The Pleistocene Maritime Migration of Modern Humans in Northern Wallacea: The Cases of Topogaro in Sulawesi and Bubog in Mindoro

Abstract

Around 50,000 years ago, early modern humans migrated from Island Southeast Asia and via the Wallacean islands into the continent of Sahul in Oceania by several sea crossings. The Wallacean archipelago can be broadly divided into northern Wallacea comprising the northern Indonesian islands and most of the Philippine islands, except Palawan, and southern Wallacea with the southern Indonesian islands and Timor. This chapter focuses on the cases of early modern human maritime migration and resource use in northern Wallacea during the late Pleistocene. Of the Pleistocene sites in northern Wallacea, two important sites are presented and discussed: the Goa Topogaro cave complex in central Sulawesi, Indonesia, now dated to 42 ka, and the Bubog rock-shelter sites in Mindoro, Philippines, dated to at least 35 ka, based on our excavations. We discuss both localities in their regional context, as well as their role in maritime migration and seafaring in Wallacea and Sahul, and provide a comparison with other cases in the Ryukyu Islands, adjacent to northern Wallacea.

1. Introduction

Modern humans (Homo sapiens) are thought to have originated in Africa and migrated into the tropical and subtropical regions of Asia during the late Pleistocene and MIS 3 [1, 2, 3]. By around 50 ka, if not earlier, they extended their sea voyages across Island Southeast Asia and towards the continent of Sahul and Near Oceania (Figure 1). Such undertakings required long-distance travels of over 50–80 km across the open sea from the eastern Wallacean islands such as Timor Island where the oldest dated archeological sites in the region are located [4]. Currently, the oldest traces of modern humans in the region of the Sunda shelf are represented by human skeleton remains from Niah Caves in Borneo and Tabon Cave in Palawan [5, 6]. At Niah, direct U-series dating of a cranial bone and AMS dating of associated charcoal has obtained an age of c. 45–39 ka for the so-called “Deep Skull” and 37 ka for its stratigraphic context, respectively [7], while the U-series dates for the Tabon fossils range from 16 to 47 ka, albeit with large standard errors [8]. Somewhat younger have been more recently obtained C14 dates on charcoal from associated hearth features in Tabon with an age between 39 and 33 ka [9, 10].

Figure 1.

Location of Sunda land, Wallacea, Sahul land, and major Pleistocene sites with possible modern human migration routes (Basemap: GEBCO_2023 NOAA NCEI visualization).

On the other hand, if we consider much older dates for Pleistocene Sahul reported from sites such as Madjedbebe in Australia with OLS dates of over 60 ka [11], the first arrival of modern humans in the Sunda, Wallacea, and Sahul regions may have occurred substantially earlier than 50 ka. However, while the OSL dates from Madjedbebe are still controversial, the currently oldest C14 dates for New Guinea from the Kosipe Mission site in the New Guinea Highland are between 49 and 44 ka [12]. If we focus on the early C14 dates in Australia and New Guinea, we can safely assume that modern human reached Sahul by around 50–47 ka and at about the same time as for Sunda, or shortly thereafter. Most importantly, this early human migration into the Sahul region required sea crossings via the Wallacean islands.

Traces of early human occupation in Flores and Luzon, such as in Mata Menge, Kalinga, Liang Bua, and Callao, indicate that Homo sapiens was not the first human species to reach to Wallacea [13, 14, 15, 16]. However, they seem to have disappeared by 50 ka, while the arrival of modern humans becomes visible across Wallacea at around the same time. In contrast, there is no evidence to date for the presence of early hominin species in Sahul. From this understanding, one of the most significant objectives for the investigation of the timeline and routes of early modern human migration into Wallacea and Sahul is aimed at the maritime technology and the capacity for island adaptation that enabled those early islanders to successfully colonize Wallacea and Sahul.

Within this paleogeographic context, our ongoing research on the islands of Sulawesi and Mindoro aims to address questions about the timeline, routes, and technology of early human migration to the region of Wallacea and beyond. In this chapter, we first review the early modern human maritime migration by water crossing in Wallacea particularly in its northern region and, second, present two cases of maritime adaptation and resource exploitations by early modern human in the region, coming from excavations at (1) Goa Topogaro complex site in Central Sulawesi, East Indonesia, and (2) sites in Ilin Island and Sta. Teresa, Mindoro, in the Philippines. Both sites revealed stratigraphic sequences of human occupation ranging from over 40,000 years ago to the late Holocene. While Goa Topogaro produced the currently oldest C14 dates for Sulawesi, the Bubog 1 rock-shelter site on Ilin Island delivered the earliest evidence for modern human presence in the oceanic part of the Philippines to date. Finally, we discuss possible seafaring techniques and strategies of early modern humans in Wallacea.

2. Two possible migration routes by modern human in Wallacea

“Wallacea” is named after Alfred Russel Wallace who proposed a zoogeographic boundary separating Asian and Oceanic animal species that passes between Lombok and Bali, Borneo/Kalimantan, and Sulawesi [17] and is known as the Wallace Line. This boundary was later revised by Thomas Henry Huxley and extended northwards, between Borneo/Palawan and the other Philippines islands, and is known as Huxley Line [18]. It separates the Sundaic islands of Southeast Asia, which were part of the mainland during the Pleistocene, from the oceanic islands of Wallacea, which were never connected to the Asian continent and required sea crossing to reach.

