Plant, grasshopper and spider food web responses to bird exclusion

Ecosystems are frequently considered to be controlled by predation
(top-down). Experiments examined this in four
bird/spider/grasshopper/prairie habitats over 34 years, employing in each
habitat three 100 m2 bird exclosures and controls (121 habitat/year cases)
where plant, grasshopper and spider abundances were measured. 
Top-down control (plants decrease and grasshoppers increase with bird
exclusion) was observed in only 13.2% of cases, while plants increased and
grasshoppers decreased in 33.1% of cases, plants decreased and
grasshoppers decreased in 25.6% of cases, and plants increased and
grasshoppers increased in 28.1% of cases. Therefore, top-down control was
not common and system responses were not constant, but varied among sites,
years and directionally over time with climate change. This diversity of
responses is expected given the variety of underlying processes in complex
ecosystems.  For example, Decision Tree/Discriminant Analysis
found that plant decreases and increases with bird exclusion were
correctly identified in 78.3% of cases by grasshopper hatchling abundance,
plant cover and annual net primary production (ANPP), while grasshopper
decreases and increases with bird exclusion were correctly identified in
76.7% of cases by edible plant biomass per grasshopper hatchling,
grasshopper hatchling abundance and large grasshopper abundance. Analysis
of other system-wide terrestrial trophic experiments indicates that the
variety of responses observed by us over time and space may be common, so
that system-wide trophic responses may, in general, be more variable than
either top-down or bottom-up as often considered.