Invisible Roads: How Archaeologists Reconstruct Ancient Networks Across Mountains and Continents

Twenty-five feathers from a pre-Inca tomb at Pachacamac have allowed archaeologists to follow Amazonian parrots across the Andes using ancient DNA, stable isotopes and spatial modelling. The case captures a larger transformation in archaeology: objects, animals, plants and people can now be traced through networks whose traders, transactions and roads have otherwise disappeared.

In 2005, archaeologists excavating in front of the Painted Temple at Pachacamac uncovered an intact Ychsma tomb containing thirty-four funerary bundles. Five of the largest were adorned with clusters of brilliantly coloured feathers attached behind the artificial heads of the bundles. The ornaments belonged to an elite mortuary setting on Peru's desert coast, but many of the birds from which they had been taken could not have lived there.

That much was suspected from their colour. The difficult question was where, precisely, they had come from, and what their presence at Pachacamac implied about the societies capable of obtaining them.

A study published in Nature Communications in March 2026 approached the problem with an unusually broad set of methods. Twenty-five archaeological feathers from eight locations within the burial assemblage were subjected to ancient-DNA analysis. Four tropical parrot species were identified: scarlet macaw (Ara macao), red-and-green macaw (Ara chloropterus), blue-and-yellow macaw (Ara ararauna) and mealy Amazon (Amazona farinosa). A white feather proved to belong instead to Sabine's gull, a migratory seabird available on the Peruvian coast. The parrots, by contrast, belong naturally to lowland tropical environments east of the Andes. Radiocarbon measurements from feathers, considered alongside dates from textiles, human remains and associated ceramics, place the tomb broadly between about AD 1000 and 1400.

The finding does more than document an exotic possession. It offers a useful demonstration of how archaeologists now reconstruct long-distance circulation when there is no account book, merchant's archive or itinerary to read.

A parrot on the wrong side of the Andes

The DNA analysis did not simply label the feathers "macaw". Targeted enrichment recovered mitochondrial sequences from the minute feather fragments, enabling comparison with modern and ancient reference material. The sequences placed the archaeological specimens among South American populations associated with regions east of the Andes. Three scarlet-macaw samples also showed considerably greater mitochondrial diversity than is known from some archaeologically documented captive-breeding populations in the American Southwest. On that evidence, the authors argue that the Pachacamac birds were drawn from genetically diverse wild populations rather than descended from a small breeding colony maintained on the Peruvian coast.

There is a qualification worth making. Three scarlet macaws are not a population history. Their diversity is good evidence against one particular model, a small and genetically restricted local breeding stock, but it cannot describe the entire system by which parrots were captured and supplied.

The stable isotopes introduce another part of the birds' biographies. Carbon and nitrogen isotope ratios measured in five feathers differ from those of modern wild Amazonian parrots. They are compatible with a diet rich in C4 plants, most plausibly including maize, while elevated nitrogen values may reflect food grown under conditions influenced by marine-derived fertilizer such as seabird guano. Because a feather records diet during the period in which it grew, the isotope measurements imply that at least these birds were already eating very differently from wild rainforest parrots while producing the sampled plumage.

Put the two results together and a persuasive sequence emerges. The birds probably originated in wild populations east of the Andes; they were transported alive; and some were subsequently maintained in a coastal environment long enough for new feathers to form under a different feeding regime. No cages, parrot eggshells or breeding assemblage have been identified at Pachacamac. The safest conclusion is therefore not that DNA has "found a trade route", but that several independent observations make live transport from tropical South America to the coast considerably more likely than local breeding. Comparable evidence from the Atacama Desert shows that Amazonian parrots were being kept far outside their natural range elsewhere in pre-Columbian western South America.

The 2026 study then goes one step further. Species-distribution models were used to estimate suitable habitats for the parrots around AD 1000, while landscape-resistance modelling assessed movement between Ychsma territory and selected contemporary centres. Elevation, rivers and possible maritime travel were among the variables. The models generated plausible northern and central corridors linking the Pacific coast with areas from which tropical birds might have been obtained.

Here the evidence becomes less straightforward than a route map may suggest. GIS can calculate routes that are more or less costly under stated assumptions. It cannot show footprints. A modelled corridor is not equivalent to an excavated road, still less evidence that a particular macaw travelled along it. The birds may have passed through several communities and several kinds of exchange before reaching Pachacamac. Direct procurement, barter, elite gift-giving, pilgrimage, tribute and redistribution can leave very similar archaeological distributions.

