Dryad DOI: https://doi.org/10.5061/dryad.dncjsxm9x

Main file: PhenologySynthesis.zip

Note that this package contains temperature data from different sources and R codes required for calculating forcing change, expected response, budburst temperature, and spring warming, examining statistical variations among different research approaches, species origins, climate types, and growth forms, identifying climatic, phenological, biological variables that strongly influence plant phenological responses in spring, and generating tables (Tables 1 and 2) and figures (Figure 1).

• Forcing change (degree-days above 0 °C), expected response (days), budburst temperature (°C), and spring warming (change in average spring temperature, °C) are calculated using daily mean temperature (°C) by individual study due to variations in temperature format among different data sources and the R codes used are in the folder "R codes for individual studies".
• Temperature data are listed by sources in three folders: European data, USA data, and other data.
• Phenological data for six Chinese cities, BaoDing, BeiJing, HarBin, HohHot, ShenYang, and XiAn, come from China’s National Earth System Science Data Center and are listed in the folder "other data". Variables "Leaf bud opening and Flower bud opening" are used for leafing and flowering, respectively. Empty cells represent missing values in the original observations

Four supplementary data files are provided.

• TableS1_Study references contains references for 66 studies used in this synthesis.
• TableS2_Study summary contains detailed information on each study, including Study Reference, Region (country), Altitude (m a.s.l.), Latitude range, Longitude range, Latitude centre, Longitude centre, Growth form,
  Research approach, Years, Species (n), MAT, Spring event (leafing, flowering, or both), Spring warming, Source of weather data, Number of weather stations used, Lon (Latitude range), and Lat (Latitude centre).
• TableS3_MetaData contains all variables used in the analysis, including Study (reference), Location, Study area, Research approach (observational or experimental), Precipitation (mm), MAT (°C), Altitude (meters above sea level),
  Latitude, Climate (boreal or temperate), Origin (native or exotic), Growth form (tree, shrub, herb, or grass), Event (leafing or flowering), Species, Spring phenology (date of spring event with baseline climate), Forcing change
  (degree-days above 0°C), Budburst temperature (°C), Observed response (days), Expected response (days), Phenological lag (days), and Spring warming (change in average spring temperature °C).
• TableS4_Weather data lists sources of temperature data for each study.

I. Calculations of forcing change, expected response, budburst temperature, phenological lag, and spring warming for each observed response, using Study1 as an example

Because of variations in temperature data formats among weather stations and periods of baseline and warmer climates among studies, the calculations of forcing change, expected response, budburst temperature, phenological lag, and spring warming are coded by individual study with R scripts differentiated by study reference number (i.e., Study1).

Each R script contains 6/7 steps, and steps 1-3 vary with individual studies in order to extract temperature data in different formats.

Step 1: Read daily temperature data. Fahrenheit is converted to Celsius. Empty cells represent missing values in original temperature observations.

	For example, temperature data for Study1 near Washington (USA), Data1.csv, came in CSV format. 

Step 2: Calculate daily mean temperature by Julian day for baseline and warmer climates (treatments/periods).

	For study 1, the baseline climate is 1970-1980, and the warmer climate is 1990-1999, as defined in the original study. Occasionally, baseline and warmer climates are not identified in observational studies; observational data are split by the start of apparent warming in the 1990s or into early (baseline) and later (warmer) periods.  

Step 3: Calculate cumulative degree days of baseline and warmer climates above threshold temperature (zero Celsius) since January 1.

Step 4: Calculate forcing change as the difference in cumulative degree days at the date of species events with baseline climate.

	For Acer negundo in study1, the date of flowering with baseline climate is 94 (Julian day) and the associated species forcing change is 80, which is calculated from the difference in cumulative degree days between baseline and warmer climates on day 94 (V2.1 and V.1 in Matrix T4).
          
	Similarly, the date of flowering with baseline climate for Acer rubrum is 71 (Julian day), and the associated species forcing change is 74 (see Matrix T4).      

Step 5: Determine expected response by the number of days in difference between baseline and warmer climates in reaching the threshold cumulative degree days of baseline climate.

	For Acer negundo in study 1, the warmer climate reaches the threshold level on day 86, 8 days before the baseline climate on day 94 (V3 in T4). Thus, Acer negundo flowering is expected to advance 8 days, assuming no changes in phenological constraints by insufficient chilling, photoperiod, or stresses.
	
	Similarly, the expected response for Acer rubrum is 15 days.   

Step 6: Calculate budburst temperature as the average temperature within the window of expected response with a warmer climate. For Acer negundo in study1, budburst temperature is 10.0 °C and calculated from the forcing change (80) divided by the expected response (8).

	Similarly, the budburst temperature for Acer negundo is 4.9 °C.   

Step 7: Calculate spring warming as the difference in average spring temperature between baseline and warmer climates if average warming is not provided in original studies.

	For study1, the reported increase in December-May daily minimum temperature was 1.2 and 0.2 Celsius at two weather stations within the area of observations. Our calculation with long-term weather data available from a different station indicated an increase of 0.72 °C in January-June daily mean temperature. As the area of observations is large, an average of 1.1 spring warming is adopted, slightly lower than 1.2 °C suggested by the authors and the average spring warming of observational studies compiled in this synthesis (see Table 2).     

Phenological: The difference between the expected response and the observed response. This calculation can be done in Excel.

lag: Phenological lag is 2.7 days for Acer negundo and 2.2 days for Acer rubrum, i.e., observed responses are smaller than expected based on forcing change and budburst temperature.

II. Statistical variations and influencing variables (Data analysis.R in the folder "R codes for individual studies")

Statistical variations are examined separately by potential sources of research approach, species origin, climate, and growth form, for each of the species events (leafing and flowering) using linear mixed-effects models. The codes for statistical testing are also included.

Influencing climatic, phenological, and biological variables are identified through multiple regression by species events (leafing and flowering), and variable influences are evaluated by relative AIC. The results of the final models are listed in TABLE 1. The codes for statistical testing are also included.

IV. Associated data

Weather data: Temperature data from the United States is in the folder "USA data," and other areas, including Europe and China, are in the folder "Other data."

R codes: R codes used for individual studies are listed in the folder "R codes for individual studies".

Supplementary data: All compiled and calculated data are listed in the file "Supplementary File 1".

For additional information, please send an email to rongzhou.man@ontario.ca.