A guide to null models for animal social network analysis.
01 Oct 2017-Methods in Ecology and Evolution (John Wiley and Sons Inc.)-Vol. 8, Iss: 10, pp 1309-1320
TL;DR: It is shown that permutations of the raw observational (or ‘pre‐network’) data consistently account for underlying structure in the generated social network, and thus can reduce both type I and type II error rates.
Abstract: Null models are an important component of the social network analysis toolbox. However, their use in hypothesis testing is still not widespread. Furthermore, several different approaches for constructing null models exist, each with their relative strengths and weaknesses, and often testing different hypotheses.In this study, I highlight why null models are important for robust hypothesis testing in studies of animal social networks. Using simulated data containing a known observation bias, I test how different statistical tests and null models perform if such a bias was unknown.I show that permutations of the raw observational (or 'pre-network') data consistently account for underlying structure in the generated social network, and thus can reduce both type I and type II error rates. However, permutations of pre-network data remain relatively uncommon in animal social network analysis because they are challenging to implement for certain data types, particularly those from focal follows and GPS tracking.I explain simple routines that can easily be implemented across different types of data, and supply R code that applies each type of null model to the same simulated dataset. The R code can easily be modified to test hypotheses with empirical data. Widespread use of pre-network data permutation methods will benefit researchers by facilitating robust hypothesis testing.
Citations
More filters
TL;DR: It is concluded that using calibration data is an important step when constructing animal social networks, and that in their absence, researchers should use a simple estimator and explicitly consider the impact of this on their findings.
127 citations
Cites background from "A guide to null models for animal s..."
...Thus, even if good calibration data can be obtained to estimate accurate relationship strengths, it will always be important to use null models when conducting hypothesis testing with animal social networks (Farine, 2017)....
[...]
TL;DR: How network representations are constructed from underlying data, the variety of questions and tasks on these representations over several domains, and validation strategies for measuring the inferred network’s capability of answering questions on the system of interest are examined.
Abstract: Networks represent relationships between entities in many complex systems, spanning from online social interactions to biological cell development and brain connectivity. In many cases, relationships between entities are unambiguously known: are two users “friends” in a social network? Do two researchers collaborate on a published article? Do two road segments in a transportation system intersect? These are directly observable in the system in question. In most cases, relationships between nodes are not directly observable and must be inferred: Does one gene regulate the expression of another? Do two animals who physically co-locate have a social bond? Who infected whom in a disease outbreak in a population? Existing approaches for inferring networks from data are found across many application domains and use specialized knowledge to infer and measure the quality of inferred network for a specific task or hypothesis. However, current research lacks a rigorous methodology that employs standard statistical validation on inferred models. In this survey, we examine (1) how network representations are constructed from underlying data, (2) the variety of questions and tasks on these representations over several domains, and (3) validation strategies for measuring the inferred network’s capability of answering questions on the system of interest.
122 citations
Cites background from "A guide to null models for animal s..."
...R DN Farine [58], Whitehead and James [193]...
[...]
...Farine et al. [59] evaluates networks inferred from geo-location data of individual olive baboons (Papio anubis) within a troop....
[...]
TL;DR: It is hypothesised that by shaping the decisions about when and where to move, physical features of the environment will impact which individuals more frequently encounter one another and in doing so the overall social structure and social organization of populations.
Abstract: Animal societies are shaped both by social processes and by the physical environment in which social interactions take place. While many studies take the observed patterns of inter-individual interactions as products and proxies of pure social processes, or as links between resource availability and social structure, the role of the physical configuration of habitat features in shaping the social system of group-living animals remains largely overlooked. We hypothesise that by shaping the decisions about when and where to move, physical features of the environment will impact which individuals more frequently encounter one another and in doing so the overall social structure and social organization of populations. We first discuss how the spatial arrangement of habitat components (i.e. habitat configuration) can shape animal movements using empirical cases in the literature. Then, we draw from the empirical literature to discuss how movement patterns of individuals mediate the patterns of social interactions and social organization and highlight the role of network-based approaches in identifying, evaluating and partitioning the effects of habitat configuration on animal social structure or organization. We illustrate the combination of these mechanisms using a simple simulation. Finally, we discuss the implications of habitat configuration in shaping the ecology and evolution of animal societies and offer a framework for future studies. We highlight future directions for studies in animal societies that are increasingly important in widely human-modified landscapes, in particular the implications of habitat-driven social structure in evolution. There is now clear evidence that simple processes can generate apparent complex patterns of social structure. However, while studies such as those on collective behaviour and social networks have been focused on processes involving individual decision-making, broader patterns of social structure and social organization can also be shaped by factors that have more fundamental impacts on the movements of animals. One set of those factors is related to the amount and spatial arrangement of both biotic and abiotic components of the habitat in which animals live. Examples include the configuration formed by habitat patches connected through corridors, by the presence of hard boundaries between habitat types or by the uneven distribution of resources, mates and competitors across space. In this contribution, we highlight the potential effects of these, which are becoming increasingly important as studies start being able to track populations spanning larger landscapes.
