Tropical-temperate comparisons in insect seed predation vary between study levels and years.

The biotic interaction hypothesis, which states the species interaction becomes stronger in the tropics, is deeply rooted in classic ecological literature and widely accepted to contribute to the latitudinal gradients of biodiversity. Tests in latitudinal insect-plant interaction have emphasized leaf-eating insects on a single or a few plant species rather than within an entire community and mixed accumulating evidence, leaving the biotic interaction hypothesis disputed. We aimed to test the hypothesis by quantifying insect seed predation in a pair of tropical and temperate forest communities with similar elevations. We applied a consistent study design to sample predispersal seeds with systematically set seed traps in 2019-2020 and examined internally feeding insects. The intensity of seed predation was measured and further applied to tropical versus temperate comparison at two levels (cross-species and community-wide). Our results showed every latitudinal pattern associated with different study levels and years, that is, negative (greater granivory in the tropics in community-wide comparison in 2020), positive (less granivory in the tropics in community-wide and cross-species comparison in 2019), and missing (similar level of granivory in the tropics in cross-species comparisons in 2020). The cross-species level analyses ignore differences among species in seed production and weaken or even lose the latitudinal trend detected by community-wide comparisons. The between-year discrepancy in tropical-temperate comparisons relates to the highly variable annual seed composition in the temperate forest due to mast seeding of dominant species. Our study highlights that long-term community-level researches across biomes are essential to assess the latitudinal biotic interaction hypothesis.