CONE WARMING TREATMENT AND TEMPERATURE MEASURES
Over the course of three sequential growing seasons (2014, 2015, and 2016), cone-warming treatments were deployed in tree canopies to develop, evaluate, and refine the cone-warming method presented here. In each deployment year, SWWP seed cones were passively warmed during the periods of fertilization and seed maturation, i.e. the full final growing season in the 27-month seed cone production cycle . During the 2014 deployment, bagging materials were compared for their ability to warm seed cones in canopies, methods were developed for quantifying the warming effect, and warmed and unwarmed (control) seeds were collected for use in a common garden demonstration of seedling responses. During the 2015 deployment, the best bagging material based on performance during the 2014 season was evaluated for the temperature effect achieved by the cone-warming treatment and control groups at five new stands. The 2016 deployment was conducted to compare the effect of the cone-warming treatment on the temperature of seed cone tissues and the air surrounding seed cones, and to determine whether seed cone temperatures could be reliably deduced from measures of air temperature in canopies and at the ground level.
During the 2014 deployment (n = 20 trees in three stands throughout the San Francisco Peaks in northern Arizona), three to five controls and three to five cone-warming treatments were deployed in tree canopies. Each control and cone-warming treatment contained at least two seed cones. The cone-warming treatment in 2014 compared the efficacy of two materials for warming seeds: (1) a non-porous, insulative bag composed of translucent plastic bubble-wrap packaging material (Fig. 2A) inside of a low-airflow fine porous polyester pollination bag (Fig. 2B), and (2) a glassine bag. Bags were affixed to branches with Velcro tape. The warming effects of the two materials were not statistically different, and the bubble-wrap bagging material was preferred due to its greater durability. No bagging material was placed over control-group seed cones in 2014. Air temperature was measured inside and outside cone-warming treatments using HOBO loggers (ONSET© HOBO V2 TidbiT Temperature Logger, Part # UTBI-001), suspended from the middle of a 10 cm long segment of white 2.54 cm diameter PVC tubing to shade loggers from direct insolation (Fig. 3), and hung from a branch with PVC tubes positioned laterally. In one of the three stands studied in 2014, two trees were affixed with one HOBO inside treatment bags (n = 2) and one HOBO outside treatment bags to record ambient air temperature (n = 2).
In 2015, cone-warming treatments and controls were deployed with HOBO loggers to quantify the effect of the treatment on air temperature at five additional stands (n = 1 treatment and n = 1 control per stand). The bubble-wrap material, which was found to be the most durable cone-warming bag type in the 2014 deployment, was the sole type of warming bag used in the 2015 deployment. Loss of treatment bags from branches during the 2014 deployment prompted us to use cable ties in 2015. Control and cone-warming treatment bags were loosely fitted around the cones, and bags were affixed to tree branches proximal to the cones by plastic cable ties placed over a ~5 cm segment of polyethylene foam pipe insulation used to increase the tree branch surface area affected by the cable tie (Fig. 1). Small branches and needles that spanned the pipe insulation barrier ensured channels for gas exchange. Whereas the 2014 deployment did not include a bag for the control, we included a control treatment bag from 2015 onward due to changing to the use of cable ties in order to ensure that the pressure that was exerted on branches was similar across cone-warming treatments and controls. The control treatment consisted of a high-airflow porous mesh nylon bag (Fig. 2C), while the cone-warming treatment consisted of the combined non-porous, insulative bubble-wrap packaging material (Fig. 2A) inside of a polyester pollination bag (Fig. 2B), as described above. Paired logged data (i.e. data from one cone-warming treatment and one control in a single tree) were retrieved from three of the five stands, whereas data from the fourth stand could only be retrieved from the control group and data from the fifth stand could only be retrieved from the cone-warming treatment due to loss of loggers during the course of the experiment (n treatment = 4, n control = 4).
We conducted a final experiment during the 2016 growth season to assess whether increased air temperatures inside cone-warming bags also increased the temperature of cone tissues. In contrast, only the temperature of air surrounding seed cones was measured during the 2014 and 2015 deployments, and not the temperature of seed cones themselves. In late May of 2016, cone-warming treatments and controls and two types of sensors were deployed in three P. strobiformis tree canopies 110 m from a weather station at Hart Prairie Preserve near Flagstaff, Arizona (35°21'06.0"N, 111°44'05.0"W). This experiment enabled evaluation of the effect of the cone-warming treatment on the temperature of seed cones using thermocouples, and to determine whether canopy air temperatures (measured with HOBO loggers) or air temperatures near the ground (measured with a thermistor 1.5 m aboveground) could be used to reliably estimate seed cone temperatures. In the canopies of three pines, three cone-warming treatment replicates and three control replicates were deployed. Each replicate contained at least two seed cones. A thermocouple was inserted into one cone within each control and cone-warming treatment bag to evaluate the effect of the cone-warming treatment on seed cone tissue (n treatment = 3, n control = 3). Thermocouple wires were inserted approximately 2 cm deep into seed cones. Each treatment and control bag in each tree contained one HOBO to evaluate the effect of the cone-warming treatment on air temperature within the bag, except in one of the three trees which received one HOBO in a cone-warming treatment. We obtained n treatment = 6 and n control = 4 HOBO data streams. Thermocouples logged temperature at five-minute intervals, and HOBOs logged temperature at hourly intervals. Temperature data were recorded from July – September.