West of Wallace-Huxley’s Line, Sundaland was a large subcontinent that connected today’s Palawan, Borneo/Kalimantan, Bali, Java, and Sumatra to the Asian continent during glacial periods and lower sea levels in the Pleistocene, which would allow to early modern humans to reach them via land routes (Figure 1). Wallacea is geographically located east of Huxley’s line and west of New Guinea and Australia as part of the Pleistocene Sahul continent in Oceania. It includes most of the Philippine islands, except Palawan, and the East Indonesian islands including Timor. The area and coastlines of the Wallacean islands that existed during the Pleistocene are more or less unchanged since they are surrounded by relatively deep sea, which it is the most significant geological character of Wallacea. It also means that sea crossing was always necessary for early humans migrating into Wallacea as well as into the Sahul region in Oceania to the east.

Two major migration routes have been suggested in this context, a “northern” route from Sulawesi to Maluku Islands into the region of New Guinea and a “southern” route along Flores, Alor, and Timor Island leading into northern Australia (see Figure 1) [19, 20]. Along the southern routes, archeological studies in Timor-Leste discovered some early sites possibly by modern human dating back to 44 and 42 ka [4, 21]. The possible appearance of Homo sapiens at Liang Bua in Flores could date back to 50–48 ka [18, 22], though there is no clear evidence of modern human in this period, yet. Since the site was occupied by Homo floresiensis, we need more solid data to confirm when exactly this early human species disappeared or was replaced by modern human in Flores.

On the other hand, Jerimalai (now Asitau Kuru) Cave site located at the eastern coast of East Timor is considered as one of the oldest modern human sites in Wallacea [21, 23]. Although there is no direct fossil evidence of modern humans, the site produced the oldest evidence for the exploitation of fast-swimming pelagic fish as early as 42–38 ka and one of the world’s oldest shell-made fishhooks with an age of 23–16 ka [21].

The northern route can be further divided into two possible segments, the Mindanao Island–Talaud Islands–Maluku Islands route and the Sulawesi Island–Maluku Islands route. Along the route’s segments are several archeological sites where excavations have been conducted, in Sulawesi, Talaud Islands, and Northern Maluku Islands. The possible gateway for both routes is Palawan and Borneo/Kalimantan on the Sunda shelf (see Figure 1). Along these two routes, the oldest modern human traces were found in several cave sites in south Sulawesi in the form of rock paintings that were dated by U-series of speleothem covering the rock art to c. 43 ka [24, 25], considerably earlier than the currently oldest C14 dates in south Sulawesi of up to 36 ka [26]. Other sites older than 30 ka in Wallacea along the northern routes are Golo Cave site in Gebe Island dated to c. 36 ka [27, 28], Leang Sarru in Talaud Islands dated to 35 ka [29, 30, 31, 32], Bubog I in Mindoro Island dated to >35 ka [33, 34, 35], and Goa Topogaro site in Central Sulawesi, which was previously dated to >30 ka [36, 37, 38, 39]. All these sites produced no early modern human fossil remains. However, if the U-series dates of the rock painting are correct, they can be the evidence of early modern human appearance in Sulawesi since rock art is recognized as typical practice of Homo sapiens.

Considering intervisibility between islands, however, the northern route would have provided an easier path for early modern humans to reach Sahul [40, 41]. However, many islands in Wallacea could have been visible along both possible migration routes and the estimated distances between each island ranged from around 10 to 50 km. Geographically, islands along the southern route seem to locate next to each other until Timor Island at the eastern end of the route. On the other hand, some islands are more isolated along the northern routes. For example, the Talaud Islands are located in about 100 km distance from both Mindanao and Sulawesi and even over 120 km away from the Northern Maluku Islands (Figure 1); thus, if early modern humans reached Sulawesi or Maluku islands from the Philippines via the Talaud group, they need seafaring and navigating skills to cross over 100 km distance of open sea. Currently, the oldest archeological traces of human appearance in Leang Sarru date back to 35 ka [30, 32] and indicate that while the Mindanao–Talaud–Sulawesi/Maluku routes may not have been among the earliest routes, they provide evidence for open seafaring and long-distance interaction in Wallacea during the late Pleistocene.

On current archeological evidence, early modern human migrations were possible via both the northern and southern routes. Along the northern routes, the oldest dates are reported from Sulawesi Island located in western edge of Wallacea and next to Boneo/Kalimantan Island in Sunda land [24, 25]. Although no early dates over 36 ka have been retrieved in Maluku Islands so far, 14C dates around 49 to 44 ka were reported from New Guinea, which may indicate that this was the destination of the northern routes in Oceania [12]. Along the southern routes, the large islands like Sumatra and Java contain numbers of old dated sites even before the appearance of modern human. Liang Bua Cave on Flores Island provided fossil evidence of Homo floresiensis between 200 and 60 ka and possible traces of modern human after 50 ka [15, 22]. On the same island, the fossil and cultural remains of an even earlier hominin were retrieved in Mata Menge and Wolo Sege, dated to c. 800–1000 ka [13, 42]. In the eastern part of Wallacea, both the islands of Alor and Timor have yielded early dates of modern human presence between 44 and 42 ka, and these islands lie in close proximity to northern Australia as part of Sahul [43, 44, 45].

