Ecological dynamics of moa extinctions reveal convergent refugia that today harbor flightless birds.

doi:10.21203/rs.3.rs-3273867/v1

Download PDF

Article

Ecological dynamics of moa extinctions reveal convergent refugia that today harbor flightless birds.

https://doi.org/10.21203/rs.3.rs-3273867/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 24 Jul, 2024

Read the published version in Nature Ecology & Evolution →

Version 1

posted

You are reading this latest preprint version

Human settlement of islands across the Pacific Ocean was followed by waves of faunal extinctions that occurred so rapidly that their dynamics are difficult to reconstruct in space and time. These extinctions included large, wingless birds endemic to New Zealand called moa. We reconstructed the range and extinction dynamics of six genetically distinct species of moa across New Zealand at a fine spatiotemporal resolution, using hundreds of thousands of process-explicit simulations of climate-human-moa interactions, which were validated against inferences of occurrence and demographic change from an extensive fossil record. This statistical-simulation analysis revealed important interspecific differences in the ecological and demographic attributes of moa that influenced the timing and pace of their geographic and demographic declines following colonization of New Zealand by Polynesians. Despite these interspecific differences in extinction dynamics, the spatial patterns of geographic range collapse of moa species were similar. The final populations of all moa species persisted in suboptimal habitats in cold, mountainous areas that were generally last and least impacted by people. These isolated refugia for the last populations of moa continue to serve as sanctuaries for New Zealand’s remaining flightless birds, providing novel insights for conserving endemic species in the face of current and future threats.

Biological sciences/Ecology/Biogeography

Biological sciences/Ecology/Ecological modelling

conservation biogeography

distribution

New Zealand

range collapse

spatially explicit population models

Oceanic islands hold a large proportion of the world’s biodiversity¹, but are hotspots of extinction and endangerment². While species introductions, habitat loss and habitat degradation are the main drivers of current and future risk of extinction on islands^3,4, a wider combination of threats is likely to have contributed to earlier extinctions and the overall erosion of ecological resilience of insular biotas⁵. Over-exploitation by first human colonists triggered population declines and extinctions of many island endemics⁶, in turn causing declines in vital ecological processes and the functional impoverishment of island ecosystems across the globe^5,7.

These anthropogenic transformations in taxonomic and functional diversity were particularly severe on islands in the Pacific Ocean, which harbored one of the most diverse and distinctive assemblages of endemic vertebrates in the world⁸. Human expansion across the Pacific began in the late Holocene, with Austronesian-speaking people embarking on extraordinary sea voyages - ultimately settling some of the most isolated landmasses on the planet⁹. This rapid expansion of people across a third of Earth’s surface¹⁰ caused wide-scale population declines and extinctions of native insular biotas, disproportionately impacting endemic species, including hundreds of species of flightless birds¹¹. The fraction of species that survived the initial waves of human colonization tend to persist today only as small and highly threatened populations in tiny fragments of their once more expansive, native ranges¹².

The last Pacific islands to be colonized by people were the islands of East Polynesia⁹. The largest and most ecologically diverse islands of this region are those of New Zealand, which were settled in the mid-13th Century C.E.¹³. Habitat change, introduced species, and hunting following the arrival of Polynesian colonists in New Zealand transformed ecosystems and their services to nature^14,15, giving rise to one of the largest and most rapid collapses of native biota in the Pacific^16,17. Included among New Zealand’s lost species were enormous raptors¹⁶, unique flightless species¹⁸, and giant wingless birds called moa (Dinornithiformes)¹⁹. Based on fossil chronology, it is estimated that these ecological and evolutionarily unusual birds went extinct within two centuries of first human contact^19,20.

Despite extinctions of moa occurring only hundreds of years ago, there is remarkably little known regarding ecological and demographic differences between the nine moa species, and how these differences may have caused variation in the spatial and temporal dynamics of their declines. Here we integrate process-driven ecological models with extensive paleontological information and detailed reconstructions of the dynamics of Polynesian colonists to disentangle and compare the mechanisms of range collapse and extinction of different species of moa across New Zealand at fine spatiotemporal resolutions²¹. Pattern oriented modelling was used to ensure that the structure, parameterization and dynamics of models accurately simulated inferences of range collapse from paleo-archives²². This verified ecological modelling approach²³ allowed us to identify key interspecific differences (and commonalities) in the demographic and life-history traits, ecological preferences, range dynamics and extinction trajectories of moa. The modelling protocol also provided a rare opportunity to address two questions fundamental to conservation biology: (i) whether range and population collapses follow the same paths in geographically co-distributed species; and (ii) whether understanding the dynamics of historical extinctions can inform likely pathways of decline in extant, but threatened species^24,25.

Using more than three hundred thousand different process-explicit simulations of climate-human-moa interactions, validated against inferences of demographic and distributional change from a well-represented fossil record, we reconstructed the annual structure and population dynamics of the geographic ranges of six (including two subspecies) species of moa over two millennia. We show that key differences in the ecology and demography of moa ultimately affected the speeds of range collapse and timings of extinction, but not the locations of their final populations. By comparing these final refugia of moa to the distributions of today’s flightless birds, many of which are critically endangered, we demonstrate how integrative research on past extinctions can provide novel insights for conserving endemic species in the face of current and future threats in New Zealand and other oceanic islands as well.