That distinction, between provenance and network, is fundamental.

Finding a source is only the beginning

Archaeologists have been sourcing stone for decades, and obsidian provides perhaps the clearest demonstration of the method. Volcanic glasses formed in different geological settings often possess distinctive combinations of trace elements. Techniques including X-ray fluorescence and neutron activation analysis can therefore compare an archaeological flake with samples from known geological outcrops.

At Olorgesailie in Kenya, geochemical sourcing contributed to the identification of non-local obsidian in Middle Stone Age contexts more than 300,000 years old. The material had travelled over distances well beyond the immediately available stone resources. The association of such transport with other changes in technology and material use became part of a wider argument for expanding social connections among early Homo sapiens.

Yet an obsidian source is not a social explanation. If a tool found at settlement A came from volcanic source B, the geochemistry establishes a geological relationship between object and source. It does not say whether the toolmaker visited B, obtained a prepared core from neighbours, acquired it through marriage partners, or received a piece of stone that had already passed through five communities.

Distribution adds another layer. If obsidian from one source occurs at a chain of well-dated settlements, with quantities or artefact sizes changing with distance, archaeologists can begin testing models of direct access, redistribution and "down-the-line" exchange. But those models remain interpretations of patterns rather than recovered transactions.

Amber creates a similar problem with very different chemistry. Fourier-transform infrared spectroscopy, or FTIR, can distinguish varieties of fossil resin by their molecular absorption spectra. A recent study of twenty-one supposed amber objects from north-eastern Iberia identified eighteen beads as Baltic succinite. Some may date as early as the later fourth millennium BC, despite the existence of amber sources much closer to the peninsula. The same analysis also exposed a useful archaeological warning: one unusual object thought to be amber was probably gum, while two others proved to be mineral pigments.

The result gives us a source region, but again not a caravan itinerary from the Baltic to Iberia. When the distribution of finds is added, the authors argue that contacts through southern France and across the Pyrenees provide a plausible framework for the arrival of Baltic material in north-eastern Iberia. Even here they are careful about chronology, because collective burial sites can remain in use for long periods and an individual bead may be difficult to associate with a narrow phase.

Lapis lazuli presents an even older version of the same archaeological temptation. The brilliant blue stone found at Bronze Age sites across western and southern Asia has often been linked rather automatically with the famous deposits of Badakhshan in Afghanistan. Modern mineralogical and elemental studies make the picture more demanding. A 2025 analysis of lapis-lazuli working debris from Shahr-i Sokhta in eastern Iran combined optical microscopy, cathodoluminescence, scanning electron microscopy and ion-beam techniques with geological reference material. Provenance becomes a process of excluding incompatible sources and evaluating matches rather than attaching a familiar place-name to every blue stone.

This is why reference collections matter so much. An archaeological object can only be matched against geological, biological or isotopic variation that has actually been characterized. An unknown source does not conveniently disappear because it is absent from the laboratory database.

Networks are made of more than objects

The most consequential development has been the extension of provenance research to living things, including human beings.

Strontium isotope analysis is a good example. Different geological terrains contribute different strontium isotope ratios to soils, plants, animals and ultimately human food. Tooth enamel forms at particular stages of childhood and does not subsequently remodel in the way bone does. Its strontium composition can therefore be compared with the biologically available strontium around the place where an individual was buried.

The result is not a birthplace written in chemistry. Several regions may share similar isotope values, food can be imported, and local baselines need careful construction. What the method can often establish is that an individual is compatible or incompatible with the local isotopic environment.

In Bavaria's Lech River valley, Corina Knipper and colleagues combined strontium and oxygen isotopes with ancient DNA from Final Neolithic and Early Bronze Age cemeteries. Many adult women had non-local isotope signatures, while most men appeared local. Genetic evidence added information about biological relationships. The recurring pattern was interpreted as consistent with female exogamy and predominantly patrilocal residence over many generations.

That kind of evidence changes what a prehistoric exchange network can mean. Objects were not necessarily passed between stationary communities by specialist traders. People themselves moved between households and regions, taking knowledge, obligations, identities and perhaps material connections with them. Marriage, fosterage and kinship can create durable routes along which things subsequently circulate.

Ancient DNA adds still another scale. It can reveal biological kinship within a cemetery, population affinities between regions and episodes of substantial demographic movement. But genes do not identify ethnicity, language or occupation, and a genetic connection between populations does not prove that two particular settlements traded. The strength comes from comparison: graves, artefacts, isotopes, settlement archaeology and genomes asking related questions independently.

