1. Most resource subsidies are temporally variable, dynamically affecting the consumer populations, community structures, and ecosystem functions of recipient ecosystems. Temporally variable resource subsidies are characterized by the duration, magnitude, timing, and frequency of resource subsidy inputs. These different characteristics may have different mechanisms by which to affect recipient ecosystems.

2. Few studies have examined the duration of resource subsidy inputs on recipient ecosystems, although there exist previous studies focusing on magnitude, timing, and frequency.

3. We provide the first experimental test of the effects of subsidy duration on a stream ecosystem by using an outdoor mesocosm experiment, in which we directly manipulated the subsidy duration (pulsed vs. prolonged) of terrestrial invertebrate input into the mesocosm.

4. Given the same overall amount of terrestrial invertebrate subsidy were added, a prolonged subsidy allowed large-stage fish to effectively monopolize the subsidy over small-stage fish, which led small-stage fish to maintain their predation pressure on in-situ prey, i.e., benthic invertebrates. On the other hand, a pulsed subsidy allowed small-stage fish to increase their feeding rate of the subsidy and to become away from foraging in-situ prey. Consequently, weaker indirect positive effects on in-situ benthic prey and leaf breakdown rate were found with the prolonged versus pulsed subsidy. However, these indirect effects varied by the dominant benthic prey species, which differed in edibility for fish. Such predator-specific vulnerability of benthic prey can be important in mediating trophic cascades in detritus-based stream food webs.

5. Phenological events that generate temporal subsidies (e.g., salmon spawning run and arthropod emergence) can be synchronized (pulsed) or desynchronized (prolonged) within and among species, depending on the degree of spatial and temporal environmental heterogeneity. The effects of subsidy duration would thus be important to better understand ecological processes in spatially and temporally coupled ecosystems.

All data were collected in the field mesocosm experiment that was done in this study.

All raw data are provided:

Fish and fishcv files are for the all analyses related with fish growth, body size and its CV.

Benthic file is for the analyses related with the abundance of benthic invertebrates.

Litter file is for the analyses related with the leaf break-down rate.

R codes were not included because all statistical analyses specified in the methods are originally implemented in R.