1Analyzing spatial nesting patterns within managed hives of small local solitary bees and wasps (Hymenoptera: Colletidae, Megachilidae, Vespidae, Sphecidae). P.E.Hallett, Departments of Physiology and Zoology, University of Toronto, c/o 144 Hendon Avenue, Willowdale, Ontario, Canada M2M 1A7
Telephone 416-221-4844 (mssgs)
Infrequent access FAX 416-946-7422 (mark “Please phone Prof Hallett at 416-221-4844)
—In the 2001 season in the Dundalk Highlands (S. Ontario), overwintered species emerged to make new nests within stands of multispecies hives loaded with easily inspected trapnests. Maps were made of old and new nest positions within the hives. A harvesting or management issue is where new nests tend to occur relative to the old. Accordingly the maps were analyzed for new nests made within the set of old nest sites, and within sets of increasingly remote empty sites. Statistical significance came from randomizing the positions of the natural clusters of new nests within the maps.
There were two replicated, statistically significant, nesting patterns. The new nest densities of identified species of Osmia
(Panzer) [Megachilidae], Symmorphus
Wesmael [Vespidae] and Passaloecus
Shuckard [Sphecidae] were highest at the old nests, declining with increasing distance
, while the Hylaeus
Fabricius [Colletidae] species were fairly evenly dispersed. These patterns were due to natives of the hives as new immigrants were very scarce. Competition and management are briefly discussed.
Hallett, P.E. 2003. Analyse des configurations spatiales des nids dans ruches controlées des petites abeilles et guêpes solitaires et locales (Hymenoptera: Colletidae, Megachilidae, Vespidae, Sphecidae). The Canadian Entomologist
—Aux printemps 2001 dans les Dundalk Highlands (S. Ontario), des plusieurs espèces surgit pour faire des nids nouveaux dans des étalages de ruches chargées avec des
piege-nids facilement examinés. Des cartes on fait des positions des nids nouveaux et vieux dans les ruches. La moisson ou la gestion demande où les nids nouveaux ont tendance avoir lieu relatif aux vieux. En conséquence les cartes sont analysées pour nids nouveaux à l’intérieur d’ensemble de nids vieux ou aux intérieurs des ensembles des nids vides de plus en plus lointain. La signification statistique est venue de permutations randomisées des batteries des nids nouveaux dans les cartes. Cettes moyens ont indiquées deux configurations répétées, statistiquement significatives, des nids. Les densités des nids nouveaux des espèces identifiées d'Osmia (Panzer) [Megachilidae], de Symmorphus Wesmael [Vespidae] et de Passaloecus Shuckard [Sphecidae] etaient les plus hautes aux nids natals, diminuant avec l'augmentation de la distance, alors que les espèces de Hylaeus Fabricius [Colletidae] étaient dispersées plus également. Les configurations des nids étaient duées aux indigènes des ruches car les nouveaux immigrés étaient très rares. La competition et la gestion sont brièvement discutées.
An increasingly urban existence, continued population expansion, and accelerated global change (Science1997) suggest the need for better methods of manipulating pollinators, predators and other beneficial insects —immediately for teaching of all ages and for research, and within a human generation or so for conservation, restoration, farming and perhaps even gardening. A number of female solitary bee and wasp species will nest within borings in opaque sticks or blocks and are easily collected. This serves museum studies well (Fye1965ab; Krombein,1967) but is unsuited to repeated inspection and measurement, as in conservation research or experimental farming. A tested new method has been described, including carpentry, and protections against rain, parasitoids, ants, porcupines and waste accumulation (Hallett 2001a-d). Conspicuous open-faced hives are stacked with 35-55 grooved wooden nest blocks closed by transparent lids to provide blind-ended U-section holes, or bores, as potential nest sites. Each block has 4-8 bores, depending on the borewidth. The 140-440 bores in a hive can vary from 3.2 to 9.6 mm in borewidth and are 150 mm long, standardizing Krombein’s dimensions. As nest construction and provisions are diagnostic of the local taxa (Danks1971) —and immediately determinable visually— one can easily map the locations of the different types of nests in a hive stand in the field. But this raises the need for an appropriate method to recognize pattern in a map in the presence of ‘spatial autocorrelation’ or clustering (Dale1999; Legendre & Legendre1998; Fortin, Drapeau & Legendre1989).
