348
among the individual rod pairs located 8 m apart.
To evaluate this assumption, they constructed semivariograms (Turner et al. 1991) based on TDR measurements of soil water content at a 0.15-m spacing
along both vertical and horizontal transects within
the TDE area and demonstrated that beyond 4 to 5
m the individual soil water measurements could be
considered independent.
Collection of Adequate
Weather Data
All large-scale manipulation studies must collect
adequate weather data to allow for the evaluation
of impacts of the treatment infrastructure on uncontrolled environmental variables (i.e., air temperature, humidity). A complete set of variables to
be monitored should include the following: incoming rainfall, irradiance, photosynthetic photon flux
density, air temperature, and relative humidity or
dewpoint. Understory climate data should also include photosynthetically active radiation (PAR) at
-1.5 m, air temperature, soil temperatures (multiple depths), and relative humidity. Collecting
weather data at an hourly resolution is recommended, as many current process models of forest
stand or ecosystem function demand this level of
time resolution of weather input data sets.
Confounding Issues
Any experimental design involves compromises
between ideal conditions and logistical reality.
Gundersen (1995) discusses some of the common
unintended consequences of large-scale manipUlation studies in detail. Gundersen concludes that
large manipulations often lead to modified microclimatic conditions (i.e., light, temperature, and humidity) due to the presence of experimental structures. Those studies using complete understory
roofs (near 100% interception) do show temperature differences between roofed and control plots
(Gundersen et al. 1995), but the sign of the differences is seasonally dependent. In the TDE, Hanson
et al. (1995, 1998) originally anticipated that the
presence of the troughs, gutters, and pipes might
act as a radiation barrier or thermal blanket during
Paul J. Hanson
periods of incomplete canopy leaf coverage and
lead to changes in the understory microclimate.
However, they found little evidence of any changes
in air or soil (10 cm) temperatures during the growing season when the canopy is fully leafed out.
Large gaps between collection troughs in the study
design of the TDE probably allowed for adequate
turnover and circulation of air around the trough
infrastructure. Fay et al. (in press) found that soil
temperatures under their grassland roofs were elevated 1.2 to 1.8°C over ambient conditions, and attributed the temperature increase to reduced nighttime radiative loss. The grassland roofs also
reduced radiation reaching the plant canopy, and
unroofed control plots were incorporated into the
experimental design to assess the long-term impact
of the altered light environment.
Unintentional modifications of rainfall chemical
inputs also often accompany large-scale water manipulations. Chemicals derived from atmospheric
dry and wet deposition and from canopy leaching
are normally transferred to the forest floor in
throughfall. In the attempt to modify the water budget of forest stands using understory troughs (Hanson et al. 1995) or roofs (Beier et al. 1995; Fay et
al., in press; Gundersen et al. 1995; Rasmussen et
al. 1995), unavoidable chemical transfers are also
made. Rasmussen et al. (1995) showed that treatments in which 100% of the intercepted throughfall
was extracted also removed significant potassium
from the site, and they concluded that such a change
would accumulate and become biologically significant over time. Hanson et al. (1998) stated that
some changes in chemical inputs due to ± 33%
throughfall alterations on their upland oak site were
taking place, but concluded that the biological impact for their system would be minimal. Even if the
anticipated level of biological response to small
changes in chemical inputs/outputs resulting from
an experimental manipulation is expected to be
small, some measurements of the level of chemical
transfers taking place should be included.
Plot Size and Edge Effects
Small-scale water manipulation studies having manipulated areas less than 50 m 2 (Breda et al. 1993,
1995; Fay et al., in press; Holstener-Jj1irgensen
1994) often use belowground barriers to root and
among the individual rod pairs located 8 m apart.
To evaluate this assumption, they constructed semivariograms (Turner et al. 1991) based on TDR measurements of soil water content at a 0.15-m spacing
along both vertical and horizontal transects within
the TDE area and demonstrated that beyond 4 to 5
m the individual soil water measurements could be
considered independent.
Collection of Adequate
Weather Data
All large-scale manipulation studies must collect
adequate weather data to allow for the evaluation
of impacts of the treatment infrastructure on uncontrolled environmental variables (i.e., air temperature, humidity). A complete set of variables to
be monitored should include the following: incoming rainfall, irradiance, photosynthetic photon flux
density, air temperature, and relative humidity or
dewpoint. Understory climate data should also include photosynthetically active radiation (PAR) at
-1.5 m, air temperature, soil temperatures (multiple depths), and relative humidity. Collecting
weather data at an hourly resolution is recommended, as many current process models of forest
stand or ecosystem function demand this level of
time resolution of weather input data sets.
Confounding Issues
Any experimental design involves compromises
between ideal conditions and logistical reality.
Gundersen (1995) discusses some of the common
unintended consequences of large-scale manipUlation studies in detail. Gundersen concludes that
large manipulations often lead to modified microclimatic conditions (i.e., light, temperature, and humidity) due to the presence of experimental structures. Those studies using complete understory
roofs (near 100% interception) do show temperature differences between roofed and control plots
(Gundersen et al. 1995), but the sign of the differences is seasonally dependent. In the TDE, Hanson
et al. (1995, 1998) originally anticipated that the
presence of the troughs, gutters, and pipes might
act as a radiation barrier or thermal blanket during
Paul J. Hanson
periods of incomplete canopy leaf coverage and
lead to changes in the understory microclimate.
However, they found little evidence of any changes
in air or soil (10 cm) temperatures during the growing season when the canopy is fully leafed out.
Large gaps between collection troughs in the study
design of the TDE probably allowed for adequate
turnover and circulation of air around the trough
infrastructure. Fay et al. (in press) found that soil
temperatures under their grassland roofs were elevated 1.2 to 1.8°C over ambient conditions, and attributed the temperature increase to reduced nighttime radiative loss. The grassland roofs also
reduced radiation reaching the plant canopy, and
unroofed control plots were incorporated into the
experimental design to assess the long-term impact
of the altered light environment.
Unintentional modifications of rainfall chemical
inputs also often accompany large-scale water manipulations. Chemicals derived from atmospheric
dry and wet deposition and from canopy leaching
are normally transferred to the forest floor in
throughfall. In the attempt to modify the water budget of forest stands using understory troughs (Hanson et al. 1995) or roofs (Beier et al. 1995; Fay et
al., in press; Gundersen et al. 1995; Rasmussen et
al. 1995), unavoidable chemical transfers are also
made. Rasmussen et al. (1995) showed that treatments in which 100% of the intercepted throughfall
was extracted also removed significant potassium
from the site, and they concluded that such a change
would accumulate and become biologically significant over time. Hanson et al. (1998) stated that
some changes in chemical inputs due to ± 33%
throughfall alterations on their upland oak site were
taking place, but concluded that the biological impact for their system would be minimal. Even if the
anticipated level of biological response to small
changes in chemical inputs/outputs resulting from
an experimental manipulation is expected to be
small, some measurements of the level of chemical
transfers taking place should be included.
Plot Size and Edge Effects
Small-scale water manipulation studies having manipulated areas less than 50 m 2 (Breda et al. 1993,
1995; Fay et al., in press; Holstener-Jj1irgensen
1994) often use belowground barriers to root and