Primates301: Primate life history and social learning data

In rmcelreath/rethinking: Statistical Rethinking book package

Primate life history and social learning data

Description

Life history data, social learning data, and phylogenetic distance matrix for 301 primate species. These data were assembled by and analyzed in Street et al 2017 (see references).

Usage

data(Primates301)
data(Primates301_distance_matrix)
data(Primates301_vcov_matrix)

Format

Primates301 is a data.table with elements:

1. name: Full taxonomic name of species

2. genus : Genus of species

3. species : Species name within genus

4. subspecies : Sub-species designation, if any

5. spp_id : Unique ID for species

6. genus_id : Unique ID for genus

7. social_learning : Count of mentions of social learning in literature

8. research_effort : Size of literature on species

9. brain : Brain volume (endocranial volume) in cubic centimeters

10. body : Body mass in grams

11. group_size : Average social group size

12. gestation : Length of gestation (days)

13. weaning : At at weaning (days)

14. longevity : Maximum lifespan (months)

15. sex_maturity : Age of sexual maturity (days)

16. maternal_investment : Period of maternal investment (days) = gestation + weaning

Primates301_distance_matrix is a matrix with species on the margins and phylogenetic distances in the cells. Primates301_vcov_matrix is a matrix with species on the margins and variances-covariances in the cells.

References

Street SE, Navarrete AF, Reader SM, Laland KN (2017) Coevolution of cultural intelligence, extended life history, sociality, and brain size in primates. PNAS https://doi.org/10.1073/pnas.1620734114

Arnold C, Matthews LJ, Nunn CL (2010) The 10kTrees Website: A New Online Resource for Primate Phylogeny. Evol Anthropol 19(3):114-118.

Reader SM, Hager Y, Laland KN (2011) The evolution of primate general and cultural intelligence. Philos Trans R Soc B-Biological Sci 366(1567):1017-1027.

Isler K, et al. (2008) Endocranial volumes of primate species: scaling analyses using a comprehensive and reliable data set. J Hum Evol 55(6):967-978.

Jones, Kate E, et al. (2009) PanTHERIA: a species-level database of life history, ecology, and geography of extant and recently extinct mammals. Ecology 90:2649.

Examples

data(Primates301)
plot( log(brain) ~ log(body) , data=Primates301 )

data(Primates301_distance_matrix)
image(Primates301_distance_matrix)

# Gaussian process phylogenetic regression
# prep variables
d <- Primates301
d$name <- as.character(d$name)
dstan <- d[ complete.cases( d$social_learning, d$research_effort , d$body , d$brain ) , ]
# prune distance matrix to spp in dstan
spp_obs <- dstan$name
y <- Primates301_distance_matrix
y2 <- y[ spp_obs , spp_obs ]
# cbind( sort(spp_obs) , sort(colnames(y2)) )
# scale distances
y3 <- y2/max(y2)

mP301GP <- ulam(
    alist(
        social_learning ~ poisson( lambda ),
        log(lambda) <- a + g[spp_id] + b_ef*log_research_effort + b_body*log_body + b_eq*log_brain,
        a ~ normal(0,1),
        vector[N_spp]: g ~ multi_normal( 0 , SIGMA ),
        matrix[N_spp,N_spp]: SIGMA <- cov_GPL2( Dmat , etasq , rhosq , 0.01 ),
        b_body ~ normal(0,1),
        b_eq ~ normal(0,1),
        b_ef ~ normal(1,1),
        etasq ~ exponential(1),
        rhosq ~ exponential(1)
    ),
    data=list(
        N_spp = nrow(dstan),
        social_learning = dstan$social_learning,
        spp_id = 1:nrow(dstan),
        log_research_effort = log(dstan$research_effort),
        log_body = log(dstan$body),
        log_brain = log(dstan$brain),
        Dmat = y3
    ) , 
    control=list(max_treedepth=15,adapt_delta=0.95) , 
    sample=FALSE , iter=400 )

