Primate sympatry shapes the evolution of their brain architecture

The main hypotheses on the evolution of animal cognition emphasise the role of conspecifics in affecting the socio-ecological environment shaping cognition. Yet, space is often simultaneously occupied by multiple species from the same ecological guild. These sympatric species can compete for food, which may thereby stimulate or hamper cognition. Considering brain size as a proxy for cognition, we tested whether species sympatry impacted the evolution of cognition in frugivorous primates. We first retraced the evolutionary history of sympatry between frugivorous primate lineages. We then fitted phylogenetic models of the evolution of the size of several brain regions in frugivorous primates, considering or not species sympatry. We found that the evolution of the whole brain or brain regions used in immediate information processing was best fitted with models not considering sympatry. By contrast, models considering species sympatry best predicted the evolution of brain regions related to long-term memory of interactions with the socio-ecological environment, with a decrease in their size the higher the sympatry. We speculate that species sympatry, by generating intense food depletion, might lead to an over-complexification of resource spatiotemporality that counteracts the benefits of high cognitive abilities and/or might drive niche partitioning and specialisation, thereby inducing lower brain region sizes. In addition, we reported that primate species in sympatry diversify more slowly. This comparative study suggests that species sympatry significantly contributes to shaping primate evolution.


INTRODUCTION
for small samples (AICc). The model weight depicts how well the model fits the observed data compared 2 4 5 with the other tested models (i.e. a model set).

4 6
To account for uncertainty (in the data, Supplementary Figure S3, and fit), we fitted and compared 2 4 7 these models (within each model set) 10 times for 10 different ancestral reconstructions of primate 2 4 8 biogeography and diet considering all thresholds. Thus, for each brain region, we fitted all the models 10 2 4 9 (uncertainty on diet/biogeography ancestral reconstructions x 10 (uncertainty in brain/diet data) x 2 2 5 0 (biogeographic overlap threshold) x 2 (frugivory threshold) x 2 (folivory threshold) = 800 times. We 2 5 1 nonetheless stopped computations when the calculation of the likelihood was excessively long (> 1 week) 2 5 2 and the final sample size was 730 model sets. Diversity-dependent models of trait evolution inform whether species sympatry has impacted brain size 2 5 5 evolution by increasing or decreasing the tempo of trait evolution. Nonetheless, they do not tell the 2 5 6 directionality of the effect (i.e. do brain sizes increase or decrease with sympatry?). To determine whether 2 5 7 species sympatry positively or negatively affected the sizes of brain regions, we independently fitted 2 5 8 Gaussian Pagel's lambda phylogenetic regressions for each brain region of extant frugivorous species.

5 9
This model is a derivative of the BM model, where the phylogenetic variance-covariance matrix has all 2 6 0 coefficients, but its diagonal ones, multiplied by lambda (therefore relaxing the hypothesis of BM). To fit 2 6 1 these models, we used a frequentist-based approach with the "phylolm" function from the phylolm 2 6 2 package (Ho and Ane 2014). We considered the least stringent frugivory assessment, with the frugivory 2 6 3 threshold fixed at 20% and the folivory threshold fixed at 40%.

6 4
The response variable was the log-transformed relative size of each brain region relative to the 2 6 5 body mass or the whole-brain size. Doing so controlled for allometry while being in line with the 2 6 6 phylogenetic models of trait evolution, which cannot account for additional variables and are thus limited different datasets and assessed the sensitivity using non-averaged values (see Supplementary Material,

6 9
Model stability). We used as covariates (i.e. continuous predictors) two explicit measures of the level of 2 7 0 species sympatry for each extant frugivorous species: (1) the number of frugivorous sympatric species 2 7 1 (square-rooted to reach symmetrical distribution to limit leverage effects) and (2) the average percentage 2 7 2 of the overlapping current range (assessed based on IUCN data) with other sympatric frugivorous species.

7 3
For a given species A, sympatry with another species B was considered when at least 10% of the range of 2 7 4 species A overlaps with the range of species B. This was done to reduce the noise induced by coarse 2 7 5 identification of species range.

