the expectation that mole cricket eggs might serve as diet.Survival was very poor: most larvae roamed for their entire life span (see below), frequently walking over the eggs, but did not feed, and then died.We off ered instead eggs of T. ni, pieces of T. ni larvae, eggs of Gryllus sp., and small pieces of cucumber.Th e few larvae that did begin to feed would almost invariably survive and develop, but only on a diet of mole cricket eggs.We enclosed the Petri dishes in aluminum foil to exclude light, to no avail.We fi lled the Petri dishes with sand except for a shallow central depression, to no avail.We tried using plenty (30) of mole cricket eggs from the outset because of a suggestion by Habu andSadanaga (1965, 1969) that larvae could recognize that small numbers were inadequate for their development, and refuse to feed, to no avail.We speculated that initiation of feeding relied upon dual cues of burrowing through sand and consequent arrival at mole cricket eggs, so we devised columns of sand of various depths up to 30 cm in Plexiglas® tubes over plastic chambers containing mole cricket eggs on paper towel.None of this improved initiation of feeding so it is not reported in detail.
Ultimately, we adopted a variant of the method used by Habu and Sadanaga (1969) for rearing S. jessoensis.Th ey used real and artifi cial mole cricket eggs chambers constructed with mud.We placed sand into a plastic vial (4 cm diameter × 6.5 cm height) to a depth of ~ 5 cm.An "artifi cial mole cricket egg chamber" was scooped from the sand.Mole cricket eggs (≥30) were placed into the chamber, and the top of the chamber was covered with broken pieces of wooden tongue depressors, which were covered by more sand.Th en, a neonate larva was dropped onto the sand surface.Usually, it then burrowed to the eggs, fed on them, and developed to the adult stage.Th e method worked well, but it denied us the ability to observe attack by the neonate larva on the eggs and subsequent development.Much later, by accident and after the culture of S. jessoensis had been terminated, we discovered that the egg chambers do not have to be covered to exclude light: many larvae will develop without this step.Th is allowed some observation of development of the larvae, although they had to be observed at the bottom of a pit ~ 2 cm deep; the small larvae were diffi cult to see among a pile of mole cricket eggs.

Larval and pupal development
By using records from individuals that survived when reared in plastic Petri dishes under daily observation, we compared development times of the F 1 immature stages of S. jessoensis and P. aequinoctialis when larvae were provided with a diet of mole cricket eggs.We obtained specimens of the developmental stages of P. aequinoctialis and S. jessoensis for taxonomic description.

Tests of larval prey specifi city
We compared survival of P. aequinoctialis on various diets, albeit initially under inadequate conditions, and later in pits in sand within vials.

Descriptions of immature stages
In most cases, setae are numbered on the left side of illustration and pores are lettered on the right side of illustration according to their ancestral positions (Bousquet and Goulet 1984); additional setae and pores are numbered and lettered sequentially beyond that presented in Bousquet and Goulet (1984), where appropriate.Habu andSadanaga (1965, 1969) were the fi rst to describe in detail the immature stages of Stenaptinus jessoensis (Morawitz), at about the same time Erwin (1967) described in detail the immature stages and way of life of the new world species Brachinus pallidus Erwin.Below, we will briefl y compare and contrast immature stages of Brachinus with our newly described immature stages of Pheropsophus aequinoctialis (Linné) and Stenaptinus jessoensis (Morawitz).

Field behavior of P. aequinoctialis adults
Notes provided by our collectors give hints on the habitat of adult P. aequinoctialis.All collectors agree with Reichardt (1971) that they are nocturnal and are most readily collected with the aid of a fl ashlight, while they are moving at night.In Brazil, they were seen at night on sandbars in Amazonian rivers (K.Zinner), running at night among clumps of grasses by an artifi cial pond providing water to cattle (J.H. Frank), in Uruguay, running on the soil surface (A.Silveira-Guido), in Bolivia, on a riverbank, often under driftwood or stranded dead fi sh during the day (F.D. Bennett and C.J. Pruett).In the western Amazon Basin, they are nocturnal on the alluvial and sandy banks of large rivers (Fig. 1) running together with the tiger beetles Phaeoxantha aequinoctialis (Dejean) and P. klugii (Chaudoir) and the galeritine carabid beetle Trichognathus marginipennis Latreille, all of which share similar coloration and color pattern, likely forming a Mullerian mimicry complex (Erwin 1991).