It should be also noted that the past northern and western coasts of Sahul continent were visible from many locations with relatively higher altitude of these islands including Alor and Timor during the late Pleistocene [40, 41, 46], and intentional sea crossing by modern humans through sight navigation could be possible [47]. Bamboo raft is currently considered as the most potential method of sea crossing in Wallacea and into Sahul continent [19, 48, 49, 50]. In fact, the experiment navigation by use of bamboo raft made by stone tools from the western coast of Timor Island to the past coastline of Sahul (currently Australia) confirmed its possibility as the raft reached the target (90 km away from Timor coast) in 5 days in December when strong wind basically blows from north to southward [51, 52]). This suggests the existence of developed nautical knowledge and seafaring technology soon after, if not already at the time of the arrival of early modern humans in Wallacea.

By around 35 ka, the archeological traces on remote islands like the Talaud Islands that were located over 100 km from the nearest islands in Wallacea appeared. In Oceania, the human traces on Manus Island that was located about 230 km away from the coast of New Guinea were dated to 20 ka [53]. These traces are still very sparse and fragmentary for the Pleistocene, so it is still unclear whether such sea crossings were intentional or just accidental drifting, what kind of vessels were used, and how great the risks of such crossings and the survivability of those early seafarers were. However, the availability of bamboo as boat-building material and the relatively warm temperature of the tropical sea, even during the Pleistocene, certainly had a favorable influence on the success of such ventures in Wallacea and the West Pacific.

3. Early modern human occupation in Northern Wallacea

As examples of early settlement along the northern migration routes, we present the cases of Goa Topogaro in Central Sulawesi dated to between 42 and 12 ka based on the results of our recent study [54, 55] as well as several rock-shelters on Ilin Island and St. Teresa in Mindoro. Both locations have provided significant timelines for the arrival of modern humans in the northern part of Wallacea and showcase successful human adaptation to maritime environments and their capacity long-distance open seafaring as well as strategies for efficient exploitation of various marine and terrestrial resources.

3.1 Goa Topogaro complex in Sulawesi Island

Goa Topogaro complex site is located in c. 3.5 km distance from the current coast in Morowali district of Central Sulawesi and elevated about 75 m above the current sea level (Figure 2). Goa Topogaro is composed of three large caves (Topogaro 1–3) as well as three rock-shelters (Topogaro 4–7) located along the wall of the upper doline at 90 m above sea level. The Pusat Arkeologi Nasional Indonesia (National Center of Archeological Research) and Ono (National Museum of Ethnology, Japan) have conducted excavations since 2016. Among them, Topogaro 1 is the largest cave (about 500m² of floor area) as 24 m wide and 25 m deep with a maximum height of about 20 m and faces northwest. Over 30 broken wooden coffins with human skeletal remains were found on the cave floor with fragments of prehistoric pottery, Chinese and European ceramics, chert flakes, including finely retouched tools, and shells. Topogaro 2 is located southwest of Topogaro 1, and both caves are connected to each other by a narrow passage. The Topogaro 2 is 15 m wide and 24 m deep width with maximum height of about 12 m and faces north. The floor size of Topogaro 2 is about 360 m² [37, 38, 39, 54, 55].

Figure 2.

Location and views of Topogaro 1 and 2 in Goa Topogaro complex.

The excavation in Topogaro 1 (10 m² in total) confirmed three cultural layers down to about 1 m depth; the acquired C14 dates indicate they belong to the Holocene: Layer 3–2 as early Holocene during 10 to 8 ka (Table 1), Layer 1 as late Holocene with burial human remains, pottery, glass beads, and metal materials after 600 BP. On the other hand, the excavations along the east wall (Sector A) and west wall (Sector B) of Topogaro 2 (10 m² in total) confirmed eleven cultural layers down to about 300 cm depth in Sector A by 2018 (Figure 3). The results of C14 dating from the Sector A indicate four occupation phases: 1. late Pleistocene (c. 29–26 Ka/Layer 9–11), 2. post Last Glacial Maximum (LGM) (c. 16 Ka/ Layer 6–8), 3. early Holocene (c. 10 Ka/ Layer 3–4), and 4. early Metal Age (c. 2300 cal BP/Layer 1–2). Our latest excavation in 2019 opened two 1 × 1 m squares in Square A-6 down to 4 m (Spit 80) and Square A-4 down to 5.1 m (Spit 102) in Sector A along the east wall of Topogaro 2 Cave [54, 55]. Two charcoal samples retrieved from Layer 15 and 16 (Spit 82 and 85) were C14-dated to around 42 ka, while charcoal samples from Layer 11 to 14 are mainly around 30 ka (Table 1). No charcoal was found in Layer 18 and 19 at Square A-7 at depth between 4 and 5 m from the surface, though both layers produced some chert artifacts as well as Anoa (Bubalus sp.) bones, which can be associated with human activities (Figures 3 and 4). It is yet to be determined whether there are more traces of human occupation below Layer 19 [54, 55]. Detailed analysis of excavated artifacts and further excavations will be conducted in 2024.

