Supporting Information

Wilmshurst et al. 10.1073/pnas.0801507105 SI Text separate areas described by Worthy (2) and named 1: Test pit; 1. Stratigraphic Provenance of from South Island Sites. We 2: South Excavation; 3: Baulk (material between the Test Pit and (J.M.W. and T.F.G.H.) sampled over 1,600 seeds from peat or South Excavation and added to the Test Pit sample); 4: Enlarged sandy deposits preserved in five former swamp forests which Test Pit (lower faunal layer of units 1 and 3); 5: North Excavation have been drained, cleared, and converted to pasture over the (which contained a single infilled tuatara (Sphenodon sp.) or last 100 years. One of these sites, Nguroa Bay (172°38ЈE; 40°32’S) petrel burrow, not considered to have experienced recent or contained rat-gnawed as well as intact and bird-cracked seeds. continuous bioturbation (3); and 6: Against Cliff (mostly de- Waitoetoe-B (1), Wharariki (172°40ЈE; 40°30’S), Cannibal Bay scribed as ‘‘disturbed sediments’’, the lower faunal layer of this (169°44ЈE; 46°27’S), and Long Beach (170°38ЈE; 45° 45’S) con- unit was added to the Enlarged Test Pit sample). The relative tained only intact and bird-cracked seeds. All sites were about positions of these units are shown in Fig. S1. 1–2 km from the coast and at all sites except Nguroa Bay and The first sequence of bone AMS radiocarbon determinations Taranaki seeds were preserved in stratigraphic order and by age. published from these excavations (4) included three Pacific rat All species of seeds dated from the peat deposits are eaten by rats (Rattus exulans) bones and two bird bones, the latter were today including taxifolia, P. ferruginea, Elaeocarpus reported as being from the same depths and were used as ‘‘paired hookerianus, and E. dentata. We selected a subsample of seeds samples’’ to provide independent support for the age of the rats. from each site for AMS radiocarbon dating by the Oxford However, the ages and depths of these determinations were Radiocarbon Accelerator Unit and Waikato Radiocarbon Dat- published with very little information regarding their exact ing Laboratory (the latter undertake chemical pretreatments, position (4). In response to criticism about the veracity of the old CO2 production and graphitization, and the graphite is analyzed rat bone ages, an attempt was made to validate the results by at Rafter). All dates and details of stratigraphy are pre- presenting three more rat bone dates after the 1995–1996 period sented in Table S2. (which were all less than 700 years BP), and two OSL dates from At Nguroa Bay, a reworked sandy deposit 2.25 m long, 35 cm the enclosing sediments (3). Although additional information wide, and Ϸ50 cm deep on the bank of a natural creek running was given about the stratigraphy, the exact locations of the bones along the length of a drained valley swamp was excavated in were still not made clear in all cases. Anderson & Higham (5) 10-cm spits and sieved for seeds (Ϸ300 collected). The highly critically reviewed the bone dating and stratigraphy in detail, and localized sand lenses were embedded in peat and overlay organic highlighted the problems with assumptions and stratigraphy. peats which contained wood. Seeds were most abundant in the 2.2 Reexcavation of Earthquakes #1 in 2006. Our (J.M.W., A.J.A., sand lenses, particularly in the upper 10 cm. The sand deposit T.F.G.H., and T.H.W.) reexcavation of Earthquakes #1 had thinned away from the side of the creek to less than 5 mm on the three components (Fig. S1): surface of the drained swamp. Sand lenses had poorly defined 1. A working pit. This consisted of the ‘‘North’’ excavation and boundaries suggesting the sand had been reworked by high- part of the ‘‘Test pit’’ excavation (2), plus Ϸ0.2mby0.3mtothe rainfall events, and during these periods the seeds have been west of the North excavation. All this appeared to be fill from the entrained and buried in the sand. Rat-gnawed seeds made up 1995 excavation and it was spaded out. 13% of the total seeds collected. Three seeds were also dated 2. The result was a 1-m-long by 0.7-m-deep northern wall of from test pits 600 m upstream at 10–20 cm depth (Nguroa Bay material undisturbed by excavation which was component 2. Road). North of this our main excavation, called ‘‘North Extension’’, At Wharariki, 113 seeds were collected, from a peat deposit extended 0.5 m north to south. buried under 50 cm of clay topsoil. The site was a valley peat 3. Immediately east of the working pit was an area of the which had deposited behind dunes and was at least 4m deep. ‘Against Cliff’ excavation, but east of it again was a strip of Seeds were sieved from 10-cm spits of peat, but were only found previously unexcavated material under a low limestone overhang at 50 cm and below. We selected seeds for dating from 50–60 cm, which we excavated as ‘East sampling area’. The East sampling where they were also most abundant. area and the North Extension were excavated by observed layers At Cannibal Bay, Ϸ300 seeds were collected from a 60-cm- and 10-cm spits. All material was sieved to 1 mm and everything deep organic peat deposit overlaying sterile sand. The peat retained for sorting and identification of faunal bones. deposits had accumulated in a depression between