Open-top chambers for temperature manipulation in taller-stature plant communities

Abstract

1. Open-top chambers simulate global warming by passively increasing air temperatures in field experiments. They are commonly used in low-stature alpine and arctic ecosystems, but rarely in taller-stature plant communities because of their limited height.
2. We present a modified International Tundra Experiment (ITEX) chamber design for year-round outdoor use in warming taller-stature plant communities up to 1.5 m tall. We report a full year of results for the chambers' effects on air and soil temperature, relative humidity, and soil moisture in a northern hardwood forest clearing and a southern early successional grassland site located in Michigan, USA. Detailed construction plans are also provided.
3. The chambers elevated daytime air temperatures at 1 m height by 1.8°C above ambient levels, on average over an entire year, at both the northern and southern site. The chambers did not affect relative humidity at either site. The chambers did not alter average soil temperature or moisture at the northern site and reduced soil temperatures and soil moisture at the southern site. The chambers increased variability in soil freeze/thaw cycles at both sites.
4. The chambers achieved predicted levels of warming for mid-century (2046–2065) scenarios consistent with the majority of representative concentration pathways in the International Panel on Climate Change Fifth Assessment Report, with minimal experimental artefact. This design is a valuable tool for examining the effects of in situ warming on understudied taller-stature plant communities and creates the opportunity to expand future comparisons across a diversity of systems.

1 INTRODUCTION

Climate warming can have profound impacts on ecological communities (Parmesan & Yohe, 2003; Root et al., 2003). Warmer temperatures affect individual growth and metabolism, shift development times, and alter survival and reproductive success (Brown, Gillooly, Allen, Savage, & West, 2004). Individual responses can lead to community level changes through shifts in population dynamics and species distributions (Walther, 2010), and the combination of these responses can be difficult to predict. Therefore, it is critical to collect empirical observations to develop a deeper understanding of ecological community responses to warming essential to inform the sustainable management of natural systems.

Several experimental tools exist that allow researchers to manipulate air and/or soil temperature of in situ plant communities (Aronson & McNulty, 2009). Active warming methods (e.g. overhead infrared lamps, heat resistance cables) are used to achieve a pre-determined level of warming (Harte et al., 1995; Norby et al., 1997; Pelini et al., 2011; Peterjohn et al., 1994). These methods have been useful for examining above and below-ground ecological responses to warming; however, they often require extensive energy demands, start-up costs, and maintenance that are often not feasible for remote and budget limited research projects (Aronson & McNulty, 2009).

Passive warming methods such as open-top chambers (OTCs) are often used to warm above-ground systems, especially in arctic and alpine ecosystems (Marion, 1996; Walker et al., 2006). Their simple, cost-effective design makes them ideal for field experiments that require replicate plots across environmental gradients or in remote areas. Passive OTCs also allow natural levels of precipitation, light, and gas exchange (Marion, 1996). Further, they enable winter warming, a phenomenon that is likely to impact plant community dynamics and demands further research (Bokhorst et al., 2010; Kreyling, 2010). Passive OTCs provide a low-maintenance and cost-effective method for researchers to examine the effects of year-round warming on plant communities (Arft et al., 1999).

Despite their utility, the limited height of passive OTCs (c. 0.4 m) has restricted their use to low-stature plant communities, early life stages, and low productivity plant systems (Elmendorf et al., 2012). While some taller-stature OTCs exist, they are typically only used during the growing season and not in harsh winter conditions (Chiba & Terao, 2014). Thus, there is a gap in understanding in situ warming effects on communities such as prairies, savannahs, grasslands, and scrublands (Settele et al., 2014).

Here, we provide a durable, passive, OTC design that warms taller-stature (≤1.5 m height) plant communities. We monitored the effects of 24 chambers on the abiotic environment over 1 year in two different temperate ecosystem types (forest clearing and old-agriculture field) spanning approximately 3° of latitude. We report the chambers' effects on air and soil temperature, relative humidity, and soil moisture during different timeframes: year-round, in the growing season vs. the dormant season, and during daytime vs. nighttime hours. We aim to facilitate the use of passive OTCs across understudied taller-stature systems and expand knowledge of climate warming effects in a wider array of natural ecosystems.

