1. Evolutionary Biology
  2. Neuroscience
Download icon

Quantitative uniqueness of human brain evolution revealed through phylogenetic comparative analysis

  1. Ian F Miller  Is a corresponding author
  2. Robert A Barton
  3. Charles L Nunn
  1. Princeton University, United States
  2. Duke University, United States
  3. University of Durham, United Kingdom
Research Article
Cite this article as: eLife 2019;8:e41250 doi: 10.7554/eLife.41250
5 figures, 6 tables and 4 additional files

Figures

Figure 1 with 1 supplement
OU Model of ECV Evolution in Primates Panel.

A shows the location of the selection regimes identified in an OU model of ECV predicted by body mass. Panel B shows the corresponding optimum regression lines representing the various selection regimes, along with body mass and ECV data. Data are colored by their corresponding selection regimes. All results are from the un-weighted predictor OU model in the bayou analysis using the alternate hominin phylogeny. Only the great ape clade is shown; selection regimes across the entire primate phylogeny are show in Figure 1—figure supplement 1.

https://doi.org/10.7554/eLife.41250.006
Figure 1—figure supplement 1
OU Model of ECV Evolution in Primates Results are shown for the un-weighted predictor OU model of ECV predicted by body mass.

Figure 1 displays the same results, but only for great apes. Panel A shows the location of the selection regimes. Panel B shows the optimum regression lines representing the various selection regimes, along with body mass and ECV data. Data in panel B are colored according to the corresponding regimes shown in panel A.

https://doi.org/10.7554/eLife.41250.007
Figure 2 with 1 supplement
BayesModelS predictions of ECV in hominins.

Panel (A) shows a scatter plot of primate ECV and body mass data. Panel (B) shows the topology of the great ape portion of the hominin phylogeny used in the BayesModelS analyses of hominin ECV. Panel (C) shows the posterior distributions of predicted ECV values generated by BayesModelS for hominin species with body mass used as the predictor variable. Vertical lines indicated observed values.

https://doi.org/10.7554/eLife.41250.008
Figure 2—figure supplement 1
BayesModelS predictions of ECV in hominins.

Panel (A) shows a scatter plot of primate ECV and body mass data. Panel (B) shows the topology of the great ape portion of the alternate hominin phylogeny used in the BayesModelS analyses of hominin ECV. Panel (C) shows the posterior distributions of predicted ECV values generated by BayesModelS for hominin species with body mass used as the predictor variable. Vertical lines indicated observed values. The observed value for H. sapiens s exceeded the mean value predicted by BayesModelS by more than seven standard deviations. All hominin species were strongly supported positive outliers, with >99.9% of predictions falling below the observed values for ECV.

https://doi.org/10.7554/eLife.41250.009
Figure 3 with 1 supplement
Accelerating Evolution of Brain Size Deviation in Hominins.

(A) Brain size deviation was calculated as the difference between the mean BayesModelS prediction (made while excluding all hominin data from analysis and using the hominin phylogeny) and the observed value. Phylogenetic distance was measured as time since the shared ancestor of hominins and Pan at 7.43 mya. (B) Hominin clade in the hominin phylogeny after δ transformation, with δ = 8.36 following the directional acceleration model.

https://doi.org/10.7554/eLife.41250.013
Figure 3—figure supplement 1
Accelerating Evolution of Brain Size Deviation in Hominins (alternate hominin phylogeny).

(A) Brain size deviation was calculated as the difference between the mean BayesModelS prediction (made while excluding all hominin data from analysis and using the alternate hominin phylogeny) and the observed value. Phylogenetic distance was measured as time since the shared ancestor of hominins and Pan at 9.28 mya. (B) Hominin portion of the alternate hominin phylogeny after δ transformation, with δ = 3.745 following the directional acceleration model. Among the PGLS models fit to this data, the directional acceleration model (AICc = −23.88) was favored, as it outperformed the the Brownian (AICc = −15.71), directional (AICc = −22.12), and accelerating (AICc = −22.38) evolution models. This model gave evidence for both evolution towards larger brain volume relative to body mass (slope = 0.06) and for accelerating evolution (δ = 3.745).

https://doi.org/10.7554/eLife.41250.014
Figure 4 with 1 supplement
Human Outlier Status for Brain Traits Predicted distributions of trait values generated by BayesModelS are show as histograms.