Plants can travel through networks too. Archaeobotanical work in Central Eurasia has documented wheat, barley, millet and other crops in Bronze Age pastoralist settings, helping researchers reconstruct the movement of agricultural packages across mountain systems and ecological zones. A charred grain fixes a crop at a place and approximate time; regional sequences of such finds reveal pathways of dispersal that may accompany pastoral mobility without requiring a modern-style merchant economy.

Shells, ivory and animal remains can be approached in related ways. Taxonomy may narrow their ecological origin; isotopes can refine geographical possibilities; proteins or DNA can identify species; elemental or molecular analysis can distinguish source areas. Each technique answers a somewhat different question. The error comes when those questions are collapsed into a single confident arrow on a map.

The tin problem

Metals make the limitations particularly visible because they can be smelted, alloyed, recycled and mixed.

Tin is crucial here. Bronze Age societies required tin to make tin bronze, yet major concentrations of tin ore are unevenly distributed. Establishing where the metal in Mediterranean bronze came from has consequently been one of archaeometallurgy's persistent problems.

In 2022, Wayne Powell and colleagues analysed tin ingots from the Late Bronze Age Uluburun shipwreck using tin and lead isotopes together with trace-element chemistry. They argued that part of the cargo could be linked to Central Asian sources, including the Mushiston region of Tajikistan, while another component was associated with Anatolia. If correct, the Central Asian attribution would imply commodity connections extending across a large part of Eurasia before the ship sank off the coast of Anatolia in the late fourteenth century BC.

The interpretation did not go uncontested. Daniel Berger, Kai Kaniuth, Gerhard Brügmann and Ernst Pernicka subsequently argued that the isotopic evidence cannot uniquely identify Mushiston. Tin-isotope ranges within ore deposits can be broad and, more importantly, different deposits overlap. Combining tin isotopes with lead isotopes and trace elements can reduce the possible source regions, but it does not always produce a unique answer. Their reassessment proposed other possibilities, including south-west England for some of the material, while leaving parts of the cargo unresolved.

The disagreement is instructive precisely because both sides use sophisticated analytical evidence. Scientific provenance is not a machine into which an artefact is inserted and from which a mine name emerges. Geological variability, incomplete source sampling, metallurgical contamination and recycling all intervene. The appropriate result may be a set of source probabilities rather than a point on a map.

This also explains why the most convincing archaeological networks rarely rest on a single spectacular measurement.

From provenance to invisible roads

Seen in that light, the Pachacamac feathers are unusually informative because different kinds of evidence constrain different stages of the same biography.

Archaeological context establishes where the objects ended their useful lives and how they were socially valued. Radiocarbon dating places them in time. DNA identifies the birds and relates them to wider biological populations. Genetic diversity bears on whether local captive breeding is plausible. Stable isotopes record part of the birds' diet while particular feathers were forming. Ecology places the species on the opposite side of the Andes. GIS then asks where movement between those ecological and cultural worlds might have been feasible.

None of those measurements alone demonstrates a Ychsma-controlled commercial system. Together they make certain histories much harder to sustain than others.

The distinction matters because the word trade easily imports assumptions for which archaeology has little evidence. Long-distance goods need not have travelled through markets or professional merchants. A macaw could be captured by one community, exchanged to another, kept alive for months, moved again as part of a political obligation, and finally have its feathers removed by specialists who never met the people who caught it. The archaeological endpoint preserves the result of a chain, not necessarily the chain itself.

Something similar applies to obsidian, Baltic amber, lapis lazuli or tin. A raw material can travel a thousand kilometres without any individual making a thousand-kilometre journey. Indeed, extensive networks often work precisely because people do not traverse them from end to end.

Nor should "global" be taken literally in the modern economic sense. The prehistoric and early historic systems reconstructed by archaeology were overlapping networks of very different scale and organization. Some joined neighbouring valleys; some crossed mountain ranges and seas; a few linked chains of communities over continental distances. They were not a single integrated world economy.

What modern archaeology can increasingly recover is the architecture of those connections.

The strongest reconstructions begin with secure context and chronology, establish provenance through independent scientific methods, and then examine how distributions relate to settlements, geography and human mobility. Network analysis and GIS can test whether proposed connections make spatial sense. Ancient DNA and isotope analysis can establish that animals and people moved as well as objects. Materials science can identify geological sources inaccessible to typology alone. None abolishes uncertainty; rather, their value lies in making different uncertainties explicit.

