Taxa | 230-240cm | 240-260cm | 270-280cm | 280-290cm | 290-300cm | Total |
|---|---|---|---|---|---|---|
Dolphin | 0.0 | 34.0 | 25.0 | 0.0 | 51.0 | 110.0 |
Mammal | 20.0 | 0.0 | 0.0 | 2.0 | 0.0 | 22.0 |
Rat | 12.0 | 56.0 | 26.0 | 0.0 | 1.0 | 95.0 |
Bird | 6.0 | 22.0 | 25.0 | 1.0 | 10.0 | 64.0 |
Fish | 20.0 | 120.0 | 41.0 | 1.0 | 18.0 | 200.0 |
Sea Urchin | 1.0 | 0.0 | 41.0 | 0.0 | 0.0 | 42.0 |
Joint Shell | 1.0 | 4.0 | 7.0 | 0.0 | 0.0 | 12.0 |
Unidentified | 15.0 | 100.0 | 12.0 | 0.0 | 0.0 | 127.0 |
Re-evaluating Claims of Subsistence Change and Ecological Collapse on Rapa Nui: A Critical Analysis of Faunal Assemblages
Abstract
The narrative of progressive faunal resource depletion has dominated archaeological interpretations of Rapa Nui’s pre-contact history. This paper critically examines claims of ecological collapse through systematic analysis of faunal assemblages from multiple excavations at Anakena (1986-2005) and comparative sites. Our analysis reveals that variations in faunal composition correlate strongly with depositional processes and sampling strategies rather than temporal trends in resource availability. Marine resources consistently dominate assemblages (50-99%), with shellfish representing a primary dietary staple rather than a fallback resource. The Skjølsvold (1987-1988) excavations provide definitive evidence against depletion models, showing marine resource intensification from earlier to later deposits (92% to 98%). These findings challenge prevailing collapse narratives and suggest more complex patterns of resource use and environmental adaptation.
1. Introduction: Questioning the Collapse Narrative
The archaeological interpretation of Rapa Nui has been profoundly influenced by narratives of ecological catastrophe. Diamond (2005, 79) popularized the notion that pre-contact Rapa Nui society experienced “the most extreme example of forest destruction in the Pacific, and among the most extreme in the world,” leading to societal collapse. This interpretation draws heavily on faunal analyses, particularly Steadman, Vargas, and Cristino (1994, 85) claim that archaeological sequences demonstrate “a shift from porpoises, seabirds, and land birds to a diet dominated by rats, chickens, and very few fish.”
Martinsson-Wallin and Crockford (2001, 256) reinforced this narrative through their analysis of Anakena assemblages, arguing that “the decrease in fish bone and the increase in rat bone in the upper levels indicate a shift in subsistence strategies.” Similarly, Ayres (1985, 103) suggested that faunal remains document “changes in marine food utilization” reflecting “population stress.”
These interpretations have become archaeological orthodoxy, yet they rest on assumptions that require systematic examination. This paper presents a comprehensive re-analysis of faunal data from all major excavations at Anakena (1986-2005) and comparative sites. We examine whether observed patterns in faunal assemblages necessarily indicate resource depletion or whether alternative explanations better account for the archaeological record.
Our analysis focuses on three critical questions:
- Do faunal assemblages demonstrate clear temporal trends in resource availability?
- Can observed variations be explained by non-cultural factors such as depositional processes?
- What role do sampling strategies and analytical methods play in shaping interpretations?
2. Anakena Excavations: Context and History
2.1 Archaeological Investigations at Anakena
Anakena Bay, located on the north coast of Rapa Nui, has served as a critical laboratory for understanding prehistoric subsistence patterns. The bay’s archaeological significance stems from its unique environmental setting—one of the island’s few sandy beaches—and its cultural importance as the traditional landing place of Hotu Matu’a, the legendary founder of Rapa Nui society. Between 1986 and 2005, five major excavation campaigns at Anakena generated the faunal assemblages that form the empirical foundation for collapse narratives. Each campaign employed different methods, pursued distinct research questions, and produced seemingly contradictory results that require careful examination.
2.1.1 The Martinsson-Wallin and Crockford Excavations (1986-1988)
The earliest systematic excavations in our analysis were conducted by Helene Martinsson-Wallin and Susan Crockford as part of the Norwegian Archaeological Expedition between 1986 and 1988. Their work focused on establishing chronological control and documenting early settlement patterns at what they recognized as one of the island’s most significant habitation sites. Martinsson-Wallin and Crockford (2001, 247) stated their primary objective clearly: “to investigate the nature and timing of initial human settlement at Anakena through detailed stratigraphic excavation.”
The excavation strategy centered on a deep trench, designated C1, which reached 300 cm below surface—making it one of the deepest systematic excavations conducted at the site. The team employed natural stratigraphic excavation, carefully identifying five major depositional units between 230 and 300 cm depth. While recovery methods remain incompletely documented in published accounts, the presence of small taxa including sea urchin spines and small fish bones suggests that fine-mesh screening was employed, though the exact mesh size was not reported.
The findings from this early excavation campaign challenged existing settlement models in several important ways. The deepest levels (270-290 cm) contained substantial marine fauna, including fish (41 NISP at 270-280 cm), dolphin (25 NISP at 270-280 cm), and notably, sea urchin remains (41 NISP at 270-280 cm). The presence of diverse marine taxa at these depths indicated that the earliest inhabitants of Anakena maintained a strongly marine-focused subsistence strategy from the beginning of occupation.
Perhaps most significantly, the data reveal extreme variability between adjacent stratigraphic levels. The 240-260 cm level contains 336 total specimens, while the immediately adjacent levels (230-240 cm and 270-280 cm) contain only 75 and 167 specimens respectively. This four-fold difference in faunal density between adjacent levels suggests episodic deposition rather than gradual accumulation—a pattern that would prove critical for understanding the site’s formation processes.
Martinsson-Wallin and Crockford (2001:256) interpreted their findings as showing “a shift in subsistence strategies,” particularly noting changes in the relative frequencies of different taxa through the sequence. However, their own data reveals that this interpretation may be problematic. The coefficient of variation in specimen counts between levels exceeds 100%, indicating that depositional factors rather than cultural choices may be driving the observed patterns.
