Generation and applications of simulated datasets to integrate social network and demographic analyses

Social networks are tied to population dynamics; interactions are driven by population density and demographic structure, while social relationships can be key determinants of survival and reproductive success. However, difficulties integrating models used in demography and network analysis have limited research at this interface. We introduce the R package genNetDem for simulating integrated network-demographic datasets. It can be used to create longitudinal social network and/or capture-recapture datasets with known properties. It incorporates the ability to generate populations and their social networks, generate grouping events using these networks, simulate social network effects on individual survival, and flexibly sample these longitudinal datasets of social associations. By generating co-capture data with known statistical relationships, it provides functionality for methodological research. We demonstrate its use with case studies testing how imputation and sampling design influence the success of adding network traits to conventional Cormack-Jolly-Seber (CJS) models. We show that incorporating social network effects into CJS models generates qualitatively accurate results, but with downward-biased parameter estimates when network position influences survival. Biases are greater when fewer interactions are sampled or fewer individuals observed in each interaction. While our results indicate the potential of incorporating social effects within demographic models, they show that imputing missing network measures alone is insufficient to accurately estimate social effects on survival, pointing to the importance of incorporating network imputation approaches. genNetDem provides a flexible tool to aid these methodological advancements and help researchers testing other sampling considerations in social network studies.

Predicting primate-parasite associations using exponential random graph models

Ecological associations between hosts and parasites are influenced by host exposure and susceptibility to parasites, and by parasite traits, such as transmission mode. Advances in network analysis allow us to answer questions about the causes and consequences of traits in ecological networks in ways that could not be addressed in the past. We used a network-based framework (exponential random graph models or ERGMs) to investigate the biogeographic, phylogenetic and ecological characteristics of hosts and parasites that affect the probability of interactions among nonhuman primates and their parasites. Parasites included arthropods, bacteria, fungi, protozoa, viruses and helminths. We investigated existing hypotheses, along with new predictors and an expanded host-parasite database that included 213 primate nodes, 763 parasite nodes and 2319 edges among them. Analyses also investigated phylogenetic relatedness, sampling effort and spatial overlap among hosts. In addition to supporting some previous findings, our ERGM approach demonstrated that more threatened hosts had fewer parasites, and notably, that this effect was independent of hosts also having a smaller geographic range. Despite having fewer parasites, threatened host species shared more parasites with other hosts, consistent with loss of specialist parasites and threat arising from generalist parasites that can be maintained in other, non-threatened hosts. Viruses, protozoa and helminths had broader host ranges than bacteria, or fungi, and parasites that infect non-primates had a higher probability of infecting more primate species. The value of the ERGM approach for investigating the processes structing host-parasite networks provided a more complete view on the biogeographic, phylogenetic and ecological traits that influence parasite species richness and parasite sharing among hosts. The results supported some previous analyses and revealed new associations that warrant future research, thus revealing how hosts and parasites interact to form ecological networks.

Social Support and Network Formation in a Small-Scale Horticulturalist Population

Evolutionary studies of cooperation in traditional human societies suggest that helping family and responding in kind when helped are the primary mechanisms for informally distributing resources vital to day-to-day survival (e.g., food, knowledge, money, childcare). However, these studies generally rely on forms of regression analysis that disregard complex interdependences between aid, resulting in the implicit assumption that kinship and reciprocity drive the emergence of entire networks of supportive social bonds. Here I evaluate this assumption using individual-oriented simulations of network formation (i.e., Stochastic Actor-Oriented Models). Specifically, I test standard predictions of cooperation derived from the evolutionary theories of kin selection and reciprocal altruism alongside well-established sociological predictions around the self-organisation of asymmetric relationships. Simulations are calibrated to exceptional public data on genetic relatedness and the provision of tangible aid amongst all 108 adult residents of a village of indigenous horticulturalists in Nicaragua (11,556 ordered dyads). Results indicate that relatedness and reciprocity are markedly less important to whom one helps compared to the supra-dyadic arrangement of the tangible aid network itself.

Permutation tests for hypothesis testing with animal social network data: Problems and potential solutions

Permutation tests are widely used to test null hypotheses with animal social network data, but suffer from high rates of type I and II error when the permutations do not properly simulate the intended null hypothesis.

Two common types of permutations each have limitations. Pre‐network (or datastream) permutations can be used to control ‘nuisance effects’ like spatial, temporal or sampling biases, but only when the null hypothesis assumes random social structure. Node (or node‐label) permutation tests can test null hypotheses that include nonrandom social structure, but only when nuisance effects do not shape the observed network.

We demonstrate one possible solution addressing these limitations: using pre‐network permutations to adjust the values for each node or edge before conducting a node permutation test. We conduct a range of simulations to estimate error rates caused by confounding effects of social or non‐social structure in the raw data.

Regressions on simulated datasets suggest that this ‘double permutation’ approach is less likely to produce elevated error rates relative to using only node permutations, pre‐network permutations or node permutations with simple covariates, which all exhibit elevated type I errors under at least one set of simulated conditions. For example, in scenarios where type I error rates from pre‐network permutation tests exceed 30%, the error rates from double permutation remain at 5%.

The double permutation procedure provides one potential solution to issues arising from elevated type I and type II error rates when testing null hypotheses with social network data. We also discuss alternative approaches that can provide robust inference, including fitting mixed effects models, restricted node permutations, testing multiple null hypotheses and splitting large datasets to generate replicated networks. Finally, we highlight ways that uncertainty can be explicitly considered and carried through the analysis.

The performance of permutations and exponential random graph models when analyzing animal networks

Social network analysis is a suite of approaches for exploring relational data. Two approaches commonly used to analyze animal social network data are permutation-based tests of significance and exponential random graph models. However, the performance of these approaches when analyzing different types of network data has not been simultaneously evaluated. Here we test both approaches to determine their performance when analyzing a range of biologically realistic simulated animal social networks. We examined the false positive and false negative error rate of an effect of a two-level explanatory variable (e.g., sex) on the number and combined strength of an individual’s network connections. We measured error rates for two types of simulated data collection methods in a range of network structures, and with/without a confounding effect and missing observations. Both methods performed consistently well in networks of dyadic interactions, and worse on networks constructed using observations of individuals in groups. Exponential random graph models had a marginally lower rate of false positives than permutations in most cases. Phenotypic assortativity had a large influence on the false positive rate, and a smaller effect on the false negative rate for both methods in all network types. Aspects of within- and between-group network structure influenced error rates, but not to the same extent. In "grouping event-based" networks, increased sampling effort marginally decreased rates of false negatives, but increased rates of false positives for both analysis methods. These results provide guidelines for biologists analyzing and interpreting their own network data using these methods.