Conservation paleobiology on Minami-Daito Island, Okinawa, Japan: anthropogenic extinction of cave-dwelling bats on a tropical oceanic island

Skeletal remains in the caves of Minami-Daito Island

There are countless sinkholes (dolines) and caves on Minami-Daito Island, some of which preserve the remains of extirpated bats. We collected vertebrate skeletal remains and bat guano-like deposits from two of these caves, which are a public cave called Hoshino Cave (also called Hoshinod$\bar{o}$ or Hoshinodo Cave, as “d$\bar{o}$” means “cave” in Japanese) and a privately-owned cave informally called “Cave A” (Fig. 1A). Fieldwork was conducted in January 2016 and January 2018 because the CO₂ concentration levels in the caves are lowest during the winter. All fieldwork was conducted outside national parks or restricted areas designated by the Nature Conservation Act.

Cave A is accessible via several spots where the ceiling has collapsed, but we entered through the main natural entrance, which is located in the Hagu Shita basin near the basin-side edge of the Hagu Ue rim (Fig. 1C). Because there is no published map for Cave A, we made a simple route map from the main entrance to our sample locations (Fig. 1D). Skeletal remains of insectivorous bats were collected from a small hall about 60 m north of the main entrance. Generally, skeletal remains are scattered, suggesting that these individuals were not transported by water flow (Figs. 2A–2B). In Cave A, we observed that at least four individuals were almost completely articulated and embedded by thin crusts of flowstone, which have served as protection against dissociation and bone decay (Fig. 3). Two of these individuals were collected (Fig. 2D). For others, bones are disarticulated and selectively preserved with a bias toward long bones (radius, metacarpal, etc.), tympanic bullae, and jaws. In Hoshino Cave, we entered from a natural entrance, which is not open to the public, and found an incomplete skeleton of one individual of Rhinolophus sp. several meters below a paved commercial route (Fig. 1E).

Fragile bones were reinforced in the lab by coating with a solution of 5% Paraloid B-72 dissolved in acetone. They are stored at the National Museum of Nature and Science, Tokyo (NMNS; Tsukuba, Ibaraki, Japan). A complete list of collected bones is provided in Table 1, and a detailed taxonomic study of the specimens is in progress.

Some well-preserved fragments without secondary calcite crystal growth or acid etching were consumptively sampled for radiocarbon dating; however, none preserved collagen (i.e., no signal was detected for nitrogen using Elemental Analysis [EA] on a bone fragment), so these materials were not useful for direct dating. Because we were unable to determine the age of any individual fossil, we instead examined guano-like deposits accumulated in the same caves for time-averaged ¹⁴C ages.

Guano-like deposits on Minami-Daito Island

Guano-like deposits were sampled at four localities in Cave A (Fig. 1D) and one locality in Hoshino Cave (Fig. 1E). These deposits are accumulated in piles on the floors of the caves that can be washed downslope. They are soft muddy sediments brownish-black (10YR2/2 to 10YR2/3) to black (10YR2/1) in color. Mud particles are loosely coalesced to form sand-sized granules, which are often contaminated with dolomite fragments. They are also present in local small depressions above the floor. Flowstones in the caves are often colored black from pollution in areas where the Japanese military used them as headquarters during World War II (WWII; “Headquarter Hall” on the map) and where groundwater drains. We collected guano-like deposits from large piles (Fig. 4A) or handpicked coalesced particles using forceps (Fig. 5A). The largest pile of guano-like deposits (8∼10 m wide on the surface) was found in Hoshino Cave, which is probably the deposit noted by Shimojana (1978). The thickness of the pile is over 50 cm at a random spot where we collected samples from its bottom, a mid-part, and its top. There is no sedimentological difference observed except that the bottom sample contained more carbonate lithic fragments. In Cave A, a thin crust under a bat-lying flowstone was also sampled (Figs. 5B, 5D). The sampled deposits were dried in a convection oven at 38 °C and were sieved at 850 µm, 425 µm, and 250 µm. Samples between 250 µm and 425 µm were carefully handpicked using forceps for black particles to avoid carbonate contamination prior to further analyses.

