Title from second page of PDF document. Includes bibliographical references (p. 145-173).

Culture is an important means by which both human and non-human animals transmit useful behaviours between individuals and generations. Amongst animals, chimpanzees live particularly varied cultural lives. However, the processes and factors that influence whether chimpanzees will be motivated to copy an observed behaviour are poorly understood. In this thesis, I explore various factors and their influence on social learning decisions in chimpanzees. In turn, the chapters examine the influence of (i) rank-bias towards copying dominant individuals, (ii) majority and contextual influences and finally (iii) individual differences in proclivity for social learning. In my first experiment, I found evidence that chimpanzees are highly motivated to copy the behaviour of subordinate demonstrators and innovators in an open-diffusion puzzle-box paradigm. In contrast, behaviours seeded by dominant individuals were not transmitted as faithfully. This finding has important implications for our understanding of the emergence of novel traditions. In my second experiment, I found that some chimpanzees are highly motivated to relinquish an existing behaviour to adopt an equally rewarding alternative if it is consistently demonstrated by just one or two individuals within a group context, but not in a dyadic context. This contrasts with prior studies which argue that chimpanzees are highly conservative and may hint at a hitherto unrecognised process by which conformity-like behaviour might occur. Finally, I performed a novel type of ‘meta' analysis on 16 social learning studies carried out at our research site to determine whether individuals demonstrated consistency in their social learning behaviour across experimental contexts. Strong evidence for individual differences in social information use was found, with females more likely to use social information than males. No effect of age, research experience or rearing history was found. This presents a promising new method of studying individual differences in behaviour using the accumulated findings of previous work at a study site.

abstract: Identifying the ecological role, or niche, that a species occupies within their larger community elucidates environmental adaptability and evolutionary success. This dissertation investigates the occupied niche of chimpanzees (Pan troglodytes schweinfurthii) living in an open, dry savanna-woodland environment by examining patterns of resource use and interspecific interactions. Data were collected October 2010--November 2011 at Issa, in the Ugalla region of western Tanzania, which is one of the driest, most open, and seasonal habitats inhabited by chimpanzees. Unlike most primatological studies which employ methods that include focal follows, this study focused instead on observing 'resource patches' for chimpanzees. Patch focals allow for the observation of all animals within a study area; capture resources that are not used by the study species; and are particularly well suited for unhabituated communities. In order to better understand relationships between environment and behavior, data collected at Issa are compared with published data from other chimpanzee populations. Issa chimpanzees were expected to have broader resource use than forest chimpanzees, as well as increased competition with other fauna, due to fewer available resources. However, in contrast to the assumption of food scarcity in dry habitats, dietary resources were available throughout the year. Like other populations, the diet of Issa chimpanzees consisted of mostly fruit, but unlike at other sites, the majority of plants consumed were woodland species. Additionally, although chimpanzees and other fauna shared spatial and dietary resources, there was only nominal overlap. These results point to extremely low levels of indirect competition between chimpanzees and other fauna. Despite extensive study of forest chimpanzees, little is known about their role within their faunal community in open, dry habitats, nor about how greater seasonality affects resource use. This project addresses both of these important issues and fosters novel approaches in anthropological studies, especially in reference to chimpanzee ecology and evolution. Understanding current chimpanzee behavioral relationships with their environments shapes hypotheses about their pasts, and also informs predictions about behaviors of similar taxa in paleo-environments. Lastly, examining the ecological role of chimpanzees within their larger communities will influence the formation of, as well as evaluate, conservation strategies. / Dissertation/Thesis / Ph.D. Anthropology 2013