Parasites and phoretics of adults
Many of the fi eld-collected adult Pheropsophus were infested with nematodes, mites, and Laboulbeniales.Nematodes and mites were provided to specialists who told us they were non-pathogenic.Smart and Nguyen (1994) described a new species of Rhabditis (Nematoda: Rhabditidae), and H.A. Denmark (pers.comm.)identifi ed a large mite (Echinomegistus sp., Paramegistidae) from beetles from Potenji.Other mites remained unidentifi ed.Pinned adult beetles retain specimens of Laboulbeniales, which we will provide upon request to specialists.Use of the revised rearing methods suppressed these contaminants.

Laboratory behavior of adults
A direct contrast between the two species showed large diff erences in diel behavior.
Although it has been stated that adult P. aequinoctialis are nocturnal, this is not entirely true (Table 1).
Mean daily food consumption by pairs of P. aequinoctialis fell from 2.34 T. ni larvae in July to 1.23 in October.In comparison, that of S. jessoensis fell from 2.27 in July to 0.99 in October (Table 2).

Oviposition
Th e number of eggs produced per female by fi eld-caught S. jessoensis declined from June-July 1986, but by November had once more begun to increase, and in February-March 1987 was at least as high as it had been at the outset (Table 3).Th e 31 adult female S. jessoensis were of unknown age when received in June.About half of them survived at least a year.Average monthly oviposition by the surviving group had declined to 0 by September, but then it increased somewhat, and increased much more after hours of artifi cial lighting were increased in January, and by March was at least as great as it had been at the outset.Th e initial June-July oviposition matches the report (Habu andSadanaga 1965, 1969) of annual oviposition in those months, but the observed increase in oviposition beginning in November and peaking in February-March does not do so; perhaps the increase in illumination in January 1987 advanced it.We learned that oviposition is not confi ned to June-July.If there is one annual ovipositional peak as suggested by Habu andSadanaga (1965, 1969) and Li (1988), its timing changes under ambient conditions.
Despite constant laboratory conditions, the number of eggs laid per female per day varied from one to 31 (Fig. 2).
Th e major diff erence from the trial with S. jessoensis is that only fertile eggs, those from which larvae eventually hatched, were recorded.Many infertile eggs were produced but are not recorded.Th ese females were brought to observation from the southern hemisphere autumn at the end of April and were immediately exposed to a northern hemisphere daylight regime.Th en, the apparent peak of oviposition was in January.However, females laid fertile eggs during every months of the year.Th ey were of unknown age when recording began.Just as with S. jessoensis, the number of eggs laid daily by female P. aequinoctialis varied (Fig. 3 ).Although the number 5, and perhaps harmonics of it (10, 15, 20) in Fig. 3, and perhaps Fig. 2, has a high frequency, we can think of no biological explanation, and we assume this occurred by chance.Most eggs were laid singly, but some were clustered in groups.Group sizes ranged up to 13 for S. jessoensis, up to 17 for P. aequinoctialis; these group sizes, too, may occur by chance.

Egg fertility
Th e number of eggs produced (fecundity) fl uctuated widely.Furthermore, fertility of eggs was often low but fl uctuated widely.Fertility had no obvious relation to season, nor would we necessarily have expected a relationship to season because rearing was carried out under constant temperature and light.Fluctuations sometimes resulted in absence of hatchling larvae from our culture, but the longevity of the adults and their resumption of oviposition of fertile eggs prevented loss of the culture.Wolbachia bacteria were present in our P. aequinoctialis culture (A. Jeyaprakash, pers. comm.).Raising the temperature to a sublethal level has been known to eliminate Wolbachia from other insects (Werren 1997).We briefl y explored this possibility.Our attempt to improve the proportion of fertile eggs, by eliminating bacteria, by raising the ambient temperature of an incubator in which part of the culture was housed at 35°C for 24 hours was unsuccessful.Th is heat treatment of adult beetles resulted in total cessation of oviposition for several weeks.When they began to oviposit again, they still produced a large proportion of infertile eggs.
After oviposition, by day 8 the pigmented larval mandibles are visible through the thin chorion of viable eggs.