raised dunes. In summary, we excavated a sample of Earthquakes #1, which The seeds were sieved from the peat in 20-cm spits, from the contains stratigraphic units that are the continuation of Layers surface down to 60 cm. We dated a subsample of seeds from 0–20 1 and 3 as designated in the original excavations, and which might cm and 50–60 cm. Two seeds were also dated from a forest soil include Layer 2 as an unfossiliferous band that is unrecognizable under a P. ferruginea canopy in Cannibal Bay (20–30 cm depth). by color variation. Our excavation showed that bioturbation is At Long Beach, Ϸ500 seeds were sampled from a peat deposit extensive across the deposit (Figs. S2 and S3) and reaches from at least 90 cm deep which had formed in a depression behind the surface to at least 85 cm in depth, which is well below the sand dunes. Seeds were sieved from the peat in 10-cm spits from fossiliferous units of interest. As North Extension is located the surface down to 90 cm, but were only found at 20 cm and adjacent to the earlier North excavation, the extensive rabbit below. Seeds were most abundant at 30–40 cm. We dated a burrowing recognized across North Extension must have ex- subsample of seeds from 20–30 cm. tended at least some way into the North excavation seriously compromising the original stratigraphy from the top to the 2. Stratigraphic Provenance of Bones from Earthquakes #1. 2.1 Sum- bottom. Below is a detailed description of our excavation of the mary of previous excavation and radiocarbon dating. The faunal re- North Extension sampling area. mains found in a laughing owl (Sceloglaux albifacies) midden at Our observed stratigraphy of the North Extension is broadly the base of a small overhanging cliff at Earthquakes #1 rock similar to that recorded in the earlier excavations at Earthquakes shelter in North Otago were originally excavated in 1995 and #1 (Fig. S1). The general pattern in the original work consisted described in detail (2). The excavation was complex with six of 13–18 cm of Layer 1 (friable, fossiliferous, reddish silts), 8–12

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 1of13 cm of pale unfossiliferous silts (Layer 2) overlying the reddish by rabbits (Figs. S2 and S3). These burrows contained mixed fossiliferous silts of Layer 3. Our excavation has the same total assemblages of bones including those from recently introduced depth of fossiliferous silts (excluding in areas of bioturbation) fauna (rats, rabbits, mice, skylark, blackbird), and species now and our Layer 3 is very comparable in material and extent to the extinct from the mainland (e.g., tuatara – one of which was dated original Layer 3. Our Layer 1, although we saw it as yellow-gray to 1029 Ϯ 25 yrs BP: OxA-15197), suggesting both recent and to yellow-brown rather than reddish, is otherwise comparable in significant bioturbation, which is contrary to the original inter- material and extent to the original Layer 1, but it is deeper pretation (3). because we saw no stratigraphic evidence of the earlier Layer 2. We dated bones of taxa considered to be rat-vulnerable (i.e., The color differences may reflect different weathering regimes extinct well before European arrival, or extant only on rat-free as our North Extension excavation was more sheltered under the islands in ) from the surface, and at 0–10 cm and rock overhang than most of the previous excavated area, which 10–26 cm below the surface, including from within the burrow in consequence was likely wetter more often. The three main fills (Table S1). Rat bones were also dated from the same levels. units we recognized in the North Extension stratigraphy (Fig. S1) Our dates from the rat-vulnerable fauna did not overlap with are described in detail below. dates on rat bones excavated from the same layers (apart from Layer 1: At the top, beneath about 5 cm of very loose silt and long- lived Tuatara – Table S1), and there was no chronological recent sheep droppings (which were swept off the surface and order with depth for any of the bones dated (Table S1). Apart sieved), was a compact silt containing some sand and fine from the rat-bone-free undisturbed yellow silty unit (Layer 4), limestone gravel with occasional larger clasts of limestone. It was the rest of the profile, particularly the North excavation where looser at the top and pale yellow-gray grading to pale yellow or the old 1995–1996 series rat bones were originally sampled from, light brown below. This unit was generally 25–27 cm thick, and was riddled with rabbit burrows. This has resulted in the sloping to the west. It was crossed by a substantial burrow thorough disruption of the chronology of the site, and the observed as coarser and much looser material of the same color previous argument for a secure stratigraphic context for the 1996 which could be scooped out without excavation. series of old rat bone ages (3) at Earthquakes #1 can now be Layer 3: Beneath Layer 1, in the same matrix of compact dismissed. Earlier concerns raised about bioturbation at such yellow-brown silt and limestone clasts, were several lenses of sites going unnoticed are confirmed (2). reddish-brown silt and fine gravel, labeled Layer 3. These extended to about 40 cm depth and lay upon Layer 4 which 3. Stratigraphic