Non-verbal acoustic communication is of high importance to humans and animals: Infants use the voice as a primary communication tool. Animals of all kinds employ acoustic communication, such as chimpanzees, which use pant-hoot vocalizations for long-distance communication. Many applications require the assessment of such communication for a variety of analysis goals. Computational systems can support these areas through automatization of the assessment process. This is of particular importance in monitoring scenarios over large spatial and time scales, which are infeasible to perform manually. Algorithms for sound recognition have traditionally been based on conventional machine learning approaches. In recent years, so-called representation learning approaches have gained increasing popularity. This particularly includes deep learning approaches that feed raw data to deep neural networks. However, there remain open challenges in applying these approaches to automatic recognition of non-verbal acoustic communication events, such as compensating for small data set sizes. The leading question of this thesis is: How can we apply deep learning more effectively to automatic recognition of non-verbal acoustic communication events? The target communication types were specifically (1) infant vocalizations and (2) chimpanzee long-distance calls. This thesis comprises four studies that investigated aspects of this question: Study (A) investigated the assessment of infant vocalizations by laypersons. The central goal was to derive an infant vocalization classification scheme based on the laypersons' perception. The study method was based on the Nijmegen Protocol, where participants rated vocalization recordings through various items, such as affective ratings and class labels. Results showed a strong association between valence ratings and class labels, which was used to derive a classification scheme. Study (B) was a comparative study on various neural network types for the automatic classification of infant vocalizations. The goal was to determine the best performing network type among the currently most prevailing ones, while considering the influence of their architectural configuration. Results showed that convolutional neural networks outperformed recurrent neural networks and that the choice of the frequency and time aggregation layer inside the network is the most important architectural choice. Study (C) was a detailed investigation on computer vision-like convolutional neural networks for infant vocalization classification. The goal was to determine the most important architectural properties for increasing classification performance. Results confirmed the importance of the aggregation layer and additionally identified the input size of the fully-connected layers and the accumulated receptive field to be of major importance. Study (D) was an investigation on compensating class imbalance for chimpanzee call detection in naturalistic long-term recordings. The goal was to determine which compensation method among a selected group improved performance the most for a deep learning system. Results showed that spectrogram denoising was most effective, while methods for compensating relative imbalance either retained or decreased performance.:1. Introduction 2. Foundations in Automatic Recognition of Acoustic Communication 3. State of Research 4. Study (A): Investigation of the Assessment of Infant Vocalizations by Laypersons 5. Study (B): Comparison of Neural Network Types for Automatic Classification of Infant Vocalizations 6. Study (C): Detailed Investigation of CNNs for Automatic Classification of Infant Vocalizations 7. Study (D): Compensating Class Imbalance for Acoustic Chimpanzee Detection With Convolutional Recurrent Neural Networks 8. Conclusion and Collected Discussion 9. Appendix / Nonverbale akustische Kommunikation ist für Menschen und Tiere von großer Bedeutung: Säuglinge nutzen die Stimme als primäres Kommunikationsmittel. Schimpanse verwenden sogenannte 'Pant-hoots' und Trommeln zur Kommunikation über weite Entfernungen. Viele Anwendungen erfordern die Beurteilung solcher Kommunikation für verschiedenste Analyseziele. Algorithmen können solche Bereiche durch die Automatisierung der Beurteilung unterstützen. Dies ist besonders wichtig beim Monitoring langer Zeitspannen oder großer Gebiete, welche manuell nicht durchführbar sind. Algorithmen zur Geräuscherkennung verwendeten bisher größtenteils konventionelle Ansätzen des maschinellen Lernens. In den letzten Jahren hat eine alternative Herangehensweise Popularität gewonnen, das sogenannte Representation Learning. Dazu gehört insbesondere Deep Learning, bei dem Rohdaten in tiefe neuronale Netze eingespeist werden. Jedoch gibt es bei der Anwendung dieser Ansätze auf die automatische Erkennung von nonverbaler akustischer Kommunikation ungelöste Herausforderungen, wie z.B. die Kompensation der relativ kleinen Datenmengen. Die Leitfrage dieser Arbeit ist: Wie können wir Deep Learning effektiver