Peony Bloom Date Project

Current Bloom Date Project, Part II

By Michael Denny
    In the last CPS Newsletter, an article described the history of the Bloom Date Project and the method used to collect data. In this issue, I will discuss recent developments and the interpretation of the bloom date evidence.
    The project has a web site ( ) that contains several articles about the project and the data. There are two data files that contain 11,000 observations on approximately 1100 cultivars. These files contain the same evidence. However, the difference lies in that one file sorts by the cultivar name, while the other sorts by the bloom offset. For each cultivar, there are columns showing the description, the observations and the offset. The description shows the colour but this is very elemental at the moment and so almost the only colours cited are red, white and pink. (Better cultivar descriptions are found in other sources.)
    The crucial data in this project are the observations and the offset. The observations indicate how many inspections have been collected on the cultivar’s bloom date. This data is an indication of the reliability of the evidence. The offset measures how many days the cultivar first blooms before or after the peony Red Charm. The offset data for any cultivar varies and I report the average offset. If there are only a few observations, the average offset may be unreliable. As more observations are collected, the average offset may change considerably.  Currently there are five or more observations on over 600 cultivars.  This is encouraging; but it indicates that for about 500 cultivars, there are only four or fewer observations.
    One of the reasons to continue to collect data is to increase the number of observations on those cultivars, which currently have relatively few observations. At the moment, I still have another 1100 observations on 500 cultivars that have not been included in the web site data. These should be added this coming winter and hopefully will push some additional cultivars beyond the five observations threshold.  Please note that you should not assume that the information on cultivars, which have only a few observations, is wrong. The information may be correct.  However, it is a safer indication when there have been many observations.
How to Use the Data
    Individuals should use the offset evidence with some caution. Here is how I think about this data.  Suppose I am considering three cultivars, peony A, peony B and peony C. The offsets are 7, 9 and 2 for peony A, peony B and peony C respectively. The offsets for cultivars A and B are close. So, I would consider them to bloom in my garden at the same time. It may even be that A will bloom after B in some years. However, they should bloom at about the same time. 
      Cultivar C has a lower offset and the difference is large enough that I would assume C would bloom before the other two cultivars. It may not bloom precisely five days before the others in my garden but it should bloom earlier. 
    The basic lesson is that the detailed data can be misleading; but when used with caution, it can be very helpful. One can easily choose cultivars that will bloom early in their garden as well as those which will bloom very late. 
    To my way of thinking, the data provides a ranked list of blooming for peony cultivars. At any given location and in different years the time between the blooms for any given pair of cultivars can vary. If the weather is very warm, the time shortens; if it is mild, the time expands.
    There are definite limitations to the data in our project. First, we are dependent on many volunteers to observe and report data. As a result, the following problems may arise: the identification of the cultivar may be wrong; local soil, rain and the number of hours of sunshine can alter the information. These I cannot control.  The major limitation arises from temperature variations both between sites and across years. 
    At any site, the calendar dates on which a cultivar blooms and the time between blooming for different cultivars is altered by temperature patterns. The use of offsets from Red Charm represents an attempt to address this. The offset adjustment is very useful; but it is far too simple to completely adjust for temperature variations across sites and across the years, even at the same site.  There is no simple alternative to our adjustment. In my judgement, the project provides us with a large volume of useful information but we just need to be careful in its use. 
    In recent years, there have been discussions about the creation of a new system of bloom information based on weeks. This would build on the detailed project information but create six categories that would cover all the cultivars. These efforts will be put forward in a final article in the next issue.

Bloom Date Project, Part 1

By Michael Denny

One of the important characteristics of peonies is the difference in the bloom dates of different cultivars. This variation extends the peony season and assists in the identification of cultivars. Peony catalogues, books and registration lists provide rough guidelines to the time at which a cultivar will bloom. There is no single source for this information and different sources often provide conflicting information.

When lactifloras were the dominant peonies the problem was not severe. Most lactifloras bloom over a very short period of time. As hybrids became more important, the bloom period lengthened from five to seven weeks and the value of correct information increased. Each nursery chooses its own
method of describing the bloom period. Most are variants of Early (E), Middle (M) and Late (L).

However, a cultivar that is listed as Early in one catalogue is often listed as Middle in another. The situation is even more confusedwhen some, but not all, nurseries expand their categories to include Very Early (VE) and Very Late (VL).

One could not be sure which cultivars listed as (E) in one catalogue belonged with those listed as (VE) in another catalogue.  The Bloom Date Project is an effort to improve the current situation for all peony lovers.

History From 1963-75, the Reverend F. Miller recorded the date of the first bloom for about 200 peony cultivars in Fergus Falls, Minnesota. For a given cultivar, the date varied each year depending on the spring weather. Rev. Miller chose the most common date from the 13 years, which is similar to choosing the median date. The results are published on pages 191-2 of the American Peony Society publication, APS, The Best of 75 Years.

In 1999, I entered the Miller data into a spreadsheet and posted the file on the peony group web site. Currently, this is the Yahoo peony group; but it had earlier names. The posting of the Miller data led to an extended discussion on the usefulness of this type of data. A group composed of Jim Langhammer from Michigan, Leon Pestell from Kansas, Bob Johnson from Oregon, Tom La Bron of  Oklahoma and I began investigating the possibilities of extending the Miller data to a larger number of cultivars.

