As the atmosphere warms, precipitation events become larger, but less frequent. Yet, there is fundamental disagreement about how increased precipitation intensity will affect vegetation. Walter’s two-layer hypothesis and experiments testing it have demonstrated that precipitation intensity can increase woody plant growth. Observational studies have found the opposite pattern. Not only are the patterns contradictory, but inference is largely limited to grasslands and savannas. We tested the effects of increased precipitation intensity in a shrub-steppe ecosystem that receives >30% of its precipitation as snow. We used 11 (8 m x 8 m) shelters to collect and redeposit rain and snow as larger, more intense events. Total annual precipitation was the same in all plots, but each plot received different precipitation event sizes ranging from 1 mm to 18 mm. Over three growing seasons, larger precipitation event sizes increased soil water availability, sagebrush (Artemisia tridentata) stem radius, and canopy greenness, decreased new root growth in shallow soils, and had no effect on herbaceous plant cover. Thus, we found that increased precipitation intensity can increase soil water availability and woody plant growth in a cold semi-arid system. Assuming that stem growth is positively correlated with shrub reproduction, establishment and spread, results suggest that woody plant encroachment observed around the world in the past 50 years may be explained in part by increasing precipitation intensity. Further, continued atmospheric warming that is likely to increase precipitation intensity may also potentially increase shrub encroachment in the future

Author ORCID Identifier

Karen H. Beard

Andrew Kulmatiski



Document Type




File Format

.txt, .csv

Publication Date



Utah Agricultural Experiment Station


Utah State University


In June 2015, 14 plots, each 8 m x 8 m in size were established in a grid with at least 15 m between plots. Pre-treatment vegetation surveys and soil moisture measurements were taken until January 2016, when treatments were assigned randomly and applied through July 2018. Three plots were shelter-free controls and used to describe shelter effects. The remaining 11 plots were covered with 8 m x 8 m x 2.5 m (w x l x h) shelters. To allow a regression of vegetation responses across a wide range of precipitation event sizes, seven plots were assigned to different treatment levels (described below; Smith et al. 2014). Treatment levels were designed to create precipitation event sizes that could be expected with temperature changes from -1 to +10 °C relative to current temperatures.This resulted in minimum precipitation event sizes of 2, 3, 4, 8 and 18 mm for hypothetical temperature increases of 1 °C, 2 °C, 3 °C, 5 °C and 10 °C. To further expand our inference, one treatment designed to reflect precipitation intensity associated with -1 °C temperature change was added. In this treatment, irrigation was triggered manually multiple times per growing season depositing additional 1 mm events (hereafter referred to as the 1 mm treatment). All these treatments received the same total precipitation, and only differed in event size and frequency. Soil moisture, and above ground and below ground vegetation responses were measured in plots. When regressions were conducted the data was first averaged across plots by treatment, however, in the files provided here the plot level data is provide.

Start Date


End Date



Hardware Ranch Wildlife Management Area, UT. 41° 36’ 53” N, 111° 34’ 1” W



Code Lists

See the README.txt file.


File name: soil-moisture_diviner.csv; Description: Soil moisture measured in all plots. Measurements of soil moisture were taken roughly every two weeks in every plot during the growing season using capacitance sensors in PVC access tubes which were installed in June 2015, before treatment applications (Diviner 2000, Sentek Pty Inc., Stepney, Australia).

File name: soil-moisture_229.csv; Description: Daily soil moisture measured in one control and one treatment (4 mm treatment) plot (229L heat dissipation sensors, Campbell Scientific, Logan, UT, USA)

File name: plant-cover.csv, Description: Cover of plants by functional type, based on visual surveys made in June.

File name: stem-growth.csv; Description: Daily radial stem growth of sagebrush as measured by automatic point dendrometers (spring return linear position sensor BEI 9605, BEI Sensors, Thousand Oaks, CA, USA). Values are daily means taken across measurements from three shrubs in a plot.

File name: NDVI_twice-monthly.csv; Description: NDVI measured roughly twice monthly in all plots during the growing season (SRS-NDVI Sensor, Meter Group, Inc., Pullman, WA, USA).

File name: NDVI_continuous.csv; Description: Daily NDVI measured in one control plot and one 4 mm treatment plot (SRS-NDVI Sensor, Meter Group, Inc., Pullman, WA, USA).

File name: roots.csv; Description: Root area and number of new roots. Images taken roughly twice monthly during the growing season with a video microscope camera (Bartz Technology Co, Carpinteria, CA, USA).


Environmental Sciences | Natural Resources and Conservation | Water Resource Management


Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.



Additional Files

README.txt (9 kB)
MD5: bfb83586bda39324213afbf1317c39f3

soil-moisture_diviner.csv (323 kB)
MD5: e62ef3b43e02858c569af131109e09f9

soil-moisture_229.csv (404 kB)
MD5: 24d157576d9355d54d83f9a3b91e7c9b

plant-cover.csv (3 kB)
MD5: f230619ceb9e98b41c573e7bdaf29571

stem-growth.csv (421 kB)
MD5: e340e02fc596ad45857f467c7a6c6d73

NDVI_twice-monthly.csv (11 kB)
MD5: cc9e8d388ea628fe8babb903ea5bb6fe

NDVI_continuous.csv (68 kB)
MD5: 6a93d386412a64ccc4fb86ea93a59723

roots.csv (72 kB)
MD5: 1e9a11f1cbc63e3ab9923fd54b8c7da5