112 citations
Cites background or methods from "A guide to null models for animal s..."
...…to quantify complexity in animal populations (see Weiss et al. 2019, topical collection on Social complexity), which can easily be compared to spatially explicit null models (see Farine 2017) to quantify the effects of habitat configuration on animal social complexity (see also Aplin et al. 2015b)....
[...]
...…of social behaviour and habitat configuration to social complexity by using permutation tests for network hypothesis (Croft et al. 2011; Farine 2017), generalised affiliation indices for extracting affiliations from network data (Whitehead and James 2015), randomisation of animal…...
[...]
TL;DR: This article details several multilayer methods, which can provide new insights into questions about animal sociality at individual, group, population and evolutionary levels of organization, and gives examples for how to implement multilayers methods.
111 citations
Cites background from "A guide to null models for animal s..."
...Just as in monolayer network analysis (Farine, 2017; Fosdick, Larremore, Nishimura, & Ugander, 2018; Newman, 2018c), it is vital to tailor the use of null models in multilayer networks in a context-specific and question-specific way....
[...]
TL;DR: The contributions of movement ecology in disease research are synthesised, with a particular focus on studies that have successfully used movement-based methods to quantify individual heterogeneity in exposure and transmission risk.
Abstract: Though epidemiology dates back to the 1700s, most mathematical representations of epidemics still use transmission rates averaged at the population scale, especially for wildlife diseases. In simplifying the contact process, we ignore the heterogeneities in host movements that complicate the real world, and overlook their impact on spatiotemporal patterns of disease burden. Movement ecology offers a set of tools that help unpack the transmission process, letting researchers more accurately model how animals within a population interact and spread pathogens. Analytical techniques from this growing field can also help expose the reverse process: how infection impacts movement behaviours, and therefore other ecological processes like feeding, reproduction, and dispersal. Here, we synthesise the contributions of movement ecology in disease research, with a particular focus on studies that have successfully used movement-based methods to quantify individual heterogeneity in exposure and transmission risk. Throughout, we highlight the rapid growth of both disease and movement ecology and comment on promising but unexplored avenues for research at their overlap. Ultimately, we suggest, including movement empowers ecologists to pose new questions, expanding our understanding of host-pathogen dynamics and improving our predictive capacity for wildlife and even human diseases.
94 citations
Cites background from "A guide to null models for animal s..."
...Observed association patterns in social networks are often compared to expected patterns in null models (e.g. ideal gas model) or randomised networks to test hypotheses about the mechanisms underlying social structure (Farine 2017; Silk et al. 2017b)....
[...]
...ideal gas model) or randomised networks to test hypotheses about the mechanisms underlying social structure (Farine 2017; Silk et al. 2017b)....
[...]
References
More filters
TL;DR: Seven major types of sampling for observational studies of social behavior have been found in the literature and the major strengths and weaknesses of each method are pointed out.
Abstract: Seven major types of sampling for observational studies of social behavior have been found in the literature. These methods differ considerably in their suitability for providing unbiased data of various kinds. Below is a summary of the major recommended uses of each technique: In this paper, I have tried to point out the major strengths and weaknesses of each sampling method. Some methods are intrinsically biased with respect to many variables, others to fewer. In choosing a sampling method the main question is whether the procedure results in a biased sample of the variables under study. A method can produce a biased sample directly, as a result of intrinsic bias with respect to a study variable, or secondarily due to some degree of dependence (correlation) between the study variable and a directly-biased variable. In order to choose a sampling technique, the observer needs to consider carefully the characteristics of behavior and social interactions that are relevant to the study population and the research questions at hand. In most studies one will not have adequate empirical knowledge of the dependencies between relevant variables. Under the circumstances, the observer should avoid intrinsic biases to whatever extent possible, in particular those that direcly affect the variables under study. Finally, it will often be possible to use more than one sampling method in a study. Such samples can be taken successively or, under favorable conditions, even concurrently. For example, we have found it possible to take Instantaneous Samples of the identities and distances of nearest neighbors of a focal individual at five or ten minute intervals during Focal-Animal (behavior) Samples on that individual. Often during Focal-Animal Sampling one can also record All Occurrences of Some Behaviors, for the whole social group, for categories of conspicuous behavior, such as predation, intergroup contact, drinking, and so on. The extent to which concurrent multiple sampling is feasible will depend very much on the behavior categories and rate of occurrence, the observational conditions, etc. Where feasible, such multiple sampling can greatly aid in the efficient use of research time.
12,470 citations
"A guide to null models for animal s..." refers background in this paper
...However, there is no well-defined routine for constructing null models using raw data generated from successively following focal individuals and recording their interactions with others, despite this being a common sampling method (Altmann 1974)....
[...]
TL;DR: It is concluded that the new specifications of exponential random graph models increase the range and applicability of the ERGM as a tool for the statistical analysis of social networks.