Ecological and demographic attributes

Our statistical-simulation approach that integrated spatially explicit population models (SEPMs²⁶) of climate-human-moa interactions with detailed fossil chronologies, revealed important inter-specific differences in traits of moa that govern geographical range and extinction dynamics. Parameter and structural tests using pattern-oriented modelling with Approximate Bayesian Computation²⁷ showed that species-specific ecological and demographic traits were needed to closely reconcile interspecific differences in the occurrence, distribution and extinction dynamics of moa as inferred from the fossil record (Fig. 1.; Table S1).

Models with species-specific parameter values (Table S1) were highly accurate in reconstructing spatiotemporal occurrence at fossil sites and inferences of location and timing of extinction (Figure S1). A principal component analysis of estimates of six ecological (niche marginality, carrying capacity and area of occupancy) and demographic (population growth, proportion of birds dispersing and dispersal distance) attributes from the ‘best’ SEPMs (0.25% of all SEPMs), showed three significant axes of corelated trait variation for moa, explaining 73.8% of variation (Table S2).

Trait probability densities²⁸ revealed large inter-specific variation in occupancy of this 3-dimensional trait space, showing that each species occupied < 20% of the total volume of moa trait space (Figure S2). Dinornis robustus and Euryapteryx curtus curtus occupied the smallest trait spaces (12.7 and 13.2% of total trait space) and Megalapteryx didinus the largest (19% of total trait space). Moreover, we show that there was little overlap between the trait probability densities of most moa (Table S3), indicating significant ecological specialization²⁹. The exception was for the two species of Dinornis, where D. robustus was likely to have occupied a trait space that fell almost entirely within the envelope of trait space for D. novaezealandiae (Table S3).

Extinction dynamics

Our validated simulations show that overlap between moa and people is likely to have been much longer than previously thought^19,20, with extinctions occurring over a period of three centuries following initial human settlement (Fig. 2). We show that population sizes of all six species of moa were stable for centuries prior to human occupation of New Zealand (Fig. 2). This is because of relatively constant climatic and environmental conditions during the late-Holocene³⁰, as indicated by ‘best’ projections of habitat suitability (Figure S3). However, following colonization of New Zealand by Polynesians, anthropogenic hunting of birds and harvesting of eggs caused rapid population declines and range contractions of moa (Fig. 2). Despite the rapidity of these declines, our statistical-simulation analysis pinpoints important interspecific differences in paces of geographic and demographic declines, which reflect species’ attributes interacting with human threatening processes.

Our simulations show that Pachyornis geranoides was the first species to go extinct (Fig. 2). Its model-averaged estimate of extinction of 1354 CE (Fig. 2) was approximately 100 years after human arrival in New Zealand (Fig. 2). A small area of occupancy, low population abundance and low growth rate (Fig. 1) are likely to have made P. geranoides particularly vulnerable to exploitation by humans. Extinctions of D. robustus, P. elephantopus and D. novaezealandiae occurred in quick succession nearly a century later, followed by extinctions of E. curtus gravis, M. didinus and E. curtus curtus (Fig. 2). Megalapteryx didinus is likely to have persisted longest on the South Island by virtue of its high population abundance, medium-to-high growth rate, a large area of occupancy and a particular ecological affinity to high-altitude habitats. Tests of niche marginality using outlier mean index³¹ show that M. didinus preferred habitats that were environmentally distinct, being found in highest densities at high altitudes (Fig. 1). Prolonged persistence of E. curtus curtus reflects its harvest dynamics, where it became harder (and perhaps more energetically costly) to hunt E. curtus curtus at low densities. Model validation required harvest of E. curtus curtus to follow a strongly sigmoidal Type III functional response, where harvest success is small at low densities (Fig. 1). This can result from prey adaptation (evolved or learned behavior), prey switching by hunters or prey being located in refugia³². Our models show that the collapse of E. curtus curtus to rugged areas in the southeast of the North Island (Fig. 2), and their low dispersal capabilities (Fig. 1) facilitated reduced contact between E. curtus curtus and Polynesian colonists in their last populations.

By analyzing spatiotemporal patterns of population extirpation (accounting for species and island differences), using general additive models (GAM³³), we show that the spatial pattern of geographic range collapse of moa followed similar paths for co-distributed species. Moa tended to contract their ranges to the latitudinal center of their pre-human geographic distributions on the North and South Islands (Fig. 3, Figure S4), with their final populations persisting south and west of these locations, respectively (Fig. 2). These areas of prolonged persistence (Fig. 3) were some of the most isolated from human impacts, with elevation and year of human arrival also being important predictors of time of extirpation (Table S4).

