Tree seed dispersal among forest fragments: II. Dispersal abilities and biogeographical controls

Abstract

AimTo investigate the medium to long‐distance dispersal abilities of temperate hardwood tree species and the ecological controls on dispersal, including distance to and connectivity with seed sources, seed source strength, and species dispersal mechanism and seed size.LocationA fragmented forest system in the Long Point region, Southern Ontario.MethodsPine plantations were the `seed traps' in which seedlings of hardwood species were enumerated to indicate past dispersal events. The influence of distance to seed sources, dispersal mechanism and seed mass on the probability of dispersal to plantations were evaluated using logistic analysis. Regression analysis was used to determine the effect of seed source strength (mature tree abundance within 150 m of plantations) on seedling density in plantations. Connectivity was assessed by comparing the strength of correlations between seedling abundance in plantations and the abundance of mature trees around plantations in connected vs. unconnected source areas.ResultsSeedling presence in plantations decreased significantly with distance from the nearest potential seed source for species grouped according to dispersal mechanism. Probabilities of seedling presence were ≥0.8 at 25 m distances, decreasing to under 0.3 at 175 m distances. While twenty‐seven of twenty‐nine species were present in at least one plantation with a seed source within 25 m, only thirteen of twenty‐seven species occurred in plantations with a seed source ≥100 m away, and only nine of twenty‐four species in plantations ≥150 m from seed sources, indicating limited potentials for interfragment migration. Seed source strength was significantly related to seedling density in plantations for twelve of fifteen species tested, indicating the importance of species commonness to interfragment migration. Connectivity was not related to dispersal frequency in the system, but this finding applies to a relatively well‐connected system of forests and plantations and is expected to differ for systems with greater patch isolation. In addition to these general controls, dispersal was related to species dispersal mechanism. Seed size was negatively related to dispersal frequency, but only within the rodent dispersal spectra. Bird‐dispersed species appeared to have superior interfragment dispersal abilities, closely followed by lighter seeded rodent‐ and wind‐dispersed species. Large‐seeded rodent‐dispersed species (Juglans spp., Quercus macrocarpa) and species lacking well‐developed adaptations for dispersers were infrequent in plantations generally, and with the exception of Q. macrocarpa, were absent from plantations more than 50 m from seed sources. Species dispersal abilities were ranked according to dispersal mechanism and seed size such that: bird > lighter‐seeded rodent = wind > larger‐seeded rodent = unspecialized.Main conclusionsDistance to seed sources appears to be a key determinant of patch colonization. Infrequent dispersal over distances of >100–150 m for most species in this system raises concerns about the abilities of tree species to be sustained in fragmented forests. For some species, distances of as little as 50 m appear to be isolating, and these, in particular, risk regional extinctions over time scales depending on their local population persistence. Artificial introductions may be needed to maintain fragmented tree populations, particularly for rare species and those with poor dispersal indicated in this study.