Provenance of Bones from Predator Cave. 3.1 Summary extended down to the apparent floor of the shelter at about 75 of previous excavation and radiocarbon dating. Detailed descriptions cm in the western part of the excavation, and to an unknown have been reported for the stratigraphy and cave environment depth below 100 cm in the eastern part. (7) and faunal remains (8, 9) of a laughing owl deposit at In the eastern part of North Extension, substantial bioturba- Predator Cave, North West Nelson. These are summarized tion extended from lower Layer 3 to 45 cm down into Layer 4. briefly below. The laughing owl roost was on a protruding rock As in Layer 1, Layer 4 consisted of very loose material, including ledge within the cave entrance (Fig. S4). The middens were fine pale gray silt, gravel, dried and large airspaces and scattered on a boulder directly below (covering an area Ϸ50 cm filled burrows. The material was mostly the same color as the wide and 20 cm deep) and had fallen down through spaces layer matrix, just faintly darker in places, and it could be scooped between the boulders into a roofed chamber Ϸ300 cm long by 60 out without excavation (Figs. S2 and S3). At least three burrows cm wide (Figs. S5 and S6). Four main ‘‘layers’’ of fossiliferous were recognized, two of them coalescing, and they extended to cave sediment were identified from the original excavation based unknown levels above, below and behind the recorded section. on color and texture, which broadly reflected the decomposition The Layer 4 burrow diameters are about 18–20 cm and they are and mineralization of organics (8, 9). Fossil bones in each of almost certainly the work of the introduced rabbit (Oryctolagus these layers were analyzed separately (7). The uppermost cuniculus) as they contained juvenile rabbit bones. As such they ‘‘black’’ layer (20 cm thick) contained abundant charcoal and must have originated at or near the surface of the deposit and organics. This overlay a less organic rich ‘‘brown’’ layer (15 cm crossed through all stratigraphic units down to about 85 cm from thick), which in turn overlay a layer of white microcrystalline the surface. calcites which thinned away from the direction of the roost site. 2.3 New radiocarbon dating at Earthquakes # 1. The undisturbed This white layer was divided into the ‘‘upper white’’ (1–3 cm yellow sandy sediment unit (Layer 4) contained only native thick) and the more compact ‘‘lower white’’ layer (3 cm thick). faunal elements, and no Rattus exulans bones. This sediment unit Below the four layers were nonfossiliferous gray silts. Bones was originally dated using two optically stimulated luminescence from extinct native fauna, including rat-vulnerable taxa, were dates (1530 Ϯ 150 yrs BP and 1980 Ϯ 150 yrs BP) and then used found along with Rattus exulans bones in all four layers, whereas by extrapolation to support the oldest rat bone dates originally Mus musculus (introduced Ϸ1850s) bones were recovered in the reported from this site (3), all of which were assumed to be in upper two layers (brown and black). This mixture of bones from undisturbed sediments (4–6). These OSL dates and the absence introduced rodents in the profile suggested there was some of any serious bioturbation were key points in support of the 1996 downward reworking of bones over time. series of old rat bone ages from this site (ref. 3, p. 493). However, A single robin (Petroica australis) bone excavated from the our reexcavation has revealed that this extrapolation cannot be base of the lower white layer during the original excavation was supported. Firstly, while the two OSL dates were indeed taken dated at 10,879 Ϯ 83 yrs BP (NZA 3065), suggesting bones had from the same depths as two paired and dated rat and bird bones, been accumulating at the site since at least the early Holocene they were also taken from an undisclosed lateral distance (9). Dates on six Rattus exulans bones from the four layers ranged (perhaps 40 cm) away from the location of the paired rat and bird from 913 Ϯ 69 yrs BP (NZA 5345) to 2155 Ϯ 130 yrs BP (NZA bones. The layers were assumed to be contiguous (3), but we 4916), the latter being the oldest rat bone dated in New Zealand found the layers to slope across the site, a feature also noted by (4). The presence of Rattus exulans bones at the bottom of the Worthy (2), making association of layers by depth alone spuri- profile along with extinct early Holocene fauna indicates the ous. Secondly, one of these pairs of rat and bird bones was taken considerable mixing of more recent material throughout the from a burrow which was assumed to have been made by a sediments at Predator Cave (9). tuatara or petrel and the rat and bird bones encased within the 3.2 Reexcavation of Predator Cave in 2005. We (J.M.W., A.J.A., and burrow deposited before its construction (3). We found that this T.H.W.) returned to Predator Cave in 2005, and collected more burrow was in fact one of many interconnecting burrows made material from these same four layers to provide more bones that

