Access to this full-text is provided by Wiley.
Content available from Crop Science
This content is subject to copyright. Terms and conditions apply.
Received: 1 June 2024 Accepted: 26 February 2025
DOI: 10.1002/csc2.70053
Crop Science
ORIGINAL ARTICLE
Special Section: 2025 International Turfgrass Research Conference
Recovery of five cool-season turfgrasses following long-term ice
encasement
Eric Watkins1Dominic P. Petrella2Trygve Aamlid3Dominic C. Christensen1
Sigridur Dalmannsdottir4Andrew P. Hollman1Gary Deters1
1Department of Horticultural Science,
University of Minnesota, St. Paul,
Minnesota, USA
2Agricultural Technical Institute, The Ohio
State University, Wooster, Ohio, USA
3Norwegian Institute of Bioeconomy
Research, Landvik Research Center,
Grimstad, Norway
4Norwegian Institute of Bioeconomy
Research, Tromsø Research Center, Tromsø,
Norway
Correspondence
Eric Watkins, Department of Horticultural
Science, University of Minnesota, St. Paul,
MN 55108, USA.
Email: ewatkins@umn.edu
Assigned to Associate Editor Samuel
Trachsel.
Funding information
National Institute of Food and Agriculture,
Grant/Award Number: 2021-51181-35861;
Norges Forskningsråd, Grant/Award
Number: 310090
Abstract
Ice encasement is a major concern for turfgrass managers in cold climates; however,
there is a lack of data about both which turfgrasses are best suited for survival under
these conditions and the reasons behind the superior recovery of some grasses from
long-term ice encasement. In this study, we encased golf course putting greens-height
field plots of creeping bentgrass (Agrostis stolonifera L.), velvet bentgrass (Agrostis
canina L.), annual bluegrass (Poa annua L. var. reptans Hausskn.), Chewings fescue
(Festuca.rubra L. ssp. commutata Gaudin), and slender creeping red fescue (F.rubra
L. ssp. littoralis (G. Mey.) Auquier) with ice for 90–120 days with the inclusion of
CO2,O
2, and temperature sensors at 2.5 and 12.5 cm depth to better understand envi-
ronmental conditions under ice and factors related to winterkill. Velvet bentgrass had
the best overall performance and recovery, while annual bluegrass did not survive.
Differences in recovery among turfgrass taxa may have been affected by the length of
the ice encasement period, higher CO2levels (>40,000 ppm), and lower O2values,
particularly in the second experimental run. During the recovery period in both years,
photochemical efficiency values began increasing 5–10 days before percent green
cover, suggesting that visual performance of the turf surface is a lagging indicator of
recovery. Overall, recovery from ice encasement was annual bluegrass <Chewings
fescue <creeping bentgrass =slender creeping red fescue =velvet bentgrass. These
results can guide turfgrass managers in making species selection decisions in areas
where long-duration ice encasement is a risk.
Plain Language Summary
Turfgrasses on golf course greens in cold climates can be severely damaged or even
die from ice encasement. Little is known about this stress, including why certain
grasses can survive longer. As a first step to learn more about this problem, we tested
Abbreviations: Fv/Fm, maximum quantum efficiency of photosystem II; NDVI, normalized difference vegetation index; PAR, photosynthetically active
radiation.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original
work is properly cited.
©2025 The Author(s). Crop Science published by Wiley Periodicals LLC on behalf of Crop Science Society of America.
Crop Science. 2025;65:e70053. wileyonlinelibrary.com/journal/csc2 1of11
https://doi.org/10.1002/csc2.70053
2of11 WATKINS ET AL.
Crop Science
five different turfgrasses for their ability to survive under ice. The study was done
during two separate winters in Minnesota under field conditions, resulting in 98 days
of ice in 2021–2022 and 112 days of ice cover in 2022–2023. Annual bluegrass died
completely during both experimental runs, while Chewings fescue suffered some
injury in the first year and did poorly in the second year. Velvet bentgrass was the
best grass in both years. Under the longer duration of ice cover in the second year,
carbon dioxide levels were very high, while oxygen gas levels slowly declined over
the course of the ice encasement period.
1INTRODUCTION
With a changing climate, northern latitudes are now more
commonly faced with unstable snow cover and formation of
ice (Cohen et al., 2018). Long-term duration of ice cover is
known to reduce the growth potential of grasses and cause
winterkill (Merewitz, 2021). This is an increasing problem
on golf course putting greens and results in delayed open-
ing of golf courses and economic loss because of frequent
re-establishment (Kvalbein et al., 2017). The effect of ice
encasement on turfgrasses has received scant attention by
researchers. A lack of new knowledge in this field is concern-
ing, especially in the context of golf greens, which are the
most important surfaces on a golf course and necessary for an
economically viable, fully functioning operation.
Early work on ice encasement effects on turfgrasses began
in the mid-20th century. Beard (1964) found that Toronto
creeping bentgrass (Agrostis stolonifera L.) could survive
under 5.1 cm (2 in.) of ice for at least 51 days in East Lans-
ing, MI. In a follow-up experiment, Beard (1965b) harvested
cold-acclimated plugs of creeping bentgrass (Toronto) and
annual bluegrass (Poa annua L.) from the field and found that
creeping bentgrass had far superior survival under multiple
ice-formation scenarios, including the most severe treatment
of fully encasing the plug in ice: Toronto had 100% survival
after 60 days and still had 10% survival after 90 days; con-
versely, annual bluegrass had no survival after just 15 days.
Later, Beard (1965a) found few differences for ice encase-
ment tolerance among creeping bentgrass cultivars available
at the time of his study. Together, these studies have been the
basis for estimating how long each of these species can sur-
vive under ice. Golf course superintendents in cold climates
have, for decades, preferred creeping bentgrass to annual blue-
grass when ice encasement is a risk; however, due to the
competitive ability of annual bluegrass biotypes adapted to
greens-height conditions, most golf course budgets likely do
not allow for long-term control. Plant breeders have developed
perennial-type annual bluegrass cultivars (Bonos & Huff,
2013); however, to our knowledge, there has not been any
selection work done by turfgrass breeders specifically for ice
encasement tolerance.
