Reprint Series13 April 1979, Volume 204, pp. 207-209 S C I E N C E

Honey Caches Help Female Paper Wasps (Pdk~e~ anndark)Survive Texas Winters

Joan E. Strassman

Copyright @ 1979 by the American Association for the Advancement of Science

Honey Caches Help Female Paper Wasps (f%Zktes annularis)

Survive Texas Winters

SCIENCE, VQL. 204, 13

Abstract. Polistes annularis females store honey in their nests in autumn. Theyreturn to their nests on warm winter days, eat honey, and defend it from non-sisters.Honey deprivation decreases numbers surviving the winter; females that do survivewithout honey build smaller spring nests.

Polistes annularis exhibits a newlydiscovered behavior, winter honey cach-ing. Previously, only large perennial col-onies of social insects have been report-ed to make and store honey for use asfood during periods of drought or coldweather (1, 2). Females of P. annularisabandon their nests after caching winterstores of honey and retreat to hibernatein more protected places. On warm win-ter days females leave hibernacula, re-turn to natal nests, and feed on their hon-ey. At this time they will defend itagainst non-sisters. Dependence on hon-ey stored overwinter has a decisive im-pact on the social biology of P. annul-aris, for it necessitates proximity of hi-bernacula to nests and continued contactand cooperation between sisters. This fa-cilitates springtime recognition of sisterswho cooperate in building new nestsnear the natal nest.

Although winter honey caching is pre-viously unreported in polistine wasps,honey manufacture and storage of smalldroplets in the nest in cells occupied byeggs or small larvae is common (2-5).Honey droplets have been seen in nestsat all seasons, as long as the wasps areon the nests.

Winter honey caching was observed ina population of P. annularis along a 15-mhigh limestone cliff overlooking a reser-voir, Lake Travis, 26 miles west of Aus-tin, Texas. Nests were extraordinarilycommon along this particular west-fac-ing cliff; more than 1000 nests werefound along a 200-m section. In 1976, 40nests were observed, and all autumn re-productive males and females weremarked with enamel (6). Reproductivefemales were distinguished from workersby their lack of wing wear (7). Honeyfirst began to appear in quantity in emptycells in September. By November, all

nests still occupied had some honey, andmost had all cell walls coated with athick layer of honey similar in taste tothat of honeybees, but much more vis-cous. In November, females began en-tering a crack in the cliff and abandonedmud cliff swallow nests, searching forplaces to hibernate. When nights werecooler than 5’C, wasps spent them in hi-bernacula, returning to their nests onsunny warm days. On the nest they fedon honey and repelled all intruders notmarked as sisters. Commonest among in-truders were other wasps from the samepopulation, often bearing marks in-dicating that they were born on othernests. These were fought off vigorouslyand chased (8). In addition, other insectstried to steal honey. Several Vespula sp.were successful, although they werechased off immediately ‘when discov-ered. Three low nests were knockeddown by a raccoon that shredded them inthe process of consuming honey.

Only a small amount of honey loss wasendured by P. annularis and several fac-tors appeared to be contributory. Mostof the nests are inaccessible to mammalsin that they cannot walk up overhangingrock faces. Wasps defend the nests onwarm days so that other insects havelittle chance of stealing successfully.Nest predation by birds for honey doesnot occur (9).

Winter visits to the study site revealedconditions necessary for females to leavetheir hibernacula. Females returned totheir nests on days that reached at least21°C with at least 110 minutes of sun,and on overcast days when the temper-ature was at least 26’C. Warm days asdefined above were counted for everyU-day period in November through Feb-ruary for the past 13 years (10). Each ofthe 15-day periods examined had a mean

APRIL 1979 0036~8075/79/0413-0207$00.50/0 Copyright @ 1979 AAAS 207

months. Since deficits in stored fat canbe compensated for by honey, waspswithout honey might be expected to bemore variable relative to these indices ofquality. Since sisters hibernate togetherand may control temperature fluctua-tions in the hibernaculum to some ex-tent, winter survivorship was comparedbetween nests that initially had the samenumber of females, with and withouthoney. It was possible to form 11 suchpairs of equal numbers. Significantlymore females survived the winter fromnests with honey (Wilcoxon rank sumtest, P < .OL N = I I).