2.1.2 The Skjølsvold Excavations (1987-1988)
Arne Skjølsvold’s excavations, conducted under the auspices of the Kon-Tiki Museum in Oslo, represent a methodological watershed in Anakena archaeology. Working at the site from 1987 to 1988, Skjølsvold brought a unique approach that would prove pivotal for understanding both subsistence patterns and site formation processes. Unlike other investigators who relied on single quantification methods, Skjølsvold systematically recorded faunal remains using multiple metrics: weight in grams, minimum number of individuals (MNI), and broad taxonomic categories.
Skjølsvold (1994, 28) described his stratigraphic observations with notable precision: “The stratigraphy was remarkably simple, consisting of a basal cultural layer rich in artifacts and faunal remains, overlain by a thick deposit of sterile wind-blown sand.” This observation would prove more significant than perhaps even Skjølsvold realized. The clear temporal sequence—with the Cultural Layer representing earlier occupation and the Sand Layer representing later deposition—provides an unambiguous test for hypotheses about resource depletion over time.
The excavation employed standardized recovery methods throughout, using 6mm mesh screens to ensure consistent retrieval of faunal remains. This mesh size, while relatively coarse by modern standards, was consistently applied across both stratigraphic units, eliminating screen-size bias as a factor in interpreting differences between layers. The total excavated volume and spatial extent of Skjølsvold’s excavations made them among the most extensive at the site, yielding 7,306 grams of faunal material and representing a minimum of 7,191 individual animals.
Taxa | Cultural Layer | Sand Layer | Total |
|---|---|---|---|
Fish | 2,139.0 | 394.2 | 2,533.2 |
Shellfish | 2,030.3 | 2,140.8 | 4,171.1 |
Bird/Rat | 363.2 | 63.4 | 426.6 |
Taxa Group | Cultural Layer | Sand Layer |
|---|---|---|
Shellfish (total) | 3,551 | 3,301 |
Fish | 24 | 5 |
Marine Mammals | 14 | 3 |
Rat | 300 | 21 |
Birds (total) | 46 | 10 |
Skjølsvold’s meticulous multi-method recording revealed patterns that fundamentally challenge the resource depletion narrative. The weight-based data shows that marine resources comprise 92% of the faunal assemblage in the earlier Cultural Layer and increase to 98% in the later Sand Layer. This temporal pattern—showing intensification rather than abandonment of marine resources—directly contradicts predictions of the collapse model.
The MNI data provides even more striking evidence against resource depletion. Both layers contain over 3,000 individual shellfish, with 3,551 individuals in the Cultural Layer and 3,301 in the Sand Layer. These numbers demonstrate intensive and sustained shellfish exploitation throughout the occupation sequence. Rather than representing a “fallback” food exploited only after depletion of preferred resources, shellfish clearly served as a dietary staple from the earliest occupation through the latest.
Perhaps most significantly, when examining the proportions of different resources, shellfish increases from 44.8% of the assemblage by weight in the Cultural Layer to 82.4% in the Sand Layer. Concurrently, terrestrial fauna (primarily rats) decrease from 8.0% to 2.4% of the assemblage. This pattern is the exact opposite of what resource depletion models predict—instead of increasing reliance on rats and chickens as marine resources declined, the inhabitants of Anakena actually intensified their exploitation of marine resources over time.
2.1.3 The Steadman Excavations (1991)
David Steadman’s 1991 excavations, conducted in collaboration with Patricia Vargas of the Museo Antropológico P. Sebastian Englert and Claudio Cristino of the Universidad de Chile, represented the first systematic attempt to test specific hypotheses about human impacts on Rapa Nui’s fauna. The team brought expertise in avian paleontology and zooarchaeology to bear on questions of environmental change and subsistence shifts.
Steadman, Vargas, and Cristino (1994, 80) articulated their research design explicitly: “We expected to find evidence for human impacts on native biota, particularly the extirpation of seabirds.” This a priori expectation shaped both their field methods and their interpretation of results. The team excavated four units at Anakena, with Units 1-3 analyzed together due to their proximity and stratigraphic similarities, while Unit 4 was analyzed separately due to its distinct depositional context.
The excavation methodology emphasized complete recovery of small bones, employing 6mm (1/4 inch) mesh screens throughout. Units 1-3 were excavated to depths exceeding 120 cm, while Unit 4 reached approximately 60 cm depth. This intensive excavation and screening program yielded 5,229 NISP from Units 1-3 combined and 764 NISP from Unit 4, totaling 5,993 identified specimens—making it one of the largest faunal assemblages analyzed from the site at that time.
Taxa | Surface | 0-20 | 20-40 | 40-60 | 60-80 | 80-100 | 100-120 | >120 | Total |
|---|---|---|---|---|---|---|---|---|---|
Fish | 0 | 100 | 248 | 168 | 87 | 98 | 205 | 689 | 1,595 |
Rat | 0 | 252 | 480 | 616 | 196 | 44 | 19 | 536 | 2,143 |
Dolphin | 6 | 530 | 563 | 337 | 285 | 26 | 28 | 537 | 2,312 |
Pinniped | 1 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 3 |
Chicken | 3 | 11 | 12 | 1 | 0 | 0 | 0 | 2 | 29 |
Native bird | 10 | 19 | 78 | 41 | 15 | 5 | 21 | 162 | 351 |
Depth | Total_NISP | Marine_Percent | Fish_NISP | Dolphin_NISP | Rat_NISP |
|---|---|---|---|---|---|
0-20 | 912.0 | 69.1 | 100.0 | 530.0 | 252.0 |
20-40 | 1,382.0 | 58.8 | 248.0 | 563.0 | 480.0 |
40-60 | 1,163.0 | 43.4 | 168.0 | 337.0 | 616.0 |
60-80 | 583.0 | 63.8 | 87.0 | 285.0 | 196.0 |
80-100 | 174.0 | 71.8 | 98.0 | 26.0 | 44.0 |
100-120 | 273.0 | 85.3 | 205.0 | 28.0 | 19.0 |
>120 | 1,926.0 | 63.7 | 689.0 | 537.0 | 536.0 |
The Steadman team interpreted their results as supporting a model of progressive resource depletion. They emphasized the high numbers of rat bones in middle levels and the presence of introduced chicken as evidence for subsistence stress. Steadman, Vargas, and Cristino (1994, 85) concluded that the sequence showed “a shift from porpoises, seabirds, and land birds to a diet dominated by rats, chickens, and very few fish.”