More recent work has confirmed Beard’s general findings
(Beard, 1964,1965a,1965b). Tompkins et al. (2004) com-
pared annual bluegrass and creeping bentgrass for response
to ice encasement and found that annual bluegrass was killed
after 90 days of ice cover in controlled environment condi-
tions, while creeping bentgrass only began to lose hardiness
at that point and was able to survive beyond 150 days. Labora-
tory studies showed that annual bluegrass was more sensitive
to anoxia (low O2, like found under ice) than creeping bent-
grass (Castonguay et al., 2009). In a field study in central
Norway, Waalen et al. (2017) found that the tolerance to ice
encasement of either 98 or 119 days decreased in the order
starting with velvet bentgrass (Agrostis canina L.) as the most
tolerant >creeping bentgrass >Chewings fescue (Festuca
rubra L. ssp. commutata Gaudin), slender creeping red fes-
cue (F.rubra L. ssp. littoralis (G. Mey.) Auquier) ≥colonial
bentgrass (Agrostis capillaris L.) >annual bluegrass.
The concentrations of gases change under ice encasement
(Andrews, 1996), with CO2increasing and O2decreasing.
This change can be both directly detrimental to turfgrass
survival and indirectly damaging due to causing production
of and exposure to other toxic metabolites/volatiles such as
ethanol and lactic acid. Castonguay et al. (2009) exposed
annual bluegrass to various combinations of CO2and O2
concentrations under low temperature and showed that gas
concentration was important for survival. However, we still do
not know how CO2and O2concentrations change over time
in the soil of a putting green encased in ice. Such data may
be able to help build prediction models for turfgrass mortal-
ity and recovery under ice encasement, helping golf course
superintendents make better decisions on if/when to remove
ice.
When evaluating survival from winter-related damage such
as from ice encasement or from direct low-temperature kill,
researchers have typically used visual assessment methods.
The most common being rating plants as either dead (0) or
alive (1), or rating the percentage of surviving tillers on a 1–9
scale to estimate whole plant survival (Hoffman et al., 2012;
Hulke et al., 2008). While these methods are simple, the rat-
ings are not always straightforward, and results may vary due
WATKINS ET AL.3of11
Crop Science
to the person collecting visual data due to interpretation of
what is dead or alive. The use of digital image analysis and
the evaluation of the amount of green tissue cover in a pic-
ture taken of a turfgrass plot could remove interpretation of
survival. Additionally, measurements of chlorophyll fluores-
cence parameters such as the maximum quantum efficiency
of photosystem II (i.e., Fv/Fm) could determine the health of
survived tissue and distinguish mortality of organs that appear
dead to a human eye or ordinary camera. Recent research
has used Fv/Fm data to predict mortality in plants exposed to
drought (Guadagno et al., 2017; Rico-Cambron et al., 2023)
and could be used to evaluate survival from ice encasement.
A better understanding of turfgrass species differences for
ice encasement tolerance and recovery would be beneficial
for turfgrass managers seeking recommendations for turfgrass
stands in climates where long-term ice encasement is a risk.
Therefore, we compared five cool-season turfgrass taxa that
are used on golf greens in cold climates for survival and recov-
ery from long-term ice encasement under field conditions in
Minnesota. Our objective was to identify turfgrasses that can
both survive this stress and recover fully to allow for regu-
lar functioning of the turf surface. Furthermore, we aimed to
measure physiological and environmental parameters to help
better understand the mechanisms associated with ice encase-
ment stress responses, and to help develop better methods
to assess ice encasement survival and recovery compared to
traditional methods.
2MATERIALS AND METHODS
2.1 Experimental design and management
Research plots were established at the University of Min-
nesota Turfgrass Research, Outreach and Education Center
in Falcon Heights, MN (lat: 44.995353, long: −93.185347)
on a sand-based, United States Golf Association–specified
research green with an original sand layer depth of 30.5 cm.
Turfgrass plots of 1 m2were arranged as a randomized com-
plete block design with four replications. The experiment was
performed during the winter and spring of 2021–2022 and
replicated in 2022–2023. Average air temperatures for the first
experimental year were −5.6˚C, −14.3˚C, −11.8˚C, −2.0˚C,
and 3.7˚C for December, January, February, March, and April,
respectively. For the second year, monthly average tempera-
tures were −9.4˚C, −7.1˚C, −7.8˚C, −3.5˚C, and 5.9˚C for
December, January, February, March, and April, respectively
(Minnesota Department of Natural Resources, 2024). Prior to
each run, individual plots were seeded (July 17, 2021, and
June 7, 2022) with one of the following turfgrass cultivars:
Radar Chewings fescue (Mountain View Seeds), Cezanne
slender creeping red fescue (DLF Seeds), Luminary creeping
bentgrass (Landmark Seed Company), Nordlys velvet bent-
Core Ideas
∙Annual bluegrass did not survive under ice encase-
ment of at least 90 days.
∙Velvet bentgrass, slender creeping red fescue, and
creeping bentgrass showed good recovery from ice
encasement.
∙Photochemical efficiency was a good predictor of
turfgrass recovery from ice encasement stress.
∙Environmental sensing of microenvironments can
improve our understanding of ice encasement
tolerance.
grass (Graminor AS), or Two-Putt annual bluegrass (P. annua
L. var. reptans Hausskn) (DLF North America). Chewings
fescue and slender creeping red fescue were seeded at 50 g
m−2, creeping bentgrass and velvet bentgrass at 7.5 g m−2, and
annual bluegrass at 10 g m−2. Cultivars were chosen based on
previous use in cold-climate research studies. Single cultivars
were used for each species due to space and logistical limita-
tions; in previous studies on turfgrass winter survival, species
has been a more important factor than cultivar (Beard, 1965a),
so this limitation was considered minor.