7 6
Body mass and sympatry 2 7 7 Body mass may itself be affected by species sympatry. Yet, as body mass was used to compute some 2 7 8 relative brain size, to control for the potential confounding effect due to a relationship between sympatry 2 7 9 and body mass, we repeated all model fitting (models of trait evolution and PGLS) with body mass as the 2 8 0 output variable.

8 2
Next, we also investigated the correlations between primate diversification rates and brain sizes or levels 2 8 3 of sympatry to better understand the impact of cognition and interactions between primates on their 2 8 4 evolutionary success. To do so, we inferred how primates diversified over time and across lineages.

8 5
Lineage-specific net diversification rates (defined as speciation minus extinction rates) were estimated  diversification rate over time, for further explanations). We extracted the mean diversification rates 9 2 against the net diversification rates. Because assumptions for a frequentist-based approach were unmet, we 2 9 3 performed Bayesian-based inferences instead. We used the "MCMCglmm" function of the MCMCglmm 2 9 4 package (Hadfield 2010). Each chain had a burn-in period of 5000 iterations, a total length of 50,000 2 9 5 iterations, and was sampled every 50 iterations. We used the least informative priors. Fixed priors were set 2 9 6 to default values (Gaussian distribution of mean 0 and variance 1 0 ଼ ). Again, we used the mean of the 2 9 7 brain trait values for the main model and assessed the sensitivity by re-running the model several times 2 9 8 using non-averaged values.

9 9
To determine whether species sympatry was associated with lower or larger diversification rates,

1 1
The dataset we gathered contained between 34 to 182 frugivorous primate species (depending on the brain 3 1 2 region considered). When considering the sizes relative to the body mass of the different brain regions, we 3 1 3 observed ample variations ( Figure 3). For instance, the lemuriformes, which are known to prioritize smell

DISCUSSION
species sympatry, the latter nonetheless induced a change in the relative sizes of the hippocampus and the 3 9 8 striatum.

9 9
Our results were similar whether the relative sizes were obtained by weighting by body mass or 4 0 0 whole-brain size. Theoretically, the two weighting methods are expected to give insights into differences 4 0 1 in resource allocation between body tissues or differences in resource allocation within the brain tissue, 4 0 2 respectively. Nonetheless, we observed no differences in the response to sympatry by weighting by body 4 0 3 mass or whole-brain size, despite weak to moderate correlations between sizes relative to the body mass 4 0 4 and sizes relative to the whole-brain size (see Supplementary Figure S5). Therefore, our results suggest 4 0 5 that species sympatry mostly affected within-brain energy partitioning (as weighting by body mass can 4 0 6 depict reallocation changes towards the whole brain or only within the brain), and in particular 4 0 7 reallocations to two relatively small brain regions: the hippocampus and the striatum.

0 8
The hippocampus and the striatum are brain regions involved in individual-based and social-based 4 0 9 information processing, pinpointing that these two components might be under strong selection in

1 6
The fact that the hippocampus, particularly relevant to processing and memorising spatiotemporal 4 1 7 information, is negatively sensitive to sympatry, is consistent with the idea of an effect of sympatric and thus smaller brain region sizes.

3 5
In contrast to direct/indirect competition, potential indirect facilitation between species due to 4 3 6 "social" cues (Hypothesis 2), is ruled out by the absence of an effect of sympatry on brain regions 4 3 7 involved in immediate sensory information processing (e.g. cerebellum or neocortex). This absence of  Given the context-dependence of the direction of selection (towards bigger sizes when sympatry is low, 4 5 4 smaller sizes otherwise), there is no surprise that we do not observe a correlation between the net 4 5 5 diversification rate and the three brain regions affected by species sympatry. Surprisingly however, we 4 5 6 found no positive association between the net diversification rate and the EQ, the cerebellum, or the 4 5 7 neocortex, which were insensitive to species sympatry. By contrast, a positive association between brain 4 5 8 size and diversification was also found in birds (Sayol et al. 2019) given that bigger brains act as a buffer The use of brain size as a proxy for cognition is a central debate with no optimal solution (see grounded alleviate biases stemming from brain size analysis, but this will take time to generate large-enough 4 8 0 datasets. In the meanwhile, brain size is a proxy much appreciated in practice, because of its easy taken into account by considering the brain as a mosaic of singular and independent regionalised areas that