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 2of13 could be radiocarbon dated. We found ‘‘brown’’ sediments in bone of Aegotheles novaezealandiae from the surface layers had previously overlooked crevices, e.g., in areas R and S (Fig. S6), a radiocarbon date age of 12,940 Ϯ 80 yrs BP (OxA-15238), and further ‘‘lower white’’ sediments in infilled cracks between whereas a Mystacina robusta bone from the lowest crystalline limestone clasts in areas B–M (Fig. S6). This latter was from white sediments was also 12,930 Ϯ 65 yrs BP. A Pachyplichas below any excavation unit of the previous excavation, yet rat yaldwyni bone from the brown surface layers returned an age of bones were found in all our excavation areas. Moreover, there 5,535 Ϯ 27 yrs BP (OxA-15234). Reworking by surface water was no evidence of water action having redeposited sediments flows is the most likely cause for this mixing of sediments and between the original excavations (12 August and 1 September bones. As with Earthquakes #1, Predator Cave does not offer a 1992) and our new ones. As with Earthquakes #1, Trevor secure stratigraphy by which to analyze temporal faunal trends, Worthy was involved in the original excavation of Predator Cave, and the old suite of rat bone dates cannot be reproduced. which ensured we returned to exactly the same place to reexca- vate. The four layers were treated separately, and sediments were 4. Stratigraphic provenance of bones from Hawkes Cave. 4.1. Summary wet sieved to separate out all bone remains larger than 1 mm. of previous excavation and new radiocarbon dating. The stratigraphy, Rattus exulans were found in all four layers, and a subsample of diet and biology of the laughing owl site at Hawkes Cave, North these, and bones of rat-vulnerable taxa were selected from each West Nelson (Ϸ1 km from Predator Cave), has been described layer for radiocarbon dating (Table S1). At Predator Cave, we in detail (7, 9). Radiocarbon-dated bones of native fauna from found rat-vulnerable fauna throughout the sedimentary profile, the original excavation revealed Hawkes Cave has been accu- which consisted of four thin layers distinguished by variable mulating bones from 29,000 yrs BP to the present. We (J.M.W., organic sediment. We also found bones of rat-vulnerable fauna A.J.A., and T.H.W.) returned to Hawkes Cave and collected mixed with Rattus exulans bones throughout the sediment pro- more sediments from the Shield Area, Area 10 and the Roost file. Site (7) from which we dated seven rat-vulnerable species that 3.3. New radiocarbon dating at Predator Cave. Our dates from rat- ranged in age from 2,429 to 12,500 yrs BP (Table S1). Two Rattus vulnerable fauna and Pacific rat bones (Table S2) in each layer exulans bones selected from the Shield Area were also dated: one reveal that the former can be much older by thousands of years, from our new excavation, the other from Trevor Worthy’s and that there is no chronological order with depth in either set, collection from the original excavation, and both were no older contrary to what was previously implied (4, 6). For example a than 350 yrs BP (Table S2).

1. Wilmshurst JM, Higham TFG, Allen H, Johns D, Phillips C (2004) Early Maori settlement 7. Holdaway RN, Worthy TH (1996) Diet and biology of the laughing owl Sceloglaux impacts in northern coastal Taranaki, New Zealand. NZ J Ecol 28:167–179. albifacies (Aves: Strigidae) on Takaka Hill, Nelson, New Zealand. J Zool London 2. Worthy TH (1998) Quaternary fossil faunas of Otago, South Island, New Zealand. J Roy 139:545–572. Soc NZ 28:421–521. 8. Worthy TH, Holdaway RN (1996) Taphonomy of two Holocene microinvertebrate 3. Holdaway RN, Roberts RG, Beavan-Athfield NR, Olley JM, Worthy TH (2002) Optical deposits, Takaka Hill, Nelson, New Zealand, and identification of the avian predator dating of quartz sediments and accelerator mass spectrometry 14C dating of bone responsible. Historical Biol 12:1–24. gelatin and moa eggshell: a comparison of age estimates for non-archaeological 9. Worthy TH, Holdaway RN (1994) Quaternary fossil faunas from caves in Takaka Valley deposits in New Zealand. J Roy Soc NZ 32:463–505. and on Takaka Hill, northwest Nelson, South Island, New Zealand. J Roy Soc NZ 4. Holdaway RN (1996) Arrival of rats in New Zealand. Nature 384:225–226. 24:297–392. 5. Anderson AJ, Higham TFG (2004) The age of rat introduction in New Zealand: further 10. Bronk Ramsey C, Higham T, Bowles A, Hedges R (2004) Improvements to the pretreat- evidence from Earthquakes #1, North Otago. NZ J Archaeol 24:135–147. 6. Holdaway RN (1999) A spatio-temporal model for the invasion of the New Zealand ment of bone at Oxford. Radiocarbon 46:155–163. archipelago by the Pacific rat Rattus exulans. J Roy Soc NZ 29:91–105.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 3of13 Fig. S1. (Upper) Earthquakes #1 plan view. (Lower) Section of stratigraphy in the North Extension Excavation. A–A1 represents string line. Bar to left of excavation is stratigraphy after Worthy (2). See Table S1 for details of bones dated from excavation areas.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 4of13 Fig. S2. View of the vertical section of the Earthquakes #1 North Extension outlining area of disturbed gravely sediments in yellow-gray, yellow-brown silts caused by rabbit burrowing at North Extension. Rabbit burrow Ϸ18–20 cm wide. Rattus exulans bone dated from both within and to the side of these burrows returned similar ages (see Table S1).