zur automatischen Erkennung nonverbaler akustischer Kommunikation verwenden? Diese Arbeit konzentriert sich speziell auf zwei Kommunikationsarten: (1) vokale Laute von Säuglingen (2) Langstreckenrufe von Schimpansen. Diese Arbeit umfasst vier Studien, welche Aspekte dieser Frage untersuchen: Studie (A) untersuchte die Beurteilung von Säuglingslauten durch Laien. Zentrales Ziel war die Ableitung eines Klassifikationsschemas für Säuglingslaute auf der Grundlage der Wahrnehmung von Laien. Die Untersuchungsmethode basierte auf dem sogenannten Nijmegen-Protokoll. Hier beurteilten die Teilnehmenden Lautaufnahmen von Säuglingen anhand verschiedener Variablen, wie z.B. affektive Bewertungen und Klassenbezeichnungen. Die Ergebnisse zeigten eine starke Assoziation zwischen Valenzbewertungen und Klassenbezeichnungen, die zur Ableitung eines Klassifikationsschemas verwendet wurde. Studie (B) war eine vergleichende Studie verschiedener Typen neuronaler Netzwerke für die automatische Klassifizierung von Säuglingslauten. Ziel war es, den leistungsfähigsten Netzwerktyp unter den momentan verbreitetsten Typen zu ermitteln. Hierbei wurde der Einfluss verschiedener architektonischer Konfigurationen innerhalb der Typen berücksichtigt. Die Ergebnisse zeigten, dass Convolutional Neural Networks eine höhere Performance als Recurrent Neural Networks erreichten. Außerdem wurde gezeigt, dass die Wahl der Frequenz- und Zeitaggregationsschicht die wichtigste architektonische Entscheidung ist. Studie (C) war eine detaillierte Untersuchung von Computer Vision-ähnlichen Convolutional Neural Networks für die Klassifizierung von Säuglingslauten. Ziel war es, die wichtigsten architektonischen Eigenschaften zur Steigerung der Erkennungsperformance zu bestimmen. Die Ergebnisse bestätigten die Bedeutung der Aggregationsschicht. Zusätzlich Eigenschaften, die als wichtig identifiziert wurden, waren die Eingangsgröße der vollständig verbundenen Schichten und das akkumulierte rezeptive Feld. Studie (D) war eine Untersuchung zur Kompensation der Klassenimbalance zur Erkennung von Schimpansenrufen in Langzeitaufnahmen. Ziel war es, herauszufinden, welche Kompensationsmethode aus einer Menge ausgewählter Methoden die Performance eines Deep Learning Systems am meisten verbessert. Die Ergebnisse zeigten, dass Spektrogrammentrauschen am effektivsten war, während Methoden zur Kompensation des relativen Ungleichgewichts die Performance entweder gleichhielten oder verringerten.:1. Introduction 2. Foundations in Automatic Recognition of Acoustic Communication 3. State of Research 4. Study (A): Investigation of the Assessment of Infant Vocalizations by Laypersons 5. Study (B): Comparison of Neural Network Types for Automatic Classification of Infant Vocalizations 6. Study (C): Detailed Investigation of CNNs for Automatic Classification of Infant Vocalizations 7. Study (D): Compensating Class Imbalance for Acoustic Chimpanzee Detection With Convolutional Recurrent Neural Networks 8. Conclusion and Collected Discussion 9. Appendix

While the gestural communication of apes is widely recognised as intentional and flexible, their vocal communication still remains considered as mostly genetically determined and emotionally bound. Trying to limit the direct projections of linguistic concepts, that are far from holding a unified view on what constitute human language, this thesis presents a detailed description of the pant-grunt vocalisation usage and development in the chimpanzees (Pan troglodytes schweinfurthii) of the Budongo forest, Uganda. Pant-grunts are one of the most social vocalisations of the chimpanzee vocal repertoire and are always given from a subordinate individual to a dominant. The question of how such a signal is used and develops is critical for our understanding of chimpanzee social and vocal complexity in an ontogenetical and phylogenetical perpective. Results suggest that pant-grunt vocalisations can be used in a flexible way, both in their form and usage within a social group. More specifically, chimpanzees seemed to take into account the number and identity of surrounding individuals before producing these vocalisations. At the acoustic level, pant-grunts seem to be very variable vocalisations that corresponded to different social situations commonly encountered. Grunts are one of the first vocalisations produced by babies but they are not first produced in social contexts. Although some modifications of the social grunts form and usage could not entirely be attributed to maturation only, the role of the mother seemed to be restricted. Her direct influence was perhaps more visible in the rhythmic patterns of chorusing events. Taken together, this thesis suggests that chimpanzee vocalisations are more flexible in their usage, production and acquisition than previously thought and might therefore be more similar to gestural communication.