Initiation of larval feeding
Presentation to neonate larvae of eggs of T. ni, pieces of T. ni larvae, eggs of Gryllus sp., and small pieces of cucumber on paper towel in small Petri dishes seemed to elicit no feeding response except to pieces of cucumber.Neonate larvae were observed to imbibe liquid from cucumber, but then they blackened and died.Only mole cricket eggs elicited a feeding response, and only sometimes, that led to development of larvae to the pupal stage.Dozens of trials failed because not even the control treatment, mole cricket eggs, was successful.

Larval and pupal development times
Pheropsophus aequinoctialis had shorter development in instar I and longer in the pupal stage, and it consumed more prey eggs relative to S. jessoensis (Table 5).Eggs of all four mole cricket species were used as diet for S. jessoensis, and the species offered seemed to make no diff erence in development time.Survival of P. aequinoctialis was achieved only on eggs of N. hexadactyla and Sc.borellii, but this was because of the poor experimental conditions; subsequent routine rearing on Sc. abbreviatus eggs shows they are an adequate diet; again the specifi c identity of the eggs did not seem to infl uence development time.

Tests of larval prey specifi city
Initiation of feeding by neonate larvae was largely unsuccessful until arenas were changed from Petri dishes to artifi cial mole cricket egg chambers.
A count at 14 days showed no survivors on the diet of T. molitor pupae, but almost all of the latter decomposing; this suggests that the neonate P. aequinoctialis had injured the mealworm pupae.When, in an immediate add-on trial, 10 T. molitor pupae were placed into such cells without P. aequinoctialis larvae, eight survived to the adult stage; the other two molded, supporting that viewpoint.A count at 14 days showed no more than one survivor in each cell initially supplied with two larvae; this suggests fratricide, because at that point numerous prey eggs remained.
Tests were also conducted to detect whether P. aequinoctialis larvae, having developed to instar II on Sc. abbreviatus eggs, could be switched to P. molitor pupae and would develop.If successful, this could lead to reduced rearing costs.Twenty fi ve artifi cial egg chambers were constructed.Into each were placed 5 Sc.abbreviatus eggs and one neonate P. aequinoctialis larvae.After 5 days, 11 beetle larvae were alive in instar II, the uneaten mole cricket eggs in each chamber were removed and replaced with one T. molitor pupa.None of the beetle larvae survived to the adult stage.
Another set of tests used 30 artifi cial mole cricket egg chambers.Th irty Sc. abbreviatus were placed into each of 10, 100 Acheta domesticus eggs were placed into each of 10, and a tiny cube of cucumber weighing 0.2 g was placed into each of 10.Th e piece of cucumber was replaced at day 7.A check for surviving beetle larvae was made at day 14, at which time there were eight survivors, and all of them had been provided with Sc. abbreviatus eggs.Among the various diets provided, only mole cricket eggs proved adequate.