During both establishments, starter fertilizer was applied
at seeding to provide 10.7 kg N ha−1, 21.3 kg P ha−1, and
13.4 kg K ha−1, and again 2 weeks later to provide 26.9 kg
Nha
−1, 21.3 kg P ha−1, and 49.1 kg K ha−1. Plots were
fertilized again approximately monthly with 4.88 kg N ha−1
at each application: 2021 plots were fertilized in late July,
late August, and late September; 2022 plots were fertilized
in late August and late September. The experimental area was
maintained at a 5.6-mm height of cut with a reel mower and
top-dressed with United States Golf Association specification
sand every 2 weeks. Fungicides were applied for preventa-
tive disease control. In 2021, for the first experiment, granular
mefenoxam was applied once at 0.38 kg ai ha−1after seeding.
In 2022, for the second replication of the experiment, granular
mefenoxam was applied once at 0.38 kg ai ha−1after seeding
and then weekly during establishment for a total of four appli-
cations. In 2022, a liquid application of mefenoxam (0.763 kg
ai ha−1) was made on July 14 for further pythium disease
(Pythium aphanidermatum) prevention, and a combination
of fluazinam (0.81 kg ai ha−1) and azoxystrobin (0.32 kg
ai ha−1) was applied on September 14 to prevent anthrac-
nose (Colletotrichum cereale). Fungicides were applied for
snow mold (Microdochium nivale;Typhula spp.) prevention
for the first experiment on November 3, 2021, using a com-
bination of fluazinam (0.809 kg ai ha−1), thiophanate methyl
(4.29 kg ai ha−1), tebuconazole (0.82 kg ai ha−1), and ipro-
dione (3.05 kg ai ha−1) and for the second replication of the
experiment on November 11, 2022, using a combination of
4of11 WATKINS ET AL.
Crop Science
fluazinam (0.81 kg ai ha−1), thiophanate methyl (3.43 kg ai
ha−1), tebuconazole (0.82 kg ai ha−1), and iprodione (3.43 kg
ai ha−1). Weed pressure was very low and did not require
control measures.
2.2 Sensor installation and ice encasement
Plots were subjected to ice encasement and monitored through
winter and spring recovery. Prior to ice development, custom
solar-powered sensor nodes, v2.0.5.B systems as described
in Runck et al. (2024), were installed near the plot area,
with individual sensors for measuring environmental vari-
ables buried in the soil. In the first trial (2021 seeding), sensor
nodes were installed to monitor two blocks (repetitions) of
the ice-encased plots, and two additional sensor nodes mon-
itored the area outside of the plot area (creeping bentgrass
maintained at the same height as the experimental area and
not encased in ice). For the trial seeded in 2022, a sensor
node was installed to monitor a single block of the ice-encased
plots, and a second sensor outside of the ice-encased plot area
monitored conditions in the single replication of plots that
served as a control without ice encasement. Environmental
parameters could not be monitored in additional blocks due to
sensor availability. Each node regularly measured rootzone O2
concentration, CO2concentration (Schulz et al., 2024), and
temperature. All sensor measurements were recorded at two
depths: 2.5 and 12.5 cm.
Each block (replication) was 1 m ×5 m and consisted of
five individual 1-m2plots. Aluminum barriers were created
with multiple 15.2 cm by 3 m lengths of aluminum sheet and
used to prevent water movement during ice formation and
melt. The barrier was installed around each block in late fall
to a depth of 5 cm below the soil surface and was placed such
that there was a 50 cm distance between the edge of the block
and the aluminum barrier. Water was initially sprayed over the
plots to create a thin layer of ice and prevent it from percolat-
ing through the soil. Over a 2-week period, further additions
of water were added multiple times each day, allowing each
to freeze before more was added, until each block was cov-
ered by 10.2 cm of ice. In 2021, the soil had frozen prior to
the ice-formation process. In 2022, the plot area was covered
with snow that had fallen earlier in the winter and late fall;
snow was removed from the plots designated for ice encase-
ment to allow for the soil to freeze but was left on the control
plots and surrounding areas (a single replication of control
plots was present only during the second run of the study).
In both years of the trial, snow was removed from the ice-
encased plots throughout the study, and water was periodically
added to the surface to maintain ice depth. The ice remained
over the plots for at least 90 days and was then allowed to
melt naturally. Ice formation (10.2-cm ice depth) for the first
experiment began on December 16, 2021, and complete ice
melt was reached on March 25, 2022, for a total of 98 days
of ice encasement. Ice formation for the second experiment
began on December 19, 2022, and reached complete ice melt
on April 10, 2023, for a total of 112 days of ice encasement.
There was no ice formation on any of the areas surrounding
the ice-encased plot area including the control plots that were
present during the 2022–2023 winter.
2.3 Data collection
In addition to the environmental data recorded by the sen-
sor nodes, turfgrass responses were collected after ice melt
each year. To measure physiological plant stress, a mobile
chlorophyll fluorescence imaging system (IMAGING-PAM
M-Series Maxi Version; Heinz Walz Co.) with a zoom and
focus adjustable lens was used to collect data on the maximum
quantum efficiency of photosystem II (Fv/Fm; variable fluo-
rescence to maximal fluorescence; photochemical efficiency)
and photosynthetically active radiation (PAR) absorptivity
data on an area that was approximately 10 cm ×13 cm
(absorptivity =1 – [red light remission/near-infrared light
remission]). Plots were dark-adapted for 5–18 min using
blackout cloth drapery as previous research has indicated that
15 min is a good length of time to dark adapt for measur-
ing Fv/Fm (Kalaji et al., 2014). Based on our experience
measuring Fv/Fm in turfgrass systems, dark adaptation times
between 5 and 30 min results in no significant change in
Fv/Fm data, and dark adaption in our study was achieved in 5
min.
Next, LED intensity and gain were adjusted in Walz Imag-
ingWin software so that steady-state fluorescence (Ft)was
between 0.16 and 0.18 prior to capturing maximum fluores-
cence (Fm) data. One sample per plot was taken 2, 3, 4, and
10 days after complete ice melt and weekly for the follow-
ing 5 weeks. In 2022, data were collected at 7:00 a.m., 11:00
a.m., and 6:00 p.m. to determine optimum sampling time,
but an analysis of photochemical efficiency and absorptivity
showed no significant main or interactive effect among sam-
pling times, so 11:00 a.m. was chosen for convenience and
used in the analyses for both years (Table S1).