Les SIV infectant les chimpanzés et les gorilles sont les précurseurs des virus de l'immunodéficience humaine de type 1. Les quatre groupes du VIH-1 sont le résultat de quatre transmissions virales des grands singes à l'Homme. Des méthodes non invasives ont permis d'identifier le réservoir des VIH-1 M et N dans deux communautés de chimpanzés (Ptt) au Cameroun et de montrer que les gorilles (Ggg) sont infectés par un SIV proche des VIH-1 O et P. Si le SIVgor n'a jamais été détecté chez les chimpanzés, la phylogénie montre que les Ptt ont transmis ce virus aux gorilles. Par une méthode pluridisciplinaire, nous avons étudié les caractéristiques de l'infection SIVgor en milieu naturel. Nous avons prospecté 13 sites au Cameroun et 2 en RCA. Au total, 2120 fèces de gorilles et 442 de chimpanzés ont été collectées. L'infection SIVgor a été détectée dans 3 sites Camerounais et les prévalences varient entre 3,2% et 4,6%, résultats plus faibles que ceux retrouvés chez les chimpanzés. Nous avons ensuite montré que plusieurs groupes sociaux de Ggg dont les domaines vitaux se chevauchent sont infectés et que les prévalences SIV dans les groupes peuvent dépasser 25%. Les virus touchant les gorilles du même groupe sont génétiquement proches montrant des liens épidémiologiques. Enfin, un suivi de l'infection réalisé de 2004 à 2009 sur un site a permis de découvrir un foyer d'infection, 2 cas de séroconversions et de retrouver une femelle gorille infectée à 5 ans d'intervalle. Dans ce site, la prévalence SIV est stable et le nombre de femelles infectées est plus important que le nombre de mâles. La structure sociale des gorilles et leur comportement peuvent alors expliquer en partie la répartition et la prévalence du SIVgor, ainsi que les différences avec l'infection chez les chimpanzés.Cette étude multidisciplinaire montre la faisabilité du suivi de l'infection SIV chez les gorilles en milieu naturel. Si le SIVgor est pathogène, le suivi pourra s'avérer essentiel chez cette espèce menacée d'extinction. / SIV infecting chimpanzees and gorillas are the precursors of the Human Immunodeficiency Virus type 1. The four groups of HIV-1 are the results of four different viral transmissions from apes to humans. Using non invasive methods we discovered the reservoir of HIV-1 M and N in two communities of chimpanzees (Ptt) in Cameroon and found that Gorillas (Ggg) are infected by a SIV close to HIV-1 O and P. While SIVgor has not yet been detected in chimpanzees, phylogeny shows that Ptt transmitted this virus to Ggg. Using a multidisciplinary approach, we studied the characteristics of the infection in wild living gorillas. We prospected 13 sites in Cameroon and 3 in CAR. 2120 fecal samples of gorillas and 442 of chimpanzees were collected. SIVgor infection was detected in 3 sites in Cameroon and the prevalence ranges from 3.2% to 4.6%, lower than in chimpanzees. Several social groups of gorillas with overlapping home-ranges were infected and the prevalence within group could exceed 25%. Viruses of the same group are genetically close, showing epidemiologic links. In a follow up study between 2004 and 2009 on one site, we discovered a focus of infection with 2 cases of seroconvertion and we re-sampled one infected female 5 years after. In this site, the prevalence of SIVgor is stable and the number of infected females is higher than the males. The social structure of gorillas and their behavior can partly explain for the repartition and prevalence of SIVgor, as well as the differences with the infection in chimpanzees. This multidisciplinary study proves the feasibility of a follow up study in wild living gorillas. If SIVgor turns out to be pathogenic, a follow up will be essential for this endangered species.