Three-dimensional plots of GAM response surfaces³⁴ showed that the last refugia for moa were invariably at higher elevations with relatively cool and wet climates, making them suboptimal habitats. We found no evidence that the ranges of moa contracted inwardly (as predicted by theory³⁵) to optimal habitats where their abundances were greatest, which for most species was in lowland areas (Figure S4). Pre-human spatial patterns of abundance, reconstructed across the entire geographic distributions of moa, showed that lowland areas tended to host higher population densities of moa (Figure S4). On the South Island, these were to the north and south of the latitudinal center of the distribution of D. robustus, E. curtus gravis, M. didinus and P. elephantopus. On the North Island, these areas were located east and west of the latitudinal center for D. novaezealandiae, E. curtus curtus and P. geranoides.

Conservation implications

By comparing contemporary ranges of New Zealand’s extant flightless birds with the extirpation patterns of moa, we show that the spatial patterns of extinction for moa predict geographic pathways of decline for threatened birds. Distributional data for extant flightless birds in New Zealand prior to large-scale conservation reintroductions and restorations³⁶, showed that populations of flightless birds are found with higher frequency in or near to critical final refugia of moa species (Fig. 4). This comparative analysis showed that flightless birds today are retreating to the same final sanctuaries of moa, and the spatial concordance is strongest for extant birds under the highest levels of threat: those in the latter stages of geographic range collapse (Fig. 4). Thus, regardless of their initial and once expansive historic ranges¹², our modelling revealed that New Zealand’s flightless birds are contracting their ranges to isolated and high elevation refugia last impacted by humanity as they progress towards extinction (Fig. 4).

Human colonization of the islands of the Pacific Ocean was followed by the extinction of thousands of endemic species^4,37. Most of these declines occurred so rapidly that it has proven intractable to reconstruct their population and distributional collapses at high spatiotemporal resolution with existing approaches. Our advances in process-driven ecological modelling, complemented by extensive paleontological information, provide invaluable insights into the diverse ecology and demography of moa of New Zealand and the different dynamics underpinning their extinctions.

Reconstructions of population decline, range collapse and extinction of six species of moa, informed and validated by the fossil record, revealed substantial specialization among species, with each species occupying < 20% of the total (across-species) volume of ecological and demographic trait space. Megalapteryx didinus was the most ecologically and demographically distinct species, with the least amount of overlap in trait space with any other species. This was driven largely by having a niche with unusually high marginality, resulting from a particular ecological affinity to high-altitude habitats. Divergence of this most basal moa lineage happened ~ 6 million years ago, during a period of rapid mountain uplift³⁸, and its preference for steep, mountainous habitats³⁹ meant that M. didinus populations were the last and least impacted by expanding human populations on the South Island. Pachyornis geranoides also occupied a relatively unique area of moa trait space characterized by very low population growth rates, distinguishing it from its sister species P. elephantopus (Figure S2).

Dinornis novaezealandiae, Euryapteryx curtus curtus and Euryapteryx curtus gravis emerged as having ecological and demographic attributes that were typical of moa more generally. Their attributes clustered around the center of total trait space for moa, with each species exhibiting moderate overlap in trait space with at least one other moa species. Fossil evidence also points to these taxa having been generalists in their ecological preferences⁴⁰. For example, the wide distributional record of fossils for D. novaezealandiae and E. curtus curtus, suggests that both species were wide ranging on the North Island, living in environments as diverse as forests and coastal dunes. In the case of D. novaezealandiae, fossils have also been found in subalpine herb fields and grasslands in addition to forest and coastal environments ^38,41. Coprolites and gizzard content of Euryapteryx curtus gravis (and D. robustus on the South Island) indicate a broad diet that included fruit, twigs and leaves from small herbs and trees³⁹.

Our modelling suggests that the range collapse and extinction of all species of moa resulted from low but sustained harvest rates, which nevertheless exceeded the limited reproductive capacities of these species. While there has been ample speculation on rates of moa decline following colonization of New Zealand by Polynesians^19,20,42,43 our process-driven modelling provides the first evidence of substantial interspecific variation in extinction trajectories. We show that it is likely that P. geranoides went extinct within just 100 years of Polynesian colonization, and nearly 100 years before any other moa. Its high vulnerability to extinction reflects its comparatively low population growth rate. Other species, such as M. didinus, went extinct later, because their ecological and demographic attributes made them more resilient to human hunting. These included a high population abundance, medium-to-high growth rate, large area of occupancy and an ecological preference for high-altitude habitats, which were challenging to Polynesian colonists. A low dispersal capacity for E. curtus curtus, matched with potential behavioural adaptations that minimised human contact, is likely to have effectively isolated their final populations from Polynesian colonists, prolonging persistence for approximately 200 years.

While these ecological and demographic differences influenced timings of extinctions of different species following the arrival of Polynesians in New Zealand, they did not alter patterns in range collapse. Overexploitation by expanding Polynesian populations⁴² was so dominant that it overwhelmed pre-existing spatial patterns in densities and variations in ecological and demographic attributes of moa, causing convergence in patterns of geographic range collapse of the six species of moa that we studied. The final refugia of all species of moa were in suboptimal habitats that were most isolated from the impacts of humans.