Vertical bars represent the observed values.

https://doi.org/10.7554/eLife.41250.015
Figure 4—figure supplement 1
Human outlier status for ECV In the BayesModelS analysis of ECV with no predictor variable, humans were not detected as outliers.

Results for other species are given in Source data 1. Because BayesModelS requires a predictor variable, we assigned each species a random number for the predictor trait. This resulted in the predictor variable not being included in the PGLS model in ~98% of post burn-in MCMC samples. We discarded the remaining samples that included the predictor in the PGLS model before generating predictions.

https://doi.org/10.7554/eLife.41250.016
OU Models of Brain Structure Evolution in Primates.

(A and B) correspond to the OU weighted predictor model of neocortex volume predicted by the rest-of-brain. (C and D) correspond to the OU unweighted predictor model of cerebellum volume predicted by the rest-of-brain. (E and F) correspond to the OU weighted predictor model of the rest-of-brain volume predicted by body mass. (A, C) and (E) show the location of selection regimes on the primate phylogeny. (B, D) and (F) show the optimum regression lines associated with the selection regimes. Points show primate trait and predictor data; colors correspond to the selection regimes. Colors in (A, C) and (E) match those in (B, D) and (F).

https://doi.org/10.7554/eLife.41250.017

Tables

Table 1
Hominin ECV and body mass data details.

All values are from Robson and Wood (2008).

https://doi.org/10.7554/eLife.41250.003
SpeciesECV (mL)Sample sizeFemale body mass (kg)Sample size
Australopithecus africanus464.008307
Homo erectus969.0040574
Homo habilis609.006322
Homo rudolfensis726.003512
Homo sapiens neanderthalensis1426.0023657
Homo sapiens1478.00665736
Paranthropus boisei481.0010341
Paranthropus robustus563.002322
Australopithecus afarensis458.006304
Table 2
Human brain data.
https://doi.org/10.7554/eLife.41250.004
Brain traitValueSourceNotesDataset
ECV1478.00 mL(Robson and Wood, 2008)Composite of values from 66 fossil specimens from locations across Eurasia and africa1
Brain volume1267.65 mL(Barton and Harvey, 2000)Average of measurements of modern human brains2
Brain volume1251.85 mL(Stephan et al., 1981)Measurement of modern human brain3
Table 3
Priors for bayou MCMC analyses.
https://doi.org/10.7554/eLife.41250.005
Model parameterPrior distribution
αHalf-cauchy with scale factor 1. Fixed at 0 in Brownian model.
σ2Half-cauchy with scale factor 0.1
βNormal distribution with standard deviation = 0.5, mean = slope of linear model of trait and predictor data
θNormal distribution with standard deviation = 1, mean = intercept of linear model of trait and predictor data
Number of shifts per branchFixed at one
Branch-wise shift probabilityUniform
Number of shiftsConditional Poisson distribution* with mean = 0.1*number of edges on phylogeny and maximum = number of edges on phylogeny. Fixed at 0 in Brownian model.
Location of shift along branchUniform
  1. *Calculated using ‘cdpois’ option in bayou.

