In 1998, we established soil moisture sampling points at Archbold
Biological Station in rosemary scrub, scrubby flatwoods and
oak-hickory scrub and along man-made sandy roadsides. In the three
natural vegetation types, we sampled in recently burned (3-5 years
prior to the initiation of the study) and long-unburned (>20 years)
areas (sandy roadsides do not burn). At each sampling site, sampling
points were paired with one sampling point within the shrub matrix and
the other within a shrub-free gap. At each sampling point, tubes were
installed using schedule 40 PVC tubes measuring 100 cm in length
inserted perpendicular into the ground, with about 5 cm of tube
protruding above the ground surface. We measured soil moisture at
three depths (10, 50 and 90 cm) using a Troxler Electronics Lab Sentry
200-AP frequency domain reflectometry probe. The instrument was
inserted into the tube to the appropriate sampling depths. To
calibrate the probe, we compared instrument readings to gravimetric
calculations of soil moisture using native soils in a large plastic
garbage can with an inserted PVC tube. These data were used in a
regression to develop the following relationship: percent soil
moisture = D x 49.887 + 3039.543, where D is the resonance frequency
shift measured by the Sentry 200-AP. This equation was programmed into
the Sentry 200-AP, which then calculated percent soil moisture by
volume with each reading. During the wettest periods, ground water
intrusion into the tubes was recorded as evidence of soil saturation
at depths below the point where water was encountered by the probe.
Saturation was recorded as 40.0% soil moisture by volume.
We sampled tubes weekly from 1 October 1998 through 9 June 1999,
bi-weekly from 23 June 1999 through 3 October 2001, monthly from 17
October 2001 through 15 June 2011, and every other month thereafter.
On 27 January 2010, we reduced our sampling efforts collecting data
from only 48 tubes. Monthly samples were taken in the middle of the
month to allow for year-to-year comparisons. We completed our last
round of sampling on 16 July 2018.
Although habitat types remained stable during the 19 years of the
study, the time-since-fire and whether the point was in a shrub matrix
or a gap changed. Tubes were replaced when dislodged by falling trees
or partly melted in fires. Elevation data were acquired by lidar in
2006 and can be found (along with tube coordinates) in the elevation
CSV file or zipped shapefile. Investigators analyzing this dataset are
encouraged to test for spatial autocorrelation.
