Cardiocrinum leaf distribution  Feed back button

As I've mentioned on previous pages, the arrangement of the leaves differs on different forms of Cardiocrinum.

This section concerns attempts I've made to see if these differences can be quantified.

For the plants that flowered in 2014, the leaf positions on the stem were measured as follows,

This data was then normalised by calculating Δdn/h

The data was entered on an Excel sheet and a scatter chart displayed.

Cordatum image

The Excel TrendLine facility was then used to plot possible lines of best fit based on both linear and polynomial functions.[1]

Cordatum image

The order of the equations of best fit were recorded with their R2 values and the y-axis intercept for the best fit in the table below for giganteum and cordatum types.[2] The cell of the selected best fit order has a grey background. The plant codes are colour coded as on the map and databases pages.

 giganteum types 

PlantHeightR2 order 1R2 order 2R2 order 3R2 order 4Intercept, (%)
yunnanense24900.96870.99790.99800.998914.2
96o111100.99720.99720.99870.99887.7
giganteum19700.98260.99280.99430.995912.9
105o2 A12500.99000.99000.99880.99887.4
105o2 B13900.99640.99850.99880.9992-0.39
55o224000.99650.99730.99800.99896.4
101p0 A8050.98190.98610.99260.99286.5
101p0 B4900.98640.99280.99430.995912.9

 cordatum types 

PlantHeightR2 order 1R2 order 2R2 order 3R2 order 4Intercept, (%)
15p010300.90320.99650.99860.998229.9
90o1 A11100.97520.99710.99810.99993.7
90o1 B10500.97730.99920.99830.99982.3
44o113300.99620.99790.99830.99975.6
38o114000.93420.99070.99740.998023.2
59p017250.94460.99020.99740.998028.5
84o19150.54610.85260.97710.997825.1

 Analysis 

The first thing to make plain is that this data can't be used to identify plants. Whilst there seem to be some trends, there is nothing definite enough to give an identity.

The giganteum types do have a tendency to distributions which match a linear graph. The exceptions to this are the two uncoded plants and the pair of C. cathayanum (hort.), 101p0, which needed cubic & quadratic expressions! In the latter case, the reason may be linked to the small size of these plants. They were grown from imported, dried bulbs which only started to grow in May and then flowered in late June!

The cordatum types have a more irregular leaf pattern, leading to quadratic or higher order functions being needed to describe them. The exception to this is the C. cordatum glehnii, 44o1. However, 90o1 and 44o1 are both descendents from our original cordatum, 4p0, which shows that there are variations within a generation as these are both offsets from the original plant.

There is a page showing some of the plants, together with their graphs here.

The leaf distribution shown by 15p0, 38o1, 59p0 and 84o1 are very similar to that described by Wilson for C. cathayanum and it would be interesting to make this analysis of a cathayanum, if I could find one! This does raise doubts concerning the use of leaf distribution for the identification of C. cathayanum.

[1] Care must be taken with polynomials as, with a data set of n items, a polynomial of order n-1 will fit ALL points, (R2 = 1), without being able to make ANY prediction of a value outwith this set.
[2] The criteria for a 'best fit' was taken as an R2 value ≥0.99, (99%), and the intercept value was taken from the first function to give this value.