At Pachacamac, we can now say with considerable confidence that feathers from several Amazonian parrot species reached an elite Ychsma mortuary context on the Pacific coast; that the birds were probably taken from genetically diverse wild populations; and that at least some lived long enough outside their rainforest environment to grow feathers while eating a markedly different diet. The modelling identifies reasonable trans-Andean corridors through which such movement could have occurred.

We cannot yet identify the people who caught those parrots, count the number of exchanges involved, determine what was given in return, or assign an individual feather to one of the modelled routes. Nor can we be certain that feathers, live birds and other Amazonian commodities all passed through the same social machinery.

That boundary between what can be demonstrated and what can only be reconstructed is not a weakness of the exercise. It is where the archaeology becomes most interesting. The question is no longer whether distant societies were connected. The harder task is to work out what kind of connection an exotic object actually records: direct travel, repeated exchange, migration, political redistribution, kinship, religious pilgrimage, or several of these at once.

The invisible roads are becoming easier to locate. Their social rules remain much harder to recover.

Selected Sources and Further Reading

1. Olah, G., Bover, P., Llamas, B., Heiniger, H., Segura Llanos, R. et al. 2026. “Ancient DNA and spatial modeling reveal a pre-Inca trans-Andean parrot trade.” Nature Communications 17: 2117. https://doi.org/10.1038/s41467-026-69167-9
2. Capriles, J. M., Santoro, C. M., George, R. J. et al. 2021. “Pre-Columbian transregional captive rearing of Amazonian parrots in the Atacama Desert.” Proceedings of the National Academy of Sciences 118(15): e2020020118. https://doi.org/10.1073/pnas.2020020118
3. Brooks, A. S., Yellen, J. E., Potts, R. et al. 2018. “Long-distance stone transport and pigment use in the earliest Middle Stone Age.” Science 360(6384): 90–94. https://doi.org/10.1126/science.aao2646
4. Frahm, E. 2025. “Archaeological obsidian sourcing: Looking from the first 60 years to the next.” Journal of Archaeological Science 177: 106200. https://doi.org/10.1016/j.jas.2025.106200
5. Murillo-Barroso, M., Navero Rosales, M., González-Marcén, P. and Martinón-Torres, M. 2025. “Amber Networks in Prehistory: North-Eastern Iberia as a Case Study.” European Journal of Archaeology 28(4): 405–422. https://doi.org/10.1017/eaa.2025.12
6. Guidorzi, L., Magalini, M., Re, A. et al. 2025. “Provenance attribution of lapis lazuli rocks processed at the Bronze age archaeological site of Shahr-i Sokhta (Iran).” Journal of Archaeological Science: Reports 67: 105411. https://doi.org/10.1016/j.jasrep.2025.105411
7. Powell, W. et al. 2022. “Tin from Uluburun shipwreck shows small-scale commodity exchange fueled continental tin supply across Late Bronze Age Eurasia.” Science Advances 8(48): eabq3766. https://doi.org/10.1126/sciadv.abq3766
8. Berger, D., Kaniuth, K., Brügmann, G. and Pernicka, E. 2023. “Why Central Asia's Mushiston is not a source for the Late Bronze Age tin ingots from the Uluburun shipwreck.” Frontiers in Earth Science 11. https://doi.org/10.3389/feart.2023.1211478
9. Knipper, C., Mittnik, A., Massy, K. et al. 2017. “Female exogamy and gene pool diversification at the transition from the Final Neolithic to the Early Bronze Age in central Europe.” Proceedings of the National Academy of Sciences 114(38): 10083–10088. https://doi.org/10.1073/pnas.1706355114
10. Spengler III, R. N., Frachetti, M. D., Doumani, P. N. et al. 2014. “Early agriculture and crop transmission among Bronze Age mobile pastoralists of Central Eurasia.” Proceedings of the Royal Society B 281: 20133382. https://doi.org/10.1098/rspb.2013.3382

Suggested keywords

Pachacamac; Ychsma; archaeological provenance; exchange networks; ancient DNA; stable isotope analysis; GIS; obsidian sourcing; archaeometry; trans-Andean trade; human mobility; Bronze Age exchange

Editorial summary

The Pachacamac research is important now because it demonstrates how genetic, isotopic, archaeological and spatial evidence can be combined to reconstruct ancient connectivity while making unusually clear where scientific provenance ends and historical inference begins.