Because insectivorous bats consume arthropods (crickets, beetles, moths, etc.), bat guano contains their chitinous exoskeletons, which are resistant to decay. We utilized established signals of the chitinous exoskeletons of arthropods to determine whether or not the guano-like deposits are guano-derived by comparison with modern and fossil bat guanos as positive references and red soil as a negative reference. The red soil was collected from the bottom of thick soil deposits beneath a fissure in Cave A.

Modern and subfossil bat guano as positive references

For comparison with the guano-like deposits sampled in Hoshino Cave, we used a commercial guano fertilizer collected in Indonesia (called R1 in figures; ARK Co. Ltd., Ushiku, Ibaraki, Japan), modern bat guano of Miniopterus fuliginosus (R2) from an unnamed artificial cave of soft sandstone in Taito, Isumi, Chiba, subfossil guano deposit (R3) from Fujido Cave in Udeno, Gunma, and fecal pellets (R4) of Vespertilio sinensis in Chichibu, Saitama. Localities were selected by Y.K. for ease of access. These samples were used as “positive references” for comparison with our unknown guano-like samples from the Minami-Daito Island caves (see Article S1).

Scanning Election Microscopy (SEM) observation

In the fecal pellets of modern insectivorous bats, arthropod exoskeleton fragments are usually observed. Thus, we imaged dry samples of the guano-like deposits from Hoshino Cave and the four guano references at the National Museum of Nature and Science (NMNS; Tsukuba, Ibaraki, Japan) using a JSM-6510 (JEOL Ltd., Tokyo, Japan) scanning electron microscope (SEM) under 70x to 300x magnification. Images were taken at an acceleration voltage of 3 kV with gold sputter coating on the dry samples.

Fourier Transform Infrared Spectroscopy (FTIR)

Fourier transform infrared (FTIR) spectroscopy is a technique commonly used to characterize specific molecular structures in organic compounds. The FTIR spectra of chitin extracted from modern and fossil guano as well as those of bulk guano are well known based on previous studies (e.g., Wurster et al., 2010; Kaya et al., 2014), so we used FTIR to identify whether chitin is a major component of the guano-like deposits. The hand-picked samples of the guano-like deposits and reference materials (modern and fossil guano, red soil) were refluxed with 1.0M HCl at 80 °C for at least 8 h to remove exogenous carbonates and organic acids (e.g., fluvic acids) and were rinsed with deionized water until the solvent became a neutral pH. After drying in a convection oven below 38 °C, the samples were ground in an agate mortar with a pestle. They were then stored in a desiccator at room temperature until analysis.

Each powdered sample was diluted with ground KBr, pressed into a pellet in a stainless-steel disk, and analyzed without a vacuum using a JASCO FT/IR-6800 (JASCO Inc., Tokyo, Japan) at NMNS. Pure KBr was measured as the background under the same conditions as the samples. The infrared spectra of absorbance were measured from 400 cm⁻¹ to 4,000 cm⁻¹ by 64 scans at a resolution of 4 cm⁻¹ and are expressed as the percentage of transmittance (%T). Automatic baseline corrections were made using a second-degree polynomial fitting (x²).

Stable isotope analyses and C:N ratio of the guano-like samples

To supplement FTIR results of the guano-like samples, we measured stable carbon and nitrogen isotope values and mass ratios of organic carbon to nitrogen (C:N) for the HCl-treated samples using a FLASH 2000 CHNS/O elemental analyzer coupled with a Finnegan MAT253 isotope ratio mass spectrometer (Thermo Fisher Scientific, Massachusetts, USA) at NMNS. Isotopic ratios are expressed in delta notation (δ¹³C, δ¹⁵N) in parts per thousand (‰) and reported on the VPDB scale for δ¹³C values and the AIR scale for δ¹⁵N values. For guano-like deposits, a thick sample from Hoshino Cave and a handpicked pellet-like sample from Cave A were selected for analysis. Three internal lab standards were analyzed for each run and used for data correction. Repeated analyses of the standards were within ±0.2‰ for δ¹³C and ±0.25‰ for δ¹⁵N. Carbon isotope values of guano references are used to reconstruct isotopic signals of their diet by taking the following enrichment factors into consideration. The enrichment factor between the feces of insectivorous bats and their diet (i.e., insects) is known to be negligible (Salvarina et al., 2013), whereas larger enrichment factors are observed between the cuticles of insects and their diet of plant matter. In an experimental study, Gratton & Forbes (2006) showed that tissues of predacious beetles are more enriched by 2.2‰ relative to their aphid diet.