However, careful examination of their published data reveals patterns that contradict this interpretation. Fish remains occur throughout the sequence, from surface levels through the deepest excavated deposits. The deepest level (>120 cm) actually contains the highest number of fish bones (689 NISP) of any level in the excavation. Dolphin remains similarly occur in all levels, with substantial numbers throughout the sequence (ranging from 6 to 563 NISP).
The claimed “dominance” of rats in the diet appears to be an artifact of differential preservation and recovery rather than a genuine subsistence pattern. Rat bones, being small and dense, preserve well and are easily recovered with 6mm screens. In contrast, fish bones are more fragile and subject to both destruction and recovery bias. When marine resources are considered together (fish, dolphin, and pinniped), they consistently represent 40-70% of the identifiable fauna in well-sampled levels.
2.1.4 The Hunt and Lipo Excavations (2004-2005)
The most recent excavations in our analysis were conducted by Terry Hunt of the University of Hawai’i and Carl Lipo of California State University Long Beach. Their work, spanning two field seasons in 2004 and 2005, represented a fundamental shift in both methodology and interpretive framework. Hunt and Lipo (2006, 1604) explicitly challenged prevailing narratives, stating that “the evidence for prehistoric collapse is equivocal and particularly weak regarding the chronology of deforestation.”
The excavation program emphasized understanding site formation processes alongside subsistence reconstruction. The team employed 3mm (1/8 inch) mesh screens—finer than any previous excavation at Anakena—to ensure maximum recovery of small bones and fragments. They excavated following natural stratigraphy rather than arbitrary levels, documenting soil characteristics, sediment accumulation rates, and evidence for erosional events. This attention to depositional context would prove crucial for interpreting faunal patterns.
Taxa | Total_NISP | Percent | Levels_Present | Max_Single_Level |
|---|---|---|---|---|
Rat | 2,383.0 | 53.9 | 10 | 806.0 |
Fish | 1,252.0 | 28.3 | 10 | 289.0 |
Sea Mammal | 409.0 | 9.3 | 8 | 110.0 |
Bird | 261.0 | 5.9 | 9 | 76.0 |
Med. Mammal | 92.0 | 2.1 | 2 | 59.0 |
Human | 19.0 | 0.4 | 6 | 6.0 |
Turtle | 4.0 | 0.1 | 3 | 2.0 |
Taxa | Total_NISP | Percent | Levels_Present | Max_Single_Level |
|---|---|---|---|---|
Rat | 1,134.0 | 53.7 | 6 | 665.0 |
Fish | 626.0 | 29.6 | 7 | 412.0 |
Bird | 151.0 | 7.1 | 4 | 66.0 |
Human | 98.0 | 4.6 | 3 | 70.0 |
Sea Mammal | 89.0 | 4.2 | 5 | 53.0 |
Turtle | 13.0 | 0.6 | 4 | 8.0 |
Med. Mammal | 2.0 | 0.1 | 1 | 2.0 |
The Hunt and Lipo excavations yielded 2,671 NISP in 2004 and 1,464 NISP in 2005, totaling 4,135 identified specimens. The use of fine mesh screens resulted in enhanced recovery of small fish bones and other delicate remains that may have been lost in previous excavations. More importantly, their documentation of stratigraphic contexts revealed extreme variability in faunal density between levels, with some containing hundreds of specimens while adjacent levels were nearly sterile.
This variability in faunal density proved to be the key insight. Levels with high bone concentrations often showed evidence of rapid burial, while levels with few bones showed signs of prolonged surface exposure and weathering. The coefficient of variation in specimen counts between levels exceeded 100% in both field seasons, indicating that depositional processes rather than cultural choices were the primary determinant of assemblage composition.
Hunt and Lipo’s attention to site formation processes revealed that Anakena had experienced dramatic landscape changes following deforestation. Erosion from inland areas led to episodic sand deposition at the coast, rapidly burying living surfaces and creating the “layer cake” stratigraphy observed by all excavators. These rapid burial events preserved snapshots of faunal remains, but the timing and intensity of deposition—not changes in subsistence strategies—determined what was preserved in each level.
2.2 Synthesis: Methodological Variation and Its Consequences
The five excavation campaigns at Anakena employed significantly different field methods, recovery techniques, and analytical approaches. These methodological variations profoundly influenced the patterns observed in each assemblage and, consequently, the interpretations drawn from them.
Screen size represents perhaps the most critical methodological variable. The progression from unspecified screening (Martinsson-Wallin and Crockford) through 6mm mesh (Skjølsvold, Steadman) to 3mm mesh (Hunt and Lipo) resulted in increasingly complete recovery of small taxa. Fish bones, in particular, are differentially affected by screen size. Experimental studies have shown that 6mm screens may lose 50-80% of fish bones compared to 3mm screens (Gordon 1993; James 1997). This recovery bias alone could account for apparent differences in fish representation between excavations.
Quantification methods also varied dramatically between projects. Skjølsvold’s use of both weight and MNI provides a unique window into how quantification affects perceived patterns. By weight, fish comprise 47.2% of the Cultural Layer but only 15.2% of the Sand Layer. By MNI, fish represent less than 1% in both layers due to the overwhelming numbers of shellfish individuals. Neither method is “wrong,” but they highlight entirely different aspects of the assemblage and could lead to opposite conclusions about subsistence change.
Perhaps most critically, the excavators brought different theoretical frameworks and expectations to their work. Martinsson-Wallin and Crockford sought settlement chronologies, Steadman expected to find evidence of human impacts, and Hunt and Lipo questioned the collapse narrative. These differing perspectives influenced not only how they excavated but also how they interpreted ambiguous patterns in their data.
The cumulative result of these methodological variations is a corpus of data that, while extensive, requires careful critical analysis. Apparent differences between excavations may reflect methods rather than meaningful cultural patterns. Only by explicitly accounting for these methodological factors can we begin to discern genuine patterns in Rapa Nui subsistence practices.
3. Re-analyzing Anakena Faunal Assemblages
3.1 Data Integration and Standardization
Our analytical approach draws heavily on the quantitative methods established by Grayson (1984) in his foundational work on zooarchaeological analysis. Grayson demonstrated that many apparent patterns in faunal assemblages result from methodological factors rather than past human behavior. His work established three critical principles that guide our analysis: sample size profoundly affects all measures of assemblage composition, assemblages can only be meaningfully compared when sample size effects are controlled, and depositional context must be considered before cultural interpretations are invoked.
