A critical take on “Persistent effects of pre-Columbian plant domestication on Amazonian forest composition”

Summary: Levis et al. (SCIENCE03 MAR 2017 : 925-931) challenge the idea that Amazon forests were largely untouched by humans prior to 1492. Specifically, they show that forest plots near archeological sites are enriched with useful tree species relative to plots located farther away from these sites. This is an intriguing finding, but the idea that prehistoric gardening has broadly shaped Amazon forests is problematic: (1) Amazon forest inventory plots are spatially biased toward human settlements (both ancient and modern) so broader extrapolation is not warranted; (2) Most useful tree species have short lifetimes and their local enrichment should not persist after several hundred years; (3) Amazon people still care for these useful tree species, making it it difficult (if not impossible) to disentangle ancient and modern impacts. The simplest interpretation is that Amazon research plots are often established in forests used by rubber tappers, small-scale farmers, or other recent peoples.

A forest worker harvests patauá fruit from Oenocarpus bataua near one of the ATDN plots north of Manaus, Brazil. Photo by C. Dick, circa 1995.

Popular accounts of this paper have featured in Science, New York Times, The Atlantic, Nature, and other major outlets. The popular message is that pre-Columbian people, through their cultivation of useful tree species, left an enduring imprint on patterns of Amazon rainforest tree diversity.

Tree data was obtained from 1104 Amazonian (i.e. Amazon basin + Guiana Shield region) forest plots coordinated by the Amazon Tree Diversity Network (ATDN). ATDN contains 4962 tree species at latest tally (1), but half of the trees are concentrated in 227 “hyperdominant” species, which play an inordinate role in ecosystem function (2). The authors found that 20 domesticated tree species are among the hyperdominants and, more importantly, that they are most abundant near Pre-Columbian archeological sites. This lead the authors to conclude that “Domestication shapes Amazon diversity”. There are several potential problems.

ATDN forest plots are a nonrandom sample of Amazon forests, often located near human settlements  Forest inventory plots sample less than 0.0005% of Amazon rain forest (4) and they are not randomly distributed or necessarily representative of Amazon landscape and/or habitat diversity. In a January 2017 paper in PNAS, McMichael et al. (4) showed that the ATDN plots are spatially correlated with archeological sites, which are likely to be associated with modern human settlements. This makes intuitive sense, because ancient and recent peoples (including botanists) use similar criteria in choosing locations to settle (proximity to rivers, unflooded areas, rich soils, etc.). There are few inventory plots in the sparsely settled interfluvial regions of the Amazon basin, which show little evidence of pre-Columbian influence (5).

There are few truly domesticated Amazon trees and none are common The authors use the word “domestication” as an umbrella term that includes a spectrum of human uses. Strictly speaking, domestication implies genetic modification through breeding. Two looser categories used are: incipiently domesticated (cultivation of wild trees, possibly changing allele frequencies locally) and semi-domesticated (some evidence of anthropogenic phenotypic variation in parts of the range) (3). It is difficult to disentangle natural and anthropogenic phenotypic variation in these latter categories. In most cases, these are simply wild species that are cared for (planted or not cut down) by people.

Those who read the paper understood that their use of the word “domesticated” was shorthand, but it lead to problems in scientific communication as the news outlets conveyed only the standard meaning. Here I will use “domesticated” in its strict meaning, and “cultivated” for the secondary categories.

There are only 8 truly domesticated trees out of 4962 species in the ATDN (Table 1). The most common of these (Genipa americana) ranked 1010th out of 4962 species in abundance (mean abundance rank of the 8 species = 2539). These species clearly contribute little in terms of biomass or species diversity.

Species name Family Use Abund. Rank
Annona mucosa Annonaceae Fruit 3271
Annona muricata Annonaceae Fruit 3167
Annona squamosa Annonaceae Fruit 2385
Ilex guayusa Aquifoliaceae Stimulant 2797
Bactris gasipaes Arecaceae Fruit 1530
Crescentia cujete Bignoniaceae Gourd 4632
Bixa urucurana Bixaceae Colorant 1518
Genipa americana Rubiaceae Fruit/Colorant 1010
Table 1: The 8 domesticated ATDN tree species, human uses, and abundance rank (from Table S2 in Levis et al. 2017)

The 77 cultivated species do include 20 hyperdominants (Table 2) including the açaí palm, which is the most abundant tree species across the ATDN. The top 10 most common cultivated species include six highly edible palms, the rubber tree Hevea brasiliensis, and three chocolate-tree relatives (Theobroma cacao, T. subincanum, T. speciosum). All 20 species have been widely planted and used by post-Colonial Amazon peoples over the past two centuries (see photo).