<p>Many of our closest relatives, the primates, are endangered and could be extinct in a near future. To increase the knowledge of non-human primate genomes, and at the same time acquire information on our own genomic evolution, studies using high-throughput technologies are applied, which raises the demand for large amounts of high quality DNA.</p><p>In study I and II, we evaluated the multiple displacement amplification (MDA) technique, a whole genome amplification method, on a wide range of DNA sources, such as blood, hair and semen, by comparing MDA products to genomic DNA as templates for several commonly used genotyping methods. In general, the genotyping success rate from the MDA products was in concordance with the genomic DNA. The quality of sequences of the mitochondrial control region obtained from MDA products from blood and non-invasively collected semen samples was maintained. However, the readable sequence length was shorter for MDA products.</p><p>Few studies have focused on the genetic variation in the nuclear genes of non-human primates. In study III, we discovered 23 new single nucleotide polymorphisms (SNPs) in the Y-chromosome of the chimpanzee. We designed a tag-microarray minisequencing assay for genotyping the SNPs together with 19 SNPs from the literature and 45 SNPs in the mitochondrial DNA. Using the microarray, we were able to analyze the population structure of wild-living chimpanzees.</p><p>In study IV, we established 111 diagnostic nucleotide positions for primate genera determination. We used sequence alignments of the nuclear epsilon globin gene and apolipoprotein B gene to identify positions for determination on the infraorder and Catarrhini subfamily level, respectively, and sequence alignments of the mitochondrial 12S rRNA (MT-RNR1) to identify positions to distinguish between genera. We designed a microarray assay for immobilized minisequencing primers for genotyping these positions to aid in the forensic determination of an unknown sample.</p>

The process of evolution is of both scientific and medical interest. This thesis presents several studies using complete genomic reference sequences, comparative genomic data, and intraspecific diversity data to study the two key processes of evolution: mutation and selection. Large duplications, deletions, inversions, and translocations of DNA contribute to genomic variation both between and within species. Human chromosomes 15 and 17 contain a high percentage of dispersed, recently duplicated sequences. Examination of the relationships between these sequences showed that the majority of all duplications within each chromosome could be linked through core sequences that are prone to duplication. Comparison to orthologous sequences in other mammals allowed a reconstruction of the ancestral state of the human chromosomes, revealing that regions of rearrangement specific to the human lineage are highly enriched in chromosome-specific duplications. Comparison to copy number variation data from other studies also shows that these regions are enriched in current human structural variation. One specific region, the MAPT locus at 17q21.31, known to contain an inversion polymorphism in Europeans, was resequenced completely across both human orientation haplotypes and in chimpanzee and orangutan, revealing complex duplication structures at the inversion breakpoints, with the human region being more complex than chimpanzee or orangutan. Fluorescent in-situ hybridization analysis of human, chimpanzee, and orangutan chromosomes showed inversion polymorphisms of independent origin in all three species, demonstrating that this region has been a hotspot of genomic rearrangement for at least twelve million years. These results reveal a mechanistic relationship between sequence duplication and rearrangement in the great apes. We also generated a draft sequence of the chimpanzee genome and compared it to that of the human. Among other findings, this showed that CpG dinucleotides contribute 25% of all single base mutations, with a rate of mutation ~10-fold that of other bases, and that the male mutation rate in great apes is ~5-6 times the female rate, a higher ratio than had been observed in comparisons of primates and rodents. We detected six regions of probable recent positive selection in humans with a statistical method relying on chimpanzee sequence to control for regional variation in mutation rates. Finally, resequencing of several lines of domestic chicken and comparison to the reference chicken genome identified a number of gene deletions fixed in domestic lines and also several potential selective sweeps. Of particular interest are a missense mutation in TSHR nearly fixed in all domestic chickens and a partial deletion of SH3RF2 fixed in a high growth line. The TSHR mutation may play a role in relaxation of seasonal reproduction. A high-resolution QTL mapping experiment showed that the SH3RF2 deletion is significantly associated with increased growth. This work provides important new insights into the mechanics of evolutionary change at both the single nucleotide and structural level and identifies potential targets of natural and artificial selection in humans and chickens.

Many of our closest relatives, the primates, are endangered and could be extinct in a near future. To increase the knowledge of non-human primate genomes, and at the same time acquire information on our own genomic evolution, studies using high-throughput technologies are applied, which raises the demand for large amounts of high quality DNA. In study I and II, we evaluated the multiple displacement amplification (MDA) technique, a whole genome amplification method, on a wide range of DNA sources, such as blood, hair and semen, by comparing MDA products to genomic DNA as templates for several commonly used genotyping methods. In general, the genotyping success rate from the MDA products was in concordance with the genomic DNA. The quality of sequences of the mitochondrial control region obtained from MDA products from blood and non-invasively collected semen samples was maintained. However, the readable sequence length was shorter for MDA products. Few studies have focused on the genetic variation in the nuclear genes of non-human primates. In study III, we discovered 23 new single nucleotide polymorphisms (SNPs) in the Y-chromosome of the chimpanzee. We designed a tag-microarray minisequencing assay for genotyping the SNPs together with 19 SNPs from the literature and 45 SNPs in the mitochondrial DNA. Using the microarray, we were able to analyze the population structure of wild-living chimpanzees. In study IV, we established 111 diagnostic nucleotide positions for primate genera determination. We used sequence alignments of the nuclear epsilon globin gene and apolipoprotein B gene to identify positions for determination on the infraorder and Catarrhini subfamily level, respectively, and sequence alignments of the mitochondrial 12S rRNA (MT-RNR1) to identify positions to distinguish between genera. We designed a microarray assay for immobilized minisequencing primers for genotyping these positions to aid in the forensic determination of an unknown sample.