If females without honey are in poorercondition, they will be expected to formlarger foundress associations if largerfoundress associations represent groupsof females unable to form their ownnests. With honey, the mean number offoundresses per nest was 3.64: femalesfrom nests without honey formed associ-ations averaging 3.70. These means arenot significantly different. The distribu-tion of foundress number from nests withhoney was not different from that foundfrom nests without honey (Fig. 1, a andb) ( Komolgorov-Smirnov test, x? = 1.81,d.f. = 2: not significant).

A comparison of nest size just beforeworkers emerge at the end of April, 2months after nest initiation, was made toascertain whether females without honeybuilt smaller nests. Comparisons weremade by pairs for one, two, three, four,and six foundresses. Females who didnot have honey over winter built smallerspring nests (Wilcoxon rank sum test,P < .OX, N = 14) (Fig. lc).

To compare variability in quality be-tween females with honey and without,Kendall tau correlations were calculatedbetween variables that should be highlycorrelated if females are all of the samequality. The number marked in autumnwas correlated with the number surviv-ing the winter, the number surviving thewinter was correlated with the number ofnew nests formed, and the number offoundresses was correlated with thenumber of cells in the nest after 2 months(Table 1). Differences in correlations be-tween individuals that overwintered withhoney as opposed to those without weresignificant. All three correlations weresignificant for wasps with honey, whileonly the correlation of foundress numberand number of cells was significant inwasps without honey, and that one bare-ly so (Table 1). It is clear that in all situa-tions where the condition of the waspscould affect the outcome of the correla-tion, it is much lower for those without

’ honey .Depriving some wasps of honey is a

SCIENCE, VOL. 204, 13 APRIL 1979

.rather crude way of manipulating thecomponent of their quality or conditiondetermined by adult experience. Whilethe results show that this was detrimen- to.tal treatment to the wasps overall, theremay have been some that were able tosurvive the loss of honey without ill ef-fects. Perhaps some were able to stealhoney from other nests. These wouldhave been other nests in the study, since

*the two types of nests were intermingled.$

In addition, females with honey probably “*had different amounts depending on the 14.number of females, the time that honey 15.gathering was initiated, the nest size, andthe number of workers surviving throughthe honey gathering period.

J O AN E. STRASSMANN

Department oj’ Zoology, 16.U of’ T~WL S (it Autiu,Awtit / 78712

I.

;:

4.

5.

6

References and Notes

0. W. Richards and M. J. Richards, Trans. R.Entomol. Soc. 102, 1 (195 1): E. 0. Wilson, Theinsect Societies (Belknap, Cambridge, Mass.,1971); C. D. Michener, The Social Behavior ofthe Bees (Belknap, Cambridge, Mass., 1974).P. Rau, Bid. Bull. 54, 503 (1928).M. J. West Eberhard, The Socia Biology of Po-listine W a s p s (Museum of Zoology, Universityof Michigan, Miscellaneous Publication, AnnArbor, 1969).P . Rau, Bull. Brook. Entomol. SM. 2 6 , 1 1 1(1931): ecology 10, 191 (1929).0. W. Richards, The Social Insects, (Philosoph-ical Library, New York, 1953). p. 63.: L. Pardi,Atti Soc. Toscana Sc,i. nat. Pisa Mem. 50, 1(1941).Wasps were marked by placing a plexiglass 17.bucket on a pole over the nest. The wasps at-tacked the bucket, which contained ether, andhence were anesthetized. They were removed,placed on ice, and marked with Testor’s PLAenamel on the thorax. After a quick recovery,they flew off. A different combination of colorswas used for each nest. Laboratory and field 18.longevity studies indicated that the ether had nolong-term detrimental effects.To affirm this, 1 dissected presumed workersand presumed future queens. The large fat bod-ies and slightly developed ovaries of futurequeens always distinguished them from the yel-low fat bodies and stringy atrophied ovaries of 19.workers, and the predictions based on wingwear were always correct. 20.Sometimes the fights were so intense that bothfemales crashed to the ground and lay there forup to a minute grappling and trying to sting eachother. Although no fatalities were observed,there were wasps lying on the ground on theirbacks, alive and twitching their legs, but appar-ently unable to move otherwise: these may havebeen stung.