These convergent patterns of range collapse seem to contradict earlier assertions by macroecologists that, because abundance tends to be greatest and least variable in the center of a species’ historic distribution⁴⁴, geographic ranges of declining species should implode, with final populations persisting in the range center⁴⁵. Subsequent research, however, discovered that empirical patterns were just the opposite, with final populations of most species being located in isolated reaches of their historic ranges — those last and least impacted by the expanding waves of anthropogenic threats^24,46. Our fine spatiotemporal resolution reconstructions of the range and extinction dynamics of six species of moa are consistent with these neontological patterns and observations, indicating that range contractions for moa were largely determined by the dynamic geography of Polynesians and their commensals – disappearing first from high quality lowland habitats, then diminishing in abundance more slowly with increasing elevation and distance from the shores^25,47,48.

Moa ranges contracted towards the abundance-weighted latitudinal center of their pre-human distributions in response to Polynesian settlement and subsequent harvesting of birds and their eggs in coastal lowlands. Refugial areas on the North Island were in the Ruahine and Kaweka Ranges, while on the South Island, they were increasingly restricted to the Southern Alps. These refugia invariably were in higher elevations with relatively cool and wet climates, making them suboptimal habitats for the moa but, more importantly, not easily accessible, and less intensively exploited by humans. Similar patterns of more recent range collapses have been reported on other islands, including for widespread Hawai’ian Honeycreepers, where remaining species are restricted to high elevation refugia⁴⁹. This pattern likely reflects the geography of mountainous islands, where the geographic center of a species distribution can fall in sub-optimal, high elevation habitat, even when directly accounting for the distributional pattern of spatial abundance⁴⁴, as done here.

Our integrative modelling shows how knowledge of the spatial dynamics of ancient extinctions can inform likely pathways of decline for extant species and vice versa. We show that extant flightless birds in New Zealand are retreating to the same final sanctuaries of moa, regardless of their initial and once expansive historic ranges¹². Thus, New Zealand’s flightless birds are contracting their ranges to isolated and high elevation refugia as they decline towards extinction, like moa did many centuries ago. Again, this pattern arises not because these sites provide the best habitat, but because they are the most isolated from the effects of European colonists and the mammalian predators that they introduced⁵⁰.

There are eight remaining flightless endemic birds on the main islands of New Zealand today, seven of these are threatened with extinction. These include a flightless parrot, the kākāpō (Strigops habroptilus), two rails, the takahē (Porphyrio hochstetteri) and the weka (Gallirallus australis), and five species of kiwi (Apteryx spp.). In the 1980’s, before wide-scale conservation-driven translocations⁵¹, the kākāpō, takahē, tokoeka (Apteryx haastii) and roroa (Apteryx australis) were confined exclusively to small distributions in mountainous areas where moa last persisted. This is despite having extensive wide-ranging pre-human distributions¹⁶. Even the most widespread and least threatened of these flightless birds — the weka and the North Island brown kiwi (Apteryx mantelli) — were most frequently sighted in mountainous areas close to moa refugia (Figure S7). The North Island brown kiwi (Apteryx mantelli) has one population inhabiting lowland habitats of the Northland Peninsula (Figure S7), which are far from the last stronghold of moa on this island. This apparent anomaly reflects early implementation of in situ management programs for flightless birds on the Northland Peninsula and demographic traits that promote resilience to extinction, including a high reproductive capacity⁵².

Although current-day drivers of decline of New Zealand’s native birds differ from those that drove the collapse and loss of moa, the geographic signatures of these extinction forces are largely similar in that they reflect the spatial dynamics of anthropogenic forces. Habitat conversion by Europeans and their commensals (competitors, predators, pathogens and parasites⁵³) spread from lowland sites of initial introduction and, like the earlier waves of Polynesian expansion, remain least severe in the most isolated, mountainous regions⁵⁴. Thus, the key factor in the persistence of the remaining endemic species of New Zealand is the preservation of their geographic and ecological isolation.

New Zealand represents just one of many island systems of globally-significant biodiversity and conservation⁵⁵. The Hawai’ian archipelago lost more species of native birds following human settlement than New Zealand⁸, as did many other Pacific archipelagos, yet we have little understanding of the spatiotemporal dynamics of biodiversity loss in these unique island ecosystems. Similarly rapid extinction of elephant birds and other megafauna occurred on Madagascar following human arrival^56–58, with many other species lost on the neighboring Mascarene Islands⁵⁹, but their timing and dynamics remain obscure. Applying process-driven statistical-simulation models to these events — informed and validated by fossil, subfossil and archaeological records — holds great promise for untangling the history of forces contributing to the decline of biological diversity across the oceanic realm, providing novel insights for conserving endemic species in the face of current and future threats.

Spatially-explicit population models (SEPMs)^26,60 of climate-human-moa interactions that integrate extensive palaeoecological information were developed to reconstruct patterns and ecological processes underpinning range collapses, population declines and extinctions of six distinct species of moa and two sub-species. Pattern-oriented modelling techniques²³ and Approximate Bayesian Computation²⁷ were used to independently validate model results. A complete description of the process-driven modelling procedure and paleontological data is provided in the Supporting Methods. The computational code used to build and run models is accessible here: 10.6084/m9.figshare.21946187.