Stable carbon isotopes of bone apatite

Stable carbon isotope analysis of animal tissues is a useful technique to detect an isotopic signal of C₄ plants in their habitat. Although no C₄ plant is native to the Daito Islands, historical records (Editorial Committee of the History of Minami-Daito Village, 1990) document that the C₄ perennial grass sugarcane was introduced on Minami-Daito Island in 1899–1900 and increasingly replaced native C₃ forests on the island by the 1930s (see Introduction). We therefore hypothesize that any organic materials on the island with the isotopic signature of C₄ plants must have originated after the introduction of sugarcane in 1899–1900. Thus, stable carbon isotope analysis with correct applications of fractionation factors can serve as a time constraint for direct human influences on Minami-Daito Island. Note that no scientific study such as palynological analysis of sediment cores is available to independently validate the historical documents.

Relatively well-preserved bat bones (no acid etching, no or limited calcite crystallization) from Cave A were selected for carbon isotope analysis of bone apatite. For each bone, the surface was shaved off to remove calcite crystals and carbonate deposits as much as possible from the thin bones. The powdered samples were soaked in 2% NaOCl overnight for at least 12 h at the room temperature. After rinsing with deionized water, they were treated with buffered 0.1M acetic acid (pH = 4.2) for 2 h at 40 °C until no obvious bubbles were observed, and they were neutralized with ultrapure water and dried in an oven at 40 °C. The treated samples were sent to Shoko Science Co., Ltd. (Saitama, Japan) and analyzed for stable carbon isotope via a H₃PO₄ digestion using by a continuous flow Thermo Scientific Delta V Plus isotope ratio mass spectrometer coupled to a Thermo Scientific GasBench II. Of the eight samples analyzed in this study, seven come from individual bones, whereas one sample is composed of multiple small bone fragments.

The carbon isotope values of bone apatite were used to reconstruct isotopic signals of vegetation by applying experimentally known enrichment factors (ɛ*) and binary mixing between C₃- and C₄-endmembers. The enrichment factor applied in this study is +11.0 ± 0.1‰ between bone apatite and insectivorous diet (Podlesak et al., 2008). For isotopic spacing (δ) between insectivorous diets of bats and surrounding vegetation, we applied +0.8‰ or +3.0‰ based on an a dietary-switch experimental study by Gratton & Forbes (2006). For the end-members in binary mixing, δ¹³C values of −12.5‰ and −29. 1‰ were used for pure C₄ (Cerling et al., 1997) and C₃ vegetation, respectively. The latter was calculated based on the equation of Kohn (2010), which takes mean annual precipitation (MAP), altitude, and latitude into account. MAP of Minami-Daito Island between 1942 and 2020 was calculated as 1680 mm/year, which was applied to the equation. See Article S1 for more details.

Radiocarbon dating of subfossil guano references

We determined the ¹⁴C age of both the fossil guano reference (R3) and the guano-like samples for comparison. The same samples analyzed for stable isotopes were chosen for radiocarbon dating. Prepared samples were sent to the accelerator mass spectrometry (AMS) facility at the University Museum, the University of Tokyo, where the samples were further prepared and measured by Compact AMS System (National Electrostatics Corp., Wisconsin, USA), following their in-house protocols (Omori et al., 2017). For each sample, δ¹³C measured by AMS was used to correct for the conventional radiocarbon age (Stuiver & Polach, 1977). These radiocarbon ages were calibrated by IntCal20 curves (Reimer et al., 2020) in OxCAL4.2 (Bronk Ramsey, 2009) and are expressed as calBP (Before Present; 0 calBP = AD 1950). Preparation and analytical methods are described in Article S1.

