Data from: foraging and the importance of knowledge in Pemba, Tanzania: implications for childhood evolution
Data files
Oct 26, 2023 version files 123.44 KB
-
processed_data.RData
-
README.md
Abstract
Childhood is a period of life unique to humans. Childhood may have evolved through the need to acquire knowledge and subsistence skills. In an effort to understand the functional significance of childhood, previous research examined increases with age in returns to foraging across food resources. Such increases could be due to changes in knowledge, or other factors such as body size or strength. Here we attempt to unpack these age-related changes. First we estimate age-specific foraging returns for two resources. We then develop non-linear structural equation models to evaluate the relative importance of ecological knowledge, grip strength, and height in a population of part-time children foragers on Pemba island, Tanzania. We use anthropometric measures, estimates of ecological knowledge, and behavioral observations for 63 individuals across 370 foraging trips. We find slower increases in foraging returns with age for trap hunting than for shellfish collection. We do not detect any effect of individual knowledge on foraging returns, potentially linked to information-sharing within foraging parties. Producing accurate estimates of the distinct contribution of specific traits to an individual's foraging performance constitutes a key step in evaluating different hypotheses for the emergence of childhood.
README: Foraging and the importance of knowledge in Pemba, Tanzania: Implications for childhood evolution
https://doi.org/10.5061/dryad.c866t1gcr
Childhood is a period of life unique to humans. Childhood may have evolved through the need to acquire knowledge and subsistence skills. In an attempt to evaluate the importance of learning for the evolution of childhood, previous research examined the increase with age of returns to foraging across various resources. Any increase could be due to increases in knowledge or other factors such as body size and strength. Here, we first estimate age-specific foraging returns for two different resources. We then model the relative importance for foraging of ecological knowledge, grip strength and height in a population of part-time children foragers on the island of Pemba, Tanzania. We use anthropometric measures, such as height and strength, and estimates of ecological knowledge for more than 250 and 90 individuals respectively, in association to behavioral observations for 63 individuals across 372 foraging trips. We find slower increases in foraging returns with age for trap hunting than for shellfish collection. We do not detect any effect of individual knowledge on foraging returns, potentially because relevant information can be shared within foraging parties. We also find positive effect of individuals' height on shellfish collection. This is the first study providing accurate estimates of individual traits' contribution to foraging performance, which is important to evaluate evolutionary hypotheses for the emergence of childhood.
Description of the data
Data are provided in processed_data.RData
, a list that contains multiple dataframes. Several columns are present in more than one data frame.
Column | Description |
---|---|
anonymeID | anonymized id for individuals in the sample. Consistent across all data frames |
success | whether a shellfishing trip yielded any return (always 1 - in shells)/ how many times a trap captured anything (in traps) |
returns | weight (grams) of unprocessed shellfish collected in one trip (in shells)/total weight (grams) of all preys captured by a certain trap (in traps) |
length_min | duration of shellfish collection trip, excluding travelling and processing time (in shells) |
tide_height_m | minimum height of tide during which shellfish collecting trip was carried out (recorded daily from https://www.tideschart.com/Tanzania/Pemba-North/Micheweni/Konde/, in shells) |
tide_avg_depth | average height of tide during trip, calculated approximating tidal bulge as an inverted gaussian kernel (in shells) |
n_item_types | n of different types of shells collected by forager during trip (in shells) |
age | standardized age of individuals in the sample(in shell_ppl, trap_ppl) |
sex | sex of individuals in sample (in shell_ppl, trap_ppl) |
height | height of individuals in sample (centimeters, in shell_ppl, trap_ppl) |
grip | grip strength of individuals in sample (kilograms, in shell_ppl, trap_ppl) |
knowledge | n of items listed by individuals in sample in a freelisting task (prompt: "list all living organisms you can think of", in shell_ppl, trap_ppl) |
data | relevant source of data (either from foraging trips, anthropometric measures, knowledge interviews, informs the statistical model in treating missing data. In shell_ppl, trap_ppl) |
exposure | days from construction to dismantling/last observation of trap (in traps) |
length_day and length_hour | distance in days or hours between subsequent observations of one trap. Note that some observations are missing and hence distance in time can be incorrect (in all_traps) |
trap_ID | identification code for individual traps (in traps) |
NA datapoints
NA data appear within individual level data when the individuals' traits were not measured. In particular, there are instances where height, grip strength or ecological knowledge of individuals was not measured. These cases are marked by NA.
Additional data:
Within the main data list are contained matrixes which report the results of freelist tasks (each row is an individual, each column a possible item in the list. The matrix reports 1 in cell [i,j] if individual i named item j during their freelist, task, otherwise it reports 0).
Note that the data contain a dataframe named traps
, where each row is a single trap, installed by a single individual (this was used to produce the results shown in the paper), and an all_traps
dataframe, where each row is an observation of a trap, i.e. an instance where the status of the trap was observed by the hunters.
Code/Software
The code provided here is relative to the results presented in the main article. Running the make_file.R script will prepare the data, compile the models and produce the plots shown in figure 4 and 5 in the main text. Model fits and plots wil be saved in folders appropriately defined. All the code was written in R 4.2.2. Statistical models are fit using the Stan MCMC engine via the rstan package (2.21.0), which requires a C++ compiler. Installation instructions are available at https://github.com/stan-dev/rstan/wiki/RStan-Getting-Started. The rethinking package (2.21) is required to process fitted model outputs - installation instructions at http://xcelab.net/rm/software/.
Methods
Data were collected through structured interviews and observations during summer 2019 and through 2020. Processing included data cleaning and wrangling (non reproducible code available online), anonymization of identifiable information and dataset restructuring.