Find research datasets worth reusing

180-Bus Medium Voltage Distribution Network Data

The data describes a realistic medium-voltage distribution network with 180 buses, 90 loads (6799 consumers), 5 EV parking lots, 2 Energy Storage Systems, 42 Wind Farms, 33 PV parks, 3 biomasses, and 1 substation supplied by a transformer with 1 MVA and 30 kV. The network characteristics, such as cable resistances, reactances, and thermal limits, are available. Demand, EV demand, and PV/Wind generation profiles are presented for 64 scenarios.

Feature attention graph neural network for estimating brain age and identifying important neural connections in mouse models of genetic risk for Alzheimer's disease

Connectome, traits and behavior data for APOE234 mice. <ul> <li>1. connectome.zip: mouse brain structural connectivity matrices from diffusion MRI.</li> <li>2. FAGNN_Phenotype.csv: a sheet of trait information of mice used in the study.</li> </ul> columns: winding numbers, total distance, normalized NE time, normalized NE distance, normalized NW time, normalized NW distance, normalized SE time, normalized SE distance, normlaized SW time, normalized SW distance, island latency to first entry, island entries, normalized thigmataxis time, and normalized thigmotaxis distance <div>rows: 4 trials for each day from day 1 to day 5 with 1 probing test each at day 5 and day 8</div> <ul> <li>3. mouse_anatomy.csv: brain region information regarding the connectivity matrix.</li> <li>4. behavior.zip: behavioral data for each mouse from Morris Water Maze experiments.</li> </ul>

A root-specific NLR network confers resistance to plant parasitic nematodes - genomic sequences and annotations

Sequence and annotation data associated with "A root-specific NLR network confers resistance to plant parasitic nematodes"

Bipartite plant-pollinator diurnal and nocturnal networks for three mountain systems of the Iberian Peninsula.

Dataset generated for the article "Addition of nocturnal pollinators modifies the structure of pollination" networks" (DOI:10.1038/s41598-023-49944-y).  Diurnal and nocturnal plant-flower visitor networks (hereafter, plant-pollinator networks) were built for each site during the flowering season of 2010 (Picos de Europa) and 2011 (Sierra de Guadarrama and Sierra Nevada). To build the diurnal networks, interactions between plants and floral visitors were recorded along diurnal transects at each site, where all insects contacting the reproductive structures of the flowers were recorded. The sampled area differed between sites from 500 × 250 m in Picos de Europa to 150 × 100 m in Sierra Nevada and 100 × 60 m in Sierra de Guadarrama. These differences were dependent on the small-scale heterogeneity of vegetation. The transects were evenly distributed throughout the study area. The length of the transects varied depending on the size of the study area. Diurnal transects were performed from 10 to 18 h on sunny days with mild wind conditions for pollinator activity. Nocturnal plant-pollinator networks were built for each site by trapping moths using light traps and analysing their pollen loads. Light traps consisted of a UV light surrounded by three white triangular sheets. Moths landing on the sheets were immediately trapped and stored in individual vials with a small piece of tissue and some drops of ethyl acetate. For more details on the methodology see original publication.

Understanding trophic interactions in a warming world by bridging foraging ecology and biomechanics with network science

