VARIATION IN CELL NUMBER IN A POPULATION OF BEMBIX AMERICANA (HYMENOPTERA: SPHECIDAE) *

This paper reports the results o observations made during the summers of I97 and I972 on a small group of Bembix americana Fabricius nesting in a vacant lot in Seattle, Washington. This population was judged to be intermediate to comata Parker and spinolae Lepeletier at the subspecific level (H. E. Evans, pers. comm.). Probably no North American digger wasp has been more thor-oughly studied than B. americana (Evans I957, I966; Evans and Matthews 1968). However there has only been one report of this species making nests with more than a single cell (Parker I925). Evans (I966) notes the uniqueness of this account and the difficulty of accurately digging out nests in dense colonies. Nevertheless in the course of my work I discovered that some of the wasps i.n Seattle were making multi-celled nests. The evidence for this is presented here as well as data on the nest building behavior of marked females over a period of time.

Probably no North American digger wasp has been more thoroughly studied than B. americana (Evans I957, I966;Evans and Matthews 1968). However there has only been one report of this species making nests with more than a single cell (Parker I925). Evans (I966) notes the uniqueness of this account and the difficulty of accurately digging out nests in dense colonies. Nevertheless in the course of my work I discovered that some of the wasps i.n Seattle were making multi-celled nests. The evidence for this is presented here as well as data on the nest building behavior of marked females over a period of time.

THE NESTING AREA
Eleven species of sphecids and pompilids nested in a 13 rn section of an old path running through a grassy area in a vacant lot in Seattle. The lot, which was largely covered with blackberry bushes (Rubus laciniatum), bordered Lake Washington and was within sight ot: the University of Washington. The nesting area was relatively open and somewhat less pebbly than other parts o.f the path.
However, only a few places were truly sandy and most nests travelled in part or in whole through hard-packed gravelly earth with pebbles up to 5 cm in diameter present.
The 2 most abundant species in the aggregation were Philanthus crabroniformis Smith and B. americana; at the peak of the nesting season there may have been as many as 40 americana nests. In the summer of I97I a group of nests were marked and closely watched for periods of I-3 hr almost every day between July 24 and August *Manuscript receie;ed by the editor September .0, 1972September .0, I58 1972 dlcock Bembix americana 159 14. The duration o activity at a burrow and the excavation o selected nests showed that some t;emales constructed burrows with I,.

Table summarizes the results o observation and excavation in
1971. Two points should be made here. (I) In many cases activity at a nest was not seen until the emale had begun provisioning a ceil. This means that the over-all length o activity at some nests is underestimated by 2 or 3 days. (2) Throughout the observation period weather conditions were such that emales were able to work or a period o 2 wk or more almost without interruption because o rain.
A variety o nests are illustrated in Fig. . Judging rom the 2 cases in which a third cell was ound containing a fly and egg, emales building three-celled nests first dug a more or less straight burrow leading to an initial cell. Ater completion o the provisioning o the first cell, a second one was added by digging a side burrow of{ at a right angle (more or less) rom the main burrow.
Ater this cell was dug and ully provisioned, a third burrow was constructed off the second cell burrow. This shaft circled back toward or under the main tunnel with the cell located well under the entrance to the nest. Figure 1. A view from above of a selected sample of nests. The cells are numbered in the order in which they were built (at the time of excavation cell held a cocoon, cell 2 contained a larva, and cell 3 held a fly with an egg).

.d lcock Bembix americana 161
The fact that wasps were building I, 2 or 3 celled nests raised the question, could an individual female construct more than one type o nest or was each wasp rigidly programmed to make either I or 2 or 3 celled nests ?
In order to answer this question, a group of i6 wasps were marked on the thorax with a dot or dots of acrylic-vinyl paint a]ter they had brought a prey to a nest. Each was given a distinctive color combination. Twelve o] those marked completed one or more nests rom the time o marking (July I4-2I, I972) to August 5 when I left the Seattle area. Two wasps (E, H) that definitely made a single celled nest initially built a multi-celled nest the second time. For 3 others (I, K, L) this is probably the case. One emale (F) that made 2 single celled nests consecutively almost surely built a double celled nest the third time around. Only wasp definitely built a second nest with fewer cells than the first o.ne (C).
The pattern o building single celled nests first and multicelled nests later is supported by indirect evidence from the 197I data.
The first year, observations o americana were not begun until later in the season (July 24 as opposed to July I4 in I972). Less than I/4 o the I97I nests excavated were single-celled as opposed to 3/4 of the initial nests in I972. I strongly suspect that most of the burrows excavated in I97I were the second or third nests o the wasps, whereas in I972 I made an effort to mark an individual as soon as it was discovered at a nest.

