This dataset contains information on plant–pollinator interactions collected in Cabrera during the 2022–2023 field campaigns. Data were obtained through direct observations and automatic camera recordings across different habitats. Each row corresponds to a sampling event (census of pollinator visits to flowering plants)., This data was collected as a part of the IslandLife project (Ref.~101054177), funded by an ERC Advanced Grant to Anna Traveset, and is also partially supported by the POLINISLAS project (Ref.~2713/2021) funded by the Organismo de Parques Nacionales to Anna Traveset., Peer reviewed

Metabarcoding results for root associated fungi in roots of plant species in five small islands. Sequencing was done using Illumina MySeq 300., European Research Council
IslandLife 101054177;
Ministerio de Ciencia, Innovación y Universidades
MCIN/AEI/10.13039/501100011033, Peer reviewed

Seed dispersal by animals supports plant regeneration and ecosystem resilience, yet conservation success is still commonly evaluated using species richness rather than the organisation of ecological interactions. This is particularly problematic in human-modified ecosystems, where management interventions can increase diversity without restoring historical interaction structures or functional stability. We compared seed dispersal networks on two climatically and biogeographically similar oceanic islands in the Seychelles archipelago with contrasting management trajectories: one strictly protected and under historically oriented management, aimed at limiting introductions and maintaining historically informed community structure, and one anthropogenically diversified through long-term human occupation and species introductions. Using a multi-metric ecological network approach integrating modularity-based roles, centrality indices and motif analyses, we quantified how plants and dispersers contribute to network structure under each management context. Anthropogenic diversification increased interaction diversity and the number of dispersal events but resulted in a less cohesive network dominated by peripheral species and a narrow set of structurally central dispersers. In contrast, the historically oriented management island exhibited lower interaction diversity but greater network cohesion, with strong redundancy in plant-level dispersal pathways. Shared species occupied different structural roles across islands, indicating strong context dependence in their contribution to seed dispersal networks. Motif analyses revealed contrasting interaction niche structures across management contexts. In the anthropogenically diversified island, dispersal relied on a small set of generalist dispersers, suggesting potential vulnerability to their loss. In the historically oriented management island, dispersal was spread across more species, indicating greater redundancy. These differences imply distinct but predictable vulnerabilities arising from alternative management pathways. Our findings show that management strategies can create fundamentally different seed dispersal regimes, even under similar environmental conditions. Anthropogenic diversification reshapes network organisation into distinct structural configurations, rather than consistently enhancing overall cohesion. Incorporating interaction structure into conservation planning can help prioritise key dispersers and plants, anticipate hidden vulnerabilities, and guide management actions that better sustain seed dispersal processes in human-modified island ecosystems., European Research Council: 101054177, Agencia Estatal de Investigación: CEX2021-001198., With funding from the Spanish government through the "Maria de Maeztu Centre of Excellence" accreditation (CEX2021-001198)., Peer reviewed

All versions are available at DOI 10.5281/zenodo.17953353, Metabarcoding results for root associated fungi in roots of plant species in five small islands. Sequencing was done using Illumina MySeq 300.

The raw number of reads for each ASV in each plant species is provided.

Information on the island in which each plant species was sampled is included in the metadata sheet: BG = Berlenga Grande, DP = Dapnhe, FR = Fregate, NR = Na Redona, MC = Montaña Clara., European Research Council
IslandLife 101054177
Ministerio de Ciencia, Innovación y Universidades
MCIN/AEI/10.13039/501100011033, Peer reviewed

These data were collected during fieldwork in the Scalesia pedunculata forest on Santa Cruz Island (Galapagos, Ecuador), near the twin volcanic sinkholes “Los Gemelos” (550–600 m a.s.l.; 00°37′20″ S, 90°23′00″ W), and are associated with manuscript entitled: "The forest within the forest: restoration of Galapagos host-epiphyte networks". Between October - December 2024, twenty 10 × 10 m plots were sampled, organized as 10 pairs representing two treatments: invaded (no management) vs. restored forest, where invasive species (Rubus niveus and Cestrum auriculatum) are regularly removed. Paired plots were located 20 m apart, and pairs were at least 50 m from the nearest pair., Sampling followed the grid-point intercept method with 36 points per plot. At each grid point, all potential host plants (>1 m height or >5 cm diameter at breast height, DBH) were recorded, and the occurrence of associated vascular epiphytes (including mosses, ferns, angiosperms, vines, and one hemi-parasite) was assessed. For each host plant, species, DBH, proportion of moss cover, and all associated epiphytes were recorded., This repository contains 7 data files (provided as compressed archive DATA.zip) and 8 R scripts (provided in the compressed archive SCRIPTS.zip). All analyses were conducted in R version 4.4.2 (2024-10-31). Required packages are specified in each script. All data files should be placed in the working directory, and the incidence matrices must be stored in the folder /data_plot_new within the working directory., To facilitate reproducibility, detailed package and dependency version information is provided in the file:
R_SessionInfo_packages.docx. This file contains the output of sessioninfo::session_info() for each of the eight R scripts provided in this repository, Ecosystem restoration is increasingly implemented to mitigate biodiversity loss. Epiphytes can represent up to a third of tropical plant diversity and perform key ecosystem functions. Yet, they are highly vulnerable to deforestation and their recovery under restoration remains understudied. We evaluated the effect of invasive plant removal (Rubus niveus and Cestrum auriculatum) on facultative host-epiphyte interactions in the endemic and endangered Scalesia pedunculata forest on Santa Cruz Island (Galapagos). Using 20 paired 10 × 10 m plots (invaded vs. 11 years of restoration), we compared host–epiphyte network structure, epiphyte diversity, and the main factors influencing epiphyte richness. While network descriptors did not differ between treatments, restored plots supported higher epiphyte richness. Epiphyte richness per host increased with moss cover and host tree diameter (DBH), both key facilitators of epiphyte colonization, and was significantly higher under restoration, indicating improved host suitability. The endemic and threatened S. pedunculata was identified as a keystone host, with the highest species strength and degree. However, its lack of regeneration in invaded plots threatens long-term epiphyte persistence, as invasive hosts offer minimal epiphyte support. Our findings demonstrate that restoration enhanced epiphyte richness by fostering suitable hosts, highlighting the relevance of integrating biotic interactions into restoration planning and monitoring., Peer reviewed