2 MATERIALS AND METHODS

2.1 Study sites

We tested 24 chambers at two sites separated by 380 km and approximately 3° latitude in Michigan, USA from 1 August 2015 to 31 July 2016. The southern site, located within Kellogg Biological Station's Long Term Ecological Research site (42º24′40.11″ N, 85º22′24.46″ W, 289 m a.s.l.), is a former agriculture field last cultivated roughly 40 years prior and mowed annually until 2014. The average annual air temperature (1981–2010) is 10.2°C (Menne et al., 2012). The site's vegetation is dominated by Solidago spp., Poa pratensis, and Phleum arvense and reaches approximately 1.5 m height (K. B. Welshofer, P. L. Zarnetske, N. K. Lany, Q. D. Read, unpubl. data). The northern site, located at the University of Michigan Biological Station (45º33′40.38″ N, 84º40′46.54″ W, 239 m a.s.l.), was located within a 1.5 km² northern deciduous forest clearing created roughly 25 years prior and mowed annually until 2014. The average annual air temperature (1981–2010) is 5.5°C (Menne et al., 2012). Vegetation is dominated by Centaurea stoebe, Poa compressa, and Pteridium spp. and reaches approximately 0.5 m (K. B. Welshofer, P. L. Zarnetske, N. K. Lany, Q. D. Read, unpubl. data).

2.2 Chamber and experimental design

We elevated the hexagonal ITEX OTC design described in Marion (1996) onto 0.91 m vertical polycarbonate walls with pressure-treated wood framing (Figure 1; plans in Appendix S1). The resulting chamber dimensions were 1.57 m tall × 2.5 m diameter with a top opening large enough to contain a 1 m² plot. We used clear, 0.32 cm Lexan polycarbonate sheets without UV protective coating to allow high solar transmittance and natural ultraviolet conditions (ePlastics, San Diego, CA, Item ID: PCCLR0.125AM48X96). We elevated the polycarbonate 10 cm above the ground surface to allow migration of ground dwelling organisms. We used cold and UV resistant cable ties to attach the frame to the polycarbonate panels through drilled holes. We screwed pressure treated plywood supports to the top 120º corners of the hexagon and attached 25.4 cm metal spikes to anchor each leg in the soil for extra support during high-wind and snowfall conditions.

Figure 1

(a) Photograph of open-top chamber (OTC) in a taller-stature old agriculture field in Hickory Corners, MI, USA. (b) Photograph of OTC in an early successional northern hardwood forest clearing in Pellston, MI, USA. (c) Perspective view. (d) Aerial view

Warmed and ambient treatments were randomly distributed across 24, 1 m² plots arranged within a 25 × 36 m deer exclosure at each site. Each OTC was centred on a warmed plot and ambient control plots were left untreated, with a 1 m buffer area on each side of each plot. As a part of a longer term fully factorial experiment, plots additionally received insect and small mammal reduction treatments (Appendix S2).

2.3 Abiotic measurements

At the plot-level, we recorded hourly abiotic conditions from 1 August 2015 to 31 July 2016 with HOBO products (Onset Computer Corporation, Bourne, MA). Three chambered and three ambient plots at each site were instrumented with four-channel external U12-008 data loggers that recorded air temperature at 10 cm above the soil surface and 5 cm below the soil surface, and with Microstation H21-002 data loggers that recorded hourly air temperature and relative humidity at 1 m above the soil surface along with soil moisture at 5 cm below the soil surface. We installed data loggers (Pendant UA-002-64) to record air temperature at 1 m height in the remaining nine chambers at each site. We installed solar shields above each air temperature sensor (Appendix S2).

