Spider prey ( Araneae ) of Trypoxylon ( Trypargilum ) rogenhoferi ( Hymenoptera : Sphecidae ) in southeastern Brazil

Fifty five nests and 216 cells of Trypoxylon (Trypargilum) rogenhoferi were obtained from trap-nests (cut bamboo canes) in Santa Carlota Farm (Itaoca Section = IS and Santana Section = SS), Cajuru and on the São Paulo University Campus, Ribeirão Preto (= RP), both in the State of São Paulo, Brazil (Sept. 1993-Oct. 1995). The prey (spiders) of 40 cells from IS, 58 from SS and 39 from RP were identified. The greatest nesting frequency occurred during the hot and wet season (September to April). T. rogenhoferi preyed upon individuals of five spider families, with Araneidae (orb-weaver spiders) being the most frequent (99.6%). Alpaida aff. negro (58%) was the most frequently collected species in IS, followed by A. alto (24.8%); in SS (59.6%) and RP (64.7%) the most frequent species was A. veniliae, followed in SS by A. aff. negro (14.9%) and in RP by A. leucogramma (13.5%). The size of reproductive niches, H' = 1.25 (IS), H' = 1.30 (SS) and H' = 1.29 (RP) were not significantly different. There was a positive correlation between the reproductive niche width (H') and evenness.

One important species characteristic in Sphecidae is the species of prey that females captured.This i8 important froID botb evolu tionary and taxonomic perspectives.These wasps collect prey in places where people do not normal1y look (Genaro et al. 1989); thus, nests of these wasps can provide vast collec tions of spiders, including rarely collected species (Coville 1987).
The purpose of this paper is to present data on prey speciescollected by Trypoxylon (Trypargilum) rogenhoferi Kohl, 1884.In addition observations obtained from trap-nests placed in three s .ites.ipsoutheastem Brazil are presented to reveal sorne parameters of their reproductive niches.
Two distinct seasons are recognízed in this regíon: 1-Cold and dry season (May to August) with• monthly average temperatures ranging from 18.5 to 22.2 oC (SCF) and from 17 to 23 oC (RP), precipitation from O to 84.4 mm (SCF) and from O to 82.9 mm (RP); 2-Hot and wet season (September to April) with monthly average temperatures rangíng from 23 to 28.8 oC (SCF) and from 22.4 to 27.2 oC (RP), precipitation from O to 422.6 mm (SCF) and from 1.4 to 409.4 mm (RP) (Fig. 1).
Trap-nests: Bamboo stems (B) were used as trap-nests (TN) with a nodal septum closing one end.The internal diameter of nests ranged from 7 to 25 mm, and nest lengths ranged from 70 to 250 mm.These TN (450 in RP, 450 ín SS and 200 in IS) were tied together in batches of six to eight units and placed on the shelves of shelters built at the study locations.
The TN were inspected at least once a month during the period from September/93 to August/95 at SCF and from November/93 to October/95 in RP.During the inspections, which were performed wi th the help of an oto scope, the nests that had been recently complet ed or were being provisioned were collected and transported 10 the laboratory.Each nest was replaced by a new TN.The nests were opened in the laboratory and the prey (spiders) from cells, which contained eggs, or recently hatched lar vae were removed and preserved in 80% alco hol.At least one ímmature wasp was left in all of these nests so that it could develop fuUy, thus allowing fOÍ" determinatíon of the nesting species.AH the wasp and spider specimens were deposited in the Entomology Collection of Biology Department-FFCLRP-USP and a por tion of the spiders samples was deposited in the arachnological collection of Butantan Institute.
Data analysis: The reproductive niche (following Á lvarez et al. 1988) width was cal culated using the number of prey species as well as the Shannon-Weaver Diversity Index H ' = -l: P h' In Ph' where P h is the proportion of individuals belonging to the ht h species in the total sample (Pielou 1975).
The evenness index was calculated according 10 Pielou (1966): 1'= H' 1 Hmax, where H' is the Shannon-Weaver Index and Hmax is the logarithm (ln) of the total number of prey species in the sample.This index varies from zero to one.
The similarity in species composition among the studied habitats was calculated by the Sfirensen Similarity Quotient (Sfirensen 1948): S.Q.= 2JI a+b, where a and b are the numbers of families, genera and species of col lected prey in each of the habitats and J is the number of families, genera and species of col lected prey cornmon to all samples.The simi larity among the studied cornmunities, consid ering the number of collected spiders of each sampled species, was calculated by the method of percentage similarity (Hanski and Koskela 1977) : PS ¡ j = l: min.(P ¡ h ' P jh)' where P ¡h and Pjh are proportions of the h families, genera or species of collected prey.The "Test for differ ence between two indices" was used in the sta tistical analysis of indices (Zar 1984).