Knowing the likely patterns of nest dispersal would be useful when arranging, cleaning, sorting or harvesting large numbers of nest blocks. Management for the year 2001 used two hives of clean empty nesting materials (called the super and trap hives by analogy with honey beekeeping) to harvest nest dispersal from a third hive, the ‘brood hive’, that had been seeded
with nest blocks containing clean viable overwintered nests. The investigation of nesting patterns began with the Spring map which contained the locations of all the seeded and empty bores in a hive stand. It defined, for each nest taxon, spatial sets of bores ordered by distance
(the seeded bores, the set of adjacent bores, the set of somewhat more remote bores, the set of still more remote bores
, and so on). The Fall map recorded the crop of new nests, providing new nest densities within the spatial sets defined in the Spring. Computer-intensive resampling gave valid estimates of the significance of the nesting trends by making permutations of the Fall map that preserved the natural clusters of new nests while randomizing their positions across the spatial sets. Two distinct, replicable and statistically significant nesting patterns were recognized.
Materials and methods
A row of three hive stands (W, P, M) was fielded just before Spring 2001, facing East at 50 m spacing, in a mixture of old highland pasture and willow and balsam swamp near Flesherton in Grey County, S. Ontario (N44E13', W80E32', altitude 480m). Each stand consisted of a slatted platform raised on 1.5 m greased poles and bearing a fixed arrangement of three hives packed with nest blocks. External height by width of the hive entrances and nest blocks were 22 x 50 cm and 1.6 x 8.8 cm respectively, and there was a slot in the flooring between the brood hive and its superposed ‘super’ (Hallett 2001d). The trap hive stood by itself north of the others at a separation of 12 cm. By either average crawl or flight distance the trap was more remote from the brood hive than the super.
Only the brood hives were seeded with nests in overwintered nest blocks, at a mean density of 0.40 nests/brood hive bore. The seeded species were all relatively small and evenly spaced at the block level. In stands W
smaller species in 8-bore 3.2 mm borewidth blocks alternated with larger species in 7-bore 4.8 mm blocks. All blocks of stand P
were of 4.8 mm bores, but the larger and smaller species also alternated from block to block. Each seeded bore contained only one nest, and every cell in it had been opened in the winter and examined with a 10x handlens to exclude nest parasites or failed cells (except for the tough cocoons of Osmia tersula
Cockerell). Supers and traps were packed like their brood hives with used blocks, but all bores were initially empty. Weekly inspections allowed failed cells, parasitized cells, and emerged nests to be scraped clean. Overall emergence was 89%, failure 11%. New nests were found either at the sites of, or within, all types of seed nests (i.e., scraped and unscraped seed nests, and some seed nests apparently removed in competition), and in empty bores, at a density of 0.55 nests/bore across hives.
Each hive-stand map was a 2-dimensional semiregular arrays of bores, and bore states were essentially binary. A bore in a Spring map would show either as a named seed nest or as an empty bore, and in a Fall map as either a named first new nest or as an empty bore. A first new nest had to have at least one provisioned and completed cell to be counted. Maps of the first new nests were developed through the season and finished in Aug.-Sept. to ensure complete counts of the first generation. A few second generation nests in empty bores may have been included. Post-count nesting was negligible and by the smaller species (Hylaeus
The major local nest taxa
for mappings and counts were naked-eye or 10x magnifier identifications by nest construction and provisions. Keeping to the order of the tables and graphs, the nest taxa were (a)
a small megachilid Spring Mason Bee Osmia
for nests constructed of leaf pulp with pollen provisions in 4.8 mm bores, (b)
a small eumenid Potter Wasp Symmorphus
for mud nests provisioned with beetle larvae (chrysomelids) in 4.8 mm bores, (c)
colletid Masked Bees Hylaeus
for bee cellophane nests in 3.2 and 4.8 mm bores, and (d)
very small sphecid Aphid Wasps Passaloecus
for resin nests with aphids in 3.2 and 4.8 mm bores. Osmia
differed from the others in overwintering as the adult and finishing its nesting early. All the mud nests in 4.8 mm bores with wholly consumed provisions were mapped as Symmorphus
Given the paucity of the only confusable alternative (mud nests of Ancistrocerus
sp., Eumeninae, in which the caterpillars were often consumed), the final Symmorphus
counts were likely less than 4% too high.