A digital image of each plot was taken weekly for at least
7 weeks beginning on Day 2 after ice melt. Images were taken
with a digital camera (RX1000 III; Sony Corp.) set at a 1/30-s
shutter speed, F6.3 aperture, 200 ISO, and 5500 k white bal-
ance atop a custom 60.3 cm ×90.8 cm ×64.8 cm LED light
box emitting 90 μmol m−2s−1. Images were later uploaded
to Turf Analyzer 1.0.4 (Green Research Services, LLC) to
determine percent green cover.
2.4 Statistical analysis
Our analysis was designed to understand the effect of our
experimental design on observed outcomes (green cover,
WATKINS ET AL.5of11
Crop Science
Fv/Fm, and absorptivity) as well as to test how species recover
and respond to ice encasement. All statistics were conducted
using the program R version 4.3.1 “Beagle Scouts” (R Core
Team, 2023). Green coverage, plant absorptivity, and photo-
chemical efficiency were all proportion data bounded from
0 to 1 and were modeled using a generalized linear model
(Warton & Hui, 2011)withtheglm function with family set
to binomial using a logit transformation. Annual bluegrass
completely died under the ice in both years, and data for
that species were not analyzed. Model selection was guided
by reducing the AIC value, which penalizes additional vari-
ables in the model, but we still included repetition and year
in the model since they were experimental features. Explana-
tory variables included in the final model were repetition,
year, days after melt, species, and species-by-year interac-
tion. Model estimates are shown compared to Replication 1,
year 2022, Chewings fescue (this comparator was selected by
default based on numerical and alphabetical categorical vari-
ables). Model summary estimates are reported exponentiated.
The deviance table reports ANOVA information for each gen-
eralized linear model using a chi-squared option to calculate
pvalues. The McFadden pseudo-R2values were calculated
with reference to Pampel (2021). Prediction curves were
developed from the results of each model using the predict
function; since repetition was not significant, model predic-
tion curves showed only one repetition for each dependent
variable. Residuals and assumptions were assessed visually;
we found some heteroscedasticity in percent green cover, with
greater variability at larger values of percent cover. Disper-
sion was tested in each model, and we found no significant
effect of dispersion. Statistical differences among specific
species interactions were determined from the generalized
linear models using the emmeans function in the emmeans
package with no pvalue adjustment (Lenth, 2024).
3 RESULTS AND DISCUSSION
3.1 Ice encasement environment
Field-based ice encasement studies are difficult to replicate
from year to year since this stress is very dependent on
temperature and other environmental factors. We success-
fully encased the plot area in ice in both years; however,
we were not able to match ice encasement durations. To
address this limitation, we monitored O2and CO2concen-
trations under the ice throughout the study period for both
years using environmental sensing nodes (Figure 1and Figure
S1). Snow cover prior to ice encasement in 2022 likely con-
tributed to higher soil temperatures, which appears to have
affected CO2and O2levels, possibly due to microbial res-
piration (Clein & Schimel, 1995; Rochette et al., 2006). In
both years, CO2levels rapidly rose during ice encasement,
and in 2023 reached the maximum our sensors could detect
(40,000 ppm) (Table S2); extended durations of high levels
of CO2(above 40,000 ppm) have also been reported in ice-
encased forest soils in northern Finland (Martz et al., 2016).
Oxygen did not decline much during the first year of the study
but showed a steadier decline throughout the study period
in 2023. In 2022, there was a failure of O2sensing during
a period when the shallower sensors were under water (not
under ice); these data are not shown. We recorded higher O2
values at the deeper sensor placement depth (12.5 cm) com-
pared to the shallower depth (2.5 cm), while soil temperature
was similar at each of the two sensor depths (Figure 1; Table
S2).
3.2 Turfgrass performance
We quantified turfgrass performance by calculating the per-
centage of each plot that had green cover at a given data
collection time point. In both years, there were clear differ-
ences between species, with a wider range of responses in
2023 (Table 1; Figure 2). Since the number of days under ice
cover was different between years and we found differences in
percent green cover for some species, we kept the interaction
of year and species in the model. In 2022, creeping bentgrass
green cover at 43 days after ice melt was statistically com-
parable to velvet bentgrass and slender creeping red fescue.
When analyzed over the entire 2023 recovery period, velvet
bentgrass had the highest percent green cover and was sim-
ilar to slender creeping red fescue (p=0.37) (Figure 2). In
2023, creeping bentgrass percent green cover at 49 days after
melt was 33%, which was only significantly lower than vel-
vet bentgrass at 69% (p<0.01). The only species that died
in both years was annual bluegrass. No damage occurred for
any species in 2023 in the single replication of ice-free plots
(Figure S2).
Castonguay et al. (2009) manipulated gas levels of both
O2and CO2at high and low levels in all combinations
and surmised that the primary driver of damage in an ice
encasement–like environment was low O2, but also found that
the combination of low O2and high CO2was the most dam-
aging. Anoxic conditions under ice cause plants and microbes
to switch to anaerobic respiration, producing compounds such
as CO2, ethanol, lactic acid, formic acid, and butyric acid;
CO2is considered to be the most toxic product of anaerobic
respiration and a major contributor of cellular damage during
anoxia (Andrews, 1996).TheveryhighlevelsofCO
2in our
study, combined with a steady decline in O2, appear to have
affected annual bluegrass much more than the other grasses.
Injury to creeping bentgrass and slender creeping red fescue
in 2023, after both species were unaffected by ice encasement
the previous year, was possibly only present because O2lev-
els fell well below 2%. Our study did not monitor gases under
6of11 WATKINS ET AL.