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 5of13 Fig. S3. Extensive interconnecting network of rabbit burrows showing extensive bioturbation revealed during the excavation of Earthquakes #1 North Extension above and left of the string line A–A1.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 6of13 Fig. S4. Cave entrance of Predator Cave, looking out from laughing owl roost site.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 7of13 Fig. S5. Sediment layers beneath laughing owl roost site at Predator Cave, showing brown sediments overlying upper white and lower white sediments in Areas R and Q.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 8of13 Fig. S6. Plan section of Predator Cave re-excavation. See Table S1 for details of bones dated from excavation areas.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 9of13 Table S1. AMS determinations and associated analytical data for samples of bone collagen from the sites investigated in this article 14C Use Gelatin Sample age Std. weight yld %C Site reference Species Element OxA BP error CN ␦13C ␦ 15N (mg) (mg) yld

Earthquakes Otago Museum Rattus exulans Femur 14472 586 30 3.2 Ϫ19.9 8.2 140 16.3 41.6 #1 (OL 444: VT 774) (Pacific rat) Layers 1, 2, 3, and 0–30 cm test Against Cliff trench equate to original Otago Museum R. exulans Femur 14513 411 25 3.2 Ϫ19.6 11.3 120 12.7 42.5 layers described by (OL 444: VT 782) Worthy (5) 50–150 cm ЉAgainst CliffЉ Otago Museum R. exulans Femur 14566 534 29 3.2 Ϫ19.4 9.1 60 6 42.6 (OL 444: VT 794) 150–250 cm ЉAgainst CliffЉ Layer 1, above R. exulans R. pelvis 14646 570 25 3.1 Ϫ19.4 10.9 55.4 4.9 41.0 string Layer 1, above R. exulans Tibia 14647 389 26 3.1 Ϫ19.9 10.9 49.5 4.7 39.0 string Layer 1, above R. exulans L. pelvis 14648 369 26 3.1 Ϫ20.0 9.1 42.7 4.2 41.1 string Layer 2, 0–10 cm R. exulans R. pelvis 14649 516 26 3.1 Ϫ20.0 10.9 69.2 8.9 40.9 below string Layer 2, 0–10 cm R. exulans R. femur 14650 354 26 3.1 Ϫ19.9 11.6 99.2 9.7 39.9 below string Burrow 1, Layer 1 R. exulans L. pelvis 14651 475 26 3.1 Ϫ20.0 11.5 78.2 10.2 42.0 Burrow 1, Layer 1 R. exulans Cranial fragment 14652 549 26 3.1 Ϫ21.9 9.3 76.2 7.5 39.4 Layer 2, 10–20 cm R. exulans R. femur 14653 563 26 3.1 Ϫ21.2 11.1 137.2 13.1 41.3 below string Layer 2, 10–20 cm R. exulans R. pelvis 14654 518 26 3.1 Ϫ19.6 9.1 79.1 9.8 41.1 below string Rabbit warren R. exulans Femur 15198 100 45 3.2 Ϫ20.4 10.9 30 4.2 40.9 complex, burrows2&3 in Layer 3 Layer 3, basal Mystacina Lower mandible 15195 2240 32 3.3 Ϫ21.1 13.2 60 4.9 41.5 yellow silt robusta Rabbit warren Gallinula L. femur 15196 1249 25 3.3 Ϫ25.4 8.9 260 33.2 42.8 complex, hodgenorum burrows2&3 in Layer 3 Rabbit warren Sphenodon sp. Vertebra 15197 1029 25 3.2 Ϫ20.1 9.7 140 15.6 41.4 complex, burrows2&3 in Layer 3 Layer 3, 10–26 cm Pelecanoides L. humerus 15224 1632 25 3.2 Ϫ15.5 13.0 320 41.9 42.8 Layer 3, 10–26 cm Sphenodon sp. L. dentary 15225 1687 25 3.1 Ϫ20.3 8.6 100 8.4 40.9 Layer 2, 0–10 cm Sphenodon sp. R. maxilla 15226 1685 26 3.1 Ϫ20.2 9.6 100 11.4 41.0 below string Layer 2, 0–10 cm Mystacina Radius 15227 1644 26 3.2 Ϫ19.1 14.4 40 4.2 41.4 below string robusta Layer 2, 0–10 cm