Table 4
Predicted Hominin ECV values from BayesModelS analysis using the hominin phylogeny.
https://doi.org/10.7554/eLife.41250.010
True value (ml)Corrected prediction (ml)Difference (ml)% difference
Australopithecus africanus464.00294.73169.2757.43
Homo erectus969.00438.24530.76121.11
Homo habilis609.00306.83302.1798.48
Homo rudolfensis726.00409.63316.3777.23
Homo sapiens1478.00437.761040.24237.63
Homo sapiens neanderthalensis1426.00474.46951.54200.55
Paranthropus boisei481.00319.00162.0050.78
Paranthropus robustus563.00307.60255.4083.03
Australopithecus afarensis458.00288.52169.4858.74
Table 5
Predicted Hominin ECV values from BayesModelS analysis using the alternate hominin phylogeny.
https://doi.org/10.7554/eLife.41250.011
True value (ml)Corrected prediction (ml)Difference (ml)% difference
Australopithecus africanus464.00288.18175.8261.00
Homo erectus969.00431.04537.96124.81
Homo habilis609.00300.16308.84102.89
Homo rudolfensis726.00401.94324.0680.62
Homo sapiens1478.00431.201046.80242.76
Homo sapiens neanderthalensis1426.00468.41957.59204.44
Paranthropus boisei481.00311.41169.5954.46
Paranthropus robustus563.00299.74263.2687.83
Australopithecus afarensis458.00281.59176.4162.65
Table 6
Summary of evidence for exceptional brain evolution among non-human primates.
https://doi.org/10.7554/eLife.41250.012
Species/CladeExceptional traitEvidence
AlouattaReduced ECV relative to body massShift in OU model
Aotidae and CallitrichidaeReduced ECV relative to body massShift in OU model
Cacajao calvusIncreased ECV relative to body massOutlier Detection
CebinaeIncreased ECV relative to body massShift in OU model
Cebus albifronsIncreased cerebellum relative to body massOutlier detection
Chiropotes satanasReduced ECV relative to body massOutlier Detection
ColobinaeReduced ECV relative to body massShift in OU model
Daubentonia madagascariensisIncreased ECV relative to body massShift in OU model
Gorilla beringei*Reduced ECV relative to body massOutlier Detection
Gorilla gorilla*Reduced neocortex relative to body massOutlier Detection
LemuridaeIncreased ECV relative to body massShift in OU model
Loris tardigradusReduced medulla relative to the rest of brainOutlier Detection
Microcebus murinusReduced medulla relative to the rest of brainOutlier Detection
Nasalis larvatusReduced neocortex relative to the rest of the brainShift in OU model
Otolemur crassicaudatusReduced neocortex, cerebellum relative to body massOutlier Detection
Pan troglodytes schweinfurthiiIncreased ECV relative to body massOutlier Detection
Pan troglodytes troglodytesReduced ECV relative to body massOutlier Detection
  1. *The dataset for this analysis did not contain any other gorilla species.

Data availability

All data used in our analyses are provided as supplementary material.

Additional files

Source code 1

Representative Code.

Representative R code files for the bayou analyses ('representative bayou code.R'), BayesModelS analyses ('representative BayesModels code.R’), and pgls model fitting (‘pgls models.R’), are contained in the this file, along with the BayesModelS code (‘mult.spec.BayesModelS_v24.R’) and other necessary data files.

https://doi.org/10.7554/eLife.41250.018
Source data 1

Bayou and BayesModelS Results Details.

Bayou Results details: Diagnostic plots giving details of chain convergence are provided in the 'bayou results summary.html' file along with detailed information on all OU and Brownain motion models for each trait and predictor pair. BayesModelS Results Details: Details of the BayesModelS results and diagnostic parameters of MCMC chains are given in the 'BayesModelS.results.csv' and 'BayesModelS.results.hominins.removed.csv’ files.

https://doi.org/10.7554/eLife.41250.019
Source data 2

All data and trees used in our analyses.

Contains the following files: 1. data set 1.csv 2. data set 2.csv 3. data set 3.csv 4. consensus.tree.txt 5. tree.block.txt 6. grafted.tree.txt

https://doi.org/10.7554/eLife.41250.020
Transparent reporting form
https://doi.org/10.7554/eLife.41250.021

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)