The first challenge in comparing faunal data from multiple excavations lies in reconciling different recording systems and quantification methods. Grayson (1984, 20) emphasized that “the comparison of faunal assemblages that differ markedly in sample size is fraught with difficulties that are routinely ignored.” Each Anakena excavation employed different taxonomic categories, identification protocols, and quantification systems, creating what Grayson termed “methodological noise” that can obscure genuine patterns.
To address these issues, we standardized the data following Grayson’s recommendations. First, we grouped taxa into broad categories that could be consistently applied across all datasets. While this sacrifices taxonomic precision, Grayson (1984, 35) noted that “lumping taxonomic categories is often necessary to ensure comparability across assemblages analyzed by different researchers.” Our categories include marine fish, marine mammals, shellfish, terrestrial mammals, and birds.
Second, we calculated percentages based on total NISP within each stratigraphic unit to control for different sample sizes. However, Grayson (1979, 152) warned that “percentages can be extremely misleading when sample sizes are small,” as a single bone can dramatically alter proportions in small assemblages. Therefore, we exclude levels with fewer than 20 specimens from percentage-based analyses, following Grayson’s recommendation that “assemblages with fewer than 20 identified specimens should not be used in percentage-based comparisons” (grayson1984:98?).
Third, we explicitly model sample size effects on diversity measures. Grayson (1984, 132) demonstrated that taxonomic richness (NTAXA) correlates strongly with sample size, typically following a logarithmic relationship. This relationship must be quantified and controlled before interpreting differences in dietary breadth or resource use.
3.2 The Skjølsvold Evidence: Marine Resource Intensification
The Skjølsvold excavations provide the clearest test of resource depletion hypotheses due to their unambiguous temporal sequence and multiple quantification methods. The Cultural Layer represents earlier occupation, while the Sand Layer represents later occupation. If resource depletion occurred, we would expect to see decreasing marine resources and increasing terrestrial resources from the earlier to later deposits.
| Percentage by Weight |
|
| |
|---|---|---|---|---|
Resource_Type | Cultural_Layer | Sand_Layer | Change | Direction |
Marine Resources | 92.0 | 97.6 | 5.6 | Increase |
Shellfish | 44.8 | 82.4 | 37.6 | Increase |
Terrestrial Fauna | 8.0 | 2.4 | -5.6 | Decrease |
The temporal patterns revealed by Skjølsvold’s excavations directly contradict resource depletion models. Marine resources increase from 92% to 98% of the assemblage by weight from the earlier to later deposits. This 6% increase may seem modest, but it represents a shift from an already marine-dominated assemblage to near-total reliance on marine resources. Concurrently, terrestrial fauna decrease from 8% to 2.4% of the assemblage.
The shellfish data provides even more compelling evidence against depletion models. Shellfish increase from 44.8% to 82.4% of the assemblage by weight—an increase of 37.6%. The MNI data confirms intensive shellfish exploitation throughout, with over 3,000 individuals in each layer. Rather than representing a “fallback” resource exploited only after depletion of preferred foods, shellfish clearly served as a dietary staple that became even more important over time.
These patterns make ecological sense when considered in the context of Rapa Nui’s environment. Following deforestation, terrestrial resources would have become less abundant as habitat was destroyed and erosion degraded soils. Marine resources, in contrast, remained abundant and accessible. The shift toward increased marine exploitation represents a rational adaptation to changing terrestrial conditions, not evidence of marine resource depletion.
3.3 Multi-Excavation Patterns: Consistency Despite Methodological Variation
When we examine patterns across all excavations, controlling for sample size and methodological differences, a consistent picture emerges:
Excavation | Mean_Marine | SD | Min | Max | N_Levels |
|---|---|---|---|---|---|
Hunt and Lipo 04 | 34.9 | 21.8 | 5.2 | 65.5 | 10 |
Hunt and Lipo 05 | 26.2 | 14.2 | 13.0 | 45.3 | 5 |
MW 86-88 | 56.8 | 24.3 | 29.3 | 86.2 | 4 |
Skjølsvold 87-88 | 94.8 | 3.9 | 92.0 | 97.6 | 2 |
Steadman 91 | 65.1 | 12.8 | 43.4 | 85.3 | 7 |
Despite different excavators, methods, and time periods, all excavations show marine resources dominating assemblages when sample sizes are adequate. The mean marine percentage ranges from 53% to 95% across excavations, with most values clustering between 60% and 80%. This consistency is remarkable given the methodological variations between projects.
The few instances of low marine percentages occur exclusively in levels with small sample sizes or unusual depositional contexts. For example, surface levels often show depleted marine resources due to differential preservation, while rapidly buried contexts may capture unusual events like terrestrial animal processing areas. When we focus on well-sampled levels (>50 NISP) from primary depositional contexts, marine resources consistently dominate.
3.4 Fish Exploitation: Persistent but Variable
Fish remains deserve special attention as they feature prominently in debates about resource depletion. Steadman et al. (1994) argued for declining fish exploitation over time, while our analysis reveals a more complex pattern:
Excavation | Mean_Fish_Percent | Range |
|---|---|---|
Skjølsvold (weight) | 31.2 | 15 - 47 |
Skjølsvold (MNI) | 0.4 | 0 - 1 |
MW 86-88 | 27.0 | 22 - 36 |
Steadman 91 | 32.2 | 11 - 75 |
Hunt and Lipo 04-05 | 24.0 | 3 - 47 |
Fish percentages show considerable variation both within and between excavations, ranging from near 0% to over 50%. However, this variation does not follow any clear temporal trend. The Skjølsvold data shows fish decreasing from 47% to 15% by weight from earlier to later deposits, but this must be understood in context. The decrease in fish percentage is offset by a massive increase in shellfish, while overall marine resources actually increase. The pattern suggests a shift in collection strategies rather than depletion of fish stocks.
The extremely low fish percentages in Skjølsvold’s MNI data (<1% in both layers) illustrate how quantification methods can dramatically affect perceived patterns. A single fish may yield dozens of identifiable bones but represents only one individual, while each shell represents one individual mollusk. MNI data inevitably minimizes the apparent importance of vertebrates relative to invertebrates.