ruary, while in 1976- 1977, honey remained inthe nests until March when females permanentlycame out of hibernacula. This difference wouldbe expected from an examination of the numberof days available to feed on honey in the 2 years.in 1976 there were 12 days warm enough forhoney feeding in November and December,while in 1977 there were 32 days in the equiva-lent period. In 1976 the wasps may not have hada chance to finish the honey, while they did in1977. Which sort of year is more typical’? A com-parison with the last 13 years indicates that 1976was the coldest year of all, 12 warm days being16 days below the mean of 28 warm days for No-vember and December. The number of warmdays in 1977 was 4 days above the mean for theperiod, but is clearly much more typical. Thushoney is usually finished before spring nestfounding, but the exact date depends on thenumber of warm days as well as the initialamount of honey stored.The entire reservoir was searched for additionaldense populations of P. annularis, and only oneother colony was found; it had 40 nests. Othercliffs nearby were searched, and only isolatednests were occasionally noted. Perhaps mostimportant is that the cliff is in the shade until 3

gfmthe 40 nests marked in the fall, all femalesfrom seven of them left the area entirely in Octo-ber. All sisters disappeared on the same day;there were no signs of predation. Since femaleshave to be present to guard honey, migrationcould be a nest specific phenomenon, occurringbefore honey storage.P. Rau, Ann. Entomol. Soc. Am . 33, 617, (1940):i b i d 23, 461 (1930).I thank R. Thomas for field assistance, Y. Hir-aizumi, L. Lawlor, and W. Mueller for help withdata analysis, and L . Gi lbe r t , J. Smiley, J.Waage, and an anonymous reviewer for criti-cizing the manuscript. This research was sup-ported by grant 5-T32-GM-07126 from the Na-tional Institute of General Medical Sciences,National Institutes of Health.

P. Rau, C a n . Entomol. 73, 196 (1941). Birds are 4 April 1978; revised 28 August 1978

often major predators on polistine nests, usuallyremoving the nest and eating all the larvae.However, in this population of P. a n n u l a r i s birdpredation was ra re excep t during severedroughts.Local climatoiogical data, Austin, Texas, ob-tained from the National Oceanic and Atmo-spheric Administration, Environmental DataService, 1965-1977. These data were taken at asite 26 miles from the study site. The main dif-ference between the two locations was that thestudy site had cooler nights: warm days andcloud cover were approximately the same forthe two locations.M. J. West, S c i e n c e 157, 1584 ( 1967).W. D . Hami l ton , J. Theor. Biol. 7, 1 (1964);ibid., p. 17.P. Rau, Can hfottd. 62, 81 (1930): Ann. Ento-mol. Soc. Am . 35, 94 (1942).D. L. Gibo, J. N. Y. Entomol. Soc. 80, 1 0 5(1972).K. Eickwort, Insectes Soc. 16, 67 (1969). 1 kept15 groups of seven sisters each in a constanttemperature room at 8OC for 2 months, inter-rupted by two warm days on which they weregiven a mixture of two-thirds brown sugar, one-third hydrolyzed yeast, and water. Under theseconditions, only 3 of 105 wasps died during thecold period.This hypothesis is supported by the observationthat in 1977-1978. honev was gone in earlv Feb-

209