Nous avons étudié la latéralité gestuelle intraspécifique de chimpanzés et de gorilles captifs dans des contextes socio-écologiques proches des conditions naturelles de vie. Nous avons montré que les chimpanzés et les gorilles étudiés présentaient un biais populationnel pour la main droite pour la majorité des gestes les plus fréquents de leur répertoire. Par la mise en oeuvre d’une approche multifactorielle, nous avons montré pour la première fois que la latéralité gestuelle intraspécifique de ces primates était influencée par plusieurs facteurs et par leurs interactions: contexte de l’interaction (champs visuels de l’émetteur et du récepteur et contexte émotionnel), caractéristique du geste (modalité sensorielle, utilisation d'un outil de communication, degré de partage et durée du geste) et par certaines composantes sociodémographiques, particulièrement le rang hiérarchique de l’émetteur et son âge dans une moindre mesure. De plus, nous avons comparé la latéralité manuelle des chimpanzés lors de l'utilisation d'outil pour des actions non-communicatives et des gestes intraspécifiques. Notre analyse multifactorielle suggère que l’utilisation d’outil dans les gestes serait plus contrôlée par l'hémisphère cérébral gauche que l’utilisation d’outil dans des actions non-communicatives. Globalement, nos résultats vérifient le modèle de Ghirlanda et collègues (2009) selon lequel les biais de latéralité au niveau populationnel pourraient être expliqués par une stratégie évolutive stable basée sur les interactions intraspécifiques. Nos résultats sont également en accord avec les études mettant en évidence l'utilisation préférentielle de la main droite pour la communication gestuelle des primates non humains et suggérant que la latéralité gestuelle serait un précurseur de la spécialisation hémisphérique gauche pour le langage. En outre, nos résultats confortent l'hypothèse que certaines espèces de primates pourraient avoir un traitement spécifique de l’hémisphère gauche pour les gestes communicatifs distinct de celui des actions manuelles non-communicatives. Du point de l’évolution, nos résultats soulignent l’importance d’étudier en détail la latéralité intraspécifique en considérant des espèces de différents degrés de socialité et en prenant en compte des contextes socioécologiques proches des conditions naturelles ainsi que de multiples facteurs potentiellement influents. / We studied intraspecific gestural laterality of captive chimpanzees and gorillas in real-life social-ecological relevant contexts. We evidenced that chimpanzees (respectively gorillas) exhibited a right-hand bias at the population level for the majority of the most frequent gestures of their specific natural communication repertoire. By designing and applying a multifactorial approach, we showed for the first time that intraspecific gestural laterality of primates was influenced by several factors and their mutual intertwinement: interactional context (visual fields of both signaller and recipient as well as emotional context), gesture characteristic (sensory modality, use of a communication tool, sharing degree, and gesture duration) and by some socio-demographic components in particular signaller’s hierarchical rank, and to a lesser extent signaller’s age. Similarities but also some discrepancies between chimpanzees and gorillas may be related to the lateralization of emotional processing, to communication strategies, and to social selection pressures related to the social structure and dynamics of the study species. Moreover, we compared manual laterality of tool use by chimpanzees in both non-communication actions and intraspecific gestures. Our multifactorial analysis showed that tool-use in gestures appear to be governed more by the left cerebral hemisphere than tool-use in non-communication actions. Our findings support Ghirlanda and colleagues’ (2009) model postulating that population-level bias could be explained by an evolutionary stable strategy based on intraspecific interactions. Our results also agree with previous reports evidencing predominant right-hand use by nonhuman primates for gestural communication and suggesting that gestural laterality would be a precursor of the left-brain specialization for language. Furthermore, our results support the hypothesis that some primate species may have a specific left-cerebral system processing gestures distinct from the cerebral system processing non-communication manual actions. From an evolutionary point of view, our findings emphasize the importance to study intraspecific laterality in detail by considering species varying in their degree of sociality and taking into account real-life social-ecological contexts and multiple potentially influential factors.