Trends in Spring Flowering Dates from Churchill, Manitoba and Northern Labrador:
An Assessment of PlantWatch North Phenological Data

Introduction

NatureWatch is a citizen science monitoring program which engages Canadians in tracking environmental phenomena to identify and better respond to ecological changes. Plantwatch (www.plantwatch.ca), part of the NatureWatch suite of programs, is a national initiative to monitor changes in the timing of blooming events of key indicator plant species; a science known as plant phenology. Phenology is the observation of the timing of biological events such as the arrival of migratory birds in the spring, and the first flowering of flowers in the spring. Studying changes in plant phenology over time has been very important to better understand our ecosystems and the way they respond and adapt to changes, disturbance and stress coming from natural or human pressures.

PlantWatch relies on volunteers to observe and record first flowering dates of specific plants from their region. PlantWatch species are selected on the basis of their ease of identification, wide geographic distribution and for their early blooming characteristics. Following a standardized methodology described on the PlantWatch website (www.plantwatch.ca), observers record the first bloom, midbloom and, in some cases, leafing dates of key indicator plants as well as location and habitat information. This data is submitted on-line into a common database. Participants come from all spheres including naturalists, professional and other “citizen scientists”. Anyone is welcome to participate.

Studies have shown that climate change and other environmental stresses are having a dramatic impact on the North. PlantWatch North, an extension of the PlantWatch program, was launched in 2002 in order to better focus efforts to track the effects of these changes in northern ecosystems. It includes the following regions: Yukon, Northwest Territories, Manitoba, Nunavut, and Northern Labrador. Bloom dates, observed and recorded by citizen scientist from across the north will help track environmental changes from region to region and from season to season.

This project examines PlantWatch data collected from Churchill, Manitoba and Northern Labrador to explore whether the first bloom dates of plants are changing over time.

Methods

Data collection and sorting

Data from the Plantwatch North program were used to conduct this assessment. The data were sorted, and only those sample points containing a first bloom date for years 1998 to 2006 were kept for analysis. Although Plantwatch North first-bloom data exist for all of the Northern provinces and territories covering six ecozones, only Churchill, Manitoba, and Labrador City and North West River, Labrador (Figure 1) have continuous datasets useable for this study, covering the Hudson Plain, Boreal Shield and Taiga Shield ecozones, respectively.

Figure 1. Locations of PlantWatch North datasets used for this analysis.

The data were sorted by ecozone, species and date of first bloom. The number of sample points for each species in each ecozone is described in Table 1. Although the datasets from the Taiga Shield and Boreal Shield ecozones (Labrador region) are small in number, they form a continuous series and are collected from reliable observers.

Table 1. Plant species analyzed in each ecozone.

Data preparation and analysis

The first bloom dates from Churchill, Manitoba, and Labrador City and Northwest River, Labrador, were transformed from calendar date into Julian date, which assigns every day of the year a number, where day 1 represents January 1, day 2 represents January 2, and so on, until day 365, which represents December 31. This was done to make bloom dates from different months and years comparable and standardized. The Julian dates were then divided into Julian day and year in preparation for the regression analysis.

The data was then sorted into ecozone, species, and year. The data for a given year for each species was then averaged to provide one data-point per year per species, for each ecozone as shown in Tables 2 &3. The data for Labrador City and Northwest River were combined together and designated as the Taiga Shield Ecozone. Initial average values from the two locations showed no significant bloom date differences (<0.5 standard deviation from the mean), for the species and years observed. The Northwest River location is part of a small disjunct area of the Boreal Ecozone imbedded in the Taiga Shield Ecozone.

Table 2. Annual average Julian day for each analyzed species from the Hudson Plains Ecozone (Churchill, MB).

Table 3. Annual average Julian day for each analyzed species from the Taiga Shield (Labrador City & Northwest River NFLB).

Microsoft Excel ® was used to analyze the first bloom data. Linear regression analysis was used to plot average Julian day of first bloom (y coordinate) against year (x coordinate).

Results and Discussion

These results reflect a limited number of sample points and very short period of data collection. Most climate studies require an average of 30 years of meteorological data. As such, the results should be considered preliminary and caution must be taken when using these results to interpret climatic or environmental change.

The results of the regression analysis for the Hudson Plains Ecosystem are shown in Figure 2. The trends were similar in pattern among the species suggesting consistency in sensitivity to temperature changes and reliability of their signal.

Figure 2. Annual average date of first bloom for species in the Hudson Plains Ecozone.

The results of the regression analysis for each species are described in Table 4. The slope of the line shows the direction of the change. The r2 value indicates how well knowing x allows you to predict y with a higher r2 indicating a tighter relationship. The p value indicates how statistically significant the relationship is with a lower p value indicating higher significance. A p value of 0.05 or less is often considered significant.

Table 4. Results of regression analysis of Julian date against year for species from the Hudson Plains Ecozone (Churchill, MB).

The trend lines for four of the five species analyzed were negative, indicating a trend toward earlier spring bloom dates. Acrtostaphylos rubra and Ledum groenlandicum demonstrated the strongest trends. The Ledum groelandicum time series was very short and did not have data for the most recent years of analysis and may be less reliable. None of the trends were statistically significant at p=0.05 level.

Figure 3 shows the average annual date of first bloom for species in the Taiga Shield Ecozone The results are summarized in Table 5. Again, the graph illustrates that while species first bloom times differ, the pattern of bloom between species is very similar.

Figure 3. Annual average date of first bloom for species in the Taiga Shield Ecozone.

Table 5. Results of regression analysis of Julian date against year for species from the Taiga Plains Ecozone (Labrador City & Northwest River NFLB).

Since 1998, the slopes of the trends were very slightly positive for Cornus Canadensis, Trientalis borealis and Syringa vulgaris. These trends were not statistically significant. Clintonia borealis and ledum groenlandicum showed negative trends, though these were not statistically significant either. This suggests that there has been little change in the bloom time for the species studied during the 2000’s. However, all species appear to show a general downward trend starting since 2002, toward earlier spring onset in both the Hudson Plains and the Taiga Shield Eco-zones.

Conclusions

The PlantWatch program is in early stages of implementation. Preliminary analysis of the few sample points and short time series available for the Hudson Plains and Taiga Shield ecozones suggest little change in spring bloom dates over time. However, more recent data since 2002 and onward show a general trend toward earlier spring bloom dates. The patterns of change for all species are quite similar, suggesting that the PlantWatch species are offering similar signals. The analysis to date does not offer statistically significant results.

Over time, further collection of data with a greater number of sample points will help detect any changes in phenology for these species and, as a result, changes in spring on-set. While all PlantWatch data are useful, longer time series data will be particularly helpful in tracking ecosystem changes. Volunteer observers are encouraged to continue tracking phenological events at their chosen locations for many years.

Acknowledgements

Analysis: Kim Monson, University of Winnipeg

Contributors and editors: Liette Vasseur, Brock University; Elisabeth Beaubien, University of Alberta; Julie Borque and Marlene Doyle, Environment Canada

Plantwatch North would like to acknowledge the support from Environment Canada, the volunteer coordinators, and the volunteer observers for their support of the Plantwatch North program. The Churchill Northern Studies Centre is acknowledged for collecting the data from Churchill, Manitoba and the Plantwatch Newfoundland/Labrador program is acknowledged for providing the data from Labrador.