Moa ecological niches

The quality of radiocarbon-dates for 662 georeferenced moa fossils²⁰ were assessed, and unreliable records removed⁶¹. High quality radiocarbon-dates of moa occurrence were intersected spatiotemporally with climatic and environmental parameters that correlate with the nesting behavior and habitat preferences of moa¹⁶. Data were used to (i) characterize the ecological niches of moa using n-dimensional hypervolumes⁶²; and (ii) project habitat suitability annually over the last 12,500 years. This multi-temporal ecological niche modelling approach provided approximations of full potential ecological niches²². To maximize occurrence data in ENMs, and avoid niche truncation⁶³, full potential ecological niches were estimated for phylogenetic groups³⁸. Phylogenetic groups comprised two Dinornis spp (119 fossils), two Pachyornis spp (74 fossils), Megalapteryx didinus (44 fossils), and two subspecies of Euryapteryx curtus (110 fossils). Estimates of potential niche space were exhaustively subsampled (n ~ 1500 per species) to derive realized niches at the species level using simulation and statistical techniques described below. Subsampling was done using ‘boxes’ of different widths (from 0.6 to 1.00, every 0.05), moving throughout a three-dimensional hypervolume to generate a series of smaller niches⁶⁰. The fossil data and the ENM approach are described in detail in the Supporting Methods.

Humans

We modelled moa-human interactions using a process-driven model of the pattern and pace of human growth and migration across New Zealand⁶⁴. Validation tests show that this model can accurately reconstruct inferences of human arrival and expansion from archaeological material. The model simulates population abundances of Polynesians at a 20km resolution, responding to interactions between arrival, demographic and environmental processes. To account for parameter uncertainty in spatiotemporal projections of population abundance, we ran 10,000 different human process-driven models each with a unique combination of parameter settings (based on established upper and lower confidence bounds⁶⁴) using Latin hypercube sampling⁶⁵. This approach has been used elsewhere to reconstruct human-megafauna interactions^22,60. The model and its application are described in detail in the Supporting Methods.

Moa-human-interactions

Spatial population dynamics of moa (including dispersal and source-sink dynamics) were simulated as annual processes operating across the entire distributions of species using SEPMs. Demographic processes (population growth rate and variance, dispersal proportion and distance, Allee effect), environmental attributes (niche breadth and climatic specialization) and human interactions (exploitation of birds and of eggs) were varied across plausible ranges for each species using Latin hypercube sampling of uniform probability distributions. This process, which provided a robust coverage of multi-dimensional parameter space⁶⁶, yielded over 300,000 conceivable SEPMs (parameterizations) of extinction and colonization dynamics. Each model was run for a single replicate²². The structure and parameters for the process-driven model of moa-human interactions are described in detail in the Supporting Methods.

We used POM to evaluate different conceivable SEPMs, by cross matching simulations with inferences of underlying population dynamics from paleo-archives, using Approximate Bayesian Computation²⁷. Specifically, simulations of population and extinction dynamics were validated against a 3 parameter multi-variate target consisting of occurrence at fossil site at the correct time (¹⁴C age ± 2 SD); timing of extinction (corrected for the Signor-Lipps effect⁶⁷); and distance from the final known natural occurrence. The top 0.25% of model parameterizations were retained for each species of moa based on multiple iterations of POM, using Bayes factors to indicate when the posterior parameters had converged²¹. These ‘best’ models were used to generate ensemble average estimates of spatiotemporal abundance. Estimates were weighted by the Euclidean distance of the model from the idealized target²². More detail regarding the POM analysis is in the Supporting Methods.

Statistical analysis

Generalized additive models (GAM) were used to establish determinants of patterns of range collapse for moa. Time of extirpation in each grid cells was regressed against each grid cell’s distance to the latitudinal center of ‘natural’ (pre-human) distribution (weighted by spatial abundance⁶⁸), average temperature in July (austral winter), annual rainfall, ruggedness of the terrain, timing of Polynesian settlement, and a proxy for human habitat suitability (maximum population density). The contribution of each extirpation correlate was analyzed using a beta regression GAM using the ‘mgcv’ package for R version 4.2.2³⁴, which incorporated species identity as a random factor, and accounted for spatial autocorrelation by incorporating geolocation as a Gaussian process in the model.

To determine whether the dynamics of historical extinctions can inform likely pathways of future decline we compared standardized ages of extirpation (timing of local extinction) of moa at locations supporting populations of extant flightless birds with a random background using a priori contrasts⁶⁹. New Zealand Bird Atlas data before 1981³⁶ were used to identify the most restricted distributions of extant flightless birds — those prior to large-scale conservation intervention⁵¹ — removing all records that may have resulted from translocations. This left three populations of kākāpō (Strigops habroptilus), four populations of takahē (Porphyrio hochstetteri), 12 populations of great spotted kiwi (roroa; Apteryx haastii), 18 populations of South Island brown kiwi (tokoeka; Apteryx australis), 64 populations of North Island brown kiwi (Apteryx mantelli) and 116 populations of weka (Gallirallus australis).

