Aim: Positive relationships between island size and species richness have been found for oceanic and non-oceanic islands. However, the effect of isolation, also predicted by island-biogeographical theory, seems challenging to test. Species-specific mechanisms drive the perception of distance and space for animals, and therefore isolation patterns are masked when described at the community level. We used epiphytic plants (canopy islands) and their spider communities to study how functional group and species-specific use of space modifies the effect of island-level predictors on species richness. Location: A coffee plantation near Xalapa, Mexico. Taxon: Spiders (Araneae), vascular epiphytes (Bromeliaceae, Piperaceae, Orchidaceae, Araceae, Pteridophyta). Methods: We collected canopy islands from three trees and recorded their three-dimensional position. Afterwards, we collected all epiphyte-dwelling spiders. We used Generalized Linear Models (GLMs) to analyse the effect of island size and isolation on the species richness of spider communities and community subgroups (guilds, mobility groups and families). Isolation was calculated using two metrics (aerial and walking distances) associated with different ways of mobility in the canopy (bridging and wandering respectively). We evaluated species-level associations with island features and tested for sample-size effects on GLMs using null-model analysis of ecological networks. Results: Although unimportant at the community level, isolation was more important than island size for some spider subgroups, while having negligible effects for others. Species-specific responses to isolation and island size were responsible for the variability in species richness patterns observed for broader subgroups (families, guilds and mobility groups). Specific responses of mobility groups to isolation metrics were in line with two ways of moving between islands; one using tree branches as pathways (walking isolation) and the other using silk threads to connect points directly across the air (aerial isolation). Main conclusions: Spatial patterns of epiphyte-dwelling spiders were explained better when functional traits were considered, and the effect of isolation became apparent only when analysed per functional group, whereas it was masked at the community level. This supports the hypothesis that hunting behaviour and mobility affect the specific perception of space and distance even at the small spatial scale of canopy islands.