Site	Location	Spit	Depth	Material	Ref-No	Cal BP (95.4% probability)
TPGR1	B-6	3	30 cm	Cyrenidae sp.	TKA-18864	8763–8547
	B-6	5	50 cm	Charcoal	TMNA1–1	3373–3242
	B-6	8	80 cm	Anadara sp.	TKA-18865	11,037–10,732
	B-6	10	100 cm	Cyrenidae sp.	TKA-18866	8660–8660
	B-6	10	100 cm	charcoal	TMNA1–2	9888–9595
TPGR2	A-6	18	90 cm	charcoal	TMNA2–6	10,729–10,561
	A-4	19	95 cm	charcoal	TKA-17034	10,760–10,587 cal BP
	A-4	26	130 cm	charcoal	TMNA2–8	16,050–15,700 cal BP
	A-4	28	140 cm	charcoal	TMNA2–9	10,805–10,651 cal BP
	A-5	32	160 cm	charcoal	TMNA2–1	16,431–16,048 cal BP
	A-5	41	205 cm	charcoal	TMNA2–2	26,213–25,813 cal BP
	A-5	46	230 cm	Telescopium sp.	2017I	26,377–25,958 cal BP
	A-6	55	275 cm	charcoal	TKA-16906	28,864–28,464 cal BP
	A-6	55	275 cm	charcoal	TKA-16906	28,864–28,464 cal BP
	A-4	56	280 cm	charcoal	TMNA2–10	23,914–23,435 cal BP
	A-4	58	290 cm	charcoal	TMNA2–11	27,864–27,485 cal BP
	A-6	59	295 cm	charcoal	TMNA2–7	26,199–25,866 cal BP
	A-4	66	330 cm	charcoal	TPG2TP7–1	31,245–30,895 cal BP
	A-4	68	340 cm	charcoal	TPG2TP7–2	27,754–27,424 cal BP
	A-4	75	375 cm	charcoal	TPG2TP7–4	28,502–27,954 cal BP
	A-4	79	395 cm	charcoal	TPG2TP7–6	31,175–30,824 cal BP
	A-4	81	415 cm	land snail	TPG2TP7-S2	34,601–34,031 cal BP
	A-4	82	410 cm	charcoal	TPG2TP7–7	42,160–41,515 cal BP
	A-4	85	425 cm	charcoal	TPG2TP7–8	42,051–41,371 cal BP

Table 1.

C14 dates from Topogaro 1 and 2 caves.

Source: Refs. [37, 53].

Figure 3.

Plan of Topogaro 2. (Source: Modified from Figure 4 from Ref. [53]).

Figure 4.

Wall section of sector A (A-4&6) in Topogaro 2. (Source: Modified from Figure 3 from Ref. [53]).

The major artifacts from the Topogaro caves are variety of lithics, bone tools including projectiles, and shell and animal remains (Figure 5). Among them, bone tools and shell remains are only excavated in large number from the Holocene layers. Topogaro 2 produced bone tools and shellfish mainly from the upper Holocene layers as well. On the other hand, the late Pleistocene layers in Topogaro 2 contained lithics and terrestrial animal remains. Regarding the animal remains from the Pleistocene layers in Topogaro 2, the majority belong to murids (some unidentified rat species) and chiropteran (some unidentified bat species) with few remains of two possible species of wild pig including Celebes warty pig (Sus celebensis) and Babirousa (Babyrousa babyrussa), Anoa or dwarf buffalo (Bubalus sp), marsupials (mainly Phalangeridae), and reptiles like snakes (Serpentes) and lizards (Lacertila) [54, 55].

Figure 5.

Selected artifacts from Topogaro 1 and 2 caves. A. Chert lithic from layer 12–17 dated to 42–31 ka in TPGE2, B. Chert scraper from layer 10 dated to 29 ka in TPGR2, C. Chert flake tools from the post LGM layers in TPGR2, D. Chert flake tools and bone projectiles from early Holocene layers in TPGR1 and 2. (Source: Figure modified from Refs. [53, 54]).

Among them, wild pigs are common in both the Holocene and Pleistocene layers in Topogaro caves, while Anoa remains were mainly excavated from the Pleistocene layers. A phalanx of Anoa is associated with several chert flake tools and charcoals dated to 29 ka, while its number increased in the lower deposits dated to between 30 and 42 ka in Topogaro 2 [37, 54, 55]. For shellfish remains, both bivalvia and gastropod species were excavated mainly from the Holocene layers in Topogaro 2. The major gastropod family is Thiaridae, while Cyrenidae is the dominant bivalvia species. They are both occupying fresh and brackish water habitats in river and mangrove environments. No fish bones were excavated from the Pleistocene layers of Topogaro 2 so far, while fish bones of freshwater and marine species do appear in the Holocene layers.

The Pleistocene layers in Topogaro 2 also produced a variety of lithic tools mainly made of chert (see Figure 5). Use-wear analysis conducted by the authors confirmed that the chert flakes associated with Anoa bone showed traces of impact on hard materials, most likely animal bones. The use-wear analysis also identified tools from the layers dated to around the end of the LGM displaying flat polishing and diagonal/transversal striations and indicating prolonged contact with phytolith-rich plants through scraping [39, 56].

While large-sized mammals like Stegodon existed in some islands along the southern and northern migration routes, including Sulawesi, Flores, and Timor during the Pleistocene, many became possibly extinct before or soon after the arrival of modern humans in Wallacea. In fact, no such remains were excavated in Timor Island along the southern route or any of the late Pleistocene sites along the northern routes including Golo, Leang Sarru, and Goa Topogaro [1, 4, 23, 27, 30, 31, 37, 38, 39], while many Stegodon remains were excavated in layers associated with Homo floresiensis in Liang Bua in Flores, its upper layers after 50 ka did not produce such remains [57, 58]. Instead, archeological evidence indicates that the terrestrial animals hunted and exploited by early modern human were mainly small-sized animals including rodents and fruit bats, cuscus as marsupial, and reptiles with various monitor lizard species and snakes in Timor and Flores [4, 22, 23, 58].