Acknowledgments

We acknowledge the contextualisation of our modelling work in mātauranga Māori. D.A.F. J.J.A. and M.V.L. acknowledge funding from the Australian Research Council (DP180102392). Silhouettes of moa were reconstructed by Joseph Nicholson (Twitter: @joe99_joseph) and reproduced with permission. The efforts of several anonymous reviewers substantially improved this manuscript through the peer review process.

Author Contributions:

D.A.F., M.V.L, and J.J.A conceived the project and were awarded subsequent funding. Data was provided by J.M.W, J.R.W. and G.L.W.P. Modelling and statistical analysis was done by S.T. with assistance from S.C.B, S.H. and D.A.F. All authors were involved in the interpretation of results and drafting the manuscript.

Competing Interest Statement:

The authors have no competing interests to disclose.

Classification:

BIOLOGICAL SCIENCES; Ecology; Population Biology.

Whittaker, R. J., Fernández-Palacios, J. M., Matthews, T. J., Borregaard, M. K. & Triantis, K. A. Island biogeography: Taking the long view of nature’s laboratories. Science 357, eaam8326 (2017).
Matthews, T. J. et al. Threatened and extinct island endemic birds of the world: Distribution, threats and functional diversity. Journal of Biogeography https://doi.org/10.1111/jbi.14474 (2022).
Russell, J. C. & Kueffer, C. Island biodiversity in the Anthropocene. Annual Review of Environment and Resources 44, 31-60 (2019).
Blackburn, T. M., Cassey, P., Duncan, R. P., Evans, K. L. & Gaston, K. J. Avian extinction and mammalian introductions on oceanic islands. Science 305, 1955-1958 (2004).
Johnson, C. N. et al. Biodiversity losses and conservation responses in the Anthropocene. Science 356, 270-275 (2017).
Nogué, S. et al. The human dimension of biodiversity changes on islands. Science 372, 488-491 (2021).
Sayol, F. et al. Loss of functional diversity through anthropogenic extinctions of island birds is not offset by biotic invasions. Science Advances 7, p.eabj5790 (2021).
Duncan, R. P., Boyer, A. G. & Blackburn, T. M. Magnitude and variation of prehistoric bird extinctions in the Pacific. Proceedings of the National Academy of Sciences 110, 6436-6441 (2013).
Wilmshurst, J. M., Hunt, T. L., Lipo, C. P. & Anderson, A. High-precision radiocarbon dating shows recent and rapid initial human colonization of East Polynesia. Proceedings of the National Acadamy of Science 108, 1815–1820 (2011).
Ioannidis, A. G. et al. Paths and timings of the peopling of Polynesia inferred from genomic networks. Nature 597, 522-526 (2021).
Steadman, D. W. Extinction and biogeography of tropical Pacific birds., (University of Chicago Press, 2006).
Innes, J., Kelly, D., Overton, J. M. & Gillies, C. Predation and other factors currently limiting New Zealand forest birds. New Zealand Journal of Ecology 34, 86-114 (2010).
Wilmshurst, J. M., Anderson, A. J., Higham, T. F. & Worthy, T. H. Dating the late prehistoric dispersal of Polynesians to New Zealand using the commensal Pacific rat. Proceedings of the National Academy of Sciences 105, 7676-7680 (2008).
Holdaway, S. J. et al. Māori settlement of New Zealand: The Anthropocene as a process. Archaeology in Oceania 54, 17-34 (2019).
Towns, D. R. Eradications as reverse invasions: lessons from Pacific rat (Rattus exulans) removals on New Zealand islands. Biological Invasions 11, 1719-1733 (2009).
Worthy, T. H. & Holdaway, R. N. The lost world of the moa: prehistoric life of New Zealand., (Indiana University Press, 2002).
Perry, G. L. W., Wilmshurst, J. M., McGlone, M. S. & Napier, A. Reconstructing spatial vulnerability to forest loss by fire in pre-historic New Zealand. Global Ecology and Biogeography 21, 1029-1041 (2012).
Wood, J. R., Scofield, R. P., Hamel, J., Lalas, C. & Wilmshurst, J. M. Bone stable isotopes indicate a high trophic position for New Zealand’s extinct South Island adzebill (Aptornis defossor)(Gruiformes: Aptornithidae). New Zealand Journal of Ecology 41, 240-244 (2017).
Holdaway, R. N. & Jacomb, C. Rapid extinction of the moas (Aves: Dinornithiformes): model, test, and implications. Science 287, 2250-2254 (2000).
Perry, G. L., Wheeler, A. B., Wood, J. R. & Wilmshurst, J. M. A high-precision chronology for the rapid extinction of New Zealand moa (Aves, Dinornithiformes). Quaternary Science Reviews 105, 126-135 (2014).
Pilowsky, J. A., Colwell, R. K., Rahbek, C. & Fordham, D. A. Process-explicit models reveal the structure and dynamics of biodiversity. Science Advances, doi: 10.1126/sciadv.abj2271 (2022).
Fordham, D. A. et al. Process‐explicit models reveal pathway to extinction for woolly mammoth using pattern‐oriented validation. Ecology Letters 25, 125-137 (2022).
Grimm, V. & Railsback, S. F. Pattern-oriented modelling: a ‘multi-scope’ for predictive systems ecology. Philosophical Transactions of the Royal Society B 367, 298-310 (2012).
Channell, R. & Lomolino, M. V. Dynamic biogeography and conservation of endangered species. Nature 403, 84-86 (2000).
Britnell, J. A., Zhu, Y., Kerley, G. I. H. & Shultz, S. Ecological marginalization is widespread and increases extinction risk in mammals. Proceedings of the National Academy of Sciences 120, p.e2205315120 (2023).
Fordham, D. A., Haythorne, S., Brown, S. C., Buettel, J. C. & Brook, B. W. poems: R package for simulating species' range dynamics using pattern-oriented validation. Methods in Ecology and Evolution 12, 2364-2371, doi:10.1111/2041-210x.13720 (2021).
Csilléry, K., Blum, M. G. B., Gaggiotti, O. E. & François, O. Approximate Bayesian Computation (ABC) in practice. Trends in Ecology and Evolution 25, 410-418 (2010).
Carmona, C. P., de Bello, F., Mason, N. W. H. & Leps, J. Traits Without Borders: Integrating Functional Diversity Across Scales. Trends in Ecology & Evolution 31, 382-394 (2016).
Kondratyeva, A., Grandcolas, P. & Pavoine, S. Reconciling the concepts and measures of diversity, rarity and originality in ecology and evolution. Biological Reviews 94, 1317-1337 (2019).
Wilmshurst, J. M., McGlone, M. S., Leathwick, J. R. & Newnham, R. M. A pre-deforestation pollen-climate calibration model for New Zealand and quantitative temperature reconstructions for the past 18, 000 years BP. . Journal of Quaternary Science 22, 535-547 (2007).