Puffinus L. tarsometatarsus 15228 1591 24 3.2 Ϫ13.7 17.3 420 55.6 42.8 below string gavia Layer 1, above Sphenodon L. humerus 15229 500 100 3.1 Ϫ20.1 12.1 140 19.6 42.5 string Layer 3, basal Xenicus sp. L. tarsometatarsus 15299 1629 28 3.2 Ϫ21.6 13.2 20 3 40.8 yellow silt Burrow 1, Layer 1 Sphenodon sp. Mandible 15300 1787 28 3.1 Ϫ20.6 11.4 50 2.8 40.5 Burrow 1, Layer 1 Sphenodon sp. Skull 15301 1794 28 3.1 Ϫ20.6 11.5 50 3.9 41.5 Layer 3, 10–26 cm Mystacina Mandible 15302 1768 27 3.2 Ϫ20.3 16.5 40 3.7 40.9 robusta Hawke’s Cave Shield Area Aegotheles L. tarsometatarsus 15230 9915 50 3.3 Ϫ19.1 4.6 300 27.3 42.9 novaezealandiae Shield Area Aegotheles L. tarsometatarsus 15231 12500 50 3.2 Ϫ19.8 4.0 180 12.12 37.0 novaezealandiae

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 10 of 13 14C Use Gelatin Sample age Std. weight yld %C Site reference Species Element OxA BP error CN ␦13C ␦ 15N (mg) (mg) yld

Site 10 Mystacina Mandible 15232 12370 50 3.1 Ϫ19.1 6.9 140 12.6 39.3 robusta Site 10 Mystacina Mandible 15233 11630 55 3.2 Ϫ19.7 7.4 100 8.5 39.6 robusta Site 10 Pachyplichas R. tarsometatarsus 15234 2429 27 3.2 Ϫ18.0 6.6 80 6.1 41.1 yaldwyni Roost Site Coenocorypha Distal 15303 10910 80 3.2 Ϫ19.0 6.1 80 2.9 40.7 iredalei R. tarsometatarsus Roost Site Aegotheles R. humerus 15235 6404 34 3.2 Ϫ18.1 1.1 300 16.1 39.1 novaezealandiae Owl Deposit 1 R. exulans Mandible 14869 228 25 3.2 Ϫ19.1 2.7 31.4 2.9 42.9 (from T. Worthy original collection) Shield Area R. exulans Mandible 14870 348 26 3.2 Ϫ18.4 4.6 50.9 4.7 42.2 Predator Cave Brown layer on R. exulans R. femur 14821 156 27 3.3 Ϫ18.2 4.8 140 14.7 42.8 rockfall below roost Brown layer on R. exulans Tibia 14871 205 25 3.2 Ϫ18.1 4.6 100 7.8 47.5 rockfall below roost Lower White R. exulans Pelvis 14872 393 25 3.2 Ϫ18.4 -0.3 40 2.8 46.9 layer, Area D Lower White R. exulans Mandible 14873 170 24 3.2 Ϫ18.2 3.4 20 3.5 45.9 layer, Area E Lower White R. exulans Ulna 14874 412 25 3.2 Ϫ18.4 0.4 40 2.6 44.0 layer, Area P Lower White R. exulans Humerus 14875 213 24 3.3 Ϫ18.3 3.0 40 3.2 45.5 layer, Area R Lower White R. exulans Pelvis 14822 175 27 3.2 Ϫ18.1 3.6 80 9.5 45.0 layer, Area Q Upper White R. exulans Ulna 14876 479 25 3.3 Ϫ17.9 5.0 40 3.9 41.5 layer (Museum collection ex MNZ S32398) Black layer R. exulans Humerus 14823 382 28 3.3 Ϫ18.8 11.5 60 5.8 45.8 (Museum collection ex MNZ S32348) Lower White Mystacina robusta Mandible 15336 6115 50 3.2 Ϫ17.4 7.7 40 2.7 39.6 layer, Area D Lower White Pachyplichas R. tibia 15337 9705 70 3.3 Ϫ18.3 4.0 80 3 34.7 layer, Area D yaldwyni Lower White Aegotheles R. ulna 15236 8380 40 3.2 Ϫ18.9 4.9 80 7.8 60.0 layer, Area E novaezealandiae Lower White Mystacina Mandible 15237 12930 65 3.2 Ϫ18.8 6.9 120 6 37.7 layer, Area Q robusta Brown layer, Pachyplichas R. tarsometatarsus 15338 5535 45 3.3 Ϫ17.7 4.9 60 3 33.7 Area S yaldwyni Upper rockfall Aegotheles L. tarsometatarsus 15238 12940 80 3.2 Ϫ20.0 3.8 100 4.9 40.8 just under Black novaezealandiae layer Hanging Rock Site 3, (Museum R. exulans R. femur 14824 458 27 3.2 Ϫ20.0 10.4 60 4.7 42.6 collection ex MNZ S34398, 1235/76) Ardenest Museum R. exulans R. femur OxA- 103 29 3.2 Ϫ20.0 7.0 100 7.4 45.6 collection ex X- MNZ S33701, 2136–38 7088/441 Timpendean Owl Museum R. exulans R. femur OxA- 200 30 3.2 Ϫ18.9 12.0 160 8.49 43.1 Falcon site collection ex X- MNZ S33413 2136–39