Background Leaf-cutter ants (Atta spp. and Acromyrmex spp.) are the principal insect pest and a major ecosystem engineer throughout the Neotropics (Leal et al., 2014; Wirth et al., 2003). They harvest plant matter in the surroundings of their colonies to grow a fungus as crop, and in doing so they cut plant matter on an almost industrial scale: about 15 % of the foliar biomass in the Neotropics, or about every sixth leaf, is consumed by leaf-cutter ant colonies (Costa et al., 2008; Fowler et al., 1989; Herz et al., 2007; Wirth et al., 2003), and more than half of all woody species are attacked by them (Cherrett, 1968; Rockwood, 1976). Leaf-cutter ants are perhaps the most voracious and polyphagous herbivorous insects (Lugo et al., 1973; Wirth et al., 2003), and their foraging activity is affected by a variety of environmental conditions, including wind (Alma et al., 2016b), precipitation (Steadman et al., 2020) and barometric pressure (Sujimoto et al., 2020), all of which will be subject to variation due to climate change. Although leaf-cutter foraging is clearly a complex, multi-factorial behaviour, it has at its core a biomechanical interaction between ant consumer and plant food resource: the force the ants can apply must exceed the force required to drag the mandible through the tissue (Püffel, Roces, et al., 2023; Püffel, Walthaus, et al., 2023). The magnitude of the available bite force is determined by worker size, and the magnitude of the minimum required cutting force is determined by structural and mechanical properties of the plant leaf; consumer and resource properties interact. This mechanical competition has resulted in extraordinary adaptations in both the anatomy and physiology of the leaf-cutter ant bite apparatus: their disproportionately large heads are filled to the rim with optimally packed mandible closer muscles (Püffel et al., 2021). Both their muscle stress and size-specific bite forces are among the highest measured for any animal (Püffel, Johnston, et al., 2023; Püffel, Roces, et al., 2023), and their mandibles are close to “ideally sharp” (Püffel, Walthaus, et al., 2023). As a result, the vast majority of worker sizes can cut the majority of tropical leafs; without these adaptations, and a bite performance commensurate with their body size, only the largest workers would be able to perform this crucial mechanical task (Püffel, Roces, et al., 2023). How will a warming climate affect resource accessibility for the leaf-cutters? Temperature increases have various implications for the trophic interactions of ants, including altered search behaviour (Frizzi, 2018), and foraging site selection (Spicer et al., 2017; Traniello et al., 1984). An increase in average temperatures can also drive body size decreases in insects (Tseng et al., 2018), including ants (Molet et al., 2017)<a href="https://www.zotero.org/google-docs/?broken=QmLD4C">,</a> concomitantly reducing their available bite force (Püffel, Roces, et al., 2023; Rühr et al., 2022). Since leaf-cutter mandibles are so sharp that they already cut with a force close to the minimum dictated by cutting mechanics, the force required to cut leaves will likely be unaffected (Püffel, Walthaus, et al., 2023), and any change in body size will therefore only significantly impact bite forces. Because the relationship between bite forces and body size in the leaf-cutter is well understood mechanistically (Püffel, Roces, et al., 2023), it is possible to predict how these changes will impact trophic networks. A very rough estimate of the change in network structure serves to illustrate how network science can integrate biomechanics and foraging ecology to study the effect of climate change on trophic interactions. To demonstrate the potential of network science to integrate biomechanical and foraging data within the context of climate change, we constructed and analysed hypothetical plant-ant networks across six hypothetical temperatures.   Datasets and methods All analysis was performed in R version 4.3.1 (R Core Team, 2023), and data processed reproducibly via the ‘tidyverse’ package (Wickham et al., 2019). We compiled two datasets and some additional contextual information. Leaf-cutter ant biomass (a proxy for body size) and bite force data were taken from Püffel et al. (2023) for 248 individual ants across three colonies. Required cutting forces for 1197 individual plants representing 868 taxa available to leaf-cutter ants were taken from Onoda et al. (2011). Insect temperature-body size relationships were taken from Tseng et al. (2018); specifically, a body size decrease of 1.56 % per degree Celsius increase for museum specimens, to represent gradual long-term change. Based on these data, edgelists (i.e., pairwise lists of consumers and resources) were generated for ants and plants in which binary interaction weights were applied; where bite forces exceeded the force required to cut leaves, a weighting of 1 was given, and 0 otherwise. This edgelist was then replicated for incremental increases of 1 °C up to a 5 °C increase by adjusting bite forces based on incremental body size decreases of 1.56 %. In order to estimate the change of bite force with body mass, we used direct bite force measurements from Püffel et al. (2023), which suggest that maximum bite force in Atta vollenweideri varies with body mass m as T ~ m^0.9. Thus, if body size decreases by a factor of 0.9844 (i.e., 1.56 % decrease) with every degree Celsius temperature increase, then the maximum bite force decreases by a factor of 0.98440.9. Consequently, adjusted bite forces were calculated, and new binary edgelist weightings generated based on whether the adjusted bite force was greater than the required cutting force. Bipartite networks were constructed with consumer nodes and resource nodes representing the three ant colonies and the 868 plant taxa, respectively. All six networks were visualised using ‘ggnetwork’ (Briatte, 2021) via ‘igraph’ (Csardi & Nepusz, 2006) in a single network diagram to highlight persistence of links across temperatures using scaled red colours. Network metrics, specifically consumer degree (the number of plants ants were deemed able to interact with) and generality (the total range of plants accessible across all ants), were generated via the ‘bipartite’ package (Dormann et al., 2008) and visually compared via ‘ggplot2’ (Wickham, 2016).