Number oC nests, cells and seasonal
abundance: 216 cells of T. rogenhoferi were obtained, 69 were from IS (16 nests), 84 from SS (19 nests) and 63 from RP (20 nests).Prey was collected and identified from 40, 58 and 39 cells, respectively.The nests (19 completed ones and 36 being provisioned) were co Í lected throughout the year, except in April, with the greatest nest ing frequency occurring in the hot and w e t sea son (September 10 April), during which 85.4% of the sampled nests were collected.The months with the greatest nesting frequencies were September and Novembet in IS, January and March in SS and December in RP (Fig. 2).
CoUected prey: F ive families of spiders were found with Araneidae being the most rr e quent in all of the sites in number of genera .(61.5%), species (81.5%) and individuals (99.4%).Only Araneidae was found at all three �ites while Salticidae was found in two (SS and RP), Anyphaenidae and Tetragnathidae were only collected in IS and Theridiidae was col lected only in SS (Table 1).
Araneidae was the most frequent farnily in the samples.This suggests a strong preference (99.6%) of T. rogenhoferi for species of orb weaver spiders, Alpaida (88.4%) was the most frequen t genus (Table 1).
T. rogenhoferi captured spider juveniles, males and females of seven prey species, only juveniles of eight, only males of two and only females of six species.Juveniles and females were collected from four species and males and females from one species (Table 1).Juveniles, males and females were collected during nine months of the year, while in 8r--------------------------  October only females were present in the sam pies.In September males we r e not found.The percent of females was greater than that of juveniles in August t hrough November, while in the others months, the percentage of juve niles was greater than that of females.The per- centage of both, juveniles and females were greater than that of males.The greatest fre quencies of juveniles occurred in May (95.4% of the collected spiders), June (84.9%)and July (94.4%); for males the greatest frequency was found in February (7.8%) and August (5%) ' a nd for females in September (90.3%),October (100%) and November (85.8%) (Fig. 3).In all locations the juveniles correspond to 59.6%, females to 37.5% and males to 2.9%. the most ftequent species in January, M ay and December, A. aff .negro (collected during four months) in August and October, with A. alto (collected in three months) in March (Fig. 5).In RP, A. veniliae (collected during seven months) was preferred in January through March, June, July, November and December, with A. alto (collected in one month) in August (Fig. 6).
Parameters oC tbe reproductive niche: The monthly reproductive niche width calculated by the number of prey species collected varied from one (February in IS, August in SS and   March in RP) to 12 (December in RP), while the total niche width was 13 (IS) and 16 (SS and RP).The monthly reproductive niche width calculated by the Shannon-Weaver Index (H') varied from 0.12 (August in RP) to 1.68 (January in IS) (Table 2).The total repro ductive niche widths, H'= 1.25 (IS), H'= 1.30 (SS) and H'= 1,29 (RP) were not significantly different (P>0.05).The correlation between the indices (number of collected species and H') was positive and statistically significant (r= 0.73, P<O.Ol).
The monthly evenness of the collections ranged from 0.13 (January in SS) to 0.85 (November in RP), and the total evenness from 0.46 (RP) to 0.48 (IS).The correlation between the monthly reproductive niche width calculat ed using the number of collected prey species and evenness (1') was positive and non-signif icant (r= 0.18, 1»0.05), while the correlation between H' and l' was positive and significant (r= 0.15, P<O.OI).
The greatest similarity value at the family level among the sites was observed when com paring SS to RP (SQ= 0.80).At the genus level, the values were similar, and at the species level, the value was greatest when comparing SS to RP (SQ= 0.75).The percent age similarity was markedly high at the family and genus level, moderate at the species level when comparing SS to RP and low when com paring SS to IS and RP to IS (Table 3).