Dolédec, S., Chessel, D. & Gimaret-Carpentier, C. Niche separation in community analysis: A new method. Ecology 81, 2914-2927 (2000).
Brook, B. W. & Bowman, D. M. J. S. The uncertain blitzkrieg of Pleistocene megafauna. Journal of Biogeography 31, 517-523 (2004).
Simpson, G. L. Modelling palaeoecological time series using generalised additive models. Frontiers in Ecology and Evolution 6, 149 (2018).
Wood, S. N. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. Journal of the Royal Statistical Society B 73, 3-36 (2011).
Brown, J. H., Mehlman, D. W. & Stevens, G. C. Spatial variation in abundance. Ecology 76, 2028-2043 (1995).
Walker, S., Monks, A. & Innes, J. (ed Landcare Research Manaaki Whenua) (Dunedin, New Zealand., 2017).
Steadman, D. W. Prehistoric extinctions of Pacific island birds: Biodiversity meets zooarchaeology. Science 267, 1123-1131 (1995).
Bunce, M. et al. The evolutionary history of the extinct ratite moa and New Zealand Neogene paleogeography. Proceedings of the National Academy of Sciences 106, 20646-20651 (2009).
Wood, J. R., Richardson, S. J., McGlone, M. & Wilmshurst, J. M. The diets of moa (Aves: Dinornithiformes). New Zealand Journal of Ecology 44, 3397 (2020).
Wood, J. R. et al. Coprolite deposits reveal the diet and ecology of the extinct New Zealand megaherbivore moa (Aves, Dinornithiformes). . Quaternary Science Reviews 27, 2593-2602 (2008).
Worthy, T. H. & Scofield, R. P. Twenty-first century advances in knowledge of the biology of moa (Aves: Dinornithiformes): a new morphological analysis and moa diagnoses revised. New Zealand Journal of Zoology 39, 87-153 (2012).
Anderson, A. Mechanics of overkill in the extinction of New Zealand moas. Journal of Archaeological Science 16, 137-151 (1989).
Anderson, A. Prodigious birds: moas and moa hunting in prehistoric New Zealand., (Cambridge University Press, 1989).
Sagarin, R. D., Gaines, S. D. & Gaylord, B. Moving beyond assumptions to understand abundance distributions across the ranges of species. Trends in Ecology and Evolution 21, 524-530 (2006).
Gaston, K. J. Patterns in the geographical ranges of species. Biological Reviews 65, 105-129 (1990).
Channell, R. & Lomolino, M. V. Trajectories to extinction: spatial dynamics of the contraction of geographical ranges. Journal of Biogeography 27, 169-179 (2000).
Lomolino, M. V. & Channell, R. Splendid isolation: patterns of geographic range collapse in endangered mammals. Journal of Mammalogy 76, 335-347 (1995).
Lomolino, M. V. The ecological and geographic dynamics of extinction: Niche modeling and ecological marginalization. Proceedings of the National Academy of Sciences 120, p.e2220467120 (2023).
Freed, L. A. in Genetics and the Extinction of Species: DNA and the Conservation of Biodiversity (eds L. Landweber & A. Dobson) 137-162 (Princeton University Press, 1999).
Doherty, T. S., Glen, A. S., Nimmo, D. G., Ritchie, E. G. & Dickman, C. R. Invasive predators and global biodiversity loss. Proceedings of the National Academy of Sciences 113, 11261-11265 (2016).
Craig, J. et al. Conservation issues in New Zealand. Annual Review Ecology, Evolution and Systematics 31, 61-78 (2000).
Robertson, H. A. et al. (ed Department of Conservation) (New Zealand Government, Auckland, New Zealand, 2021).
Dellicour, S., Rose, R. & Pybus, O. G. Explaining the geographic spread of emerging epidemics: a framework for comparing viral phylogenies and environmental landscape data. BMC Bioinformatics 17, 1-12 (2016).
Holdaway, R. N. in Extinctions in Near Time: Causes, Contexts, and Consequences (eds R.D.E. MacPhee & H-D Sues) 189-238 (Springer, 1999).
Duncan, R. P. & Blackburn, T. M. Extinction and endemism in the New Zealand avifauna. Global Ecology and Biogeography 13, 509-517 (2004).
Hansen, D. M. & Galetti, M. The forgotten megafauna. Science 324, 42-43 (2009).
Anderson, A. et al. New evidence of megafaunal bone damage indicates late colonization of Madagascar. PLoS ONE 13, p.e0204368 (2018).
Michielsen, N. M. et al. The macroevolutionary impact of recent and imminent mammal extinctions on Madagascar. Nature Communications 14, 14 (2023).
Li, H. et al. A multimillennial climatic context for the megafaunal extinctions in Madagascar and Mascarene Islands. Science Advances 6, p.eabb2459 (2020).
Fordham, D. A., Haythorne, S., Brown, S. C., Buettel, J. C. & Brook, B. W. poems: R package for simulating species’ range dynamics using pattern-oriented validation. Methods in Ecology and Evolution 12, 2364-2371 (2021).
Barnosky, A. D. & Lindsey, E. L. Timing of Quaternary megafaunal extinction in South America in relation to human arrival and climate change. Quaternary International 217, 10-29 (2010).
Blonder, B. et al. New approaches for delineating n‐dimensional hypervolumes. Methods in Ecology and Evolution 9, 305-319 (2018).
Faurby, S. & Araújo, M. B. Anthropogenic range contractions bias species climate change forecasts. Nature Climate Change 8, 252-256 (2018).
Tomlinson, S. et al. Reconstructing colonization dynamics to establish how human activities transformed island biodiversity. bioRxiv 2023.02.09.526923, doi: https://doi.org/10.1101/2023.1102.1109.526923 (2023).
McKay, M. D., Beckman, R. J. & Conover, W. J. A comparison of three methods for selecting values of input variables in the analysis of output from a comptuer code. Technometrics 21, 239-245 (1979).
Fordham, D. A., Haythorne, S. & Brook, B. W. Sensitivity Analysis of Range Dynamics Models (SARDM): Quantifying the influence of parameter uncertainty on forecasts of extinction risk from global change. Environmental Modelling & Software 83, 193-197, doi:10.1016/j.envsoft.2016.05.020 (2016).
Signor, P. W., Lipps, J. H., Silver, L. T. & Schultz, P. H. in Geological Implications of Impacts of Large Asteroids and Comets on the Earth (eds L.T. Silver & P.H. Schultz) 291-296 (Geological Society of America, 1982).
Watts, M. J., Fordham, D. A., Akçakaya, H. R., Aiello-Lammens, M. E. & Brook, B. W. Tracking shifting range margins using geographical centroids of metapopulations weighted by population density. Ecological Modelling 269, 61-69 (2013).
Schad, D. J., Vasishth, S., Hohenstein, S. & Kliegl, R. How to capitalize on a priori contrasts in linear (mixed) models: A tutorial. Journal of Memory and Language 110, 104038 (2020).

