Thirty-three of the plots were in forests, while seven contained streets, paths or gardens; together they covered approximately 21.7 hectares. During surveys conducted from April to October 2025, researchers visually scanned living trees with diameters at breast height of at least 10 centimetres, from trunk base to outer crown. Geotagged photographs documented each accidental epiphyte, its host, attachment height, substrate and evidence of reproduction.

Accidental epiphytes were defined as epiphytic individuals of predominantly terrestrial species without obvious specialisation for epiphytic growth. The analysis was observational and covered only temperate coastal Europe, so it identified associations rather than experimentally establishing causes. Host-tree size emerged as one of the clearest local filters.

Bayesian mixed-effects models estimated that the probability of a forest tree serving as a host increased from less than 3% among trees 10–30 centimetres in diameter to 27% among trees exceeding 90 centimetres. In non-forest plots, nearly 50% of trees at least 60 centimetres wide and approximately 70% of those at least 90 centimetres wide were predicted to host accidental epiphytes. Weighting these predictions by the observed distribution of tree sizes produced an estimated host-encounter probability of 2% in forests and 24% in non-forest settings.

Abundance also rose sharply with diameter: relative to the smallest class, the estimated abundance was 3.4 times greater at 30–60 centimetres, 3.9 times greater at 60–90 centimetres and 25.5 times greater above 90 centimetres, with a 95% credible interval of 9.2–70.7 for the largest comparison. Tree identity and physical form further shaped the patterns. Willows, oaks, Phoenix palms, black alder and beech were among the most frequently recorded hosts, although the findings do not demonstrate strict host specificity.

Polypody ferns occurred especially often on oak, alder and beech; navelwort was associated with oak; elder and bedstraw with willow; and sow-thistle with Phoenix palms. Polypodies usually occupied stems and branches within moss mats, while many other taxa grew on pollard heads, branch forks, crotches, knotholes and cut frond bases. These sites can retain water, organic material and soil, offering anchorage and rooting opportunities for plants that otherwise grow on the ground.

The authors interpret the pattern as host bias arising from architecture and microhabitat formation: large forks, cavities, moss mats and litter accumulations may matter more than taxonomic identity alone, although bark properties and species-specific traits could also influence suitability. Human-modified landscapes contained more accidental epiphytes than forests. Standardised richness, expressed as the expected number of species per 10 surveyed trees, was generally greater in gardens and along willow-lined roads or paths than in nearby forests.

The same contrast appeared in the structured field survey, reducing the likelihood that it arose solely from uneven community-science reporting. One plausible structural explanation is that streets and gardens often retain large, solitary or managed trees with forks, cavities and pruned crowns that collect litter and moisture. These trees can function as habitat islands where disturbance, competition or other constraints limit growth on the ground.

Urban and rural species pools may also differ, while transport corridors can increase the supply of propagules. Nevertheless, the observational design could not distinguish whether the non-forest pattern primarily reflects more propagules, better hosts, a richer candidate species pool or a combination of these factors. Latitude altered the identities of canopy plants more clearly than it produced a consistent gradient in richness.

Taxonomic composition differed among latitude bands, with latitude explaining 20% of compositional variation in a permutational multivariate analysis. Taxon replacement accounted for 91.1% of the Jaccard dissimilarity between bands, whereas nestedness accounted for 8.9%. Assemblages along the Bay of Biscay were dominated by polypody ferns and ivy, while additional genera occurred farther north and in the southernmost band near the French-Spanish border.

Abundance and richness were often decoupled: some plots contained many individuals but few species, while other species-rich plots had comparatively low abundance. Changing regional species pools may help explain the turnover, but the study did not explicitly test that possibility, so the results cannot attribute the geographic differences to latitude alone. Species-specific models showed that accidental-epiphyte records were non-randomly positioned within the climatic ranges available to their taxa.

Minimum temperatures during the coldest month had positive posterior mean effects for all 19 taxa, indicating that arboreal records generally occurred toward milder winter conditions within each species’ available climatic space. Credible intervals excluded zero for 18 taxa; only fig was uncertain, with a coefficient of +0.5 and a 95% credible interval of −1.4 to 2.3. Mean temperature of the warmest quarter showed a complementary pattern, with negative posterior means for 18 taxa and credible intervals excluding zero for 17.

These are climatic associations, not evidence that experimentally changing temperature would cause accidental epiphytism to change. Moisture and wind produced weaker and more variable signals. Common polypody, blueberry and wood sorrel showed positive associations with the aridity index, placing their records toward wetter portions of their available climatic ranges, although the interval for wood sorrel slightly included zero.

Elder, sycamore maple and rowan had negative point estimates for aridity, but their intervals also slightly overlapped zero. Most taxa had uncertain wind effects. Navelwort and southern polypody were positioned toward less windy parts of their ranges, while elder had a positive wind association; wind-dispersed and animal-dispersed taxa appeared on both sides of zero.

Wind may assist propagule delivery while also increasing drying or dislodgement, so its net association can differ among taxa and landscapes. Local microhabitats may also mediate climatic exposure: moss mats, canopy soil and accumulated litter can retain water even when regional moisture effects are weak. Several limitations constrain the conclusions.

The field sample was restricted to temperate coastal Europe, and many iNaturalist observations were concentrated along the sailing transect, potentially overrepresenting coastal conditions and uneven observer effort. Incidence data were sparse for the largest diameter classes, producing wide credible intervals and occasionally high posterior means; the authors therefore presented model estimates alongside raw observations. The climate analysis included taxa with more than 20 spatially thinned records and used four coarse predictors.

It could not directly measure propagule arrival, local canopy temperature, substrate moisture, germination or survival, nor establish whether differences among settings arose through one mechanism or several. Suggested next steps include geographically balanced sampling, diaspore traps, repeated monitoring of marked plants and seedlings, and experiments designed to separate arrival, retention, germination and survival. Taken together, the coastal survey supports a hierarchical model in which accidental epiphytism arises when several filters align: a regional pool supplies potential plants, propagules reach a suitable host, tree architecture provides a stable attachment site, habitat context affects establishment, and local and macroclimatic conditions determine whether growth can persist.

The study provides a quantitative landscape-level estimate of host-encounter probability across contrasting settings and shows that accidental epiphytes can involve dozens of species rather than a few unusual outliers. By documenting facultative and often transient stages of terrestrial plant establishment on trees, these organisms may also help researchers investigate how terrestrial plants repeatedly enter canopy niches and why that transition becomes obligate in some lineages while remaining accidental or facultative in others.

Subject of Research: Ecological controls on accidental epiphytism across temperate coastal Europe

Article Title: Multiscale drivers of accidental epiphytism in temperate Europe

Article References: Hoensbroech, M., de Rivera, O. R., & Zotz, G. (2026). Multiscale drivers of accidental epiphytism in temperate Europe. Frontiers of Biogeography, 19, Article e207331. https://doi.org/10.21425/fob.19.207331

Image Credits: AI Generated

DOI: 10.21425/fob.19.207331

Keywords: accidental epiphytes, host-tree size, temperate forests, urban ecology, plant establishment, climate filtering, iNaturalist, coastal Europe, canopy ecology, host bias, Multiscale, drivers