Optical dating provides a measure of the time since minerals, such as quartz, were last exposed to sunlight5,6,7. Mud-nesting wasps gather surface sediments from the margins of streams and pools, further exposing any quartz grains to sunlight during collection and transport of the mud. Final exposure to the sun occurs during construction of the nest in a rock shelter. Quartz grains embedded in a nest are hidden from sunlight and will accumulate the effects of the nuclear radiation flux to which they are exposed. This radiation dose (the palaeodose) will increase with time and may be measured using optically stimulated luminescence (OSL). Mud nests constructed originally by two species of mud-dauber wasp, Sceliphron laetum (Smith) and Sceliphron formosum (Smith)11, and the eumenine (potter) wasp (Abispa spp.)1,4 were collected from the rain-protected ceilings and recesses of rock shelters in the northern Kimberley region of Australia. Nests were removed at night, using red-filtered (>590 nm) light, without damage to the paintings. Most of these nests were built over paintings, the relative ages of which could be estimated from a sequence of superposed motif styles9,10. Seven nests were chosen for initial dating. Three lightly cemented S. laetum nests (DR1, DR2 and DR6) had been built over parts of red-pigmented Wandjina (anthropomorphic) figures in one rock shelter. At another site, the residual stump of a heavily cemented, probably S. laetum, nest (KERC4) directly overlay the head-dress of a mulberry-coloured human figure (which, in turn, overlay a hand stencil), and this nest thickened laterally into a much larger but similarly indurated nest (KERC5). The human figure in the painting has an elongated torso, hanging arms (no decorations visible) and a narrow head from whose apex radiates a semi-circular tufted head-dress; the painting looks archaic, and may be related to the Bradshaw style9,10. Two S. laetum nests, not obviously associated with any paintings, were also collected: an especially petrified nest (DR4) in the Wandjina rock shelter, and a ‘modern’ nest (KERC9, <2 years old). Nest KERC4 was no more than 5 mm thick, whereas the other nests were 20–40 mm thick.

To examine the opacity of a mud nest, the largest nest collected (DR6) was divided into seven portions, the first portion being composed of the outermost 1–3 mm of the nest mud, with successive portions consisting of mud concealed deeper within the nest; the ‘core’ portion had been attached to the shelter wall. Each portion was optically dated, the aim being to check that the inner portions of the nest were sufficiently light-safe to retain an OSL signal. Nest DR4 was divided into an inner core and an outer shell, while only the cores of nests DR1, DR2, KERC9 and KERC5 were dated. Quartz grains were extracted from each portion, and palaeodoses were determined using the standard multiple-aliquot additive-dose procedure12, or the newly devised single-aliquot additive-dose and regenerative-dose protocols13,14,15. Nest KERC4 required a different approach because of its small size: quartz grains were extracted from the entire nest and, as there were insufficient grains for standard multi-grain analysis, individual grains were dated using single-aliquot protocols. A frequency distribution of single-grain palaeodoses is thereby obtained15,16, enabling unilluminated grains to be discriminated from grains exposed to light near the surface.

The ‘age profile’ obtained for nest DR6 (Fig. 1, Table 1) indicates that only the outermost portion of the nest contains grains that have been exposed to modern sunlight, yielding an apparent age of 110 ± 20 years, with the next four portions having a weighted mean age of 270 ± 20 years. The two innermost portions have an average age of 610 ± 40 years, implying that DR6 consists of two generations of nest. This finding is consistent with the observation1,17 that mud-dauber wasps prefer to build nests near existing nests and on the stumps of abandoned nests, rather than on a bare rock surface. The non-zero age of the outermost portion corresponds to a residual palaeodose of 0.14 ± 0.02 Gy, and similar palaeodoses were obtained from the cores of the recent nest KERC9 and nest DR1. The ages of 100–150 years for KERC9, DR1 and DR2 indicate the lower age limit that can be expected using OSL. Two further lines of evidence support a recent origin for these nests: first, the presence of 160–270% 210Pb excess (presumably from atmospheric fallout) in nests DR1 and KERC9 indicates that these nests are built from mud collected within the last 100 years; and second, AMS 14C determinations made on pollen grains8 (>5 μm in diameter) extracted from nests DR2 and KERC9 give ages that are indistinguishable from modern (Table 1), indicating that the pollen is derived from plants flowering around the time of nest construction.