2.4 Statistical analyses

We analysed the data on air temperature (1 m and 10 cm heights), relative humidity (1 m), soil temperature (−5 cm) and soil moisture (−5 cm) for each site separately. All analyses were completed, using R version 3.1.2 (R Development Core Team, 2008). We used two sided, unpaired t-tests to compare the means of the hourly data (n_chambered = 3; n_ambient = 3) during the following intervals: year-round (24 hr), daytime, nighttime, growing season, and non-growing season. We assessed variation in the warming treatment by calculating the standard deviation of hourly air temperature (1 m) across the remaining nine chambered plots. We compared the variability in winter and spring soil temperature of chambered vs. ambient plots with an F-test such that .

We obtained daily snow depth, snowfall, and maximum 5-s wind speed records at both sites during the study period of August 2015–July 2016 and from the National Centers for Environmental Information to examine the chambers' ability to withstand wind speed and heavy snow conditions (Menne et al., 2012). Data were averaged across weather stations throughout Emmet County, Michigan for the northern site, and across Kalamazoo County, Michigan for the southern site.

3 RESULTS

Throughout the year, the average air temperatures at 1 m height at both sites were warmer in the chambers than in control plots (northern site chambers: increased 0.84 ± 0.25°C SE; southern site chambers: increased 0.70 ± 0.18°C; Figure 2a). Warming at 1 m varied according to irradiance; thus, the greatest magnitude of warming was exhibited during daytime hours (+1.84 ± 0.79°C and +1.73 ± 0.19°C at the northern and southern sites, respectively), and especially during sunny days (Figure 3), as well as during the growing season (Figures 2a, 4a, b). The results of all t-tests are given in Table S2 (northern site) and Table S3 (southern site).

Figure 2

Mean and 95% confidence intervals for the difference between chambered and ambient plots for (a) air temperature at 1 m above ground, (b) air temperature at 10 cm above ground, (c) relative humidity at 1 m above ground, (d) soil temperature at 5 cm below the soil surface, and (e) soil moisture at 5 cm below the soil surface from 1 August 2015 to 31 July 2016. Significant differences from unpaired t tests (n = 6) denoted with (*). Some confidence intervals not visible due to small size

Figure 3

Time series showing variability in chamber effects on (a) air temperature at 1 m height, (b) relative humidity at 1 m height, and (c) soil moisture at 5 cm below the soil surface due to daily weather patterns during a typical week during the growing season (1–7 May 2016) at the southern site in Hickory Corners, MI, USA. The chambers warm air temperatures most during the day during clear sky conditions. The values shown are the average hourly values of the warmed plots (solid red) and the ambient plot (dashed black)

Figure 4

Frequency of hourly daytime temperatures (°C) at 1 m height in chambered and ambient plots during the growing season at the (a) Northern Site and (b) Southern Site. Chambered plots were warmer than ambient plots, especially on the hottest days of the year

Air temperatures (1 m) measured in the nine remaining chambers showed that the OTC warmed consistently across plots, with slightly more variation between treatments in the southern old-agriculture field (northern site: median hourly standard deviation of 0.02°C; southern site: 0.23°C). The chambers did not significantly warm the air at 10 cm above the soil at either site, although nighttime air temperature at 10 cm was slightly warmer in the chambers at the southern site (Figure 2b). The chambers did not significantly alter relative humidity at 1 m at either site (Figure 2b).

The chambers' effects on soil temperature and moisture varied between the two sites. The soil was cooler (−0.21 ± 0.10°C, Figure 2d) and drier (−3.66 ± 0.57%, Figure 2e) year-round in chambers vs. on ambient plots at the southern site. However, we found no significant difference in soil temperature and moisture between chambered and control plots at the northern site (Figure 2d, e). The chambered plots exhibited greater variability in soil temperature during the non-growing season at the southern site (F = 0.02, df = 2, p = .05), with two additional spring freeze/thaw cycles than in the ambient plots (Figure 5).