Data associated to the paper "Reduced basis surrogates for quantum spin systems based on tensor networks"

Within the reduced basis methods approach, an effective low-dimensional subspace of a quantum many-body Hilbert space is constructed in order to investigate, e.g., the ground-state phase diagram. The basis of this subspace is built from solutions of snapshots, i.e., ground states corresponding to particular and well-chosen parameter values. Here, we show how a greedy strategy to assemble the reduced basis and thus to select the parameter points can be implemented based on matrix-product-state calculations. Once the reduced basis has been obtained, observables required for the computation of phase diagrams can be computed with a computational complexity independent of the underlying Hilbert space for any parameter value. We illustrate the efficiency and accuracy of this approach for different one-dimensional quantum spin-1 models, including anisotropic as well as biquadratic exchange interactions, leading to rich quantum phase diagrams.

opencc-zeroDec 2023View details →

Reduced gray matter volume in the default-mode network associated with insulin resistance

Insulin resistance may lead to structural and functional abnormalities of the human brain. However, the mechanism by which insulin resistance impairs the brain remains elusive. In this study, we used two large neuroimaging databases to investigate the brain regions where insulin resistance was associated with the gray matter volume and to examine the resting-state functional connectivity between these brain regions and each hypothalamic nucleus. Insulin resistance was associated with reduced gray matter volume in the regions of the default-mode and limbic networks in the cerebral cortex in older adults. Resting-state functional connectivity was prominent between these networks and the paraventricular nucleus of the hypothalamus, a hypothalamic interface connecting functionally with the cerebral cortex. Furthermore, we found a significant correlation in these networks between insulin resistance-related gray matter volume reduction and network paraventricular nucleus of the hypothalamus resting-state functional connectivity. These results suggest that insulin resistance-related gray matter volume reduction in the default-mode and limbic networks emerged through metabolic homeostasis mechanisms in the hypothalamus.

opencc-zeroDec 2023View details →

Women's subsistence networks scaffold cultural transmission among BaYaka foragers in the Congo Basin

In hunter-gatherer societies, women's subsistence activities are crucial for food provisioning and children's social learning, but they are understudied relative to men's activities. To better understand the structure of women's foraging networks, we present data of 230 days focal-follows in a BaYaka community. To analyze these data, we develop a stochastic block model for repeat observations with uneven sampling. We find that women's subsistence networks are characterized by cooperation between kin, gender homophily, and mixed age-group composition. During early-childhood, individuals preferentially co-forage with adult kin, but those in middle-childhood and adolescence are likely to co-forage with non-kin peers, providing opportunities for horizontal learning. By quantifying the probability of co-foraging ties across age-classes and relatedness levels, our findings provide insights into the scope for social learning during women's subsistence activities in a real-world foraging population and provide ground-truth values for key parameters used in formal models of cumulative cultural change.

NPM ecology network dataset

The dataset contains the dataset of the NPM ecology network. <ul> <li><a href="../api/records/10317762/draft/files/npm_openrank_top100.csv/content" target="_blank" rel="noopener noreferrer">npm_openrank_top100.csv</a> file contains the result of top 100 OpenRank repositories in npm ecology.</li> <li><a href="../api/records/10317762/draft/files/npm_ecology.json/content" target="_blank" rel="noopener noreferrer">npm_ecology.json</a> file contains the original network of npm ecology open source repositories, use `CALL apoc.import.json('npm_ecology.json')` to import the network into Neo4j database.</li> <li><a href="../api/records/10317762/draft/files/npm_ecology.dump/content" target="_blank" rel="noopener noreferrer">npm_ecology.dump</a> file is generated by Neo4j dump command, use `bin/neo4j-admin load --from=npm_ecology.dump --database=npmecology --force` to load the data into Neo4j database. This will be much faster than loading from JSON file.</li> </ul>

Anomalous and Chern topological waves in hyperbolic networks

Data and codes for the paper "Anomalous and Chern topological waves in hyperbolic networks", by Q. Chen et al.

