Table 1.

A summary of methods for identifying the boundaries of polydomous colonies, their utility and limitations, and the knowledge required to use them

Method	Utility (example references)	Limitations	Prerequisite knowledge	Potential biases
Marked food (e.g. immunoglobulin marking)	Movement of marked resources shows that ants from those nests are moving between nests or exchanging food (Buczkowski and Bennett 2009; Hoffmann 2014)	Function of resource sharing (e.g. sharing, stealing, and appeasement) is unknown. There are also sampling difficulties, especially within populous nests	The type of resources being exchanged between nests. Ideally, also the method by which the exchange takes place and the level of resource exchange between a pair of nests	Underestimation of colony size due to: – insufficient resource marking – heterogeneous resource movement Overestimation of colony size due to: – stealing of food by neighbours
Marked workers	By marking workers and observing their movement, the behaviours associated with a social connection can be studied. If resource exchange occurs, information about the mechanism will be revealed (Rosengren 1985; Ellis and Robinson 2016)	The re-observation rate of marked workers (and therefore the number of workers that need to be marked) will depend on the population of the nests in question and the complexity of the system of behaviours involved in inter-nest resource transfer. Marked workers in nests with large populations or complex resource exchange mechanisms are unlikely to be re-observed	Durability of individual markings Probability of re-observation of a marked individual over a time-period relevant to the study, to determine required number of marked individuals	Underestimation of colony size due to: – marking an insufficient number of workers – loss of markings – heterogeneous worker movement – high nest fidelity
Direct observation of trails between nests	Gives a good quantitative overview of the structure of social connections over a whole multi-nest system; can provide quantitative data about connection strengths from trail usage (van Wilgenburg and Elgar 2007; Ellis et al. 2014)	The nature of resources being exchanged via trails is unclear; trail usage may not be a good approximation of resource exchange via trails; mechanism of exchange is unknown	Only appropriate if the species consistently forms trails between all nests that exchange resources	Underestimation of colony size due to: – failing to observe trails that are used inconsistently – failing to record underground connections
Ecological inference (e.g. changed nest strategy in response to environmental change)	Puts the social connection, and potentially resource exchange, between nests in a clear ecological context (Banschbach and Herbers 1996a; Dahbi et al. 2008)	The nature and extent of resource exchange, the quantities exchanged and the mechanism of exchange are unclear. The timescale (i.e. temporary or long-term) of the strategy are also unknown	That observed changes in the nesting strategy are not simply a short-term intermediate strategy, rather part of a long-term, and evolutionarily relevant, strategy	Misidentification of colony boundaries due to: – observer bias – inaccurate identification of the cause of nest separation
Inter-nest aggression assays	Demonstrates whether workers from a pair of nests are mutually tolerant, or mutually aggressive (Roulston et al. 2003)	There are a great variety of types of assays which can, and have, been used to investigate aggression between nests (Table 2). The efficacy and consistency of these various methods are unknown. Observer bias is problematic when subjectively identifying aggression between ants. Some species of ants are non-aggressive, even to conspecifics from distant populations	That aggression is expected between ants from different colonies, and that this aggression will be reproduced consistently in the assay being used	Overestimation of colony size due to: – low overall aggression in population/species – low motivation for aggression due to, e.g. season or context – observer bias Underestimation of colony size due to: – inappropriate testing conditions causing increased aggression – observer bias
Spatial clustering analysis	An objective technique to assess whether nests are distributed non-randomly in the environment (Sudd et al. 1977; Santini et al. 2011)	Both ecological factors and population history can produce clusters of nests in the environment. The scale at which clustering is investigated is also subjective. Methodological difficulties with defining the boundaries of the area in which clustering is to be assessed	The impact of environmental limitations on space occupancy, so that this effect can be distinguished from the effects of space sharing	Over-or underestimation of colony size due to: – failing to identify an important environmental variable
Genetic differentiation, F ST	Workers displaying significant genetic differentiation are unlikely to be within the same reproductive unit, and therefore the same colony (Elias et al. 2004; Steinmeyer et al. 2012)	Differentiation builds up over long time scales, potentially longer than colony formation, therefore lack of differentiation does not mean that two nests are within the same colony	Any evidence that colony formation is likely to be very recent, such as recent population expansions. Is genetic differentiation detectable in the population as a whole?	Overestimation of colony size due to low differentiation, caused by: – recent founding – low power
Relatedness	Highly related workers are very likely to be from the same family unit, and therefore the same colony (Pedersen and Boomsma 1999; Pamminger et al. 2014)	In highly polygynous populations, relatedness can be indistinguishable from zero. Relatedness estimates are also highly variable within a nest, therefore it may be difficult to distinguish between nests showing small differences in relatedness	The expected level of polygyny within the population	Overestimation of colony size due to: – high variability causing lack of ability to distinguish nests on relatedness
G-distance	A comparative measure of differentiation, G-distance describes how genetically different workers are (Pedersen and Boomsma 1999)	It is impossible to compare different studies, because measures are comparative within studies. There is no obvious cut off above which colony boundaries are clear		Underestimation of colony size due to: – assumptions that there are distinctions within a population, and that comparative measure will be useful
Rare genotype sisterhoods	Nests sharing rare genotype sisterhoods share common descent, which can reveal groupings within highly variable data (Pedersen and Boomsma 1999)	Only works if there are sufficiently rare alleles. Not identifying a sisterhood does not mean that two nests are within different colonies. In recently expanded populations, many different colonies may share descent and therefore share rare genotype sisterhoods	Evidence of recent population formation or bottlenecks: if present these may obscure rare alleles, because all members of population share recent descent	Overestimation of colony size due to: – population lacking sufficiently rare genotypes – recent population expansion
Bayesian clustering methods	Bayesian clustering methods allow delineation of genetic groupings without observer bias (Holzer et al. 2009; Huszár et al. 2014)	Any genetic structure in the data will be identified, not necessarily colony boundaries, e.g. a population formed by the merging of two distinct gene pools may separate by those gene pools, even though each contains many colonies	Any genetic structuring within the population that is not related to colony structure, e.g. differentiation due to a geographic barrier	Overestimation of colony size due to: – genetic groupings above the colony level being misidentified as colony boundaries Underestimation of colony size due to – genetic isolation by distance within large polydomous colonies being misinterpreted as colony boundaries
Sequencing mtDNA	mtDNA haplotypes shared between nests is evidence of shared descent (Holzer et al. 2009; Seppä et al. 2012)	Variability can be low across large areas; there may not be mtDNA variation within the population at all	Variability of mtDNA within population or region	Overestimation of colony size due to: – lack of variation within populations Underestimation of colony size due to: – multiple haplotypes within a colony leading to incorrect inference of a division

Examples of studies are included in the ‘utility’ column, but for more complete referencing refer to the text