Abstract: The most promising class of statistical models for expressing structural properties of social networks observed at one moment in time is the class of exponential random graph models (ERGMs), also known as p∗ models. The strong point of these models is that they can represent a variety of structural tendencies, such as transitivity, that define complicated dependence patterns not easily modeled by more basicpr obability models. Recently, Markov chain Monte Carlo (MCMC) algorithms have been developed that produce approximate maximum likelihood estimators. Applying these models in their traditional specification to observed network data often has led to problems, however, which can be traced back to the fact that important parts of the parameter space correspond to nearly degenerate distributions, which may lead to convergence problems of estimation algorithms, and a poor fit to empirical data. This paper proposes new specifications of exponential random graph models. These specifications represent structural properties such as transitivity and heterogeneity of degrees by more complicated graph statistics than the traditional star and triangle counts. Three kinds of statistics are proposed: geometrically weighted degree distributions, alternating k-triangles, and alternating independent two-paths. Examples are presented both of modeling graphs and digraphs, in which the new specifications lead to much better results than the earlier existing specifications of the ERGM. It is concluded that the new specifications increase the range and applicability of the ERGM as a tool for the statistical analysis of social networks.
1,356 citations
"A guide to null models for animal s..." refers methods in this paper
...ERGMs are used to generate hypotheses about what structural processes underpin the formation of social networks (Snijders et al. 2006), and function by randomly adding and removing edges to see how they change the network....
[...]
TL;DR: Ergm has the capability of approximating a maximum likelihood estimator for an ERGM given a network data set; simulating new network data sets from a fitted ERGM using Markov chain Monte Carlo; and assessing how well a fittedERGM does at capturing characteristics of a particular networkData set.
Abstract: We describe some of the capabilities of the ergm package and the statistical theory underlying it. This package contains tools for accomplishing three important, and interrelated, tasks involving exponential-family random graph models (ERGMs): estimation, simulation, and goodness of t. More precisely, ergm has the capability of approximating a maximum likelihood estimator for an ERGM given a network data set; simulating new network data sets from a tted ERGM using Markov chain Monte Carlo; and assessing how well a tted ERGM does at capturing characteristics of a particular network data set.
1,203 citations
TL;DR: It is suggested that a golden age of animal tracking science has begun and that the upcoming years will be a time of unprecedented exciting discoveries.
Abstract: BACKGROUND The movement of animals makes them fascinating but difficult study subjects. Animal movements underpin many biological phenomena, and understanding them is critical for applications in conservation, health, and food. Traditional approaches to animal tracking used field biologists wielding antennas to record a few dozen locations per animal, revealing only the most general patterns of animal space use. The advent of satellite tracking automated this process, but initially was limited to larger animals and increased the resolution of trajectories to only a few hundred locations per animal. The last few years have shown exponential improvement in tracking technology, leading to smaller tracking devices that can return millions of movement steps for ever-smaller animals. Finally, we have a tool that returns high-resolution data that reveal the detailed facets of animal movement and its many implications for biodiversity, animal ecology, behavior, and ecosystem function. ADVANCES Improved technology has brought animal tracking into the realm of big data, not only through high-resolution movement trajectories, but also through the addition of other on-animal sensors and the integration of remote sensing data about the environment through which these animals are moving. These new data are opening up a breadth of new scientific questions about ecology, evolution, and physiology and enable the use of animals as sensors of the environment. High–temporal resolution movement data also can document brief but important contacts between animals, creating new opportunities to study social networks, as well as interspecific interactions such as competition and predation. With solar panels keeping batteries charged, “lifetime” tracks can now be collected for some species, while broader approaches are aiming for species-wide sampling across multiple populations. Miniaturized tags also help reduce the impact of the devices on the study subjects, improving animal welfare and scientific results. As in other disciplines, the explosion of data volume and variety has created new challenges and opportunities for information management, integration, and analysis. In an exciting interdisciplinary push, biologists, statisticians, and computer scientists have begun to develop new tools that are already leading to new insights and scientific breakthroughs. OUTLOOK We suggest that a golden age of animal tracking science has begun and that the upcoming years will be a time of unprecedented exciting discoveries. Technology continues to improve our ability to track animals, with the promise of smaller tags collecting more data, less invasively, on a greater variety of animals. The big-data tracking studies that are just now being pioneered will become commonplace. If analytical developments can keep pace, the field will be able to develop real-time predictive models that integrate habitat preferences, movement abilities, sensory capacities, and animal memories into movement forecasts. The unique perspective offered by big-data animal tracking enables a new view of animals as naturally evolved sensors of environment, which we think has the potential to help us monitor the planet in completely new ways. A massive multi-individual monitoring program would allow a quorum sensing of our planet, using a variety of species to tap into the diversity of senses that have evolved across animal groups, providing new insight on our world through the sixth sense of the global animal collective. We expect that the field will soon reach a transformational point where these studies do more than inform us about particular species of animals, but allow the animals to teach us about the world.
1,096 citations