Crop Science
FIGURE 1 Environmental data from sensors that were under ice in both testing years. O2is shown as percentage, CO2in parts per million, and
soil temperature in degrees Celsius. The O2sensors placed at 2.54 cm in the first winter of the study malfunctioned, and data are not reported. CO2
sensors reached maximum reading of 40,000 ppm in 2023. Vertical red solid lines are the ice-out dates, and vertical black dashed lines are 40 days
after melting in each experimental year.
TABLE 1 Modeled summary results for the three response variables of percent green cover, photochemical efficiency (Fv/Fm), and
absorptivity based on each explanatory variable. Model was a generalized linear model with response variable as a function of repetition, year, days
after melt, species, and species-by-year interaction.
Green cover Fv/Fm Absorptivity
Effect pvalue Effect pvalue Effect pvalue
(Intercept) −*** −*** −**
Rep 2 −−−
Rep 3 +++
Rep 4 −++
Year (2023) +−−
Days after melt +*** +*** +*
Creeping bentgrass +*** ++
Slender crp. fescue +** + +
Velvet bentgrass +** ++
**
Year (2023) by creeping bentgrass − + +
Year (2023) by slender crp. fescue +++
Year (2023) by velvet bentgrass + + +
Note: Estimates are compared to an intercept of Replication 1, 2022, Chewings fescue. Estimates larger than 1 indicate a positive change shown as +, and estimates less
than 1 indicate a negative change shown as −.
Abbreviation: Rep, repetition; slender crp. fescue =slender creeping red fescue.
*,**,***denote significance at the 0.10, 0.05, and 0.01 probability level.
WATKINS ET AL.7of11
Crop Science
FIGURE 2 Average percent green cover 49 days after melting in 2022 (98 days of ice cover) and 43 days after melting in 2023 (112 days of ice
cover). Letters show statistical comparisons within a year. Annual bluegrass completely died and was not modeled. Slender crp. fescue refers to
slender creeping red fescue.
all ice-encased plots and, therefore, may not fully represent
the ice-encased turf environment; however, the consistency of
plant response we found between blocks (replicates) suggests
that these microenvironments were quite similar.
In both years, velvet bentgrass and slender creeping red
fescue performed in the top statistical group for all parame-
ters. Chewings fescue had less green cover after ice melt in
2022 than all species except for annual bluegrass, while in
2023, it had less green cover than both slender creeping red
fescue and velvet bentgrass. These results differed somewhat
from those of Waalen et al. (2017) who showed that slender
creeping red fescue had poorer ice encasement tolerance than
velvet bentgrass. The two studies had some similarities (ice
cover duration, same slender creeping red fescue cultivar),
but differences in environmental parameters in Norway com-
pared to our study in Minnesota may have had a greater effect
on the slender creeping red fescue than other grasses; this
warrants further exploration and suggests that the microenvi-
ronments under which ice forms are critical to understanding
plant response.
3.3 Plant physiological responses
Each response variable was modeled as a function of repli-
cation, year, days after melting, species, and species-by-year
interaction (Tables 1and 2). Velvet bentgrass had greater PAR
absorptivity than Chewings fescue but was similar to creep-
ing bentgrass and slender creeping red fescue in 2022, while in
2023, velvet bentgrass was greater than all three other species.
PAR absorptivity has a linear relationship with normalized
difference vegetation index (NDVI) data especially when soil
and other background data are excluded from NDVI, but PAR
absorptivity is a better estimator of potential plant growth
(Asrar et al., 1992). These data indicate recovery of vegeta-
tive tissues in plots after ice encasement and that species such
as velvet bentgrass recover green tissue at a faster rate than
Chewings fescue.
Photochemical efficiency was similar among turfgrass
species in 2022, though in 2023, velvet bentgrass was greater
than Chewings fescue, and velvet bentgrass was similar to
creeping bentgrass and slender creeping red fescue. When
the 2 years were combined, creeping bentgrass photochemical
efficiency was greater than Chewings fescue (p=0.049).
Photochemical efficiency values were increasing about 5–
10 days prior to percent green cover increases (Figure 3). This
suggests that we may be able to predict potential turfgrass
recovery through the measurement of fluorescence parame-
ters. Chlorophyll fluorescence parameters such as Fv/Fm are
widely used to monitor both biotic and abiotic stress non-
invasively over time (Gorbe & Calatayud, 2012;Murchie
& Lawson, 2013). Declines in dark adapted Fv/Fm (0.00–
1.0) from the theoretical maximum of 0.84 indicate stress,
with further reduction in Fv/Fm values connected to severity,
potential mortality, and poor recovery. We monitored Fv/Fm
to assess mortality and recovery following ice encasement,
and results show that while some plots may have appeared
dead following ice melt (no green cover), Fv/Fm signals
indicated turfgrass plants were alive with potential for recov-
ery. While these data do not directly indicate the survival of
crowns on plants, increases in Fv/Fm over time as demon-
stratedinFigure3indirectly indicate whole plant survival of
8of11 WATKINS ET AL.
Crop Science
TABLE 2 Deviance table showing the relative benefit or variance explained of each modeled explanatory variable for each response variable.
The model was a generalized linear model with response variable as a function of repetition, year, days after melting, species, and species-by-year
interaction. The null value represents the individual model degrees of freedom and total deviance.
Green cover Fv/Fm Absorptivity
df Deviance pvalue df Deviance pvalue df Deviance pvalue
Null 367 229 319 55 318 32
Rep 3 1 0.90 3 0 0.95 3 0 0.96
Year 1 19 <0.0001 1 5 0.03 1 4 0.04
Days after melt 1147 <0.0001 128 <0.0001 1 3 0.09
Species 3 23 <0.0001 3 9 0.03 3 17 <0.001
Year ×Species 3 9 0.04 3 2 0.65 3 2 0.66
Note: Remaining deviances for green cover, Fv/Fm, and absorptivity are 31, 10, and 6, respectively, and McFadden pseudo-R2values for each model are 0.86, 0.81, and
0.82.
Abbreviation: Rep, repetition.