3.2 Caves and rock-shelters in St. Teresa and Ilin Island, Mindoro

Several caves and rock-shelters in southern Mindoro Occidental have delivered a combined chronostratigraphic record from the late Pleistocene to the Holocene and until the beginning of the colonial period in the Philippines (Figure 6). Surveys in the karstic landscapes of the municipalities of Magsaysay and San Jose and particularly in the adjacent islands of Ilin and Ambulong have recorded over 40 potential cave and rock-shelter sites. On Ilin Island, the site of Bubog 1 (Figure 7A) is characterized by a stratified shell midden connected to an AMS chronology between c. 4 and 35 ka cal. BP [33, 34, 35]. Underneath the shell midden deposits, another c. 2 m of terrestrial silty deposits was unearthed. Although no radiocarbon dates have been successfully obtained so far, the deposits contained several lithic artifacts and the remains of pelagic fishes (but no marine shells) indicating a considerably earlier occupation of Mindoro. The absence of marine shells in these deposits suggests that the small island of Ilin was connected to the Mindoro mainland due to lower sea level and that the shell habitat had not yet formed.

Figure 6.

Map of Mindoro and Ilin Island with locations of Bubog 1 and 2, Bilat Cave, and Cansubong.

Figure 7.

A: Connecting the former treasure hunter pits at Bubog 1; B: Bubog 2, trench 3; C: Shell midden in the first chamber of Bilat Cave.

Another effect of lower sea levels during the late Pleistocene was possibly a hiatus in the chronostratigraphy of Bubog 1, observed in the lower shell midden layer 9, where no AMS dates have been retrieved between c. 27 and 12 ka. It is likely that the location of the cave, now at c. 35 m asl, was simply too far uphill when sea levels were much lower and other caves or rock-shelters much closer to the Pleistocene coastline were better accessible and available for occupation.

While the site of Bubog 1 was already disturbed by looters at the time of the archeological survey, the neighboring rock-shelter of Bubog 2 at c. 400 m distance had an undisturbed surface with the remains of several hearths and few potsherds close to the surface, followed by shell deposits similar to Bubog 1, albeit less dense and extensive (Figure 7B). Excavations were, however, hampered by rockfall increasingly appearing after c. 1.50 m from the surface. Bubog 2 currently has a stratigraphic record from the 16th century AD down to the Pleistocene–Holocene boundary and c. 11–12 ka and the currently lowest layer [34, 59]. Excavation was ongoing until the time of the Covid-19 pandemic and consequent nationwide lockdown in the Philippines, and there is still the possibility that older deposits can be reached in successive excavations.

Nearby Bilat Cave (Figure 7C) is located on the mainland of Mindoro at c. 8 km distance from Bubog. The sites can be spotted from each location. The site is at present situated directly on the coast, just 2–3 m above sea level and with two of its three entrances opening onto the Ilin Channel. Its first chamber (from the land side) is mostly covered by a shell midden. Unlike Bubog 1, the midden is not very dense and followed soon by terrestrial sediments. Here, AMS radiocarbon dates of 13–14 ka and 21–22 ka were retrieved and provide evidence for human occupation for the period during and after the LGM that is currently missing in the Bubog 1 record [34, 35].

While the mollusk fauna dominates the faunal assemblages of the Mindoro sites, a variety of reef and pelagic fish remains are associated with the shell midden deposits and continue in the strata below the midden [60, 61, 62]. In a recent study, Boulanger et al. have been furthermore able to establish the systematic exploitation of toxic Diodontidae or pufferfish for Bubog 1 and 2 and Bilat Cave, based on the presence of numerous dermal spines and other skeletal remains, and the extraction of their poison, probably for hunting terrestrial animals, at least since 13 ka [62]. Among the terrestrial fauna appears mainly wild pig (Sus oliveri), Tamaraw (Bubalus mindorensis), an endemic bovine, and two species of deer (Rusa marianna and Cervus alfredi), as well as the nowadays extinct Ilin cloud rat Crateromys Paulus, and lizards [34, 63]. No remains of domesticated animals appeared in any of the sites, including the subrecent upper layers of Bubog 2 and Bilat Cave. In 2018 and 2019, Cansubong II, another cave site close to shore, was investigated. Similar to Bilat Cave, Cansubong II is located just 5 m above mean sea level and opens to the northern coast of Ilin Island. The cave is characterized by heavily disturbed burials. Initial test pits were opened to explore the stratigraphy of the cave, and a “forensic sweep” of the cave floor was conducted to reconstruct the original location of the burials. Due to the Covid-19 pandemic, excavations were halted but are planned to resume in 2024.

Overall, the material culture is rather sparse in the Mindoro sites (Figure 8). At Bubog 1 and 2, unmodified igneous beach pebbles and cobbles are the dominant lithic artifacts within the shell midden deposits and only few chert flakes were retrieved. The pebbles and cobbles were utilized in several ways but mainly as hammers to open the larger marine shells such as Strombus, Trochus, and Lambis for consumption, indicated by their pitted surfaces that were caused by recurring blows (Figure 8A–C) [33, 34]. Some of the hammerstones were later reused as weights for fishing nets or fish traps (Figure 8D,E) [34, 61], while use-wear traces on several hammerstone fragments with sharp edges indicated that the fragments were not immediately discarded, but used as tools for the processing of harder organic materials (Figure 8F–K) [59].

Figure 8.