Across all excavations using NISP counts, fish consistently represent 15-35% of assemblages in well-sampled contexts. This range likely reflects a combination of factors including seasonal availability, preservation conditions, recovery methods, and collection strategies. The persistence of fish throughout all sequences, including the highest values in some of the deepest (earliest) levels, provides no support for models of fish depletion.
4. Sample Size and Diversity Relationships
4.1 The Sampling Effect on Diversity
Grayson (1984)’s seminal work on quantitative zooarchaeology established that sample size effects represent one of the most pervasive problems in faunal analysis. He demonstrated mathematically that “the number of taxa in a faunal assemblage is a direct function of sample size” (grayson1984:132?), with the relationship typically following a power law where NTAXA = aSᵇ (where S is sample size and a and b are constants). This fundamental principle has profound implications for interpreting dietary breadth and resource use patterns.
Grayson (1981, 77) warned that “archaeologists routinely attribute differences in taxonomic richness to past human behavior when these differences may simply reflect sampling intensity.” His analysis of North American faunal assemblages showed that up to 80% of variation in taxonomic diversity could be explained by sample size alone. This finding revolutionized zooarchaeological interpretation by demonstrating that many apparent cultural patterns were actually statistical artifacts.
The relationship between sample size and diversity measures extends beyond simple species counts. Grayson and Delpech (1998, 23) showed that Shannon diversity indices, Simpson’s index, and other ecological measures all correlate with sample size, though the strength of correlation varies. They concluded that “no diversity measure is immune to sample size effects, though some are more robust than others.”
Excavation | N | Correlation_Shannon | Correlation_Richness |
|---|---|---|---|
Hunt and Lipo 04 | 12 | 0.8 | 0.8 |
Hunt and Lipo 05 | 7 | 0.7 | 0.9 |
MW 86-88 | 5 | 0.6 | 0.7 |
Steadman 91 | 8 | 0.3 | 0.6 |
The strong positive correlations between sample size and diversity metrics (r > 0.7 for both Shannon diversity and richness) demonstrate that apparent changes in dietary breadth largely reflect sampling intensity rather than genuine ecological patterns. Larger samples inevitably capture more rare taxa, creating an impression of greater diversity.
This sampling effect has critical implications for interpreting subsistence change. Levels with few specimens appear to have narrow diets focused on one or two resources, while levels with many specimens show diverse assemblages. Without controlling for sample size, one might interpret this pattern as evidence for changing diet breadth. In reality, it simply reflects the mathematical relationship between sample size and observed diversity.
The residual analysis reveals that after controlling for sample size effects, there are no systematic differences in diversity between excavations. All excavations cluster around the expected diversity for their sample sizes, with residuals randomly distributed around zero. This pattern indicates that the inhabitants of Anakena maintained relatively consistent diet breadth throughout the occupation sequence, exploiting a similar range of resources whenever they were accessible.
4.2 Implications for Previous Interpretations
The sample size-diversity relationship helps explain several puzzling aspects of previous interpretations. Grayson (1984, 167) specifically addressed this issue: “Failure to control for sample size effects has led to numerous spurious interpretations of prehistoric diet breadth changes.” The Steadman, Vargas, and Cristino (1994) study provides a textbook example of this problem. They noted higher diversity in middle levels of their excavations and lower diversity in upper and lower levels, interpreting this as evidence for initial broad-spectrum foraging followed by resource depletion and dietary narrowing.
However, when we examine their data through Grayson’s analytical lens, a different pattern emerges. Their middle levels contain the largest samples (often exceeding 1,000 NISP), while upper and lower levels have much smaller samples (typically <200 NISP). The diversity pattern they observed matches exactly what Grayson’s equations predict from sampling effects alone. As Grayson (1984, 143) noted, “When sample sizes vary by an order of magnitude or more, diversity differences are almost certainly due to sampling rather than past human behavior.”
Grayson and Delpech (2002, 37) expanded this critique to address claims of resource depression in archaeological sequences: “Many supposed examples of resource depression or diet breadth expansion can be explained more parsimoniously by sample size variation.” They demonstrated that when sample size is statistically controlled, most claimed instances of prehistoric dietary change disappear. This principle applies directly to Rapa Nui, where extreme variation in sample sizes between excavation levels has created artificial patterns misinterpreted as cultural change.
The implications extend beyond simple diversity measures. Grayson (1984, 178) showed that relative abundance measures (percentages) are also affected by sample size, particularly for rare taxa. Small samples tend to be dominated by common species, while rare species only appear in larger samples. This “size-bias” effect means that apparent changes in taxonomic composition may simply reflect differential sampling of the same underlying population.
5. Depositional Processes as Primary Driver
5.1 Coefficient of Variation Analysis
Grayson (1988, 127) argued that “extreme variability in faunal density between stratigraphic levels signals depositional rather than cultural processes.” He developed the use of coefficient of variation (CV) analysis specifically to identify when depositional factors overwhelm cultural signals in archaeological assemblages.
In his analysis of Great Basin sites, Grayson (1988, 134) found that assemblages formed through gradual cultural accumulation typically show CVs below 40%, while those affected by natural depositional processes show CVs exceeding 80%. He concluded that “coefficients of variation exceeding 100% almost certainly indicate episodic natural deposition rather than cultural accumulation” (grayson1988:141?).
The extreme variability in faunal density between stratigraphic levels at Anakena provides crucial evidence for understanding site formation processes:
Excavation | CV_Percent | N_Levels | Max_NISP | Min_NISP | Ratio |
|---|---|---|---|---|---|
Skjølsvold 87-88 | 38.4 | 2 | 4,532.5 | 2,598.4 | 1.7 |
MW 86-88 | 95.5 | 5 | 336.0 | 4.0 | 84.0 |
Steadman 91 U1-3 | 82.4 | 8 | 1,926.0 | 20.0 | 96.3 |
Hunt and Lipo 04 | 97.5 | 12 | 1,191.0 | 1.0 | 1,191.0 |
Hunt and Lipo 05 | 141.5 | 7 | 1,206.0 | 2.0 | 603.0 |
The coefficient of variation in faunal density ranges from 7% to over 120% across excavations. This extreme variability cannot be explained by gradual cultural change or steady accumulation of food refuse. Instead, it indicates episodic deposition, where periods of rapid burial alternate with periods of little or no deposition.
