ABSTRACT Questions Seed traps, as a common tool for measuring the seed rain, are often used, but a comparative evaluation of the effectiveness is rarely addressed. Therefore, we tested four different seed trap types under two contrasting grassland managements concerning: (1) their effectiveness in catching seeds (species number, seed density), (2) characterising the seed production of the aboveground vegetation and (3) the contamination of non‐seed material (arthropods, plant parts, soil). Location Pampa grassland (Argentina). Methods We assessed the effectiveness of two ground‐type (hereafter ground) traps (funnel, pot) and two adhesive‐type (hereafter adhesive) traps (glue, sticky) in a grazed and an unmanaged natural grassland. Additionally, we described the aboveground vegetation including the estimated seed production (ESP). The latter was compared with the seed rain measured by traps to assess which trap type can best characterise it. Results In the aboveground vegetation, no significant differences in cover and species richness were found depending on the management, whilst differences for most of the investigated parameters were detected for the ESP. Adhesive traps exhibited the highest contamination with arthropods, funnel traps that with plant parts. Both ground seed traps had considerable contamination with soil. Concerning the effectiveness, the highest seed density was provided by ground traps and the highest species richness by pot traps. Likewise, ground traps more closely resembled the ESP according to the qualitative Jaccard and quantitative Bray–Curtis similarity index. A significant Mantel correlation between measured seed rain and ESP was found for the pot traps only. However, no seed trap type was able to account for differences of ESP depending on the management. Conclusion Overall, pot and funnel traps seem to be the most suitable tool to measure seed rain in natural grassland irrespective of the management. All traps failed to detect most differences between grazed and unmanaged areas regarding ESP.
Fundneider et al. (Wed,) studied this question.