DISCUSSION
This study shows that the emales in at least one population o. B. americana are capable of building either single, or multi-celled nests. It suggests moreover that emales tend first to build celled nests and that subsequent nests are likely to. have more than cell.
Variation in cell number is not unknown in Bembix occurring in amoena, belfragei, cinerea, and niponica (Evans 1966, p. 277, 317, 352). This is only the second report of such variation for the abundant, widespread and well-studied americana suggesting that Ecological Factors and Nest Design. The significance of variation in cell number for digger wasps has not been discussed in great detail. Presumably there are both advantages and disadvantages associated with the construction of either multi-celled or single celled nests. A nest with 3 cells may require the removal of less material in building than would 3 ,separate uni-celled nests. However the advantage o,f reduced energy expenditure in nest building may be tempered by (I) the disadvantage of prolonged activity at one nest entrance which may make the individual's larvae and prey more vulnerable to parasites, especially if the latter are ,capable of learning to exploit a specific source of prey and by (2) the disadvantage stemming from the possibility of a highly localized disaster which would eliminate all the offspring in a multi-celled nest before they could emerge (e.g. poor drainage at a particular site). I these speculations are correct one would expect multi-celled nests in areas where soil conditions or other factors make nest construction particularly expensive in terms of time and energy costs.
Evans notes (1966, p. 442) that nests with more than cell are more characteristic of species digging in compact soil than those which build in sandy soils. Judging from Evans' account (I957) of typical nesting sites for americana the Seattle colony was located in an atypically hardpacked and pebbly area. These conditions may 197:] dlcock Bembix americana I63 avor individuals that build multi-celled nests and thereby avoid having to dig through a stony compacted surface repeatedly. However the population studied by Parker nested in sand dunes where this argument cannot apply. Parker was deeply impressed by the powerful winds that regularly swept the area oten blowing emales away rom their nest entrances. Perhaps with these conditions, selecting a new site and attempting to initiate digging would be both extremely time consuming and energetically demanding. This remains conjecture.
It should be noted moreover that this hypothesis is weakened by the act that those Bembix that dig very long (up to m) burrows deep into sand dunes (e.g. occidental# and pruinosa) do not construct multiple celled nests. Although these are large and powerful species, nonetheless the costs o nest building must be great. It is ditZficult to understand why one long burrow could not accommodate several cells.
Nevertheless, assuming that there is some imperfect relationship between the expense o nest building and multiple cells, one would expect intraspecic variation in nest design to occur in species which are most flexible in the selection o nesting areas. Most North American Bembix are relatively restricted to one soil type (Evans I966,p. 352). However, amoena and americana are quite variable in nesting ecology while the requirements o the other species that make both single and multi-celled nests are uncertain.
As a (inal question, why should the emales I watched have tended to build a single celled nest ([rst and then a multi-celled one? Perhaps such a strategy insures that i a wasp is able to provision only 2 cells at least her offspring will be in 2 separate locations. But i she is able to provision more, it may then be to her advantage to reduce building costs by constructing side burrows and additional cells and to expend her time and energy in oraging and pro.visioning. Probably the middle o. the nesting season coincides with the peak abundance o prey and the best weather conditions. Thus it may be especially important not to spend this time hunting or and digging a new nest, time that could be spent provisioning cells before the onset o the all rains in Seattle (which oten begin by mid-August). Thus the nest building behavior o a emale americana in Seattle may reflect conflicting pressures to avoid placing all her eggs in one nest while at the same time making the most o the short time available to her to collect prey. Psyche September UMMARY Females o a population o Bembix americana built complete nests with , , or 3 ceils at a site in Seattle, Washington. This is only the second report o variation in cell number or this well-studied species. Marked individuals generally built single-celled nests at the start o the season with subsequent nests likely to be multicelled. Speculations on the adaptive significance o multi-celled nests and the pattern o nest building observed in Seattle are' presented. ACKNOWLEDGMENTS I am especially grateful to Dr. Howard E. Evans fo,r his kindness in identifying the wasps, suggesting ideas for research, and reading an earlier draft of this paper. Virginia Anderson made a number of useful observations at the nest sites after I left the Seattle area and I thank her for her help. This work was done while the author was supported by NSF grants GB-2874X, GB-28714XI, and GB-35269.