Species names for the above nest taxa from rearings to adults were: (a) Osmia tersula Cockerell
; (b) Symmorphus cristatus
(Saussure) ); (c) Hylaeus (H.) ellipticus
in 3.2 mm bores with both H. ellipticus
and the similar but larger H. (H.) verticalis
4.8 mm bores; (d)
the similarly sized Passaloecus cuspidatus
F. Smith, P. gracilis
and P. monilicornis
Dalbohm (different infrageneric groups of Vincent1978) in 4.8 mm bores. Continued monitoring might recognize a few more species. Ontario museums have only three specimens of Passaloecus monilicornis
(M. Buck, personal communication). Keys are cited below. Voucher specimens will be placed in the University of Guelph Insect Collection. Analyses
, models and statistical tests were written by the author in the mathematical language APL/J (www.jsoftware.com).
The literature provided no specific expectations for the spatial patterns of new nests, e.g., some Potter Wasps reuse nests (though no alternative nest sites were available in Rau & Rau1917) while the Spring Orchard Bee Osmia lignaria propinqua
Cresson is said not to (Bosch & Kemp 2001). Uniform nest dispersal was adopted as the null hypothesis and diffusive dispersal as the alternative (Southwood1966). If one examines a nest map one nest taxon at a time one can in each case classify the bores into ordered spatial sets
through which dispersal might progress. The left and central columns of Table 1 summarize complete counts of the bore sets and new nests. For example, for Symmorphus,
hive stand and 4.8 mm bores (the second case in Table 1), the five bore sets, in order of increasing distance, are (i) the 47 bores originally seeded with Symmorphus
nests, (ii) the 93 remaining bores in the Symmorphus
seeded blocks, whether seeded with other taxa or initially empty
, (iii) 14 bores in other blocks in the brood hive, whether seeded with other taxa or initially empty, (iv) the 161 initially empty bores of the super, and (v) the 154 initially empty bores of the trap. In all, 222 new Symmorphus
nests were found in the 469 bores. The Symmorphus
new nest density in each bore set is the number of first new nests divided by bores in the set —as plotted by the filled circles and line W
in Fig. 1a. A generally declining trend, with increasing distance from the seed nests, is replicated at all three hive stands, W
. Comparable trends occurred, at the single stands where they are present, for Osmia
across its bore sets (open squares Fig. 1a ) and for Passaloecus
(open circles Fig. 1b) —providing five instances of diffusive dispersal, i.e., a localized dispersal concentrated on the sites of the seed nests. By contrast Hylaeus
(filled squares Fig. 1b) was comparatively evenly scattered at all three stands. The value of spatial density
as a metric is shown by the contrast between the orderly trends of Fig. 1a, which plots densities, and the scatter of the complete counts of Table 1, from which Fig. 1 is derived. Table 1 can be used for other calculations as only 10% of all new nests are omitted —either less abundant species (Trypoxylonini and Eumeninae in 3.2 mm bores) or the few Hylaeus
nests in the 4.8 mm bores at stand W
A recurring problem for this type of data is how often perceived patterns arise by chance, so it is appropriate to explain the solution. A usual test statistic for the trend of frequencies in ordered classifications is the standard normal variate Z (Snedecor and Cochran1989). Fitted binomial regression slopes, and Z scores for the deviation of a slope from zero, appear as the last columns of Table 1. Because nest clustering reduces the degrees of freedom, using the classic critical 2.5% points of Z risks identifying trends where none exist (Fortin & Jacquez 2000). Consequently I constructed theoretical hive stands by randomly permuting the positions of either individual nests (as a check) or entire nest clusters within the Fall new nest map. The seeded nests were always left in their original positions in the Spring map, fixing the spatial sets of bores, but the numbers of new nests falling into each set changed on each permutation of the Fall map. When the individual new nests were randomly permuted (null hypothesis N0) as a check calculation 999 times, slopes computed, and the original value added
, the twenty fifth largest Z value (i.e., the 2.5% critical point) was close to the expected value of -1.96 given in classic statistical tables (range "0.14).
Because both classic tests, and permutations of individual new nests, confuse breaking the spatial association between the bore sets and the new nests with disrupting the clusters of new nests, the valid reference distribution (null hypothesis N1) is the one obtained by permuting intact clusters of new nests. This is very conveniently approximated by simply randomly switching nest blocks of the same borewidth in the Fall map. The reason is that 1- and 2-dimensional Pearson lagged autocorrelations (Manly1998; Legendre & Legendre1998) of the hive maps, checked with simulated data, show horizontally oriented linear clusters with a median width, across all hives and taxa, of about 3 nests (quartiles 2 and 5, n = 23, the widest being
), and a height of 1 nest. As the number of bores in a nest block was 7 or 8, randomly switching nest blocks broke the spatial association between new nests and bore sets without appreciably fragmenting the clusters. The resulting 2.5% critical points in the last column of Table 1 are often much larger than the N0 expectation of -1.96 because the N1 reference distribution is usually wider than the fully random one. However the original description of Fig. 1 is still supported, as the near-zero slopes for Hylaeus
fall close to or within the limits of the N1 null model, the intermediate slopes of Passaloecus
at stand W
are marginally beyond, and the steep slopes of the remaining data are very strongly deviant.