Optimization of diet
To optimize the number of mole cricket eggs required for larval development, we provided 24, 27, or 30 Sc. abbreviatus eggs, expecting that the lower numbers of eggs would aff ect the number of survivors and/or pupal weight, and this would determine an optimal diet in terms of success vs resources.Not all of the eggs provided were eaten by all of the survivors, indicating that at least a diet of 30 eggs is adequate.Although such diet (30 eggs) may not provide the fastest rate of growth or the largest pupae, it is adequate for development, and it conserves resources (mole cricket eggs).In Table 6, we found that 7 of 10 larvae survived when presented with 30 Sc. abbreviatus eggs.In the current test, 12 larvae (of 20) survived when presented with 30 eggs (6 of 10), or 15 (of 20) (7.5 of 10) survived when presented with 27 eggs -there is no signifi cant diff erence.However, when presented with only 24 eggs, only 7 of 20 (3.5 of 10) larvae survived.We were expecting reduced survival at reduced diet, and analyzed this as a 1-tailed χ 2 test with Yates' correction for small numbers, and found a signifi cant diff erence (7/20 vs 14/20, χ 2 = 3.61, df =1, P<0.05, 1-tailed).Th ere was a positive trend of eff ect of diet on resultant pupal weight.Th us, a diet of 24 eggs is suboptimal, and a diet of ≥ 30 eggs is better, at least in terms of resultant pupal weight, which may infl uence reproductive success of resultant  adults, and also in terms of survival.Th e standard length of an adult P. aequinoctialis in our culture was about 16.6 mm, whereas adults produced under restricted larval diet were as small as 10.4 mm.
Coloration.Mostly white color with slightly creamy-colored head capsule and apical abdominal segments; mandibles slightly darkened toward the tips.
Abdomen.Figs.41 -44.Tergite I (Fig. 41) with 2 stout setae medially and numerous shorter auxiliary setae in patches (not labeled), and no pores each side.Tergites II -VIII as in Tergite 1. Tergite IX with 2 stout setae each side and numerous shorter setae nearby.Tergite X (PY) with numerous short apical setae and no pores.Epipleurite IX (Fig. 44) with numerous long and stout setae on raised knob and no pores.Hypopleurite not defi ned.Segment VII and VIII sternites (Fig. 44) with numerous setae on raised knobs in rosettes each side and no pores.Segment IX sternite (Fig. 44) with subapical band of short setae each side and no pores.Segment X (PY) sternite (Fig. 35, 44) with numerous scattered short setae each side, no pores.
Legs.As in Fig. 35; reduced size and setation compared to instar I. PUPA.Form.(Fig. 45) Typical of carabid species.In addition, pygidium with fi ne short setae and dorsal surface with an array of small tubercules.
Notes on advancing an understanding of phylogenetic relationships.Much works still needs to be done in solving to infer the relationships between the Brachininae tribes Crepidogastrini and Brachinini and the subtribes Brachinina, Pheropsophina, and Masticina (Erwin 1970).Here we have added larval traits that will, in part, add information toward a more robust phylogenetic analysis in the future.Immature stages of Crepidogastrini and Masticina are, as yet unknown, and we do not even know whether they are ectoparasitoids, or specialized predators.Likely, they are one or the other, but on what taxa?Larvae of Brachinus develop in 5 instars; they also have 6 eyespots (as in other carabids), whereas Pheropsophina larvae have 3 instars and at most a single eye-spot, usually none.First instar Brachinus have no egg burster and chew their way out of the egg (Erwin 1967); Stenaptinus larvae possess a single-toothed egg burster; those of Pheropsopus have a triple-toothed egg burster.Pheropsophina larvae have pygidial hooks that aid them in attacking mole cricket egg clutches (Habu and Sadanga 1969), whereas Brachinus larvae do not.Larvae of Stenaptinus and Brachinus possess urogomphi whereas those of Pheropsophus do not.