FIGURE 3 Modeled prediction curves (lines) and averaged observed data (points) of each species within both sampling years. Three response
variables are shown of photochemical efficiency (Fv/Fm), absorptivity, and percent green cover.
the species that were assessed. Similarly, the rate and magni-
tude of the increase in Fv/Fm over time provide insight into
the health and status of the turfgrasses as they recover from
winter. For example, species such as creeping bentgrass that
had a relatively higher Fv/Fm directly following ice encase-
ment exhibited a faster rate of recovery and higher Fv/Fm
and green cover at the conclusion of the experiment com-
pared to Chewings fescue that had very low Fv/Fm after ice
encasement. That rate of recovery may be directly connected
to damage sustained during ice encasement.
While Fv/Fm increased over time, indicating survival and
recovery, these data were still relatively low compared to
healthy plants. The low Fv/Fm values may be due to low-
temperature-induced photoinhibition and being exposed to
high-intensity light following ice melt. Similar observations
have been made in Scots pine (Pinus sylvestris) where Fv/Fm
was quite low following winter damage yet slowly recovered;
trees exhibited Fv/Fm values that were higher and recov-
ered faster when shaded (Ottander & Öquist, 1991). The use
of Fv/Fm will benefit turfgrass researchers studying winter
stresses. Technological advances would, however, be needed
for wider use of this photochemical efficiency measurement
by turfgrass managers since current approaches would be too
expensive.
Chlorophyll fluorescence parameters have also been used
to predict recovery from drought stress before irrigation is
WATKINS ET AL.9of11
Crop Science
returned in Arabidopsis (Rico-Cambron et al., 2023) and to
predict grafting success in melons (Calatayud et al., 2013)
when a threshold is developed and defined. Because Fv/Fm
showed a faster increase over time compared to green cover,
future research may be able to develop more refined thresh-
olds that indicate not only turfgrass survival but also estimate
the percentage of visual recovery. It is typical to evaluate mor-
tality and recovery by harvesting samples and bringing them
indoors to grow. This can take quite a bit of time and requires
destructive sampling; a better approach may be to use Fv/Fm
to predict recovery several days to a week prior to any striking
changes in growth.
Photochemical efficiency correlated better with percent
green cover (0.77) than with absorptivity (0.55). The cor-
relation between percent green cover and absorptivity was
0.40. It is not surprising that the correlation between Fv/Fm
and absorptivity is low, as leaves can absorb light that
does not result in photochemistry but instead leads to non-
photochemical quenching or photoinhibition (Murchie &
Lawson, 2013). For example, average absorptivity was greater
in 2022 (0.43) than in 2023 (0.30) (p=0.02), and Fv/Fm was
different between years with an average of 0.28 in 2022 and
0.39 in 2023 (p=0.06) (Figure 3). Leaves were absorbing
similar amounts of light in both years, but due to other envi-
ronmental conditions such as low air temperature and high
light intensity, they were likely experiencing higher photoin-
hibition with a subsequent drop in Fv/Fm. Days after melting
had a lower relative impact on absorptivity than we would
have expected especially when compared to photochemical
efficiency and green cover (Table 2). This could be related
to an individual species’ unique absorptivity such as due to
leaf color and other traits such as trichomes that affect light
absorption and remission.
3.4 Recommendations and future directions
In cold-climate regions where pesticide inputs are more
restricted, such as the Nordic countries, the use of pure creep-
ing bentgrass stands is a potential risk since there are no
documented instances of creeping bentgrass cultivars being
resistant to snow mold pathogens (Gregos et al., 2011). There-
fore, turfgrass managers in those regions must rely on other
species as important components of turfgrass stands. Our
results suggest that slender creeping red fescue is likely a
good option for these situations, and Chewings fescues should
be acceptable if the ice encasement period is not prolonged
well beyond 100 days. To reduce risk of ice encasement
injury while balancing other turfgrass performance needs on
a golf course, managers should consider mixtures of slender
creeping red fescue, Chewings fescue, and creeping bentgrass
(Hesselsøe et al., 2022). Our results also clearly reinforce
that annual bluegrass populations should be reduced when-
ever possible on greens in cold climates; while this can be an
economically challenging undertaking, allowing mixed stands
that include annual bluegrass (when well less than 50% of
the stand) to be exposed to regular winter conditions (i.e.,
not covering greens surfaces) should help reduce populations
over time. Our conclusions are based on single cultivars of
each taxon; screening of a wider set of cultivars for some
of these species could help identify better options for turf-
grass managers to use, although there is not strong evidence
for intraspecific variation for ice encasement tolerance in the
species we studied.
Gendjar and Merewitz (2023) studied soil moisture effects
on ice encasement survival and found that when soil water
content was at 8% prior to 40 days of ice encasement, recov-
ery of the plant, as measured by green tissue, was over 70%
higher than when soil moisture was at either 12% or 16% soil
water content prior to the ice encasement. This difference not
only points to a possible management strategy for turfgrass
managers aiming to reduce ice encasement injury risk, but it
also suggests that the environmental conditions under which
ice forms may be critical to turf survival. Understanding these
environments will be important if turfgrass researchers are to
find solutions for reducing the risk of this important stress.
A future experiment, which includes different durations
of ice cover in the experimental design, could analyze the
number of ice days as a factored variable, days after melt or
growing degree days, and species as a three-way interaction
to identify if there is a significant recovery adjustment for
species among different levels of ice treatments over time.
The number of ice days is likely a proxy for environmen-
tal variables such as O2and CO2, which we hypothesize
would likely give more accurate prediction of survivability
than the number of ice days since environmental measure-
ments are accounting for thermal and atmospheric properties
within the soil. Tompkins et al. (2004) offered that ice removal
could begin at 45 days of encasement for annual bluegrass
and might not be needed for creeping bentgrass; however, our
results, along with communications from superintendents in
northern latitudes, suggest even creeping bentgrass can suc-
cumb to ice encasement stress under the right combination of
environmental factors.