Species Family Useful trait Rank
Euterpe precatoria Arecaceae Oily fruit 1
Euterpe oleracea Arecaceae Oily fruit 6
Oenocarpus bataua Arecaceae Oily fruit 7
Astrocaryum murumuru Arecaceae Oily fruit 10
Hevea brasiliensis Euphorbiaceae Nut/Latex 14
Mauritia flexuosa Arecaceae Oily fruit 23
Theobroma cacao* Malvaceae Stimulant/Fruit 29
Theobroma subincanum Malvaceae Fruit 31
Oenocarpus bacaba Arecaceae Oily fruit 33
Theobroma speciosum Malvaceae Fruit 35
Attalea maripa Arecaceae Oily fruit 59
Attalea phalerata Arecaceae Thatch/Fruit 122
Pouteria caimito* Sapotaceae Fruit 124
Astrocaryum aculeatum Arecaceae Fruit 139
Caryocar glabrum Caryocaraceae Edible nut 145
Spondias mombin* Anacardiaceae Fruit 154
Garcinia macrophylla Clusiaceae Fruit 167
Inga ynga (edulis)* Fabaceae Fruit 181
Pourouma cecropiifolia* Urticaceae Fruit 182
Bertholletia excelsa Lecythidaceae Edible nut 188
Table 2: 20 cultivated hyperdominant Amazon tree species. * Semi-domesticated according to Clements (1999). Adapted from Table S2 in Levis et al. 2017.

The ecological abundance of cultivated trees is ephemeral. It is possible that some long-lived trees (e.g. Brazil nut, Bertholletia excelsa) can linger at artificially high densities for centuries (4, 6). Most of the cultivated fruit trees, however, are early succession species (e.g. Inga, Pouteria, Pourouma) and/or poor competitors. For example, the domesticated palm Bactris gasipaes (pejibaye, pupunha) stops producing fruit as forest overshadows it (3). The most abundant species, the açaí palm Euterpe precatoria, thrives in disturbed, fragmented habitats (7). It is unlikely that the artificial high abundance of fruit trees would persist for 500 years in these highly dynamic forests.

Concluding thoughts: ATDN has produced one exciting and important paper after another, and the network shaped my understanding of Amazon forest structure and dynamics. However, the over-reaching message of this paper struck me as counterproductive. I don’t know of any ecologists who think of Amazon forests as historically pristine. In that sense the authors were attacking a straw man. The idea that one could look at abundant short-lived trees and see in them a forest structure frozen in time since 1492 — well that seems more romantic and inaccurate than the pristine forest myth. The Amazon is not pristine, nor is it primarily shaped by prehistoric gardening.

Note — for my take on pre-Columbian human impacts on some forests of the eastern U.S., see this post. Thanks to members of the Dick lab for discussion.


  1. H. ter Steege et al., The discovery of the Amazonian tree flora with an updated checklist of all known tree taxa. Scientific Reports 6, (2016).
  2. H. ter Steege et al., Hyperdominance in the Amazonian Tree Flora. Science 342, 325-+ (2013).
  3. C. R. Clement, 1492 and the loss of Amazonian crop genetic resources. I. The relation between domestication and human population decline. Economic Botany 53, 188-202 (1999).
  4. C. N. H. McMichael, F. Matthews-Bird, W. Farfan-Rios, K. J. Feeley, Ancient human disturbances may be skewing our understanding of Amazonian forests. Proceedings of the National Academy of Sciences of the United States of America 114, 522-527 (2017).
  5. M. B. Bush et al., Anthropogenic influence on Amazonian forests in pre-history: An ecological perspective. Journal of Biogeography 42, 2277-2288 (2015).
  6. G. H. Shepard, H. Ramirez, “Made in Brazil”: Human Dispersal of the Brazil Nut (Bertholletia excelsa, Lecythidaceae) in Ancient Amazonia. Economic Botany 65, 44-65 (2011).
  7. Avalos G, Otárola MF, Engeln JT. Successional stage, fragmentation and exposure to extraction influence the population structure of Euterpe precatoria (Arecaeae)  Rev Biol Trop.61(3):1415-24 (2013).

One thought to “A critical take on “Persistent effects of pre-Columbian plant domestication on Amazonian forest composition””

Leave a Reply

Your email address will not be published. Required fields are marked *