Conclusion and discussion
Adult P. aequinoctialis and many S. jessoensis burrowed in sand-fi lled containers in the laboratory.Many S. jessoensis ~ 91.4%) but few P. aequinoctialis ~ 16.7%) were active on the surface in daylight.Th is supports fi eld observations that P. aequinoctialis adults are active nocturnally.Very few P. aequinoctialis adults (1.2%) found brown paper towel on the sand surface to be as adequate a refuge as their burrows; perhaps more solid objects (as a result of photoperception or thigmoperception) would have been more acceptable as refuges.In contrast, more S. jessoensis adults (78.7%) sheltered under paper towel in daylight than sheltered in burrows, but most did not shelter at all.Th ere is a clear contrast between the mainly diurnal behavior of S. jessoensis and the mainly nocturnal behavior of P. aequinoctalis.Timing of daily activity will have an eff ect on ability to fi nd food.
All diets presented to these adults were consumed, but we did not observe cannibalism by adults.Larvae of T. ni alone sustained adult S. jessoensis of unknown age for an average 12 months.Earlier authors showed that a broad diet of animal food is acceptable to them, and neither we nor previous authors tested acceptability of plant food alone.A diet of mealworm pupae, oatmeal and raisins was avidly fed upon by adult P. aequinoctialis, and sustained them well, so they will feed on plant food, supporting an observation of feeding upon palm fruits in nature (Reichardt 1971).We did not attempt to produce an optimal diet for adults of either species.Adults of the two species produced many eggs on the diets provided.Females of both species oviposited abundantly on crumpled, moist, brown paper towel under highly artifi cial conditions.Chemicals produced by eggs or adults of mole crickets are not necessary to stimulate abundant oviposition.However, Weed and Frank (2005) found that more eggs were laid in tunnels excavated by mole crickets than in artifi cial tunnels, suggesting that perhaps allomones produced by adult mole crickets are detected by female P. aequinoctialis and infl uence placement of eggs.
Adult P. aequinoctalis oviposited in all months of the year; and adult S. jessoensis oviposited in most months of the year in the laboratory.Seasonality of oviposition in the fi eld is mutable under laboratory conditions.We suspect that neonate larvae of both species suff er high mortality because they fail to detect suitable prey and thus die.Although P. aequinoctialis and S. jessoensis are highly fecund, Th iele (1977) gave examples of high fecundity among other carabids without such a specialized life cycle.Fertility of laboratory-produced P. aequinoctialis eggs varied for unknown reasons, at some times being very low.Th e evolutionary consideration is: Why are so many infertile eggs produced?Presence of two species of Wolbachia (Bacteria: Rickettsiae) in our 1992 P. aequinoctialis stock from Bolivia has been demonstrated.A suggested heat-treatment to eliminate the Wolbachia resulted in mortality of some adults, temporarily reduced oviposition, and failed to eliminate production of infertile eggs.Th e heat treatment may, of course, have killed some essential fl ora in the digestive system.Incubation of fertile eggs of S. jessoensis took 11.4 days and of P. aequinoctialis 13.5 days on average.
Most neonate larvae of S. jessoensis and P. aequinoctialis died when presented with mole cricket eggs in Petri dishes.Th ey wandered for days until they died, almost continually in motion.Neonate larvae were presented with alternative diets including A. domesticus and Gryllus sp.eggs, eggs and pieces of larvae of T. ni, intact pupae of T. molitor, and pieces of cucumber.All neonate larvae of both species died when presented with any diet other than mole cricket eggs although imbibition of fl uid was observed from pieces of cucumber.However, a few began to feed on mole cricket eggs.Th ose eggs were of Scapteriscus abbreviatus, Sc. borellii, Sc. vicinus, and N. hexadactyla; however, replication was inadequate to determine any diff erences in survival success between these mole cricket eggs diets.At least it can be stated that S. jessoensis can survive on mole cricket eggs other than those of Gryllotalpa, in contrast to unsupported claims by Habu andSadanaga (1965, 1969).Once neonate larvae began to feed, their survival to the adult stage on the same diet was highly probable.Perhaps larvae will not begin to feed until they encounter enough eggs to complete their development (Habu andSadanaga 1965, 1969), but we have no data to support this claim.Faced with the impasse that neonate larvae would seldom develop on a diet of mole cricket eggs in a Petri dish, even when that dish was enclosed totally with aluminum foil to exclude light, we adopted a variant of the rearing method proposed for S. jessoensis by Habu and Sadanaga (1969).Using this method, an artifi cial mole cricket egg chamber is made in sand in a plastic vial, stocked with 30 mole cricket eggs, covered with sand, and a neonate larva is dropped onto the sand surface.Th e larva burrows down to enter the chamber and begins feeding on eggs.Th is resulted in high survival of larvae and pupae to the adult stage, and became our standard rearing method.Feeding and development seldom occurred in the more artifi cial conditions of a small Petri dish with mole cricket eggs piled onto a disc of paper towel, even in the dark, but we ran many feeding trials under those circumstances.Unfortunately, the method of an artifi cial egg chamber excluded frequent observation.Much later, we found by accident that the artifi cial egg chamber did not need covering with sand to exclude light.Th en, we conducted more feeding trials and confi rmed that Acheta domesticus eggs, Tenebrio molitor pupae, and pieces of cucumber are not acceptable diets.Adult P. aequinoctialis are scavengers and generalist predators.Larvae, however, so far as determined, are specialist predators on mole cricket eggs.Th ey can develop under laboratory conditions on a diet containing only eggs of Scapteriscus abbreviatus, Sc. borellii, Sc. vicinus, or Neocurtilla hexadactyla but none survived using any other diet tried.Proof of restriction of the larval diet still is inadequate.
Each larva of the carabid genus Brachinus (Neobrachinus) feeds on only one water beetle pupa and is an ectoparasitoid (Erwin 1967(Erwin , 1979) ) replacing its host in a small mud chamber constructed by the water beetle larva.In Europe, Brachinus s. str.larvae feed on the pupal stage of the carabid genus Amara (Saska and Honek 2004).However, each larva of P. aequinoctialis and S. jessoensis requires tens of mole cricket eggs as food to complete its development.Such behavior is more aptly termed predation (Van Driesche and Bellows 1996, p. 21 citing many earlier authors), so we consider Pheropsophus and Stenaptinus larvae to be specialist predators.At no time in our laboratory cultures did more than one P. aequinoctialis or S. jessoensis larva survive long on a single cache of eggs, so we believe they practice fratricide as has been noted in Brachinus (Juliano 1984).Erwin (1967) noted that fratricide did not occur in Brachinus pallidus, rather the fi rst larva that began feeding became the "owner" of the pupa and the other larvae departed in search of another pupa.Th ese larvae, when off ered a fresh pupa, developed to the adult stage.