Figure 1: Optical dates obtained from each portion of S.laetum nest DR6 (40 mm from core to surface).
figure 1

Filled circles, mean ages (±1σ uncertainties) derived using a multiple-aliquot additive-dose protocol; open circles, mean ages derived using a single-aliquot additive-dose protocol. Note the concordance between multiple-aliquot and single-aliquot age estimates for layers 3 and 7.

Table 1 Dose rates, palaeodoses and optical dates for quartz grains, and AMS 14C determinations for pollen grains

The core of nest DR4 yielded an optical date of 1,800 years which, although it is not obviously associated with any painting, demonstrates the preservation potential of large nests. This is further borne out by the remarkable antiquity of nests KERC4 and KERC5. The core of the latter yielded an age of 17,500 years, and the single grains dated from nest KERC4 suggest an age of more than 16,400 years for the underlying anthropomorphic painting. These results show that the mulberry-coloured human figure was painted at or before the Last Glacial Maximum, and that the underlying hand stencil must be older still. Independent support for a Pleistocene age is implied by the condition of equilibrium existing between 230Th and 226Ra in the mud from nest KERC5. Nests constructed in the past 600 years (DR1, DR2, DR6 and KERC9) exhibit disequilibrium between 230Th and 226Ra, owing to a 40–110% 226Ra excess associated with the sediments used to build the nests. If KERC5 and KERC4 had been constructed of mud with a similar initial 226Ra excess, it would have taken at least 8,000 years (five half-lives of 226Ra) for this excess to decay away.

Nests of S. laetum (including DR1, DR2, KERC5 and KERC9), S. formosum and eumenine wasps have been examined for pollen, spores and phytoliths. Pollen is present in nest mud chiefly as a result of wasps visiting flowering plants for nectar. Phytoliths provide a record of polymerized biogenic silica which has been deposited in the source mud after release from plant organic material. Initial investigation of a S. laetum nest and eumenine wasp nest (both undated but probably late Holocene in age) showed that both were rich in pollen, yielding 1–2 mg of pollen from 1 g of mud. A large, distinctive pollen type produced by Grevillea or Hakea was predominant in the S. laetum nest, whereas a more equal representation of pollen types (eucalypt, ti-tree, paperbark, wattle, Grevillea and Hakea) was present in the eumenine wasp nest (Table 2). A diverse mix of abundant pollen was found subsequently in nests DR1, DR2 and KERC9, but an indurated S. formosum nest seemed to contain no pollen. Pleistocene nest KERC5 showed the presence of some Myrtaceae pollen, but not enough to allow AMS 14C dating. These differences almost certainly reflect the particular foraging strategies of mud-dauber and potter wasps and perhaps some degradation of pollen with time. The types of phytolith extracted18 from the mud of four dated S. laetum nests (Table 3) include varieties of grass and dicotyledon, as well as carbon particles, starch grains and, at the Wandjina site, freshwater sponge spicules. These samples did not contain a sufficient mass of phytolith for AMS 14C dating, but other nests may prove amenable. Previous studies have used occluded carbon in phytoliths for AMS 14C dating and palaeoclimatic reconstructions19. Further investigations of fossil pollen and phytolith types should allow a detailed reconstruction of the palaeoenvironment for each interval of time represented by a nest.

Table 2 Pollen and spore types extracted from selected mud-wasp nests
Table 3 Phytolith types extracted from selected S. laetum nests

The worldwide distribution of mud-wasps and the demonstrated longevity of their nests should prove a valuable tool in archaeological and palaeoclimatic research. The rapid formation and short period of usage of a nest, and the incorporation of a variety of palaeoecological indicators, makes each mud-wasp nest a ‘snapshot’ of its late Quaternary environment. By examining nests of different ages in several localities, it should be feasible to infer sequences of prehistoric human activity and vegetation change at various geographic scales and in climatic zones where other palaeoecological records are poorly preserved. Optical dating of single grains of sand now makes it possible to date very small nests, such as those produced by S. formosum (which typically overlie the stylistically oldest paintings), and the painted stumps of nests (to obtain maximum ages for art motifs). Sediment accretions created by some other animals (such as ants, termites and mud-nesting birds) may also be suitable for optical dating if the mineral grains were fully exposed to sunlight during final emplacement.


