See full methods in Pastore et al. (2024), Ecology and Evolution
We established 3 transects in each of 3 sites: (1) Medium Valley in the Dartmouth College Second College Grant in northern New Hampshire (44.925853˚, -71.08554˚); (2) Deep Valley in Vermont’s northern Green Mountains (44.384743˚, -72.85045˚); (3) Shallow Basin in the Nulhegan Basin portion of the Silvio O. Conte National Fish and Wildlife Refuge in the Northeast Kingdom of Vermont (44.846635˚, -71.75817˚). Each transect contains 5 to 6 plots (0.04 ha each) across an elevation gradient.
We measured sub-canopy air temperatures in each plot with shielded Thermochron iButton temperature sensors (DS1922L, Maxim Integrated Products, Inc., Sunnyvale, CA, USA). Two sensors per plot were hung from separate, randomly selected trees 1.5 m above the ground surface. Temperatures were logged in 1-hour intervals over the following periods: Shallow Basin: September 25, 2021 – March 12, 2022; Medium Valley: August 21, 2023 – June 30, 2022; Deep Valley: August 21, 2021 - June 7, 2022 at DV-1 and August 1, 2021 – June 7, 2022 at DV-2 and DV-3.
In each plot, we recorded the species of all living overstory trees with diameter at breast height (DBH) >= 10.16 cm at the time of plot establishment and, within a central, circular subplot (0.01 ha), the species of all living understory trees with DBH < 10.16 cm and height > 30.5 cm. For each plot, we calculated CTI (community temperature index) separately for the overstory and understory communities as the abundance-weighted mean (by species proportions using tree counts) of preferred temperature values of all overstory or understory tree species present. To explore which tree functional groups were driving CTI patterns, we assigned species to categories based on whether they are coniferous vs. broadleaved and further divided the broadleaved group into whether their temperature preferences fall within the range of the conifers (“cold-broadleaved”) vs. above the range of the conifers (“warm-broadleaved”).
We determined soil depth (organic + mineral) by inserting a metal probe 1.8 m from plot center at 60, 180, and 360˚ azimuths until reaching a restrictive layer and averaged the three measurements per plot. We collected 8 mineral soil cores (0-10 cm) per plot using a soil corer (AMS mini soil probe; 2 cm diameter; AMS, Inc.). We combined the cores, transported them to the lab on ice, and homogenized the soil within each plot by sieving (mesh size 2 mm). We measured soil pH using a 10 g wet soil subsample (Mettler Toledo S220 SevenCompact benchtop pH/ion meter in supernatant with soil:0.01 M CaCl2 ratio of 1:3; Mettler Toledo, LLC). We determined gravimetric soil water content by drying a 10 g wet soil subsample at 60°C for at least 48 h until a constant mass was reached. Samples were ground (BioSpec Mini-BeadBeater-96) and analyzed for total nitrogen concentrations using an elemental analyzer (Elementar UNICUBE in CN mode, Hanau, Germany) at University of Vermont’s George D. Aiken Forestry Sciences Lab.