There were replicable, statistically significant, differences between the comparatively evenly scattered spatial dispersal of Hylaeus new nests and the more strongly localized seed nest-centred aggregations of the other three common nest taxa (Passaloecus, Osmia and Symmorphus). These conclusions were confirmed and extended in the 2002 season (for 25 new data sets, including new localities, species and bore sizes, a different hive packing scheme, and another statistic; Hallett, in preparation).
Several candidate factors potentially link to the distinction between the flatter and the stronger slopes in Table 1 or Fig.1. (1) Competition
Competitive nest supersedure can be defined as a nest built to the detriment of an existing nest in the same bore (e.g., in place of, or in a resistant material in front of, a viable nest). Finding two different types of new nest in a bore was a rare event, 34 of the 47 supersedures being of the most dispersed nester, Hylaeus,
in 3.2m bores by Passaloecus
or less common small species. Eviction of Symmorphus
larvae from new nests by Symmorphus
was restricted to the 4.8 mm bores of the crowded brood hive in stand M.
As both supersedures occurred after
the dispersal from the seed nests they are at best associated factors. A more direct form of competition is occupancy of another taxon’s seed-nest sites. Interestingly (Pearson r2
= 0.85, n
= 8), the flatter regressions (slopes of -0.02 to -0.27 in Table 1) appear matched by high rates of seed site occupancy by other taxa (0.34 occupancy of Hylaeus
mainly by Passoloecus
at stand W
and by Osmia
0.38 of Symmorphus
; 0.21 of Passaloecus
), the steeper regressions (-0.46 to -0.89) by low rates (0.00 Osmia
, 0.03 Symmorphus
). (2) Parasitism
In agreement with Fye (1965ab) and Krombein (1978), principal nest parasites by rearing were: the eulophid Melittobia chalybii
Asmead, the chrysidine wasp Chrysis coerulans
Fabricius, and the miltogrammine sarcophagid fly Amobia,
as the main host; a sapygine wasp in Osmia
nests; and the gasteruptiid Gasteruption assectator (L.)
. New nest parasitism potentially correlates with the present data because it was qualitatively most severe for the nests of Osmia
and least for Hylaeus
. As new nest parasitism occurs after
dispersal from the host’s seed nests (like the new nest supersedures above), any correlation with dispersal might relate to success or failure at sensing parasites (or competitors) prior to nest founding. (3) Body size
A testable prediction is that small body size (and thus absolutely small eye size and low visual angular resolution, e.g., Dafni, Lehrer and Kevan1997) leads to more homing errors and more widely scattered unfinished nests. Visual reactions to competitors or parasites should also be reduced. (4) Immigration
It is unlikely that the measured dispersal patterns are due to trapping of new immigrants from the surroundings. Direct, and very low, measures for the trapping rates of Passaloecus
come from the numbers of their new nests in the hive stands which were not
seeded with those particular species (trapping rates b2
= 0.012 and 0.008 new nests/empty bore respectively). A similar, though overfitted, estimate of b2
comes from modelling the N
new nests bred from S seeded nests and trapped by E
empty bores as
Ni = b1 Si + b2 Ei + b3 Si Ei + ,i
where i = 1-5 codes the stand and bore size, and ,i is random variation. The ordinary least squares solution is breeding b1 = 5.1 new nests/seeded nest, trapping b2 = 0.011 new nests/empty bore and interaction b3 = -0.004. Note that b1 >> b2; also b2 Ei is the smallest term in eq. (1) despite the large excess of empty bores. So, by both direct and indirect estimates, trapping by empty bores was very slight in the 2001 season, and the measured nesting patterns were due to emergences from the seeded nests. That breeding of natives within the hives (i.e., farming) considerably exceeded gains through immigration (i.e., trapping) is encouraging for both farmer and conservationist.
The brood, super and trap arrangement was easily set up and maintained, and harvesting required little more than lifting off the super and trap.