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 11 of 13 14C Use Gelatin Sample age Std. weight yld %C Site reference Species Element OxA BP error CN ␦13C ␦ 15N (mg) (mg) yld

Gordon’s Valley Site 7, (Museum R. exulans 1Lϩ 1 R tibia 14867 121 25 3.1 Ϫ19.6 7.2 89.5 6.4 42.5 collection ex MNZ S34332) Site 2a, lower R. exulans R. femur 14842 405 26 3.3 Ϫ19.4 8.9 84.1 8.2 43.8 ledge, (Museum collection ex S34193, 135/6) Site 4, upper R. exulans L. femur 14868 434 25 3.1 Ϫ20.2 9.2 60.6 5.4 43.7 layer, (Museum collection ex MNZ S34268)

All bone measured is ultrafiltered rat bone gelatin obtained using the methods outlined in Bronk Ramsey et al. [Bronk Ramsey C, Higham T, Bowles A, Hedges R (2004) Improvements to the pretreatment of bone at Oxford. Radiocarbon 46:155–163.]. Use weight is the weight of bone analysed. %C yield is yield of carbon from ultrafiltered gelatin on combustion. Gelatin yield is the yield of gelatin in milligrams after ultrafiltration and lyophilization. ␦ 13C and ␦15N are reported with reference to VPDB and AIR respectively and expressed in ‰. Measurement error for C isotope ratios is Ϯ0.2‰ and for N isotope ratios is Ϯ0.3‰. C to N atomic ratios are accepted at ORAU when they range between 2.9 and 3.5. Higher CNs can indicate the addition of exogenous carbon.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 12 of 13 Table S2. AMS radiocarbon dates of woody seed cases from New Zealand sites reported in this article Sample Species of woody State of 14CLaboratory 14C Std Site name depth (cm) seed case dated seed case code age BP error ␦13C

Nguroa Bay, NW Nelson Surface Rat-gnawed OxA-14913 223 26 Ϫ26.3 Nguroa Bay, NW Nelson 35 Prumnopitys ferruginea Rat-gnawed OxA-14935 215 28 Ϫ24.0 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Rat-gnawed OxA-14915 526 27 Ϫ24.9 Nguroa Bay, NW Nelson 10–20 Prumnopitys ferruginea Rat-gnawed OxA-14936 223 26 Ϫ25.7 Nguroa Bay, NW Nelson 20–30 Prumnopitys ferruginea Rat-gnawed OxA-14917 231 26 Ϫ23.2 Nguroa Bay, NW Nelson 30–40 Prumnopitys ferruginea Rat-gnawed OxA-14918 244 26 Ϫ23.9 Nguroa Bay, NW Nelson 0–10 Rat-gnawed Wk-17875 278 32 Ϫ25.1 Nguroa Bay, NW Nelson 0–10 Prumnopitys taxifolia Rat-gnawed Wk-17876 316 32 Ϫ24.5 Nguroa Bay, NW Nelson 0–10 Prumnopitys taxifolia Rat-gnawed Wk-17877 395 34 Ϫ24.9 Nguroa Bay, NW Nelson 0–10 Prumnopitys taxifolia Rat-gnawed Wk-17878 281 34 Ϫ26.2 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Rat-gnawed Wk-17879 296 32 Ϫ23.1 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Rat-gnawed Wk-17880 290 37 Ϫ23.9 Nguroa Bay, NW Nelson Surface Prumnopitys taxifolia Rat-gnawed Wk-17881 153 32 Ϫ23.9 Nguroa Bay, NW Nelson Surface Prumnopitys taxifolia Rat-gnawed Wk-17882 383 33 Ϫ24.7 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Rat-gnawed Wk-17883 702 32 Ϫ23.2 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Rat-gnawed Wk-17884 324 31 Ϫ24.8 Nguroa Bay, NW Nelson 35 Prumnopitys ferruginea Intact OxA-14914 166 