Selected artifacts from Mindoro. A–C. Unmodified pebbles and cobbles used as hammers from Bubog 1 and 2; D-E. Used pebbles/cobbles modified into netsinkers from the mid-Holocene deposits of Bubog 1 and Bubog 2; F–K: Pebble/cobble fragments with wear traces from secondary uses; L: Bone fishing gorge from the late Pleistocene layer 10 below the shell midden of Bubog 1; M: Polished bone point from the mid-Holocene deposits of Bubog 1; N. Edge-ground shell adze from Bubog 1; O. Edge-ground shell adze from Bilat Cave; P. Shell adze preform from Bubog 2.

A small assemblage of obsidian flakes was recovered from Bubog 1 from the lowest shell midden Layer 9 and the terrestrial deposits of Layer 10 below, dating to c. 33–28 BP, or somewhat earlier, respectively. Portable X-ray florescence analysis of the samples from Bubog demonstrated a chemical match with 11–9 ka-year-old obsidian flakes from Ille Cave as well as presumably late Holocene obsidian artifacts from Alegria in Cebu [35, 64]. No obsidian sources are known on Mindoro and Ilin Island, Palawan, or Cebu so that the appearance of obsidian artifacts in these sites provides evidence of long-distance interaction across the Philippines and with regard to the suggested potential sources perhaps as far as South Sulawesi or Melanesia, from where obsidian was already being sourced and traded by the end of the Pleistocene [64, 65, 66].

Another interesting aspect is that it seems to contradict one of the arguments brought forward in the context of missing formality of lithic artifacts and in support of the “bamboo hypothesis,” the lack of quality knappable rocks. Although the number of obsidian flakes is too small to make conclusive assessment, the fact that these flakes are unretouched and of amorphous morphology suggests that raw material quality was not the main reason for the absence of formal tool types in Wallacean assemblages.

More diverse is the organic tool technology in Mindoro. A fishing gorge made of bone was found at Bubog 1 from the currently undated deposits below the lowest shell midden layer (Figure 8L) associated with remains of several large pelagic fishes in the same layer [34, 60, 61]. Together with the base of a hafted point from Matja Kuru 2 in East Timor dated to c. 34 ka cal. BP, this is currently the earliest evidence of bone technology east of Huxley’s Line [34, 67], although similarly old bone artifacts have been reported from the Sahul region [38, 68]. Several fragments of polished bone tools, possibly another fishing gorge and bone points as well have been recovered from the mid-Holocene Layers of the shell midden in Bubog 1 (Figure 8M), dated to between 6000 and 5500 cal. BP [34].

Tools made of marine shell have been identified from the late Pleistocene and mid-Holocene layers of Bubog 1. An edge-ground Tridacna adze (Figure 8N) from the mid-Holocene shell deposits of Bubog 1 was directly dated to 7550–7250 cal. BP. Another ground shell adze was found in Bilat Cave (Figure 8O) and directly dated with a similar age of 7414–7285 cal. BP [34, 69]. From Bubog 2 stems a Tridacna adze preform, directly dated to 9115–8899 cal. BP and indicating that such Tridacna adzes were manufactured locally (Figure 8P).

The two Bubog sites delivered modified flakes made of Tridacna and Conus shell from the Mid-Holocene onwards (Figure 9A). Use-wear analysis suggests that these flakes had been utilized for similar tasks as the used pebble and cobble fragments found in the sites. Even earlier are modified flakes and fragments of the bivalve Geloina coaxans found in the lowest shell midden Layer 9 at Bubog 1 (Figure 9B). Experimental use-wear analysis of the wear traces observed on the umbo and along the edges indicates a use as chisel-like implements and for scraping and sawing [34, 70]. Direct AMS dates on Geloina shell tools from the base of the shell midden returned ages of between 31 and 28 ka cal. BP, while direct AMS dates on Conus and Strombus shells from the same stratigraphic horizon provided ages of between 33 and 31 ka cal. BP.

Figure 9.

Flaked shell artifacts from Bubog 1 and 2 A. Tridacna and conus flake tools from mid-Holocene contexts; B. Flaked and modified Geloina coaxans shell tools from the lowest shell midden layer 9, directly AMS-dated to 28–33 ka cal. BP.

The shell artifacts found in Bubog also provide evidence that tools made of shell may have substituted for lithic tools or even replaced them, unlike for the still hypothetical “bamboo industry” or other tools made of plant material. Using shell was certainly a very economical choice as larger marine shells with sufficient hardness were readily available as a raw material, while the acquisition of chert or similar knappable rocks would have required traveling to Mindoro Island. Nevertheless, evidence for the prehistoric use of a wide range of plants, particularly woody vines, was found on Ilin Island. The study of macrobotanical remains from Bubog 2 identified dried and mineralized fragments of woody vines from the families Annonaceae, Dilleniaceae, and Mimosaceae, indicating their selective use probably for the manufacture of various kinds of fibercraft such as cordage, mats, traps, baskets, and more from the mid-Holocene layers and c. 7500 years ago onwards and into the late Holocene [71]. This is consistent with the findings from the traceological analysis of the Leang Sarru and Topogaro 2 lithics that exhibited intensive working and processing of fibrous plants, likewise interpreted for the manufacture of cordage, ropes, nets, and other forms of fibercraft [29, 36, 37, 38, 55] and the traceological analysis of lithic artifacts from Tabon Cave, Palawan, which interpreted similar micro traces as related to the manufacture of basketry and cords [72].