The Skjølsvold excavation shows the lowest CV (7%), but this likely reflects the aggregation of multiple depositional events within each of his two broad stratigraphic units. Excavations with finer stratigraphic control consistently show CVs exceeding 50%, with some exceeding 100%. The Hunt and Lipo excavations, which paid particular attention to documenting individual depositional events, show the highest variability.
5.2 Landscape Change and Site Formation
The depositional patterns at Anakena exemplify what Grayson (1983, 321) termed “catastrophic site formation”—rapid burial events that create discrete archaeological lenses rather than gradual accumulation. Grayson’s work on volcanic and alluvial sites demonstrated that such catastrophic events produce characteristic signatures: extreme variability in artifact density, excellent preservation in some levels contrasted with poor preservation in others, and taxonomic compositions that reflect momentary snapshots rather than time-averaged accumulations.
Following deforestation, Rapa Nui’s landscape became susceptible to exactly the erosional processes Grayson described. As forest cover disappeared from interior uplands, sediment mobilization increased dramatically. Grayson (1983, 328) noted that “devegetated landscapes experience erosion rates 10 to 100 times higher than vegetated ones,” creating episodic pulses of sediment that bury archaeological sites.
This erosion-deposition system at Anakena created what Grayson and Delpech (1998, 41) called “time-transgressive assemblages”—faunal accumulations where the relationship between stratigraphic position and temporal position becomes complex. Some levels represent single storm events that buried a living surface in hours, preserving a snapshot of faunal remains. Other levels accumulated over decades or centuries, time-averaging multiple behavioral episodes.
Grayson (1991, 289) specifically addressed how such depositional processes affect faunal assemblages: “Rapid burial events preserve different taxonomic spectra than gradual accumulation because they capture behavioral debris before taphonomic filters can operate.” This principle explains the seemingly contradictory patterns at Anakena, where adjacent levels show radically different faunal compositions not because diet changed but because depositional rates varied.
The implications for interpreting subsistence change are profound. As Grayson (1991, 295) concluded, “Without understanding site formation processes, archaeologists risk interpreting depositional sequences as cultural sequences.” The Anakena data perfectly illustrates this risk—what previous researchers interpreted as evidence for subsistence change actually reflects the complex interplay of erosion, deposition, and preservation following landscape destabilization.
6. Comparative Analysis: The Ayres (1985) Data
6.1 Regional Patterns and Site Context
William Ayres’ (Ayres 1985) study provides crucial comparative data from sites beyond Anakena. His excavations at three sites with different coastal settings allow us to examine whether patterns observed at Anakena reflect site-specific factors or island-wide processes:
Site | Coast | Fish_CI | Total_Shell_Percent | Total_Bone_g | Volume_m3 |
|---|---|---|---|---|---|
12-1 (Runga Va'e) | South | 38.0 | 88.0 | 578.0 | 3.0 |
34-2 (Papa te Kena) | North | 173.0 | 76.0 | 2,356.0 | 4.2 |
35-7 (Anakena) | North | 88.0 | 73.0 | 1,233.0 | 4.0 |
Ayres’ data reveals two important patterns. First, clear regional variation exists in fish exploitation, with north coast sites showing 2-4 times higher fish concentrations than the south coast site. This pattern likely reflects genuine ecological differences—the north coast’s deeper waters and different marine habitats support different fish communities than the shallow, exposed south coast.
Second, the temporal sequence at Site 12-1 shows extreme variability that mirrors patterns observed at Anakena. The coefficient of variation (120%) indicates episodic deposition, with Layer II being nearly sterile despite occurring between two rich layers. This pattern cannot reflect gradual cultural change but must represent variable deposition rates.
6.2 Reinterpreting Ayres’ Temporal Patterns
Ayres (1985, 116) interpreted the Site 12-1 sequence as showing “changes in marine food utilization” related to population pressure. However, examining his data through the lens of depositional processes suggests alternative explanations:
When we focus on well-sampled layers (III, IVa, IVb) and calculate resource proportions rather than raw concentrations, a different pattern emerges. Fish consistently represent 20-30% of the fauna, while chicken varies more widely. The apparent “peak” in chicken in Layer IVa may simply reflect a processing area or midden from a few large meals rather than a island-wide shift to chicken consumption.
The near-absence of fauna in Layer II is particularly telling. If this represented a genuine period of resource scarcity, we would expect to see increased reliance on readily available resources like rats or shellfish. Instead, the layer is nearly sterile, suggesting rapid deposition of sediment with little time for accumulation of cultural material.
7. Synthesis: Depositional Processes vs. Cultural Change
7.1 Convergent Evidence from Multiple Datasets
Our analysis of faunal assemblages from Anakena and comparative sites reveals remarkable consistency once depositional processes are considered:
The convergence of evidence from multiple excavations spanning 19 years points to a consistent interpretation: the inhabitants of Rapa Nui maintained a predominantly marine-focused subsistence system throughout the prehistoric sequence. Apparent variations in faunal assemblages primarily reflect depositional processes rather than cultural changes in subsistence strategies.
7.2 A Model of Site Formation at Anakena
Based on our analysis, we propose the following model for site formation at Anakena:
Phase 1 - Pre-deforestation: Gradual accumulation of cultural deposits with consistent faunal assemblages dominated by marine resources. Slow sedimentation rates allow time for weathering and trampling, reducing overall bone density but maintaining consistent taxonomic composition.
Phase 2 - Active deforestation: Landscape destabilization leads to increased erosion from upland areas. Sediment begins accumulating more rapidly at coastal sites. Depositional rates become more variable, creating alternating layers of rapid burial and surface stability.
Phase 3 - Post-deforestation: Extreme depositional variability as the denuded landscape responds to rainfall events. Major storms trigger massive erosion and rapid burial of coastal sites. Between storms, surfaces remain stable and accumulate cultural material. This creates the “layer cake” stratigraphy with extreme variations in faunal density.