We were able to monitor environmental conditions under
ice and use a variety of methods to assess recovery after a
long ice encasement period. This study identified grasses best
suited for surviving this stress and should help practitioners in
cold climates make informed turfgrass species choices for golf
greens. Our study was not designed to answer several impor-
tant questions. At what level does CO2become too high for
annual bluegrass survival? How many days can annual blue-
grass survive at the highest CO2levels? Above what CO2:O2
ratio does creeping bentgrass get damaged or die? Each of
these questions will require a dependable environmental mon-
itoring approach so that the mechanisms underlying plant
10 of 11 WATKINS ET AL.
Crop Science
responses can be discovered and leveraged to develop new
management strategies and improved cultivars to reduce ice
encasement for turfgrass managers in cold climates.
AUTHOR CONTRIBUTIONS
Eric Watkins: Conceptualization; funding acquisition;
project administration; writing—original draft; writing—
review and editing. Dominic P. Petrella: Conceptualization;
methodology; writing—original draft; writing—review and
editing. Trygve Aamlid: Conceptualization; funding acquisi-
tion; methodology; writing—original draft; writing—review
and editing. Dominic C. Christensen: Data curation; formal
analysis; methodology; writing—original draft; writing—
review and editing. Sigridur Dalmannsdottir: Investigation;
methodology; writing—original draft; writing—review and
editing. Andrew P. Hollman: Conceptualization; data cura-
tion; investigation; methodology; writing—original draft;
writing—review and editing. Gary Deters: Investigation;
methodology; writing—original draft; writing—review and
editing.
ACKNOWLEDGMENTS
This research was supported by the Scandinavian Turfgrass
and Environment Research Foundation and Research Council
of Norway’s grant no. 310090 to the Norwegian Golf Fed-
eration’s project ICE-BREAKER. Partial support was also
provided through the USDA National Institute of Food and
Agriculture Specialty Crop Research Initiative under award
no. 2021-51181-35861.
CONFLICT OF INTEREST STATEMENT
The authors declare no conflicts of interest.
ORCID
Eric Watkins https://orcid.org/0000-0001-7225-1747
Dominic P. Petrella https://orcid.org/0000-0003-2483-
5013
Dominic C. Christensen https://orcid.org/0000-0002-
9481-5635
REFERENCES
Andrews, C. J. (1996). How do plants survive ice? Annals of Botany,
78(5), 529–536. https://doi.org/10.1006/anbo.1996.0157
Asrar, G., Myneni, R. B., & Choudhury, B. J. (1992). Spatial het-
erogeneity in vegetation canopies and remote sensing of absorbed
photosynthetically active radiation: A modeling study. Remote Sens-
ing of Environment,41(2), 85–103. https://doi.org/10.1016/0034-
4257(92)90070-Z
Beard, J. B. (1964). Effects of ice, snow and water covers on Kentucky
bluegrass, annual bluegrass and creeping bentgrass. Crop Sci-
ence,4(6), 638–640. https://doi.org/10.2135/cropsci1964.0011183İ
000400060026x
Beard, J. B. (1965a). Bentgrass (Agrostis spp.) varietal tolerance to ice
cover injury. Agronomy Journal,57(5), 513–513. https://doi.org/10.
2134/agronj1965.00021962005700050037x
Beard, J. B. (1965b). Effects of ice covers in the field on two peren-
nial grasses. Crop Science,5(2), 139–140. https://doi.org/10.2135/
cropsci1965.0011183İ000500020011x
Bonos, S. A., & Huff, D. R. (2013). Cool-season grasses: Biology and
breeding. In J. C. Stier, B. P. Horgan, & S. A. Bonos (Eds.), Turfgrass:
Biology, use, and management (pp. 591–660). ASA-CSSA-SSSA.
https://doi.org/10.2134/agronmonogr56.c17
Calatayud, Á., San Bautista, A., Pascual, B., Maroto, J. V., & López-
Galarza, S. (2013). Use of chlorophyll fluorescence imaging as
diagnostic technique to predict compatibility in melon graft. Scien-
tia Horticulturae,149, 13–18. https://doi.org/10.1016/j.scienta.2012.
04.019
Castonguay, Y., Thibault, G., Rochette, P., Bertrand, A., Rochefort, S.,
& Dionne, J. (2009). Physiological responses of annual bluegrass
and creeping bentgrass to contrasted levels of O2and CO2at low
temperatures. Crop Science,49(2), 671–689. https://doi.org/10.2135/
cropsci2008.08.0484
Clein, J. S., & Schimel, J. P. (1995). Microbial activity of tundra and
taiga soils at sub-zero temperatures. Soil Biology and Biochemistry,
27(9), 1231–1234. https://doi.org/10.1016/0038-0717(95)00044-F
Cohen, J., Pfeiffer, K., & Francis, J. A. (2018). Warm Arctic episodes
linked with increased frequency of extreme winter weather in the
United States. Nature Communications,9(1), 869–869. https://doi.
org/10.1038/s41467-018-02992-9
Gendjar, M., & Merewitz, E. (2023). Leaf, root, and crown tissue phys-
iology of annual bluegrass after cold acclimation at varying soil
moisture levels and ice encasement. Journal of the American Society
for Horticultural Science,148(3), 99–107. https://doi.org/10.21273/
jashs05288-22
Gorbe, E., & Calatayud, A. (2012). Applications of chlorophyll fluores-
cence imaging technique in horticultural research: A review. Scientia
Horticulturae,138, 24–35. https://doi.org/10.1016/j.scienta.2012.02.
002
Gregos, J., Casler, M. D., & Stier, J. C. (2011). Resistance of closely
mown fine fescue and bentgrass species to snow mold pathogens.
Plant Disease,95(7), 847–852. https://doi.org/10.1094/pdis-11-10-
0791
Guadagno, C. R., Ewers, B. E., Speckman, H. N., Aston, T. L., Huhn,
B. J., DeVore, S. B., Ladwig, J. T., Strawn, R. N., & Weinig, C.
(2017). Dead or alive? Using membrane failure and chlorophyll aflu-
orescence to predict plant mortality from drought. Plant Physiology,
175(1), 223–234. https://doi.org/10.1104/pp.16.00581
Hesselsøe, K. J., Heltoft, P., Espevig, T., & Aamlid, T. (2022). Seed
mixtures of red fescue and colonial, creeping, or velvet bentgrass for
pesticide-free management of Nordic golf greens. International Tur-
fgrass Society Research Journal,14(1), 701–707. https://doi.org/10.