As might be expected, or from sums on the 8 cases in Table 1, the fraction of new nests harvested into the clean nesting materials of the super and trap was highest (0.70, Hylaeus
) for the wider dispersals (r2
= 0.91, n
= 8). Harvesting was least for the most localized nester (0.11, Osmia
, relative to 0.51, overall). Consistent with the low immigration rates, the purity of harvest was highest for the cases with the purest seed (r2
= 0.76, n = 8
; seed and harvest purities averaged 0.58 and 0.54, and ranged 0.09-0.91 and 0.17-0.94 respectively).
M.-J. Fortin (University of Toronto) and L. Packer (York University) read drafts. I am also indebted to M. Buck (University of Guelph Insect Collection) and to Helena Hallett. This scientific research and experimental development used the facilities of Hallett Pollen Bees (Ont.).
Bosch, J., Kemp, W. 2001 How to manage the Blue Orchard Bee.
Sustainable Agriculture Network; Beltsville, MD.
Carpenter, J.M., Cumming, J.M. 1985 A character classification of the North American potter wasps (Hymenoptera: VESPIDAE; Eumeninae). Journal of Natural History, 19
Cumming, J.M. 1989 Classification and evolution of the eumenine wasp genus Symmorphus
Wesmael (Hymenoptera: VESPIDAE). Memoirs of the Entomological Society of Canada, 148.
Dafni, A., Lehrer, M., Kevan, P.G. 1997. Spatial flower parameters and insect spatial, vision. Biological Reviews, 72
Dale, M.R.T. 1999 Spatial analysis in plant ecology.
Danks, H.V. 1971 Biology of some stem-nesting aculeate Hymenoptera. Transactions Royal entomological Society of London, 122,
Finnamore, A.F. 1982. The Sphecoidea of Southern Quebec (HYMENOPTERA) Lyman Entomological Museum and Research Laboratory, Memoir 1.
Fortin, M.-J., Drapeau, P. & Legendre, P. (1989) Spatial autocorrelation and sampling design in plant ecology. Vegatatio 83
Fortin, M.-J. & Jacquez, G.M. 2000 Randomization tests and spatially autocorrelated data. Bulletin of the Ecological Society of America
Fye, R.E. 1965a Biology of Vespidae
, Pompilidae and Sphecidae from trapnests in Northwestern Ontario (Hymenoptera). The Canadian Entomologist, 92,
Fye, R.E. 1965b Biology of Apoidea taken in trapnests in Northwestern Ontario (Hymenoptera). The Canadian Entomologist, 92,
Hallett, P.E. 2001a A method for ‘hiving’ solitary bees and wasps. American Bee Journal,141,
Hallett, P.E. 2001b Three factors affecting the yields of solitary bees and wasps. American Bee Journal, 141,
Hallett, P.E. 2001c “Do-it-yourself” field trials on factors affecting the yields of solitary bees and wasps. American Bee Journal, 141,
Hallett, P.E. 2001d Building hives and observation nest blocks for solitary bees and wasps. American Bee Journal, 141,
Krombein, K.V. 1967 Trapnesting wasps and bees. Life histories and associates.
Legendre, P., Legendre, L. 1998 Numerical ecology
. Elsevier, Amsterdam.
Manly,B.F.J. 1998 Randomization, bootstrap and Monte Carlo methods in biology
. Chapman & Hall, London.
Mitchell, T.B. 1960/1962. Bees of the Eastern United States.
Vols. I & II, Tech.Bulls. 141&152, North Carolina Agricultural Experiment Station.
Rau, P., Rau, N. 1917 Wasp studies afield.
Dover; New York.
Science (1997) Human-dominated ecosystems (a special report; July 25) Science, 277
Snedecor, G.W., Cochran, W.G. 1989 Statistical methods
edition. Iowa State UP.
Southwood, T.R.E. 1966 Ecological methods with particular reference to the study of insect populations.
Chapman & Hall; London.
Vincent, D.L. 1978 A revision of the genus Passaloecus
(HYMENOPTERA: SPHECIDAE) in America north of Mexico. The Wasmann Journal of Biology, 36,
Fig. 1 New nest density versus distance. Data of Table 1. Osmia and Passaloecus in open symbols, Hylaeus and Symmorphus in closed. Hive stands indicated by letters. For species identifications see Methods. The scoring of the spatial bore sets for physical distance is not critical if the rank order is preserved.