There is NO Competing Interest.

MoaextinctionsupplementaryFORSUBMISSION.docx
AnimationS1DinornisEuryapteryxcomparisons.mp4
Animation S1 Dinornis Euryapteryx comparisons
AnimationS2DinornisPachyorniscomparisons.mp4
Animation S2 Dinornis Pachyornis comparisons
AnimationS3DinornisMegalapteryx.mp4
Animation S3 Dinornis Megalapteryx comparisons
AnimationS4PachyornisEuryapteryxcomparisons.mp4
Animation S4 Pachyornis Euryapteryx comparisons
AnimationS5PachyornisMegalapteryxcomparisons.mp4
Animation S5 Pachyornis Megakapteryx comparisons
AnimationS6EuryapteryxMegalapteryxcomparisons.mp4
Animation S6 Euryapteryx Megalapteryx comparisons
AnimationS7combinedextirpationsequence.mp4
Animation S7 combined extirpation sequence

Download PDF

Journal Publication

published 24 Jul, 2024

Read the published version in Nature Ecology & Evolution →

Version 1

posted

You are reading this latest preprint version

Ecological dynamics of moa extinctions reveal convergent refugia that today harbor flightless birds.

Status:

Journal Publication

Version 1

Abstract

Figures

Main Text

Results

Ecological and demographic attributes

Extinction dynamics

Conservation implications

Discussion

Materials and Methods

Moa ecological niches

Humans

Moa-human-interactions

Statistical analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1