Skeletal remains of locally extinct cave-dwelling bats

In a survey during the 1970s, Shimojana (1978) reported a few skeletons of Rhinolophus sp. from Hoshino Cave. In this study, we found the skeletons of single individuals of Rhinolophus sp. in both Hoshino Cave and Cave A. Additionally, we collected skeletons of Miniopterus sp. from Cave A, which represent at least 13 individuals (taxonomic study in progress). Some skeletons are articulated, but many are disconnected with associated bones closely scattered on the surface of the cave floor (Fig. 2), so we infer that they were not transported by water after death. No traces of soft tissues such as pelage are preserved. In Cave A, we observed at least four individuals of Miniopterus sp. that were embedded within a thin crust of flowstone (Fig. 3), two of which were collected. No fossil remains are permineralized, but many are encrusted with calcite. The condition of calcite crystal growth on and in bones depends on the availability of mineral-rich water flow near the bones. In many bone materials, the surface is partially covered with small calcite crystals, and/or the medullary cavity of long bones is partially infilled.

Due to the lack of bone collagen for radiocarbon dating in the specimens we sampled, it is unknown whether these bat species utilized the cave contemporarily; however, because there is no difference in bone preservation and secondary crystal growth of calcite between the two species, our paleontological interpretation is that the ages of their existences do not differ even if they may not be coeval sensu stricto.

SEM observation of guano-like samples

Fragments of insect exoskeletons are clearly visible in both modern (Figs. 4D–4E) and fossil guano references (Fig. 4G) at or above 70x magnification although the fossil guano is degraded to be humus-like and contaminated with sediments (Fig. 4F). On the other hand, the guano-like samples appear to be aggregates of clay-sized particles with no clear presence of insect remains (Figs. 4B–4C), suggesting that they may not be composed of bat guano. Nevertheless, because the preservation potential of chitinous insect remains is expected to be lower in guano deposits from tropical wet caves like those on Minami-Daito Island than in the fossil guano reference collected from a more temperate region, we proceeded with further analyses.

FTIR

None of the guano-like deposits collected in this study present spectral patterns characteristic of fresh or fossil guano (Figs. 6 and 7, see Article S1 for full description of references). In the range between 1,000 cm⁻¹ and 1,800 cm⁻¹, different spectral patterns are observed between guano references and sampled deposits. Particularly, a shoulder peak at 1,712 cm⁻¹ associated with C=O stretching of carboxylic acid (Dick, Santos & Ferranti, 2003) is present in guano-like samples but does not occur in the spectra of guano references (Fig. 6). This peak occurs in humified organic matter (Dick, Santos & Ferranti, 2003; Palumbo et al., 2018).

The FTIR spectrum of the guano-like sample Y180119g3 (Fig. 6D) is comparable to that of humic acid (Figs. 6H–6I). The absorption at 1,078 cm⁻¹ is ascribed to the stretching vibration of oxygen bonds in aliphatic ether (Zhang et al., 2015). The presence of aliphatic compounds is consistent with the interpretation that this sample is a fraction of humic substances (Méndez, 1967). The peak at 1,244 cm⁻¹ can be interpreted as antisymmetric C-O stretching and OH deformation in a carboxyl group, and -C-OH bending of phenols and tertiary alcohol in humic substances (Dick, Santos & Ferranti, 2003). As organic functional groups, carboxylic acids are contained in humic substances. We interpret that the guano-like deposits sampled in this study are forms of humic substances rather than fossil bat guano, with the exception of Y160114g1, which was identified as hydroxyapatite (Fig. 7D) by comparing characteristic spectral peaks with those of pure hydroxyapatite in Jayaweera et al. (2018).

The hydroxyapatite sample, Y160114g1, occurs as a thin crust light brown in color only partially exposed under a fracture of more recent flowstone on the slope of a large stalagmite (Fig. 5B). Interestingly, we found skeletal remains of at least two individual bats lying on a slope of the stalagmite, both of which are also covered by recent flowstone (Fig. 3B). The hydroxyapatite crust stratigraphically underlies the bat skeleton. As a calcium phosphate mineral, hydroxyapatite is commonly precipitated in caves where phosphorous-rich organic deposits such as bat guano or bones are supplied and interaction between the guano-derived leachate and calcium-rich water from carbonate host-rock can occur at a pH greater than 6 (Fiore & Laviano, 1991; Hill & Forti, 1997; Tămaş, Miheţ & Giurgiu, 2014; Giurgiu & Tămaş, 2013). Because this hydroxyapatite crust was found in association with a bat skeleton, and hydroxyapatite precipitation is inhibited in the presence of humic and fluvic acids (Inskeep & Silvertooth, 1998), this crust was likely derived from bat guano rather than humified plant material.