Implications for biological control
Th e studies reported above were initiated because of a suspicion by T.L. Erwin that Pheropsophus spp.larvae might, as had been reported for Stenaptinus, develop only on a diet of mole cricket eggs.It was eggs of invasive species of the South American mole cricket genus Scapteriscus in the southern USA that were the target of our studies.Th ese studies were initiated by J.H. Frank in the name of the University of Florida/ Institute of Food and Agricultural Sciences ' Mole Cricket Research Program (Walker 1985;Frank and Walker 2006).Early mention of the studies was made by Hudson et al. (1987).Prey specifi city of these beetle larvae was important because the native North America mole cricket Neocurtilla hexadactyla was not a target.Might the South American Pheropsophus be adapted to South American Scapteriscus mole crickets but the Old World Stenaptinus be adapted to the largely Old World genus Gryllotalpa?Habu and Sadanaga (1965) stated that S. jessoensis larvae feed only on eggs of G. africana Palisot de Beauvois.Th ey provided no evidence that they had experimented with other diets.However, G. africana does not occur in Asia, and the species of mole cricket encountered by those authors may have been G. orientalis Burmeister (Townsend 1983).We had no access to eggs of Gryllotalpa.By 1987, we had found that larvae of S. jessoensis and P. aequinoctialis would develop on a diet of eggs of Neocurtilla (Gryllotalpinae) or Scapteriscus (Scapteriscinae) mole crickets but, because of diffi culties in getting neonate larvae to initiate feeding, we had conducted scores of failed trials with these and other diets.
Little that we had studied pointed to need for chemoperception.Adults laid eggs abundantly on paper towels.Neonate larvae may have used chemoperception to detect that mole cricket eggs are food, but there was no evidence that such detection occurred except in a pit in sand.
Th e Mole Cricket Research Program then concentrated on other biological control agents, which were successful, until its funding was 'unearmarked' in 1991 (Frank and Parkman 1999).At this devastating event, to save expenses and because S. jessoensis clearly could not be a specialist of Scapteriscus mole crickets, cultures of both species were terminated.
A culture of P. aequinoctialis was reinitiated with stock from Bolivia in 1992.One reason was that an additional biological control agent that could be used in the vicinity of water bodies, on their banks in particular, could be benefi cial in integrated pest management because application of chemical pesticides is prohibited from use in such habitats.A second reason is because of egg-guarding behavior by female Neocurtilla hexadactyla.Th ese excavate two side-by-side underground cells, one of which receives the eggs, the other serves as a resting site for the female, from which she emerges fom time to time to tend the eggs (J.H.Frank and R.C. Hemenway, obs.).In contrast, each Scapteriscus spp.female excavates only one cell and then, after oviposition of a clutch of eggs, leaves and blocks the entrance to the cell (J.H.Frank and R.C. Hemenway, obs.).It might therefore be possible for female N. hexadactyla to detect and kill intruding bombardier beetle larvae.If this can be demonstrated in the laboratory, it might justify release of P. aequinoctialis in Florida.
Research will not be complete until the subject of egg-guarding by N. hexadactyla females is adequately investigated.A major problem is that we have not devised a robust method for culturing N. hexadactyla.Survival of adults and nymphs was poor, perhaps because the diet we used was inadequate.We observed that females move their eggs when they are disturbed, which we believe to be a previously unreported facet of their presocial behavior.
Finally, some objection might be made to the release of a beetle whose adults are scavengers and generalist predators, even though this habit is shared with adults of many other insects, including adults of the ~ 18 native species of Brachinus bombardier beetles in Florida.Still, population sizes must be limited by availability of mole cricket eggs, and we now have some idea of the quantity of food (≤ 2.3 large T. ni larvae per day) consumed by pairs of adult beetles.