CONE WARMING TREATMENT TEMPERATURE ANALYSES
The effect of the cone-warming treatment on cone tissue temperature was determined by fitting a linear model with the warmed cone temperature as the response variable and control cone temperature as the independent variable. We also compared the influence of the cone-warming treatment on the air temperature inside bags by fitting a linear model to HOBO logger data from inside the warming bag as the response variable and data from control bags as the independent variable. For the analysis of thermocouple data, measurements from the three cone-warming treatments and three control cones were averaged at each time point, then aggregated to daytime (7am – 7pm) and nighttime (7pm – 7am) average values. For HOBO logger measurements, replicates were averaged for each tree (n = 2 for the control, n = 3 for the cone-warming treatment), then an average value determined for all trees (n = 3). Values from HOBO loggers were also aggregated to daytime and nighttime averages. To compare measurements from thermocouples and HOBO loggers, we fit a linear model with warmed cone tissue temperature as the response variable and inside-bag air temperature as the independent variable. The passive warming treatment is most effective when incoming shortwave radiation inputs are greatest, hence models were fit separately for temperature values logged during day and night to more accurately quantify the daytime warming effect. We also calculated standard differences between average maximum monthly temperatures recorded in cone-warming treatment and control groups across all deployment years. To calculate standard treatment differences, we first estimated average maximum daily temperature per measurement and treatment type (e.g. thermocouple measurement in cone-warming treatment versus control) across all replicate measurements per year, calculated an average monthly maximum temperature from daily average maximum temperatures, and then calculated differences between the control group and warming group values.
COMMON GARDEN EXPERIMENT
Seeds collected at the end of the 2014 cone-warming deployment were used in the common garden experiment. Following cone collection, cones were bench-dried in a greenhouse and extracted seeds were weighed in five replicated sets of ten seeds to estimate an average seed mass. Seeds were sown in the greenhouse in early October 2014 with subsequent greenhouse transplanting on November 18, 2014. Seeds were sown into labeled SC10 container growth tubes (Stuewe & Sons, Inc.; 3.8 cm diameter × 21 cm deep, 164 mL volume) in a completely randomized design across populations, genetic families, and cone-warming treatments. Seedling emergence occurred between 1 - 6 weeks following sowing. Seedlings were grown in the greenhouse for five months under ambient daylight conditions plus high pressure sodium lights to achieve a consistent 15 hr. day : 9 hr. night photoperiod. Seedlings were watered every other day and fertilized twice a week with 20-20-20 NPK fertilizer. Irrigation and fertilizer solutions were brought within a pH range of 5.5 to 6.2 using food grade phosphoric acid. Seedlings were placed outside of the greenhouse, and fertilization was ceased one month before outplanting to prepare seedlings for field conditions. Seedlings were then watered to keep the soil medium consistently moist. Replicates of each seedling experimental group (population, family, and cone-warming treatment) were planted into 1.2m x 1.2m raised bed garden boxes constructed at the Arboretum at Flagstaff Southwest Experimental Garden Array site (35.1603° N, 111.7309° W). Soil medium in the boxes consisted of 50% Cornell soil mix (one-part sphagnum peat moss, one-part horticultural perlite, and one-part coarse vermiculite), and 50% volcanic cinders sourced from The Landscape Connection, Flagstaff. Just before planting, each raised garden bed was inoculated with one shovel-full of a mixture of soils gathered from all seed-source stands to include native soil microbes in the garden boxes. Eighty-one experimental seedlings were transplanted in a randomized design across both boxes in a 9 × 9 arrangement on June 6, 2015. Seedlings were spaced approximately 12 cm apart and competing vegetation was routinely removed to eliminate competition for light. Extra (i.e. non-experimental) seedlings were planted along box edges to buffer experimental seedlings from heat radiated by the sides of the raised-bed boxes during the day. Edge seedlings were clipped two years after planting to avoid unintended effects of belowground competition. An average of 8 seedlings were planted per each of the 20 seed source trees included in the common garden. Between one and 18 seedlings remained per seed source tree after the first year of growth. Each garden box was hand-watered using a spray wand fitted to a hose to apply 3.79 L of water every 7-10 days between the months of April and November.