Data for Aligning rTWT with 802.1Qbv. A Network Calculus Approach

<div>This project stores Matlab scripts to solve the equations from <a href="https://dl.acm.org/doi/10.1145/3565287.3617606" target="_blank" rel="nofollow noreferrer noopener">Aligning rTWT with 802.1Qbv: a Network Calculus Approach </a></div> <div>This information can also be found in the next link, please check it for updates: <a href="https://gitlab.netcom.it.uc3m.es/predict-6g/twtscheduler" target="_blank" rel="noopener">PREDICT 6G / Aligning rTWT with 802.1Qbv. A Network Calculus Approach · GitLab (uc3m.es)</a>. </div> <div> </div> <div>This work has been partially funded by the European Commission Horizon Europe SNS JU <a href="https://predict-6g.eu/" target="_blank" rel="nofollow noreferrer noopener">PREDICT-6G</a> (GA 101095890) Project and the Spanish Ministry of Economic Affairs and Digital Transformation and the European Union-NextGenerationEU through the UNICO 5G I+D <a href="https://unica6g.it.uc3m.es/6g-edgedt/" target="_blank" rel="nofollow noreferrer noopener">6G-EDGEDT</a> and 6G-<a href="https://unica6g.it.uc3m.es/6g-datadriven/" target="_blank" rel="nofollow noreferrer noopener">DATADRIVEN</a>.  </div> <div> </div>

opencc-by-3.0-usSep 2023View details →

Data for "Synapses without Tension Fail to Fire in an In Vitro Network of Hippocampal Neurons."

Neurons in the brain communicate with each other at their synapses. It has long been understood that this communication occurs through biochemical processes. Here, we reveal a previously unrecognized paradigm wherein mechanical tension in neurons is essential for communication. Using in vitro rat hippocampal neurons, we find that (1) neurons become tout/tensed after forming synapses resulting in a contractile neural network, and (2) without this contractility, neurons fail to fire. To measure time evolution of network contractility in 3D (not 2D) extracellular matrix, we developed an ultra-sensitive force sensor with 1 nN resolution. We employed Multi-Electrode Array (MEA) and iGluSnFR, a glutamate sensor, to quantify neuronal firing at the network and at the single synapse scale, respectively. When neuron contractility is relaxed, both techniques show significantly reduced firing. Firing resumes when contractility is restored. This discovery highlights the essential contribution of neural contractility in fundamental brain functions and has implications for our understanding of neuronal physiology.

Data from: Urbanisation and agricultural intensification modulate plant-pollinator network structure and robustness

Land use change is a major pressure on pollinator abundance, diversity, and plant-pollinator interactions. Far less is known about how land use alters the structure of plant-pollinator networks and their robustness to plant-pollinator coextinctions. We analyzed the structure of plant-pollinator networks sampled in 12 landscapes along an urbanisation and agricultural intensity gradient, from early spring to late summer 2021, and used a stochastic coextinction model to correlate plant-pollinator coextinction risk with network structure (species and network-level metrics) and landscape context. Networks in intensively managed (i.e. agricultural and urban) landscapes had a lower risk of initiating a coextinction cascade, while networks in less-intensively managed landscapes may be less robust. Network structure modulated the frequency and severity of coextinctions and species loss, while the strength of species interactions increased robustness. Urban networks were more species-rich and symmetrical due to the high diversity of ornamental plants, while intensively managed agricultural landscapes had smaller, more tightly connected, and nested networks. Network structure modulated the frequency of extinctions, which was decreased by greater linkage density, interaction asymmetry, and interaction dependence in the networks, while once an extinction occurred, nestedness and linkage density propagated the degree of the coextinction cascade and species loss. At the species level, species strength was inversely correlated with extinction risk, implying that generalist species with a high number of interactions with specialists had the lowest extinction risk.

Data from: Jeans and language: social networks and reproductive success are associated with the adoption of outgroup norms

Traditional norms of human societies in rural China may have changed due to population expansion, rapid development of the tourism economy and globalization since the 1990s; people from different ethnic groups might adopt cultural traits from outside their group or lose their own culture at different rates. Human behavioural ecology can help to explain adoption of outgroup cultural values. We compared the adoption of four cultural values, specifically speaking outgroup languages/mother tongue and wearing jeans, in two co-residing ethnic groups, the Mosuo and Han. Both groups are learning outgroup traits, including each other's languages through contact in economic activities, education and social networks, but only the Mosuo are starting to lose their own language. Males are more likely to adopt outgroup values than females in both groups. Females of the two groups are no different in speaking Mandarin and wearing jeans, whereas males do differ, with Mosuo males being keener to adopt them than Han males. The reason might be that Mosuo men experience more reproductive competition over mates than others, as Mosuo men have larger reproductive skews than others. Moreover, Mosuo men but not others gain fitness benefits from the adoption of Mandarin (they start reproducing earlier than non-speakers). This article is part of the theme issue 'Social norm change: from evolution to policy intervention'.

