Most prominent theories of food-web dynamics imply the simultaneous action of bottom-up and top-down forces. However, transient bottom-up effects resulting from resource pulses can lead to sequential shifts in the strength of top-down predator effects. We used a large-scale field experiment (32 small islands sampled over 5 years) to probe how the frequency and magnitude of pulsed seaweed inputs drives temporal variation in the top-down effects of lizard predators. Short-term weakening of lizard effects on spiders and plants (the latter via a trophic cascade) were associated with lizard diet shifts, and were more pronounced with larger seaweed inputs. Long-term strengthening of lizard effects was associated with lizard numerical responses and plant fertilization. Increased pulse frequency reinforced the strengthening of lizard effects on spiders and plants. These results underscore the temporally variable nature of top-down effects and highlight the role of resource pulses in driving this variation.
Plant_RawData_Clean
These data were collected from buttonwood (Conocarpus erectus) plants located on small islands off the coast of Great Abaco, Bahamas. A detailed description of the experimental design and the study islands can be found in the paper and supplement. Column headings are as follows: eu = island identity; trip.date = date of sampling trip; sw.trt = seaweed treatment (several small = several small seaweed additions, several large = several large seaweed additions, single large = single large seaweed addition, control = no seaweed addition); liz.trt = brown anole (Anolis sagrei) lizard treatment (1 = lizards introduced, 0 = no-lizard control); block corresponds to groupings of 8 islands based on island size (blocks are named after local birds); plant = plant identity; percent damage = percent of leaf area damaged (from a sample of ~10 leaves); shoot.growth = shoot growth in the preceding sampling period (in mm, mean of ~4 stems); growth.s = shoot growth rate in the preceding sampling period (in mm per day, mean of ~4 stems); ellipsoid.volume = plant volume (in cubic meters, calculated by measuring height, length, and width, and assuming an ellipsoid shape) ; sandy.dam = plant damage resulting from Hurricane Sandy (proportion of ~4 marked, previously viable stems found dead after the storm); eu.ht.m = island height (meters above high water line at highest point).
WebSpiders_RawData_Clean
These data were collected from on small islands off the coast of Great Abaco, Bahamas. A detailed description of the experimental design and the study islands can be found in the paper and supplement. Column headings are as follows: eu = island identity; trip.date = date of sampling trip (mmm-yy); sw.trt = seaweed treatment (several small = several small seaweed additions, several large = several large seaweed additions, single large = single large seaweed addition, control = no seaweed addition); liz.trt = brown anole (Anolis sagrei) lizard treatment (1 = lizards introduced, 0 = no-lizard control); block corresponds to groupings of 8 islands based on island size (blocks are named after local birds); number.spiders.dens = web spider density (in spiders per square meter).
SalticidSpiders_RawData_Clean
These data were collected on small islands off the coast of Great Abaco, Bahamas. A detailed description of the experimental design and the study islands can be found in the paper and supplement. Column headings are as follows: eu = island identity; trip.date = date of sampling trip; sw.trt = seaweed treatment (several small = several small seaweed additions, several large = several large seaweed additions, single large = single large seaweed addition, control = no seaweed addition); liz.trt = brown anole (Anolis sagrei) lizard treatment (1 = lizards introduced, 0 = no-lizard control); block corresponds to groupings of 8 islands based on island size (blocks are named after local birds); salt.index = salticid density (number per bowl trap plus number per tap sample).
Analyses_PlantVars_LizEffect_Final
This file contains the R script used for the analyses of plant variables (leaf damage and shoot growth). Analyses were conducted using R version 3.4.3, lme4 version 1.1-15, and emmeans version 1.2.3.
Analyses_WebSpider_LizEffect_Final
This file contains the R script used for the analyses of web spider density. Analyses were conducted using R version 3.4.3, lme4 version 1.1-15, and emmeans version 1.2.3.
Analyses_SalticidSpider_LizEffect_Final
This file contains the R script used for the analyses of salticid spider abundance. Analyses were conducted using R version 3.4.3 and permute version 0.9-4.