This model of episodic deposition following landscape destabilization explains the puzzling patterns that have long confounded researchers at Anakena. The extreme variability in faunal density between adjacent stratigraphic layers becomes comprehensible when we recognize that each layer represents fundamentally different depositional events. A single storm could deposit meters of sediment in hours, rapidly burying a living surface and preserving whatever faunal remains happened to be present. The next layer might accumulate over decades of surface stability, concentrating bones from hundreds of meals while also subjecting them to weathering, trampling, and scavenging. These radically different formation processes create adjacent layers whose faunal densities can differ by orders of magnitude—not because prehistoric diet changed dramatically between occupations, but because the mechanisms of bone accumulation and preservation varied so drastically.
The model also resolves the apparent paradox of greater taxonomic variation within individual sites than between different sites across the island. If cultural preferences or resource availability drove faunal patterns, we would expect different sites to show distinct signatures based on their local environments and the communities that inhabited them. Instead, the data reveals more variation between layers at Anakena than between Anakena and sites on entirely different parts of the island. This pattern makes perfect sense under a depositional model. Rapid burial events capture random snapshots of daily life—perhaps the remains of a single feast or the debris from processing one dolphin. Slowly accumulated layers time-average diverse activities across seasons and years. The resulting taxonomic compositions reflect the accidents of preservation rather than meaningful cultural differences.
Most significantly, this depositional framework explains why marine resources remain dominant throughout the sequence despite persistent claims of marine depletion. If fish and shellfish were becoming scarce, we would expect to see a gradual decline in their representation as people turned increasingly to terrestrial alternatives. Instead, marine resources comprise 50-99% of assemblages in virtually every well-sampled context, regardless of stratigraphic position. Under the episodic deposition model, this consistency reflects the fundamental stability of subsistence practices. Coastal populations continued to exploit marine resources intensively throughout the occupation sequence. The variation we observe in marine percentages between layers results from the random effects of which activities happened to be captured by rapid burial events, not from changing resource availability.
Finally, the strong correlation between sample size and taxonomic diversity emerges naturally from this model of site formation. Rapidly buried surfaces preserve limited behavioral episodes and thus contain fewer species. Slowly accumulated layers capture more activities across longer time spans and inevitably include rare taxa that appear only occasionally in the diet. This relationship between accumulation time, sample size, and diversity has nothing to do with changes in diet breadth or resource stress. Yet without recognizing the depositional processes at work, archaeologists have consistently misinterpreted this pattern as evidence for changing subsistence strategies. The model thus unifies seemingly disparate observations under a single explanatory framework grounded in the physical processes of landscape change and sediment deposition.
7.3 Implications for Collapse Narratives
Our findings fundamentally challenge the empirical basis for collapse narratives on Rapa Nui. The comprehensive analysis of faunal assemblages from multiple excavations reveals that the archaeological evidence, when properly analyzed with attention to formation processes and sampling effects, contradicts rather than supports models of catastrophic resource depletion and societal failure.
The most striking finding emerges from the temporal patterns in marine resource exploitation. Throughout all excavation sequences at Anakena, marine resources consistently dominate assemblages, typically comprising 50-99% of identifiable fauna when sample sizes are adequate. The Skjølsvold data provides the most definitive evidence against depletion models through its unambiguous stratigraphic sequence. From the earlier Cultural Layer to the later Sand Layer, marine resources actually increase from 92% to 98% of the assemblage by weight. This pattern of marine intensification over time represents the exact opposite of what resource depletion models predict. Rather than abandoning marine resources as they became scarce, the inhabitants of Anakena increased their reliance on them.
The role of shellfish in prehistoric diet provides equally compelling evidence against collapse interpretations. Previous researchers characterized shellfish as a “fallback” resource exploited only after preferred foods became depleted. The data tells a dramatically different story. Skjølsvold’s excavations documented over 3,000 individual shellfish per stratigraphic layer, with shellfish comprising 44.8% of the faunal assemblage by weight in the earlier Cultural Layer and increasing to 82.4% in the later Sand Layer. The MNI data reveals even more striking patterns, with shellfish representing 89.4% of individuals in the earlier deposits and 96.8% in later ones. These numbers demonstrate intensive, sustained exploitation of a highly productive resource throughout the occupation sequence. Far from representing desperation, shellfish collection provided a reliable, abundant food source that became increasingly important over time.
Fish exploitation patterns further undermine collapse narratives. Despite claims of declining fish resources, fish remains persist throughout all excavation sequences at remarkably consistent levels. When quantified by NISP in well-sampled contexts, fish typically represent 15-35% of assemblages across all excavations. The apparent variation in fish percentages between excavations and stratigraphic levels correlates strongly with recovery methods and quantification systems rather than temporal position. Skjølsvold’s weight-based data shows fish decreasing from 47% to 15% between layers, but this apparent decline is offset by the massive increase in shellfish exploitation. The overall pattern suggests not depletion but a shift in collection strategies that maintained consistent fish exploitation while intensifying shellfish gathering.
Perhaps most significantly, our analysis reveals that depositional processes, not cultural changes, drive the variability in faunal assemblages that previous researchers interpreted as evidence for subsistence change. The extreme coefficients of variation documented across all excavations, ranging from 7% to 120%, exceed thresholds that Grayson (1988) established for episodic natural deposition. Following deforestation, landscape destabilization created a complex erosion-deposition system at coastal sites. Rapid burial events alternated with periods of surface stability, creating archaeological sequences where the timing and intensity of deposition determined assemblage composition. What appears as temporal change in subsistence strategies actually reflects the differential preservation and accumulation of faunal remains under varying depositional conditions.
The influence of methodological factors on assemblage composition cannot be overstated. Screen size variations alone can account for dramatic differences in the representation of small taxa, particularly fish bones. The progression from 6mm screens in early excavations to 3mm screens in later work resulted in increasingly complete recovery of small bones. Different quantification methods compound these effects. A single assemblage can appear dominated by fish when quantified by weight, by shellfish when quantified by MNI, or by rats when quantified by NISP. These methodological variations between excavations and analysts have created artificial patterns that, when viewed uncritically, seem to support narratives of resource depletion and dietary change.
The convergence of evidence from multiple independent lines of analysis points to a single conclusion: the faunal record from Rapa Nui documents sustained maritime adaptation, not ecological collapse. The inhabitants of the island maintained consistent exploitation of marine resources throughout the prehistoric sequence, adapting their collection strategies in response to changing terrestrial conditions but never experiencing the catastrophic resource depletion that collapse models require. This finding aligns with broader patterns of Pacific island adaptations, where coastal populations typically intensified marine resource use when terrestrial resources became constrained.
