Seed dispersal distance (SDD) critically influences the survival of seedlings, spatial patterns of genetic diversity within plant populations, and gene flow among plant populations. In animal‐dispersed species, foraging behavior and movement patterns determine SDD. Direct observations of seed dispersal events by animals in natural plant populations are mostly constrained by the high mobility and low visibility of seed dispersers. Therefore, diverse alternative methods are used to estimate seed dispersal distance, but direct comparisons of these approaches within the same seed dispersal system are mostly missing. We investigated two plant species with different life history traits, Leonia cymosa and Parkia panurensis, exclusively dispersed by two tamarin species, Saguinus mystax and Leontocebus nigrifrons. We compared SDD estimates obtained from direct observations, genetic identification of mother plants from seed coats, parentage analysis of seedlings/saplings, and phenomenological and mechanistic modeling approaches. SDD derived from the different methods ranged between 158 and 201 m for P. panurensis and between 178 and 318 m for L. cymosa. In P. panurensis, the modeling approaches resulted in moderately higher estimates than observations and genotyping of seed coats. In L. cymosa, parentage analysis resulted in a lower estimate than all other methods. Overall, SDD estimates for P. panurensis (179 ± 16 m; mean ± SD) were significantly lower than for L. cymosa (266 ± 59 m; mean ± SD). Differences among methods were related to processes of the seed dispersal loop integrated by the respective methods (e.g., seed deposition or seedling distribution). We discuss the merits and limitations of each method and highlight the aspects to be considered when comparing SDD derived from different methodologies. Differences among plant species were related to differences in reproductive traits influencing gut passage time and feeding behavior, highlighting the importance of plant traits on animal‐mediated seed dispersal distance.

Increasingly large proportions of tropical forests are anthropogenically disturbed. Where natural regeneration is possible at all, it requires the input of plant seeds through seed dispersal from the forest matrix. Zoochorous seed dispersal – the major seed dispersal mode for woody plants in tropical forests – is particularly important for natural regeneration. In this study, covering a period of more than 20 years, we show that small New World primates, the tamarins Saguinus mystax and Leontocebus nigrifrons , increase their use of an anthropogenically disturbed area over time and disperse seeds from primary forest tree species into this area. Through monitoring the fate of seeds and through parentage analyses of seedlings of the legume Parkia panurensis from the disturbed area and candidate parents from the primary forest matrix, we show that tamarin seed dispersal is effective and contributes to the natural regeneration of the disturbed area.

Amazon. We compare SDD estimates obtained from direct observations, genetic identification 35 of mother plants from seed coats, parentage analysis of seedlings, and modelling approaches, 36including the combination of movement data and gut passage times and individual-based 37 modelling. 38 3. We detect differences between SDD estimates that can be linked to the processes relevant at 39 different phases of the seed dispersal loop covered by the respective approaches. Despite these 40 differences, SDD estimates for P. panurensis are consistently lower than for L. cymosa which 41 is likely related to differences in fruit characteristics and fruit abundance, factors that 42 influence gut passage time, foraging behaviour and movement p a t t e r n s of the tamarins. 43 4. Our comparisons allow setting SDD estimates from studies using different methodological 44 approaches into the seed dispersal loop context, thus improving comparability of 45 methodologically different studies and method applicability. 46

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

No abstract