C:N ratios, δ¹³C, and δ¹⁵N values of the guano-like samples

The C:N ratios, total nitrogen by weight percent (%N), δ¹³C values, and δ¹⁵N values of the guano-like samples are shown in Fig. 8 (Table 2, Table S1). In the bulk sample of the fissure-filled deposit, total nitrogen is nearly 0.1%, which is within the standard range of nitrogen content for sediments (Forbes & Bestland, 2006). The nitrogen content in the fecal pellet-like sample Y180119g2 (Fig. 5A) from Cave A is lower than the %N range of guano references (Fig. 8A). Because the FTIR spectrum of the pellet-like sample similarly traces that of the fissure-filled deposit, we interpret that the fecal pellet-like sample is a mix of sediment and organic matter. In general, C:N ratios of all analyzed samples are tightly clustered, ranging from 4.2 for fissure-filled deposit to 6.5 for the fossil guano reference with the unknown guano-like samples (4.8 on average) equivalent to those of modern guano (R1 and R2) (Fig. 8B). These values are much lower than fresh plants and coals (e.g., Schmidt & Gleixner, 2005) and superficially meet one of Forbes & Bestland’s (2006) criteria to identify fossil guano (C:N < 10). Nevertheless, in our case, the low C:N values of the samples are still consistent with the FTIR-based classification of the red soil of the Minami-Daito Islands as a lateritic soil because it is known that C:N ratios in humic substances and soil organic matter extracted from modern lateritic soil show roughly comparable values between 7 and 10 (Dick, Santos & Ferranti, 2003).

The δ¹³C values of guano references range from −28.2‰ to −19.1‰, encasing the guano-like samples (mean ± 1SD = −26.4 ± 0.22‰) (Fig. 8B). Accounting for the isotopic spacing among bat guano, their insectivorous diet, and surrounding vegetation, the guano-like deposits are well within the range of C₃ plants, which exhibit a general range from −23‰ to −37‰ in non-desert areas (Kohn, 2010). As they completely lack a signal of C₄ plants, we suggest that these deposits were formed before anthropogenic influences because sugar cane, C₄ perennial grass, became exclusively dominant by the 1930s due to heavy cultivation on the island.

The δ¹⁵N values of the referenced guano samples widely range from 3.0‰ to 11.6‰, placing the guano-like deposits from Hoshino Cave in its mid-range and the fecal pellet-like sample from Cave A close to its upper limit (Fig. 8A). Forbes & Bestland, (2006) used high δ¹⁵N values above 12‰ in association with low C:N ratios below 10 and high contents of SO₃+P₂O₅+CaO to identify a potential guano layer in accumulated cave deposits. The δ¹⁵N values of the guano-like samples from Hoshino Cave do not meet their criteria although neither do those of some guano references. Overall, as in the case of %N and C:N ratios, stable carbon and nitrogen isotopes were not good indices to identify bat guano in caves of tropical islands where lateritic soil can be developed.

Stable carbon isotopes of bone apatite

Bone apatite δ¹³C values (n = 8) show a wide range from −15.0‰ to −8.36‰ with a mean of −11.5 ± 2.38‰ (±1SD) (Table 3). Fig. 9A shows estimated isotopic signals of vegetation, calculated from carbon isotope values of bone apatite of bats. Considering that enrichment factors and isotope values in the mixing model are reasonably assumed (see Article S1), C₄-feeding insects explains more than 10% of the total diet in six out of the eight individuals analyzed in this study even with a large isotopic spacing between predatory insects and plants (open circle in Fig. 9A). On average, the estimated δ¹³C value of bulk plants is −25.4‰ (open circle), which is translated to 22% of C₄ plants in the simple mixing model. On Minami-Daito Island, C₄ plants were introduced only by humans via sugarcane plantations. Therefore, the unmistakable isotopic signal of C₄ plants demonstrates that these individuals coexisted with humans after 1900.