27 Ϫ23.1 Nguroa Bay, NW Nelson 30–40 Prumnopitys ferruginea Intact OxA-14919 196 25 Ϫ23.8 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Intact Wk-17885 497 35 Ϫ24.7 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Intact Wk-17886 170 32 Ϫ25.7 Nguroa Bay, NW Nelson 0–10 Prumnopitys taxifolia Intact Wk-17887 150 35 Ϫ24.3 Nguroa Bay, NW Nelson 0–10 Prumnopitys taxifolia Intact Wk-17888 347 28 Ϫ25.9 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Intact Wk-17889 245 29 Ϫ24.8 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Intact Wk-17890 751 29 Ϫ24.8 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Bird-cracked OxA-14912 226 25 Ϫ24.6 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Bird-cracked OxA-14916 389 26 Ϫ24.6 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Bird-cracked Wk-17892 949 28 Ϫ24.9 Nguroa Bay, NW Nelson 0–10 Prumnopitys ferruginea Bird-cracked Wk-17893 249 29 Ϫ26.3 Nguroa Bay, NW Nelson Surface Prumnopitys ferruginea Bird-cracked Wk-17891 877 29 Ϫ24.3 Long Beach, Dunedin 20–30 Elaeocarpus hookerianus Intact OxA-16117 788 24 Ϫ23.8 Long Beach, Dunedin 20–30 Prumnopitys taxifolia Intact OxA-16118 982 24 Ϫ20.1 Long Beach, Dunedin 20–30 Prumnopitys ferruginea Bird-cracked OxA-16119 1200 25 Ϫ22.8 Long Beach, Dunedin 20–30 Prumnopitys taxifolia Bird-cracked OxA-16120 1300 25 Ϫ22.8 Waitoetoe, Taranaki 40–45 Prumnopitys ferruginea Intact OxA-16109 1329 25 Ϫ23.5 Waitoetoe, Taranaki 20–25 Prumnopitys ferruginea Bird-cracked OxA-16108 374 23 Ϫ21.6 (only seed) Waitoetoe, Taranaki 40–45 Prumnopitys ferruginea Bird-cracked OxA-16110 1321 25 Ϫ27.7 Waitoetoe, Taranaki 40–45 Prumnopitys ferruginea Bird-cracked OxA-16111 1396 24 Ϫ22.6 Waitoetoe, Taranaki 40–45 Prumnopitys ferruginea Bird-cracked OxA-16112 1334 24 Ϫ23.4 Waitoetoe, Taranaki 40–45 Prumnopitys ferruginea Bird-cracked OxA-16113 1541 26 Ϫ21.5 Wharariki, NW Nelson 50–60 Intact OxA-16121 1913 25 Ϫ21.5 Wharariki, NW Nelson 50–60 Elaeocarpus dentatus Intact OxA-16122 1856 25 Ϫ27.4 Wharariki, NW Nelson 50–60 Elaeocarpus hookerianus Intact OxA-16123 1903 25 Ϫ27.5 Wharariki, NW Nelson 50–60 Prumnopitys taxifolia Intact OxA-16124 2003 26 Ϫ21.3 Cannibal Bay, Ditch, Catlins 0–20 Prumnopitys ferruginea Intact Ox-A-14920 2252 31 Ϫ26.4 Cannibal Bay, Ditch, Catlins 0–20 Prumnopitys ferruginea Bird-cracked Ox-A-14921 2512 32 Ϫ24.3 Cannibal Bay, Ditch, Catlins 50–60 Prumnopitys ferruginea Bird-cracked Ox-A-14922 2462 29 Ϫ24.5 Cannibal Bay, Forest, Catlins 20–30 Prumnopitys ferruginea Bird-cracked Ox-A-14923 1985 AD bomb curve corrected 2 Ϫ24.7 Cannibal Bay, Forest, Catlins 20–30 Prumnopitys ferruginea Rat-gnawed Ox-A-14924 1992 AD bomb curve corrected 2 Ϫ23.4 Nguroa Bay Rd, NW Nelson 20 Prumnopitys ferruginea Rat-gnawed OxA-16114 216 23 Ϫ21.6 Nguroa Bay Rd, NW Nelson 20 Prumnopitys ferruginea Intact OxA-16115 215 23 Ϫ22.7 Nguroa Bay Rd, NW Nelson 10 Prumnopitys ferruginea Intact OxA-16116 83 23 Ϫ23.6

See Table S1 legend for details of mass spectrometric measurements. All seeds were pretreated using an acid–base–acid sequence and bleaching to yield holocellulose.

Wilmshurst et al. www.pnas.org/cgi/content/short/0801507105 13 of 13