The sites in St. Teresa and on Ilin Island provided a variety of new data that supported the existence of an early migration route for modern humans from Sundaland via Borneo and Palawan into Mindoro and other Wallacean islands of the Philippines before 35 ka cal BP and likely significantly earlier. The rock-shelter sites of Bubog 1 and 2 and Bilat Cave provided furthermore solid indication for a successful adaptation to the unique environment and the exploitation of its rich aquatic and terrestrial resources. These early islanders engaged in open-seafaring and pelagic fishing, and they established maritime networks and interaction spheres for the acquisition of obsidian and new technologies already during the late Pleistocene [35, 59].

4. Discussion

4.1 Island adaptation and resource use by early modern human in Northern Wallacea

The cases of Topogaro caves in Sulawesi and the caves and rock-shelters in Mindoro show the exploitation of variety of terrestrial animals. The major middle-large sized animals from the Topogaro caves are wild pigs (Sus Celebensis and Babyrousa Babyrousa), Anoa (Bubalus sp), and marsupials (mainly Phalangeridae), while both Bubog rock-shelters and Bilat Cave produced the remains of wild pig (Sus oliveri), Tamaraw (Bubalus mindorensis), an endemic bovine, and two species of deer (Rusa marianna and Cervus alfredi). Both cases clearly indicate that the late Pleistocene people hunted and exploited most of middle-large sized animals in Northern Wallacea. The favorable location of the Mindoro sites in close proximity to mangrove, reef, and open sea habitats also allowed for the exploitation of various marine resources, particularly shellfish, as well as a variety of fish taxa, employing a diverse range of techniques for both reef and pelagic fishing, reflecting the exceptional richness and biodiversity of the area [63, 64].

On the other hand, some remote islands in Northern Wallacea like the Talaud and Maluku Islands might have limited variety of mammals and no large and middle-size mammals except some marsupial animals in the Talaud and Maluku Islands were present during the late Pleistocene. However, the current archeological evidence shows only intensive use of shellfish in these islands, while none or very few fish remain and no J-shaped fishhooks were found, so far. The current absence of J-shaped fishhooks along the northern routes, including Sulawesi, may be due to the currently still limited number of excavated Pleistocene sites and analyzed assemblages in this region, but could also have been caused by varying availability of resources on the respective islands in the late Pleistocene.

An exception might be the bone fishing gorge found in Bubog 1 in Mindoro, dated to c. >35 ka [34, 35, 61]. A considerable exploitation of various pelagic and reef fishes has also been observed in Mindoro over more than 35 ka [61, 62]. Nevertheless, no Pleistocene sites with large number of fish remains are known yet in Sulawesi, which is the largest Wallacean island carrying the largest variety of terrestrial animals including Anoa and Babirusa as its endemic species in Wallacea as the excavation of Topogaro caves show. The coasts along Sulawesi are mainly covered by mangrove forests, and such dense forest environments may have made access to the coast for fishing difficult also during the late Pleistocene.

The larger volume of terrestrial animal bones and freshwater or mangrove shellfish remains from the late Pleistocene sites in Sulawesi including Topogaro caves tentatively indicates such possibility. The exact factors of such scarcity of fish remains and fishhooks along the northern migration routes are yet unclear, although the current archeological evidence implies the possibility that different types of island adaptation by early modern humans might have developed across the vast and diverse region of Wallacea.

One indication of successful island adaptation by early modern humans in Wallacea may be the abundant shell remains at many coastal sites and shellfish may be one of the most important marine resources used by early modern humans in the late Pleistocene. In Southern Wallacea, more intensive use of marine fish resources has only been confirmed in Alor, Kisar, and Timor Island (see chapter by O’Connor et al., this volume).

Geographically, the current average humidity of these islands is lower (average annual rainfall is around 1500–1000 mm) than other islands in Wallacea (cf. annual rainfall of Sulawesi varies 4000–500 mm) and particularly dry during the dry season from June to October (average monthly precipitation between 12 and 18 mm). During the late Pleistocene, these islands could be much dryer than current times with limited forest area. The variety of terrestrial animals in these islands might be also limited, and there were no large and middle-size mammals on these islands during the late Pleistocene. The dominance of fish and shell remains from the coastal sites in these islands indicates the higher demand for marine resources by early modern human in these islands. Also, such environmental condition in these islands might have triggered the fast maritime adaptation, as indicated by the relatively early appearance and active use of J-shaped shell-made fishhooks [21, 39, 43, 44, 73].

For further investigation and discussion of such a possibility and to obtain more details on the processes of island adaptation including resource exploitation and subsistence strategies, we need more archeological research and data in both the North and South of Wallacea. Although the data and assemblages from the past and recent archeological investigations are still limited, it becomes clear now that the modern human maritime and island adaptation did appear and develop in Wallacea as well as the Ryukyu Islands as remote islands in East Asia during the late Pleistocene. We hope our ongoing and future excavations of several Pleistocene sites in both regions will discover further details in the history and new evidence for the early modern human maritime and island adaptation.

4.2 Maritime migration and seafaring into Wallacea and Sahul

The current archeological record shows that modern humans migrated into the Wallacean archipelago and Sahul between 50 and 44 ka, if not earlier. This timeline is supported by the early modern human migration also by sea crossing into the Japanese Islands, which has been dated to at least by 40 ka, and into the Ryukyu Islands by 36 ka (see [39]). In terms of seafaring by early modern human in these maritime regions, the current archeological record for Sahul and Wallacea indicates that they might have ability to cross ocean over 80 km distance by 50 to 45 ka. Although there is no evidence for reconstructing the early seafaring technology currently, the recent simulation analysis has supported that early migration into Sahul could be done by intentional sailing by some kinds of watercraft(s) rather than unintentional migration including drifting [47]. The bamboo raft is so far the most supported and potential candidate for such early modern human seafaring watercraft in Wallacea [48, 49].