Group size and estimated ages of cave-dwelling bats on the Daito Islands

Two locally extinct bat species, Miniopterus sp. and Rhinolophus sp., once existed in the lifted atoll remotely located from the closest landmass. Based on the number of skeletal remains recovered from Cave A in this study (at least n = 17 for Miniopterus sp. and n = 1 for Rhinolophus sp.), this cave was utilized by more individuals of Miniopterus sp. than Rhinolophus sp. Considering the taphonomic conditions of the Minami-Daito bats (unpermineralized, exposed on cave floor without any depositional cover, fast crystal growth around bones in tropical wet caves), we interpret that these skeletons are only a few thousand years old at most but probably even younger and that the two species occurred contemporarily in the paleontological sense.

FTIR spectra and SEM images suggest that the guano-like deposits on Minami-Daito Island are most reasonably humic substances, which are the final constituent of the physicochemical degradation and microbial decomposition of organic matter, particularly plant materials (Stevenson, 1994). The hydroxyapatite crust under the bat skeleton was possibly formed from the interaction between bat guano and limestone/dolomite. Nevertheless, we reject the null hypothesis that the “guano deposits” observed by Shimojana (1978) are fossil bat guano and deny the presence of large amounts of guano deposits in the caves, which suggests the skeletal remains are not resultant from large populations or those which utilized these caves for generation after generation.

Our humic samples from both Hoshino Cave and Cave A yielded similar ages of 4,565 calBP (on the 3-point average) and 4,640 calBP, respectively (Table S2), which we think provide the maximum age constraint for the bat species as the bat bones collected in this study were scattered nearby pellet-like humic particles in Cave A. A better age constraint was provided by stable carbon isotopes of the selected bat bones from Cave A. Isotopic signals of C₄ grass-feeding insects detected in various degrees from six out of eight individuals strongly indicate that some of the sampled bats lived after 1900 (Fig. 9B).

In summary, small groups of bats that were dispersed to Minami-Daito Island did not survive long due to strong anthropogenic stresses on the islands. Whether or not their arrivals were caused by a single sporadic event or multiple events is still inconclusive in this study. However, future taxonomic studies and analyses of morphological variation could elucidate the possible biogeographic origins of the strayed populations, which would provide a hint to understand how these species ended up co-occurring on the small island.

Extirpation of two cave-dwelling bat species on the Daito Islands

Our multiproxy approach to constrain the estimated ages of the Minami-Daito cave-dwelling bats include hearing surveys of local residents on the Daito Islands. This survey was conducted by Dr. Hidetoshi Ohta on August 22 and October 28 in 1991 for a different purpose but happened to provide an imperfect yet interesting perspective regarding the extirpation of the bats (pers. comm. with H. Ohta on December 31, 2018 by YK and DF). According to his field notes, three men (Mr. Akamine, Mr. Nakama, Mr. Sunagawa; all in their 80s) commented that they witnessed “bats smaller than the Daito flying fox” in a cave during the daytime within a few years after the end of WWII. In the second survey, three other men (Mr. Toma, Mr. Suzuki, and Mr. Okiyama; two in their 80s, one in his 90s) commented that they also saw “baby flying foxes” in caves near “Hagu (rim)” before WWII. For local residents, the Daito flying fox is a typical mammal on the island. Thus, these bats are reasonably considered as Miniopterus sp. and/or Rhinolophus sp., as the Daito flying fox roosts in the canopies of trees during the daytime, not utilizing caves. If our interpretation is correct, a remnant population of these bats persisted until sometime after WWII, and the very last survivors could have been present until the 1950s, which is consistent with our taphonomic interpretation of the bones.

