Active rainfall manipulation involving five treatments in sheltered plots (each with 6 replicates):

Ambient

+50%

-50%

Altered frequency (one rain event every 3 weeks, representing ambient rainfall during that period)

Summer drought (typically Dec-end Feb, each Australian summer)

The objectives of this long-term study are: (1) to assess the effects of directional and prolonged changes in precipitation amount and variability on population, community and ecosystem processes; (2) to elucidate the mechanisms behind those changes, and finally (3) to predict the future state of drylands functioning under novel climate conditions.

This experiment is part of the Loma Ridge Global Change Experiment, established by Michael Goulden in 2007. There are two ecosystem types, grassland and shrubland, with more studies completed in the grassland. The plots in the shrubland are 4 times the area and twice the height of the grassland. In 2014, half of the manipulated plots were designated "recovery" plots and the precipitation manipulation was ended. This experiment is also split-plot factorial with nitrogen addition (60 kg/ha/yr).

Drought manipulations as described have been carried out starting 1999 until 2017 (18 years), while 1998 was a pretreatment year. Plans for the next future (starting 2019) are to transform the active manipulation into a passive, drought-net design. This would mean that the "old" manipulation of 3 months no precipitation (an average decrease of 20% on an annual basis), would change to a year long decrease in precipitation of 20% or any other percentage (not yet decided).

Question: Are hydraulic traits in dominant coastal sage scrub shrub species plastic in response to altered precipitation regimes and does plasticity in these traits change the shrubs risk of drought induced mortality.

We use the Gherardi et al approach, passive drought shelter, collect water and pump on to irrigated plots.

We record community composition with individual-level mapping, which gives us demographic data, but means our community level data are a little "nonstandard" (some life forms represented as canopy cover, others basal cover, others just density).