DISCUSSION
The great preference of T. rogenhoferi for prey of the Araneidae family was also observed in the same three habitats by Camillo and Brescovit (1999a) in relation to T. lacti tarse Saussure, 1867 (96.6%).In these habitats Camillo et al. (1994) observed in 14 T. rogen hoferi cells that 99.4% of prey were fr om the family Araneidae.In the Central Amazon, Garcia and Adis (1995) observed that the females of T. rogenhoferi captured one species of this spider family.
T. rogenhoferi captured more juvenile prey (50.3%) than adult females (34.3%) and males (15.4%).The same fact was observed by Carnillo and Brescovit (1999a) for T. lactitarse but in different percentages (59.6%, 37.5% and 2.9%, respectively).Collection oflarge percent age of juveniles was also observedfor T. texense The reproductive niche widths (H') of T. rogenhoferi in different localities were not sig nificantly different.But when comparing reproductive niche widths with those of T. lac titarse (Camillo and Brescovit 1999a) (H'=2.24 in IS, H' = 2.72 in SS and H'=1.76 in RP) they were smaller.Significant differences were also observed in relation to evenness (1') which was smaller for T. rogenhoferi.
Reproductive niche overlap between T. rogenhoferi and T. lactitarse (Camillo and Brescovit 1999a), calculated using the percent age similarity of Hanski and Koskela (1977) was very low, 0.16 for SI, 0.06 for SS and 0.04 for RP.These results indicate that these species do not compete in any significant manner for food in the three habitats studied, a fact that is easily observed when comparing the most common prey species.
An estimate of the weight of each prey collected was obtained by dividing the total biomass of the cell group by the total number of prey.Thus, we calculated that the prey col lected by T. lactitarse had a lower biomass (22.8 mg, n=1241) than those collected by T. rogenhoferi (33.3 mg, n::;1407).This differ ence is more evident when we consider only the most common prey species.In this case, we observe that the mean individual weight was 48.5 mg (n=108) for A aff .negro and 26.6 mg (n=212) for A veniliae, values greater than 14.6 mg (n=93), the weight calculated for Eustala spl which was collected by T. lacti tarse.The difference in mean weight between Aaff .negro andA veniliae is related to the dif ferent number of prey provisioned per cell (6.7 and 13.2, respectively).
T. rogenhoferi collected heavier prey than T. lactitarse wich is directly r elated to the size of these species.According to Camillo and Brescovit (1999b), T. rogenhoferi females are larger than T. lactitarse, and since the transport of prey is carried out in flight, larger individu als are able to carry larger prey.This result clearly shows why there is no overlap of the reproductive niches.

genus
Trypargilum, are restricted to the Western Hemisphere from Canada to Argentina.The Neotropical regíon contains most of the species (Bohart and Menke 1976, Coville 1982).These wasps are solitary.Each female constructs and provisions the nest with paralyzed spiders.Sorne species construct mud nests, whereas others utilize pre-existing cavi tieso Nests consíst of linear series of cells that are subdivided by mud partitions, usually fol lowed by an empty vestibular (!ell and an end plug made of mud (Bohart and Menke 1976, Coville and CovilIe 1980, Coville 1982).

Fig. 3 .
Fig. 3. Monthly percent of juveniles, males and females of the spider prey hunted by Trypoxylon rogenhoferi in the three sites.

Fig. 4 .
Fig. 4. Monthly percent of prey species most frequently collected by Trypoxylon rogenhoferi in Itaoca Section.
The total does not correspond to the surn of tbe spiders prey collected rnonthly because sorne species are repeated in other rnonths.H': Shannon-Weaver index, 1'= Evenness index