1002/its2.105
Hoffman, L., DaCosta, M., Ebdon, J. S., & Zhao, J. (2012). Effects of
drought preconditioning on freezing tolerance of perennial ryegrass.
Environmental and Experimental Botany,79, 11–20. https://doi.org/
10.1016/j.envexpbot.2012.01.002
Hulke, B., Watkins, E., Wyse, D., & Ehlke, N. (2008). Freezing tolerance
of selected perennial ryegrass (Lolium perenne L.) accessions and
its association with field winterhardiness and turf traits. Euphytica,
163(1), 131–141. https://doi.org/10.1007/s10681-007-9631-z
WATKINS ET AL.11 of 11
Crop Science
Kalaji, H. M., Schansker, G., Ladle, R. J., Goltsev, V., Bosa, K.,
Allakhverdiev, S. I., Brestic, M., Bussotti, F., Calatayud, A.,
Dąbrowski, P., Elsheery, N. I., Ferroni, L., Guidi, L., Hogewoning,
S. W., Jajoo, A., Misra, A. N., Nebauer, S. G., Pancaldi, S., Penella,
C., .. . Zivcak, M. (2014). Frequently asked questions about in vivo
chlorophyll fluorescence: Practical issues. Photosynthesis Research,
122(2), 121–158. https://doi.org/10.1007/s11120-014-0024-6
Kvalbein, A., Waalen, W. M., Bjørnstad, L., Aamlid, T. S., & Espevig,
T. (2017). Winter injuries on golf greens in the Nordic countries: Sur-
vey of causes and economic consequences. International Turfgrass
Society Research Journal,13(1), 604–609. https://doi.org/10.2134/
itsrj2016.09.0826
Lenth, R. (2024). emmeans: Estimated marginal means, aka least-
squares means (R package version 1.10.2.) [Computer software].
CRAN. https://rvlenth.github.io/emmeans/
Martz, F., Vuosku, J., Ovaskainen, A., Stark, S., & Rautio, P. (2016).
The snow must go on: Ground ice encasement, snow compaction and
absence of snow differently cause soil hypoxia, CO2accumulation
and tree seedling damage in boreal forest. PLoS ONE,11(6), Article
e0156620. https://doi.org/10.1371/journal.pone.0156620
Merewitz, E. (2021). Current outlook on ice encasement stress and man-
agement strategies in turfgrasses. HortTechnology,31(5), 561–565.
https://doi.org/10.21273/horttech04907-21
Minnesota Department of Natural Resources. (2024). Minnesota State
Climatology Office. Minnesota DNR. https://www.dnr.state.mn.us/
climate/index.html
Murchie, E. H., & Lawson, T. (2013). Chlorophyll fluorescence analysis:
A guide to good practice and understanding some new applications.
Journal of Experimental Botany,64(13), 3983–3998. https://doi.org/
10.1093/jxb/ert208
Ottander, C., & Öquist, G. (1991). Recovery of photosynthesis in winter-
stressed Scots pine. Plant, Cell & Environment,14(3), 345–349.
https://doi.org/10.1111/j.1365-3040.1991.tb01511.x
Pampel, F. C. (2021). Logistic regression: A primer. SAGE Publications.
https://doi.org/10.4135/9781071878729
R Core Team. (2023). R: A language and environment for statistical
computing. R Foundation for Statistical Computing. https://www.R-
project.org/
Rico-Cambron, T. Y., Bello-Bello, E., Martínez, O., & Herrera-Estrella,
L. (2023). A non-invasive method to predict drought survival in Ara-
bidopsis using quantum yield under light conditions. Plant Methods,
19(1), 127. https://doi.org/10.1186/s13007-023-01107-w
Rochette, P., Dionne, J., Castonguay, Y., & Desjardins, Y. (2006).
Atmospheric composition under impermeable winter golf green pro-
tections. Crop Science,46(4), 1644–1655. https://doi.org/10.2135/
cropsci2005.05-0074
Runck, B. C., Schulz, B., Bishop, J., Carlson, N., Chantigian, B., Deters,
G., Erdmann, J., Ewing, P. M., Felzan, M., & Fu, X. (2024). Real-
time geoinformation systems to improve the quality, scalability, and
cost of internet of things for agri-environment research. arXiv.https://
doi.org/10.48550/arXiv.2403.19477
Schulz, B., Runck, B. C., Hollman, A., Piotrowski, A., & Watkins, E.
(2024). A hardened CO2 sensor for in-ground continuous measure-
ment in a perennial grass system. arXiv.https://arxiv.org/abs/2409.
06828
Tompkins, D., Ross, J., & Moroz, D. (2004). Effects of ice cover
on annual bluegrass and creeping bentgrass putting greens. Crop
Science,44(6), 2175–2179. https://doi.org/10.2135/cropsci2004.
2175
Waalen, W. M., Espevig, T., Kvalbein, A., & Aamlid, T. S. (2017). The
effect of ice encasement and protective covers on the winter survival
of six turfgrass species on putting greens. International Turfgrass
Society Research Journal,13(1), 556–565. https://doi.org/10.2134/
itsrj2016.05.0432
Warton, D. I., & Hui, F. K. C. (2011). The arcsine is asinine: The analysis
of proportions in ecology. Ecology,92(1), 3–10. https://doi.org/10.
1890/10-0340.1
SUPPORTING INFORMATION
Additional supporting information can be found online in the
Supporting Information section at the end of this article.
How to cite this article: Watkins, E., Petrella, D. P.,
Aamlid, T., Christensen, D. C., Dalmannsdottir, S.,
Hollman, A. P., & Deters, G. (2025). Recovery of five
cool-season turfgrasses following long-term ice
encasement. Crop Science,65, e70053.
https://doi.org/10.1002/csc2.70053