# non-centered, Cholesky form
mP301GPnc <- ulam(
    alist(
        social_learning ~ poisson( lambda ),
        log(lambda) <- a + g[spp_id] + b_ef*log_research_effort + b_body*log_body + b_eq*log_brain,
        a ~ normal(0,1),
        vector[N_spp]: g <<- L_SIGMA * eta,
        vector[N_spp]: eta ~ normal( 0 , 1 ),
        matrix[N_spp,N_spp]: L_SIGMA <<- cholesky_decompose( SIGMA ),
        matrix[N_spp,N_spp]: SIGMA <- cov_GPL2( Dmat , etasq , rhosq , 0.01 ),
        b_body ~ normal(0,1),
        b_eq ~ normal(0,1),
        b_ef ~ normal(1,1),
        etasq ~ exponential(1),
        rhosq ~ exponential(1)
    ),
    data=list(
        N_spp = nrow(dstan),
        social_learning = dstan$social_learning,
        spp_id = 1:nrow(dstan),
        log_research_effort = log(dstan$research_effort),
        log_body = log(dstan$body),
        log_brain = log(dstan$brain),
        Dmat = y3
    ) , 
    control=list(max_treedepth=15,adapt_delta=0.95) , 
    sample=FALSE , iter=400 )

# Pagel's lambda approach --- Not endorsed!
# This is of historical interest only
data(Primates301_vcov_matrix)
vcov_thin <- Primates301_vcov_matrix[ spp_obs , spp_obs ]
mP301L <- ulam(
    alist(
        social_learning ~ poisson( lambda ),
        log(lambda) <- a + g[spp_id] + b_ef*log_research_effort + b_body*log_body + b_eq*log_brain,
        a ~ normal(0,1),
        vector[N_spp]: g <<- L_SIGMA * eta,
        vector[N_spp]: eta ~ normal( 0 , 1 ),
        matrix[N_spp,N_spp]: L_SIGMA <<- cholesky_decompose( SIGMA ),
        matrix[N_spp,N_spp]: SIGMA <- cov_Pagel( SIGMA_raw , Plambda ),
        b_body ~ normal(0,1),
        b_eq ~ normal(0,1),
        b_ef ~ normal(1,1),
        Plambda ~ beta(2,2)
    ),
    data=list(
        N_spp = nrow(dstan),
        social_learning = dstan$social_learning,
        spp_id = 1:nrow(dstan),
        log_research_effort = log(dstan$research_effort),
        log_body = log(dstan$body),
        log_brain = log(dstan$brain),
        SIGMA_raw = vcov_thin
    ) , 
    control=list(max_treedepth=15,adapt_delta=0.95) , 
    sample=TRUE , iter=400 )

# centered version --- seems to mix better
mP301L2 <- ulam(
    alist(
        social_learning ~ poisson( lambda ),
        log(lambda) <- a + g[spp_id] + b_ef*log_research_effort + b_body*log_body + b_eq*log_brain,
        a ~ normal(0,1),
        vector[N_spp]: g ~ multi_normal( 0 , SIGMA ),
        matrix[N_spp,N_spp]: SIGMA <- cov_Pagel( SIGMA_raw , Plambda ),
        b_body ~ normal(0,1),
        b_eq ~ normal(0,1),
        b_ef ~ normal(1,1),
        Plambda ~ beta(2,2)
    ),
    data=list(
        N_spp = nrow(dstan),
        social_learning = dstan$social_learning,
        spp_id = 1:nrow(dstan),
        log_research_effort = log(dstan$research_effort),
        log_body = log(dstan$body),
        log_brain = log(dstan$brain),
        SIGMA_raw = vcov_thin
    ) , 
    control=list(max_treedepth=15,adapt_delta=0.95) , 
    sample=TRUE , iter=400 )

rmcelreath/rethinking documentation built on Sept. 18, 2023, 2:01 p.m.