Palaeodose determinations. Quartz grains 90–125 μm in diameter were extracted from the mud nests and then etched in 40% hydrofluoric acid for 45 min to remove the α-particle-dosed outer rinds. Multiple-aliquot additive-dose palaeodoses were obtained using conventional quartz dating procedures6,20,21, a 300-s preheat at 220 °C, optical stimulation (500 ± 40 nm) at room temperature, and OSL emissions detected through 2.5-mm U-340 and 3-mm UG-11 filters. Single-aliquot additive-dose palaeodoses were obtained from successive short-shine/dose/preheat cycles, using a 10-s preheat at 160–300 °C (the preheat plateau6 region was determined for each multi-grain sample and invariably included 280 °C), optical stimulation of 420–550 nm at 125 °C, and OSL emissions detected through two 3-mm U-340 filters14,15,21. A 10-s preheat at 280 °C was used for all single-grain analyses. Single-aliquot regenerative-dose palaeodoses were obtained using the same equipment, the total OSL signal from successive long-shine/dose/preheat cycles, and the 110 °C thermoluminescence signal to correct for differences in OSL sensitivity between cycles15. Laboratory 90Sr/90Y sources delivered β-particle doses at 22–48 mGy s−1. OSL ages for nest KERC4 are calculated from individual grains with palaeodoses >10 Gy. The additive-dose and regenerative-dose palaeodoses (from 21 and 15 grains, respectively) show a broad, approximately gaussian, frequency distribution, which we interpret as reflecting grain-to-grain differences in dose-response behaviour and β-particle microdosimetry15. Grains with palaeodoses >10 Gy (33 grains with palaeodoses of 14–59 Gy and three grains with additive-dose palaeodoses of 75 Gy) are presumed to be light-safe, whereas those with a smaller palaeodose (25% of the grains examined) are considered to have been exposed to modern sunlight, either on the surface of the nest or in the immediate subsurface where attenuated sunlight may penetrate sufficiently to partly empty the OSL traps. This problem does not apply to nest KERC5, from which grains were extracted only from the unilluminated core.

Dose rate determinations. The γ-ray dose is derived mostly from the local bedrock, with a minor (5%) γ-ray contribution from the nest mud. The cosmic-ray plus bedrock γ-ray dose rate was estimated by two independent methods for most nests (Table 1footnote), and averaged to calculate nest ages. The β-particle dose rate is derived entirely from the nest mud and was calculated using an attenuation factor22 of 0.93 ± 0.03. The only exceptions are layer 1 of DR6 (grains in this thin exterior layer are assumed to have received only 90% of the 4π β-particle dose rate), and nests KERC5 and KERC4 (for which a 25% contribution from bedrock (β-particle dose rate, 0.49 ± 0.03 mGy yr−1) is assumed, owing to the close contact between the sampled grains and the rock surface, giving a total β-particle dose rate 18% smaller than if the β-particle flux were derived from the mud only). The β-particle dose rates from nest mud were deduced from X-ray fluorescence and α-particle spectrometry23. The latter analyses show that the 238U decay series is in disequilibrium, with unsupported excesses of 210Pb (in DR1 and KERC9) and 226Ra (in DR1, DR2, DR6 and KERC9). A 210Pb excess does not significantly affect the dose rate, and had KERC5 or KERC4 been formed with an initial 226Ra excess of 100% then the dose rate (integrated over 20,000 years) would be <1% greater than the modern (equilibrium) dose rate used to calculate their OSL ages. The 210Pb/226Ra ratio for DR6 indicates 222Rn emanation of 18%, which is assumed to have prevailed since nest construction. Because of the paucity of mud available for KERC4, the nest-derived β-particle dose rate is assumed to be the same as that determined for its lateral extension, KERC5. An internal α-particle and absorbed β-particle dose rate22 of 0.02 ± 0.01 mGy yr−1was deduced from U and Th concentrations in single grains of acid-etched quartz (measured by laser-ablation inductively coupled plasma mass spectrometry (ICP-MS) using a pulsed ArF excimer laser operating at 193 nm) and an assumed α-particle efficiency ‘a’ value of 0.1. The dose rates require no moisture corrections as the nests are dry.