The bamboo raft experimental navigation by Bednarik and his team from Timor to the past coastline of Sahul proved its possibility to reach the target (90 km away from Timor’s coast) in 5 days [51, 52]. Particularly if their target is a large landmass like Sahul, it can be expected to reach its coast if the wind mainly blows from north to south. The experimental navigation by Bednarik was conducted in December and during the northeast monsoon season when the dominant wind direction is north to southwards. Since such monsoon seasonality likely existed as well during the late Pleistocene [46], it was certainly possible for early modern humans to select the best season to navigate to Sahul from various locations in Wallacea, such as the Maluku region, as well as Alor and Timor. It is also important to note that the Sahul continent was visible from various locations of islands in Wallacea during the late Pleistocene [40, 41, 46], thus it was possible to estimate direction and duration of a sea crossing from the relatively small islands with limited terrestrial resources in eastern Wallacea towards Sahul.

In case of the modern human migration into the Ryukyu Islands in East Asia, the recent experimental navigation by dugout canoe with six adult males and females from Taiwan (as part of continental coast during the late Pleistocene) to Yonaguni Island, which locates the nearest distance from west coast of Taiwan about 110 km, has proved that such sea crossing method could be another possible option [74, 75]. Kaifu and his team consider that the use of raft or drifting navigation should be too risky and hard to reach any islands in the Ryukyu [74, 75]. With such understanding, the paddling navigation by use of dugout canoe was tested in 2019, and this navigation succeeded to reach Yonaguni Island in about 48 hours of the voyage [75]. Archeologically, the earliest dugout canoe in Japan is dated to 7500 years BP, though there are many Pleistocene sites in Japan that produced edge-ground adzes as early as 40–30 ka. These edge-ground adzes are considered to have been used for canoe making and indicate the possible production and use of dugout canoe already during the late Pleistocene. It is also worth to note that a similar type of adze/ax was also excavated from the Madjedbebe site as currently the oldest dated site in Australia dated to around 65 ka [11]. If it was also used for dugout canoe making, the dugout canoe could be another candidate for the early modern human sea crossing in Wallacea and into Sahul.

However, as noted above, we do not find any obvious reasons why early modern humans and members of hunter-gatherer groups in the late Pleistocene and around 35–28 ka would seek to migrate to such small islands as in the Ryukyus by crossing over 100–200 km of open sea with a high risk of distress. Kaifu and others examined the trajectories of satellite-tracked surface drifting buoys deployed in the Global Drifter Program. Their data show that most of the buoys did not reach the islands of the Ryukyu, and they concluded that accidental drift could not explain maritime migration to the Ryukyu Islands [76]. However, we should also keep in mind that human cannot be simply compared to buoys and the drifting of buoys and drifting of humans on a boat or raft are completely different and can lead to very different results.

Many previous cases of 30+ day drifts and survivals, especially in the Pacific [50, 77, 78], indicate that the survival and landing rates of people in distress on boats and rafts are higher particularly in tropical and subtropical temperatures and humid marine environments (as rainfall was more frequent, and fresh water was thus available). The availability of bamboo as a sturdy and buoyant building material and the relatively warm temperature of the sea, even during the Pleistocene, certainly had a favorable influence on the outcome of such incidents in Wallacea and the West Pacific.

Therefore, unintentional migration or arrival to the Ryukyu Islands cannot be ruled out either, especially for the Southern Ryukyu Islands where the early modern human may have come from the south by effect of Black Current. In this case, the potential origin of those early seafarers could be not just the Asian continental coasts around southern China and Taiwan but also Island Southeast Asia, particularly the Philippine Archipelago where the Kuroshio Current, also known as Black Current is turning from the west side of the Pacific north towards Taiwan and the Ryukyu Islands.

Contrary to the Ryukyu Islands, the early modern human migration into Sahul from Wallacean islands was most likely intentional since (1) the past coasts of Sahul were visible from their starting point, (2) about 5 days drifting with rafts might have been sufficient to reach the Sahul coasts, and (3) such migration was done from islands in eastern Wallacea regions that are relatively small in size and with limited terrestrial resources; thus, there was at least a very clear logical reason for those early hunter-gatherers aiming at a target, which could be estimated as large island or continent from their starting locations. Although the exact watercrafts used by the early modern human are yet unclear, bamboo raft and dugout canoe can be potential candidates. The past experimental navigation by bamboo raft proved such watercraft might be able to transport several people with food and water to reach the past Sahul coasts. While navigation by dugout canoe(s) was successfully tested in the waters of the Ryuku islands, this has not been tested yet for Wallacea. It is hoped that such an experiment will be conducted as well in the near future.

Acknowledgments

This excavation of Goa Topogaro complex was supported by Cultural History of PaleoAsia (Grant Number: 16H06409) as well as by Japan Society for the Promotion of Science Research Grants to Rintaro Ono (Grant Number: 21H04368, 20 K20504). The research of Alfred Pawlik and Riczar Fuentes was supported by the University Research Council of Ateneo de Manila University and AdMU Loyola Schools Research and Creative Work (RCW) Faculty Grant.