1. Natural enemies have been implicated as agents of negative density dependence (NDD) in tropical forests, but their relative contributions to NDD, and thus to the maintenance of diversity, are largely unknown.
2. We monitored the rates of survival and relative growth rates on seedlings for ten years in tropical moist forest in Manu National Park, Peru. We then experimentally manipulated the plots to exclude fungal pathogens, insects, small mammals, and large mammals for an additional 31 months to assess the influence of these natural enemies on density-dependent interactions among tropical seedlings.
3. Fungal pathogens made the most important contribution to negative density dependence. The application of fungicide led to lower mortality rates, faster growth rates, and decreased species diversity. Other taxa of natural enemies had at most minor effects on seedling performance.
4. Synthesis. We conclude that fungal pathogens are the strongest contributors to the widely observed NDD that occurs among seedlings. Moreover, the presence of fungal pathogens augments the species diversity of seedlings, indicating their critical contribution to the maintenance of species coexistence and the structure of tropical tree communities.

Site

This study was carried out at the Cocha Cashu Biological Station (CCBS). CCBS is located in Amazonian South-East Peru in lowland tropical rain forest, at 11°51’S, 71°19’W, 350 m elevation. This seasonal forest receives a mean of 2167 mm of rain annually, and mean daily temperatures vary between 21.8ºC and 24.2º over the course of a year (Paine, 2007). The site is in a highly diverse and remote area of Manu National Park, with over 350 tree species with a diameter ≥10 cm DBH. It has experienced minimal hunting, and no logging or mining, during the last century (Hazelwood et al., 2020).

Design

Circular 1 m² experimental plots were established in a random blocked design throughout a 4 km² area of mature floodplain rain forest. 24 plots were spaced between 5 and 10 meters apart in each of 24 blocks, avoiding trails and newly fallen trees, for a total of 576 plots. Within each plot, all woody seedlings ≥10 cm and < 100 cm in stem height were identified and tagged over eight censuses between 2003 and 2017. Height was measured on all seedlings as the vertical distance from the soil to the apical meristem. All understory shrubs and lianas were excluded. Owing to the widely dispersed blocked arrangement of the plots, it was not feasible to identify the adult trees surrounding them. Unfortunately, this precluded the assessment of the effects of adult conspecific and heterospecific density on seedlings. Censuses were carried out 269 to 1566 days apart (see Paine & Harms, 2009 for details).

The experimental phase of the study began in October 2014, when we applied treatments to exclude fungi, insects, and mammals. Within each block, eight plots were randomly selected for the application of one of eight treatments: none (a control), fungi, insects, large mammals, all mammals, fungi and insects, all mammals and fungi, and all mammals and insects. The fungicide Amistar (Syngenta Ltd, active ingredient: azoxystrobin) provides a broad spectrum of protection against fungal attack, has low toxicity in non-target organisms, and was found to be effective by Bagchi et al. (2014). The insecticide Karate (Syngenta Ltd. active ingredient: lambda cyhalothrin), provides protection against a broad spectrum of insect herbivores, leaving low rates of residue and has low impact on non-target organisms. Pesticides were applied according to manufacturer’s instructions, mixing 1.25 ml of pesticide with 1 litre of water, and applying 50 ml of the mixture to 1 m² plot with spray bottles. Pesticides were applied to treatment plots every 10 to 14 days, in equal amounts over 31 months, with some treatment breaks when it was logistically impossible to apply treatments (max 1 month). Control plots were misted with an amount of water equivalent to that applied to pesticide plots.

We excluded mammals from the study plots using 2×2 m wire mesh exclosures. These were 150 cm high and included a 50 cm buffer around each plot to reduce potential germination bias from perching birds. The ‘Large mammal’ exclosures allowed the entry of small mammals through 15×15 cm holes cut into the base of the mesh. These were large enough to allow agoutis (Dasyprocta sp.) or smaller rodents to enter, but were too small for peccaries (Pecari and Tayassu spp.), deer (Mazama americana), or tapir (Tapirus terrestris). The ‘All mammal’ exclosures, on the other hand, were constructed flush to the ground and excluded all terrestrial mammals. Previous studies at the same site have shown this design of exclosures to be extremely effective at excluding terrestrial mammals (Beck et al., 2013; Paine et al., 2016).