Figure 5

Hourly soil temperatures (°C) at −5 cm beneath the soil surface during the winter to spring transition for a warmed and a control plot at the (a) Northern and (b) Southern sites. In winter, chambered plots exhibited greater temperature variability, cooler soil temperature, and increased freeze thaw cycles than ambient plots likely due to decreased snowpack insulation. In general, soils were colder, experienced greater variability in temperature, and experienced more freeze thaw cycles at the southern site than at the northern site during the winter

The chambers withstood harsh weather conditions with minimal damage to the infrastructure. During the study, the northern site experienced a maximum 5-s wind speed of 26.4 m/s and annual snowfall of 244 cm with a maximum snow depth of 63.5 cm. The southern site experienced a maximum 5-s wind speed of 24.6 m/s and annual snowfall of 127 cm with maximum snow depth of 22.9 cm.

4 DISCUSSION

Our chamber design simulated mid-century global warming scenarios in taller-stature plant communities. We observed mean daytime warming of 1.8°C at both sites, consistent with all representative concentration pathway (RCP) scenarios for 2046–2065 and three RCP scenarios for 2081–2100 predictions from the International Panel on Climate Change Fifth Assessment (Stocker et al., 2014). This amount of warming is consistent with past warming experiments using passive, low-stature ITEX chambers, and is known to change the phenology, growth, survival and reproduction of low-stature plants (Arft et al., 1999).

The increased variability in soil temperature caused by the chambers led to an increase in freeze/thaw cycles also predicted to occur with climate change, likely due to reduced or intermittent snowpack providing decreased insulation (Brown & DeGaetano, 2011; Henry, 2008). This phenomenon was more pronounced at the southern site, where there is less snowpack that is more likely to completely melt between snowfall events than at the northern site where the greater snowpack was likely decreased but not absent. Soil freeze/thaw cycles lead to the lysis of soil microbes, resulting in changes in abundance and community structure of soil bacteria (Kumar, Grogan, Chu, Christiansen, & Walker, 2013). The death of these microbes may also release nitrogen and phosphorus in the soil leading to plant uptake and response (Edwards & Jefferies, 2010).

The chambers led to overall cooler and drier soil in the warmed plots vs. ambient plots in the southern site, but not in the northern site. We suggest greater productivity in the southern site's chambered plots led to increased soil shading and transpiration, resulting in cooler and drier soil on the chambered plots as observed in Hollister, Webber, Nelson, and Tweedie (2006). At the northern site, moisture likely drained quickly from the sandy soil on all plots regardless of warming treatment (x̄_{Northern Site} = 11.0 ± 1.9%; x̄_{Southern Site} = 20.5 ± 0.5%). This regional variation highlights the importance of an affordable warming method to assess variability in warming responses across many sites.

The relatively simple and low-maintenance design of these taller-stature chambers encourages their use in conjunction with globally coordinated ecological experiments such as Nutrient Network and Drought-Net (Fraser et al., 2013). By combining experiments, researchers will gain the opportunity to investigate the interactive effects of numerous global change drivers in situ across taller-stature systems worldwide.

ACKNOWLEDGEMENTS

K.W. was supported by the National Science Foundation Graduate Research Fellowship Program and Michigan State University (MSU). P.Z. was supported by MSU and the USDA National Institute of Food and Agriculture, Hatch Project 1010055. N.L. was supported by the Arnold and Mabel Beckman Foundation and MSU. We also thank MSU's Forestry Department, the University of Michigan Biological Station, Kellogg Biological Station and their Long-term Ecological Research Program (NSF DEB 1027253). Many thanks to P. Duffy and M. Chansler for help with the experiment and to two anonymous reviewers.

AUTHORS' CONTRIBUTIONS

Conceived and designed the experiment: P.Z., K.W., N.L., L.T. Performed the experiment: K.W. Analysed data: K.W. and N.L. Wrote manuscript: K.W., P.Z., N.L. All authors contributed critically to the drafts and gave final approval for publication.

DATA ACCESSIBILITY

Microclimate data used in this manuscript are deposited in the Dryad Digital Repository https://doi.org/10.5061/dryad.2br8k (Welshofer, Zarnetske, Lany, & Thompson, 2017).