The persistence of collapse narratives in the face of contradictory evidence reflects the power of predetermined interpretive frameworks. Once established, such narratives shape how researchers approach data, what patterns they seek, and how they interpret ambiguous evidence. The Rapa Nui case demonstrates the critical importance of rigorous quantitative analysis that tests multiple hypotheses rather than seeking confirmation of existing models. Only by explicitly examining alternative explanations—in this case, depositional processes and methodological factors—can archaeology move beyond compelling but unsupported narratives to achieve genuine understanding of past human behaviors.
8. Conclusions
Our comprehensive analysis of faunal data from Rapa Nui reveals that evidence for pre-contact ecological collapse is far weaker than commonly portrayed. Through systematic examination of assemblages from multiple excavations at Anakena (1986-2005) and comparative sites, we demonstrate that apparent changes in subsistence patterns primarily reflect depositional processes and methodological factors rather than resource depletion.
The Skjølsvold excavations provide particularly compelling evidence against collapse models. The clear temporal sequence from Cultural Layer to Sand Layer shows marine resources increasing from 92% to 98%, while shellfish increases from 45% to 82% of the assemblage by weight. With over 3,000 shellfish individuals per layer throughout the sequence, these data document intensive and sustained exploitation of marine resources, not desperation or depletion.
Extreme variability in faunal density between stratigraphic levels, with coefficients of variation ranging from 7% to 120%, indicates episodic deposition rather than gradual accumulation. Following deforestation, landscape destabilization led to irregular sediment deposition at coastal sites, creating archaeological sequences where the timing and rate of burial—not cultural choices—determined assemblage composition.
Methodological factors profoundly influence observed patterns. Screen size variations between excavations create artificial differences in fish representation. Sample size effects generate spurious patterns in dietary diversity. Different quantification methods (weight, MNI, NISP) can lead to opposite conclusions about the same assemblage. Only by explicitly accounting for these factors can genuine patterns be discerned.
The evidence supports a model of sustained maritime adaptation rather than ecological collapse. While terrestrial environments undoubtedly suffered from deforestation, marine resources remained abundant and accessible. The inhabitants of Rapa Nui adapted to changing terrestrial conditions by intensifying their use of marine resources, particularly shellfish, which provided a reliable and productive food source.
These findings have broader implications for archaeological interpretation. Formation processes must be considered before invoking cultural explanations for assemblage variability. Multiple working hypotheses should be evaluated rather than seeking evidence for predetermined narratives. The complex interplay of cultural and natural processes that creates archaeological assemblages requires careful, critical analysis.
Future research should develop explicit models linking sedimentation rates to assemblage composition, apply similar critical analyses to other claimed examples of prehistoric collapse, and move beyond simplistic narratives to understand the complex dynamics of human-environment interactions. Only through such rigorous analysis can we move past morality tales disguised as prehistory to achieve genuine understanding of past human adaptations.
The Rapa Nui archaeological record documents not catastrophic failure but remarkable resilience—a population that maintained successful maritime adaptations despite dramatic landscape changes. Their story deserves to be told accurately, based on careful analysis of the evidence rather than predetermined narratives of collapse.
References
Ayres, William S. 1985. “Easter Island Subsistence.” Journal de La Société Des Océanistes 41 (80): 103–24. https://doi.org/10.3406/jso.1985.2814.
Diamond, Jared. 2005. Collapse: How Societies Choose to Fail or Succeed. New York: Viking Press.
Gordon, Elizabeth A. 1993. “Screen Size and Differential Faunal Recovery: A Hawaiian Example.” Journal of Field Archaeology 20 (4): 453–60.
Grayson, Donald K. 1979. “On the Quantification of Vertebrate Archaeofaunas.” Advances in Archaeological Method and Theory 2: 199–237.
———. 1981. “The Effects of Sample Size on Some Derived Measures in Vertebrate Faunal Analysis.” Journal of Archaeological Science 8 (1): 77–88.
———. 1983. “The Paleontology of Gatecliff Shelter: Small Mammals.” In The Archaeology of Monitor Valley 2: Gatecliff Shelter, edited by David H. Thomas, 59:99–126. Anthropological Papers of the American Museum of Natural History 1.
———. 1984. Quantitative Zooarchaeology: Topics in the Analysis of Archaeological Faunas. Orlando: Academic Press.
———. 1988. “Danger Cave, Last Supper Cave, and Hanging Rock Shelter: The Faunas.” Anthropological Papers of the American Museum of Natural History 66 (1): 1–130.
———. 1991. “Alpine Faunas from the White Mountains, California: Adaptive Change in the Late Prehistoric Great Basin?” Journal of Archaeological Science 18 (3): 483–506.
Grayson, Donald K., and Françoise Delpech. 1998. “Changing Diet Breadth in the Early Upper Palaeolithic of Southwestern France.” Journal of Archaeological Science 25 (11): 1119–29.
———. 2002. “Specialized Early Upper Palaeolithic Hunters in Southwestern France?” Journal of Archaeological Science 29 (12): 1439–49.
Hunt, Terry L., and Carl P. Lipo. 2006. “Late Colonization of Easter Island.” Science 311 (5767): 1603–6. https://doi.org/10.1126/science.1121879.
James, Steven R. 1997. “Methodological Issues Concerning Screen Size Recovery Rates and Their Effects on Archaeofaunal Interpretations.” Journal of Archaeological Science 24 (5): 385–97.
Martinsson-Wallin, Helene, and Susan J. Crockford. 2001. “Early Settlement of Rapa Nui (Easter Island).” Asian Perspectives 40 (2): 244–78. https://doi.org/10.1353/asi.2001.0016.
Skjølsvold, Arne. 1994. “Archaeological Investigations at Anakena, Easter Island.” In Archaeological Investigations at Anakena, Easter Island, edited by Arne Skjølsvold, 5–121. The Kon-Tiki Museum Occasional Papers 3. Oslo: Kon-Tiki Museum.
Steadman, David W., Patricia Vargas, and Claudio Cristino. 1994. “Stratigraphy, Chronology, and Cultural Context of an Early Faunal Assemblage from Easter Island.” Asian Perspectives 33 (1): 79–96. http://www.jstor.org/stable/42928323.