On Minami-Daito Island, the irreversible rapid change from tropical/subtropical forests to sugarcane fields was extraordinary following the arrival of early settlers/residents in 1899-1900 with heavy development by the 1930s. In addition to reducing favorable habitats for cave-dwelling bats by opening the land and filling in sinkholes, there would have been other mortality causes. Many caves were used as natural storage and air-raid shelters during WWII. Cave A was a base of the Imperial Japanese Army during WWII. There are metal scraps, glass bottles, and wood plausibly derived from that age. Surface water also transports waste into caves. We found coal-like carbon particles in the guano-like samples while handpicking in the lab, which may be related to wasted carbon derived from old steam locomotives that existed from 1917 to 1983 to transport harvested sugarcane stalks as the railway track was run near Cave A (Kada, 2009). Chemical pesticides were introduced in sugarcane plantations, and the aerial spray of pesticides was conducted at least in the early 1970s (Editorial Committee of the History of Minami-daito Village, 1990: p.1167). Considering the extremely rapid development on the small island, anthropogenic activities are undeniably responsible for the local extinctions of the last population of the cave-dwelling bat species. It is noted however that the results of our study do not completely exclude the possibility that these bats were non-viable drifting populations that naturally died out without leaving descendants. In the future, this could be tested by finding out how (and whether) other caves on the Daito Islands were utilized by the extirpated bats.

Conservation paleobiology on tropical oceanic islands

Using a multidisciplinary approach, we were able to bracket the estimated chronological age of the Minami-Daito cave-dwelling bats, ranging from a maximum constrained age of 4,640 calBP (radiocarbon age of humin closely scattered with the bat bones in Cave A) to the 1950s (hearing survey). Stable carbon isotopes of bone apatite provided unmistakable direct evidence that they existed after 1900 and reinforced the hearing survey record.

Despite rapid civilization and deforestation on the island throughout the 20th century, the Daito flying fox has been able to survive on the island. The Daito flying fox, which is designated as a critically endangered subspecies (IA) in the National Red Data List of the Ministry of the Environment, Japan (Japanese Red List, 2020) and is thus of high concern for species conservation (Vincenot, Collazo & Russo, 2017), consumes fruits and nectar of native species such as Chinese banyan (Ficus microcarpa), Fiscus superba var. japonica, Ficus virgala, and fan palm trees (Livistoma chinensis var. amanoi), but it also utilizes windbreak trees and street trees as a food source (Kinjo, 2009). Flexible food preferences may explain how the Daito flying fox has survived throughout anthropogenic damages as in the case of the neighboring subspecies Orii’s flying fox (Pteropus dasymallus inopinatus) on Okinawa Island, which utilizes planted trees in urban areas as stable food sources (Nakamoto, Kinjo & Izawa, 2007).

Regarding the presence of fossil bat species on an oceanic island, a similar case was reported for an extinct endemic bat, Synemporion keana (lava-tube bat), from the Hawaiian Islands, which are over 3800 km away from the nearest continental landmass (Ziegler, Howarth & Simmons, 2016). Synemporion keana is a vespertilionid bat slightly smaller than the extant Hawaiian hoary bat (Lasiurus cinereus semotus) that is known to have inhabited the Hawaiian Islands at least until the earliest Polynesian cultural period because a fossil of S. keana was successfully radiocarbon dated to 1,670 BP (Ziegler, Howarth & Simmons, 2016). According to that study and the references therein, those fossils were recovered from infilled deposits (soil, eolian deposit, pond deposit) or exposed on the floor of limestone sinkholes, volcanic lava tubes, a piping cave, and a volcanic-tuff cone crater. Similar to our studies, those fossils were not suitable for DNA extraction or radiocarbon dating. They estimated a chronological range of the Hawaiian fossil bat species by establishing the ¹⁴C ages of both coexisting animal skeletons such as birds and rats and cultural artifacts derived from the same stratigraphic level or nearby and the formation of volcanic caves from which the fossils were found. Another interesting case of the late Holocene (<4,000 BP) extirpation of bats is found on a tropical island in the Caribbean (Soto-Centeno & Steadman, 2015).

In this study, we revealed that two cave-dwelling bat species were extirpated from Minami-Daito Island without any zoological record when they existed. Minami-Daito Island used to be a home of three bat species, but only a single species survived anthropogenic disturbance. Geological, geochemical, and paleontological approaches are useful to bracket chronological ranges of the two extirpated species. Chronological ranges of extirpated species should be estimated with higher accuracy to deduce causality of the local extinction events and subsequently to promote conservation of biodiversity on oceanic islands, which are generally vulnerable to anthropogenic factors.