opencc-by-4.0Dec 2022View details →

NANCY SNS-JU Project - Fronthaul network of fixed topology Usage Scenario - Dataset 1

In the context of the NANCY project (https://nancy-project.eu/), this Dataset provides input data for the development of the B-RAN and attacks models for the NANCY framework, to model training and model inference functions. The data collected plays the role of ML algorithm-specific data preparation. The dataset contains time-series, collected transmitting a video content through the Italtel "VTU - video streaming and transcoding application", that can convert audio and video streams from one format to another, at multiple encodings schemes, changing resolution, bitrate, and video parameters. The data collected are related to the observation of some of the resources involved in the Usage Scenario: “Fronthaul network of fixed topology – Direct Connectivity”. In the Italtel Italian in-lab testbed, a MEC assisted 5G network scenario with a video streaming application for generating traffic is provided. Two different scenarios were set-up, related to downstream and upstream video flows. The variety of collected features ranges from radio front-end metrics to physical server operating system and network function metrics. The dataset consists of raw network traffic and extracted flow-based data captured in separate files. Each file captured is associated to a 10min video streaming of the “Big Buck Bunny” video. This video was transmitted on two different bands, N3 and N78, with different resolutions, 480p, 720p, 1080p; both in uplink (UL) and in downlink (DL); the type of protocol monitored is “HTTP protocol”; in case of N78 band, data related to the resource usage were also captured, for a total of more that 100 data files. The collected dataset is representative resource-intensive video traffic that has the greatest impact on 5G/B5G network planning and provisioning. The video streaming dataset includes data directly measured while watching the video on the mobile devices and data directly measured while generating downstream video stream traversing the gNB (i.e., downstream scenario), and vice versa (i.e., upstream scenario). In each experiment, we fixed the location of the UE and the gNB. The NANCY project has received funding from the Smart Networks and Services Joint Undertaking (SNS JU) under the European Union's Horizon Europe research and innovation programme under Grant Agreement No 101096456.

Effect of green infrastructure on restoration of pollination networks and plant performance in semi-natural dry grasslands across Europe

Agricultural intensification, afforestation and land abandonment are major drivers of biodiversity loss in semi-natural grasslands across Europe. Reversing these losses requires the reinstatement of plant-animal interactions such as pollination. Here we assessed the differences in species composition and patterns of plant-pollinator interactions in ancient and restored grasslands and how these patterns are influenced by landscape connectivity, across three European regions (Belgium, Germany and Sweden). We evaluated the differences in pollinator community assemblage, abundance, and interaction network structure between 24 ancient and restored grasslands. We then assessed the effect of surrounding landscape functional connectivity (i.e. green infrastructure, GI) on these variables and tested possible consequences on the reproduction of two model plants, Lotus corniculatus and Salvia pratensis. Neither pollinator richness nor species composition differed between ancient and restored grasslands. A high turnover of interactions across grasslands was detected but was mainly due to replacement of pollinator and plant species. The impact of grassland restoration was consistent across various pollinator functional groups, whereas the surrounding GI had differential effects. Notably, bees, butterflies, beetles, and dipterans (excluding hoverflies) exhibited the most significant responses to GI variations. Interestingly, networks in restored grasslands were more specialised (i.e. less functionally redundant) than in ancient ones and also showed a higher number of insect visits to habitat-generalist plant species. Landscape connectivity had a similar effect, with habitat-specialist plant species receiving fewer visits at higher GI values. Fruit set in S. pratensis and L. corniculatus was unaffected by grassland type or GI. However, the fruit set in the specialist S. pratensis increased with the number of pollinator visits, indicating a positive correlation between pollinator activity and reproductive success in this particular species. Synthesis and applications. Our findings provide evidence of the necessity to enhance ecosystem functions while avoiding biotic homogenization. Restoration programs should aim at increasing landscape connectivity which influences plant communities, pollinator assemblages, and their interaction patterns. To avoid generalist species taking over from specialists in restored grasslands, we suggest reinforcing the presence of specialist species in the latter, for instance by means of introductions, as well as increasing the connectivity to source populations.