Figure 3 .
Figure 3. Frequency of numbers of eggs laid daily by P. aequinoctialis females in February-April 1987 (Σ observations = 145 excluding records of zero).No. of eggs per female P. aequinoctialis per day

Figure 2 .
Figure 2. Frequency of numbers of eggs laid daily by S. jessoensis females in February 1987 (Σ observations = 145 excluding records of zero).

Table 1 .
Dispersion of adult P. aequinoctialis and S. jessoensis in sand-fi lled boxes observed daily in late morning averaged over 10 consecutive days.Twenty boxes each contained a pair of wild-caught beetles of one of the two species.Th e 20 adults of each species were recorded as being (a) in the open, (b) sheltering under a triple thickness of paper towel, or (c) by default, in a self-constructed burrow.SD = standard deviation of mean.

Table 2 .
Mean daily consumption of large T. ni larvae by pairs of wild-caught adult beetles over the months July-October 1986, based on 10 pairs of each species housed by pair in sand-fi lled boxes.SD = standard deviation of mean.

Table 3 .
Numbers of eggs laid monthly by wild-caught S. jessoensis females from June 1986 to May 1987.N = number of surviving S. jessoensis females at end of month, mean = mean number of eggs laid by survivors, SD = standard deviation of mean.

Table 5 .
Development times in days of immature S. jessoensis and P. aequinoctialis when F 1 neonate larvae were provided with eggs of Neocurtilla hexadactyla, Scapteriscus abbreviatus, Sc. borellii, or Sc.vicinus at 26 o C. TL = total larval period, TD = Total duration of immature stages, Food = number of mole cricket eggs consumed, SD = standard deviation of mean.

Table 4 .
Numbers of eggs laid monthly by 10 wild-caught P. aequinoctialis females from late April 1987 to March 1988.Th ey were housed solitarily in plastic cups with crumpled paper towel, and were given access to one large T. ni larva per week as prey.N = number of surviving P. aequinoctialis females at end of month, mean = mean number of fertile eggs laid by survivors, SD = standard deviation of mean.

Table 6 .
Survival to adult stage of P. aequinoctialis when neonate larvae were provided with a diet of 30 Sc. abbreviatus eggs or one T. molitor pupa in an artifi cial mole cricket egg chamber.

Table 7 .
Numbers surviving (out of 20 neonate larvae) and pupal weights of P. aequinoctialis when provided with 24, 27, or 30 Sc. abbreviatus eggs.SD = standard deviation of mean.