Thanks to Jim Langhammer, over 20 years of observations on different cultivars at a small number of locations had been found and transformed into spreadsheets. Leon and Jim did the transformation from paper to files. At that time, we had information on the date of first bloom for about 450 cultivars. This original data can still be found as part of the Heartland Peony Society web site.  In the last decade, I have headed the effort to collect more data. Members of the Canadian Peony Society have been important partners in this task. By the end of 2012, we will have over 10,000 observations on the bloom dates of about 1000 different peony cultivars.

Method The data is collected from many individuals in different locations. At each location, the  calendar date on which the first bloom opens is recorded for each cultivar. This is not a scientific collection process.  We all know that weather impacts the time at which peonies bloom.

For example in my garden, ‘Red Charm’ has bloomed as early as May 26th and as late as June 13th in the past thirteen years.  Most of the bloom dates are clustered around June 3rd to 6th.  At any location, there will be variations in the bloom dates based on the weather at that location in that year.  Gardens that are located in warmer climates than mine will have their dates for ‘Red Charm’ shifted to an earlier period. We are confronted with variations in bloom date of ‘Red Charm’ at different locations and at the same location over time. Two methods have been used to reduce these problems. First we have converted the calendar dates into the number of days before or after ‘Red Charm’ blooms. This is called the ‘offset’. As the calendar date for ‘Red Charm’ changes, the offset for all the other peonies will change.

Second, we take averages of the offsets. This is an important step because it will tend to eliminate the extreme observations. To make averaging work, we would like to have at least five observations for each cultivar.  These two adjustments are not perfect solutions to the variation in bloom times. It is difficult to imagine a better system that is as simple. The issues are not unique to our efforts. Anyone who wants to establish accurate data on the bloom periods will confront the same problems.

Bloom Date 2004

Taken from The Peony - C to C
May 2005 Vol 8 Issue 2

by Michael Denny

     It was a very good year for the bloom date project and I want to thank all of those who participated. In 2004, we collected over 1900 observations on the bloom dates of 600 different cultivars.  This brings the overall totals to over 7200 observations on about 950 cultivars.

     Not all of the observations received this year were for observations taken in 2004.  There were several treasure troves of date from earlier years.  The publicity in the APS Bulletin helped to locate several individuals who had been carefully measuring their bloom dates in earlier years.

     It is important to have many observations on each cultivar.  I am pleased that we currently have 10 or more observations on bloom dates for 321 cultivars.  For another 154 cultivars, we have five to nine observations.  The goal is to have 500-600 for which there are ten or more observations.  It should be feasible to reach this goal in the next three years.  There is nothing unique about ten observations.  In my judgement, this amount of data allows us to have considerable confidence in the bloom date.

     There are still many cultivars, 251, with only one observation.  I have kept these in the tables but one should view the data for these cultivars with considerable caution.  As we obtain more observations on these cultivars, the bloom dates may change significantly.  There will always be many cultivars that are not widely grown for which we can not obtain numerous bloom date observations.  However if we keep collecting data we should be able to provide information on all of the widely grown cultivars.

     Mrs. Anne Oveson from Wallowa, Oregon sent me a hand-written list of over 700 bloom date observations.  Mrs. Oveson bred and registered the peonies 'Caroline Rose' and ' Mary Gretchen'.  She collected the data from her own garden and from her sister's garden in Walla Walla, Washington.  The data are from the years 1999 to 2002.  Each sister had from 100 to 150 cultivars and many were common to both gardens.  This provided four to eight new observations on the cultivars that they grew in common.

     Mrs Oveson's data are the main reason why we have twice as many new observations in 2004 than we had last year.  While we can not expect such sources to appear every year, we can continue to add more observations.

     Another historical data set was provided by Bruce Powers of Wisconsin who has data from 1996 - 2004.  To these we can add an unidentified source from Cut Bank, Montana who supplied data from 2001.  These three sources provided one-half of the observations this year.  I am trying to obtain data from another Ohio source who may have data for 50 cultivars from 1993 - 2004.

     The project depends heavily on a core set of Canadian Society contributors.   Val Ames, Brian Porter, Lindsay D'Aoust, Linda Goh and Nick Visser have provided data for three or more years.  Another dozen individuals contributed data for one or two years and we thank them all.  We can always use new contributors.

     As more data are collected there are a number of extensions to the project.  First, we would like to introduce a standard system of describing the bloom period of peonies.  This will avoid the current confusion that arises when different nurseries and different peony books describe the same cultivar inconsistently.  One possibility is to classify peonies by the week in which they bloom with week one being the very early ones and week seven being the latest.

     We know that variations in the weather cause most of the variations in bloom dates at a given location as well as the differences between locations.  We hope to look more formally at this process.

     There is more information about the  project at the we site  The data
 are available and may be downloaded and there are several articles.

     If you want to collect data this year, we welcome your help.  The process is simple.  You should only collect data for cultivars whose identity is known and that are mature plants growing in relatively full sun.  Record the calendar date of the first open bloom.  Keep a list and send them to me at

No comments:

Post a Comment