opencc-by-4.0Jun 2023View details →

Neural network ensembles and FEFF spectra for multi-modal small molecule chemical motif prediction

Data <ul> <li>22-12-05-data: original molecular XANES data created from <a href="https://doi.org/10.1103/PhysRevResearch.5.013180">Ghose et al.</a></li> <li>23-04-26-ml-data: machine learning-ready data which is prepared in the format required by <a href="https://github.com/matthewcarbone/Crescendo">Crescendo</a>.</li> <li>23-05-03-hp: hyper-parameter tuning results from 23-04-26-ml-data.</li> <li>23-05-05-ensembles: ensemble results from 23-04-26-ml-data.</li> <li>23-05-11-ml-data-CUTOFF8: a special machine learning-ready dataset constructed by a unique partitioning: only molecules with less than or equal to 8 atoms/molecule are used for training/validation, the rest are used for testing.</li> <li>23-12-06_torch_models: torch only models which can be easily used independently of our ML helper repository, Crescendo. Instead, it can be used with a few lines of code found in multimodal_molecules/core.py, in our <a href="https://github.com/AI-multimodal/multimodal-molecules">GitHub respository</a>.</li> </ul> Funding This research is based upon work supported by the U.S. Department of Energy, Office of Science, Office Basic Energy Sciences, under Award Number FWP PS-030. This research also used theory and computational resources of the Center for Functional Nanomaterials, which is a U.S. Department of Energy Office of Science User Facility, and the Scientific Data and Computing Center, a component of the Computational Science Initiative, at Brookhaven National Laboratory under Contract No. DE-SC0012704.

Flower-visitor and pollen-load data provide complementary insight into species and individual network roles

Most animal pollination results from plant-insect interactions, but how we perceive these interactions may differ with the sampling method adopted. The two most common methods are observations of visits by pollinators to plants and observations of pollen loads carried by insects. Each method could favour the detection of different species and interactions, and pollen load observations typically reveal more interactions per individual insect than visit observations. Moreover, while observations concern plant and insect individuals, networks are frequently analysed at the level of species. Although networks constructed using visitation and pollen-load data have occasionally been compared in relatively specialised, bee-dominated systems, it is not known how sampling methodology will affect our perception of how species (and individuals within species) interact in a more generalist system. Here we use a Diptera-dominated high-Arctic plant--insect community to explore how sampling approach shapes several measures of species' interactions (focusing on specialisation), and what we can learn about how the interactions of individuals relate to those of species. We found that species degrees, interaction strengths, and species motif roles were significantly correlated across the two method-specific versions of the network. However, absolute differences in degrees and motif roles were greater than could be explained by the greater number of interactions per individual provided by the pollen-load data. Thus, despite the correlations between species roles in networks built using visitation and pollen-load data, we infer that these two perspectives yield fundamentally different summaries of the ways species fit into their communities. Further, individuals' roles generally predicted the species' overall role, but high variability among individuals means that species' roles cannot be used to predict those of particular individuals. These findings emphasize the importance of adopting a dual perspective on bipartite networks, as based on the different information inherent in insect visits and pollen loads.

opencc-zeroFeb 2024View details →

Social network shrinking is explained by active and passive effects but not increasing selectivity with age in wild macaques

Evidence of social disengagement, network narrowing, and social selectivity with advancing age in several non-human animals challenges our understanding of the causes of social ageing. Natural animal populations are needed to test whether social ageing and selectivity occur under natural predation and extrinsic mortality pressures, and longitudinal studies are particularly valuable to disentangle the contribution of within-individual ageing from the demographic processes that shape social ageing at the population level. Data on wild Assamese macaques (Macaca assamensis) were collected between 2013 and 2020 at the Phu Khieo Wildlife Sanctuary, Thailand. We investigated the social behaviour of 61 adult females observed for 13,270 hours to test several mechanistic hypotheses of social ageing and evaluated the consistency between patterns from mixed-longitudinal and within-individual analyses. With advancing age, females reduced the size of their social network, which could not be explained by an overall increase in the time spent alone, but by an age-related decline in mostly active, but also passive, behaviour, best demonstrated by within-individual analyses. A selective tendency to approach preferred partners was maintained into old age but did not increase. Our results contribute to our understanding of the driver of social ageing in natural animal populations and suggest that social disengagement and selectivity follow independent trajectories during ageing.