Seedlings were grown for four summers until harvesting during the spring of the fifth season, on May 2, 2019. Traits measured in the common garden included (1) plant growth above-ground (measured annually) and below-ground (measured once during transplanting and once post-harvest), (2) multispectral and thermal indices via an unpiloted aircraft system (UAS) measured during the summer in 2017 and 2018, as in , and (3) morphotypic mycorrhizal nodulation (measured post-harvest in 2019). Plant growth traits included plant height measured as the distance from soil level to the top of the top-most bud on the central stem, diameter at root collar (DRC) measured as seedling stem diameter at soil-level, full shoot length measured as the distance from root collar to the top-most bud, full root length measured during transplanting to raised-bed garden boxes before the first summer of growth, root and shoot dry-mass measured post-harvest, and dates of bud development. Calculated plant growth traits included mean annual height and DRC growth increments (mean change (∆) in measure each year for both height and DRC), root-to-shoot length and mass (root measure divided by shoot measure, completed for both length and mass-based measures), yearly slenderness (shoot length divided by DRC), and full ∆ height and ∆ DRC (final measure minus initial measure, divided by initial measure). Multispectral and thermal infrared sensors carried by UAS recorded spectra at one timepoint at midday on May 18, 2017 and again at midday on June 2, 2018. Near infra-red spectra were used to estimate seedling crown temperatures, corresponding leaf-to-air temperature differences, and spectral indices indicative of plant health including the normalized difference vegetation index (NDVI), green NDVI (GNDVI), normalized difference red edge index (NDRE), triangular greenness index (TGI), and green-red vegetation index (GRVI). Post-harvest and before roots were dried, mycorrhizal fungi on seedling roots were assessed to the level of morphotype to determine whether cone-warming affected mycorrhizal assemblages. Mycorrhizal assemblages can affect plant performance , and mycorrhizal fungal species richness can be estimated by assessing mycorrhizal morphotypes . Percent ectomycorrhizal fungal (EMF) colonization and EMF diversity were estimated on up to 100 root tips per seedling, noting (1) dead root tips, (2) live root tips, (3) dead EMF tips, and (4) live EMF tips. Each living EMF tip was assigned a morphotype designation based on color, texture, shape, and external hyphal characteristics following .
COMMON GARDEN STATISTICAL ANALYSES
Multivariate and univariate models were used to investigate statistical relationships between the cone-warming treatment and response variables. Effect sizes of cone-warming on responses were then estimated as described below. All analyses were conducted in R (version 3.5.1, R Core Team 2018). Seedlings in the common garden demonstration from each type of cone-warming bag (glassine versus plastic bubble-wrap packaging material, both inside of a polyester pollination bag) were treated the same because there was no statistically significant difference between the effect of the two types of cone-warming bags on seedling traits. Multivariate models were built using both principal component analysis (PCA; via the function prcomp) and permutational multivariate analysis of variance (PERMANOVA; via the function adonis) separately for the following three categories of response variables: (1) plant growth traits including bud phenology, (2) foliar spectra, and (3) mycorrhizal assemblages. The first principal component of the PCA generated from variables, belonging to one response category at a time, was used as the response in linear mixed effect models. Next, PERMANOVAs were executed using Euclidean distance matrices composed of aggregated response variables for each of the three response categories (plant growth, spectra, and mycorrhizae), separately, specifying seed source tree as a random effect. Univariate linear mixed effect models were also fitted for all response variables, specifying seed source tree nested within stand, and raised-bed box (where models would allow), as random effects using functions from the R package lme4. In both multivariate and univariate models, seed mass was tested for inclusion as a covariate via AIC comparisons. Models with the smallest AIC were favored, and when models competed with AIC values within 2 AIC units of the smallest AIC, the simplest model structure with the least predictors was selected for subsequent ANOVAs. Satterthwaite approximation of denominator degrees of freedom was specified for all omnibus F-tests of fixed effects as well as type III sum of square ANOVAs for models that included interactions between seed mass and warming treatment. Type II sum of square ANOVAs were specified for models that included seed mass as an added covariate. The magnitude of the variance explained by the cone-warming fixed effect was estimated by Cohen’s Local f 2, which is suitable for use with mixed models for which denominator degrees of freedom must be approximated, and is suitable for use with unbalanced experimental designs [34,35]. Input for the calculation of Cohen’s Local f 2 includes marginal R2 goodness-of-fit values both from models with and without the factor of interest, as follows: