Plant Species of the Borrego Desert: Fabaceae:
Dalea mollis and D. mollissima

Table of Contents

Analysis of the Leaf Margins


Dalea mollis and D. mollissima seem to be well-separated at times, but not-so-well-separated other times. At those other times I start wondering what percentage of plants can be reliably determined, and even if they are actually two separate species. This page was motivated by my latest episode of confusion, stimulated by discussions with others who were also confused.

My first goal was to try to understand how well-separated the species are on their leaves alone, since we often encounter plants without flowers, and try to use the leaf characteristic in the key to distinguish them. The Jepson Manual key for the leaves is:

2. ... leaflet margin generally entire ... D. mollis
2'. ... leaflet margin shallowly lobed or wavy ... D. mollissima

There are two sources of trouble in this key.

First, there is that weasel-word generally. Does it mean only 55% of the leaves are entire, and the rest are shallowly lobed or wavy? Or does it mean 99% of the leaves are entire, and only 1% are not? Does it mean that most plants have entire leaves, but that some plants have non-entire leaves? Or does every plant have entire leaves, but some plants also have some leaves that are not entire?

Second, what does "shallowly lobed or wavy" mean exactly? If the lobes are shallow, and the leaves of D. mollis are only generally entire, is there actually a continuum in how deep the lobes are, so that many plants can't be determined with accuracy? How shallow can a lobe be and still be clearly assigned to D. mollissima?

To try to resolve this trouble, I decided to measure the depth of lobes, and/or the width of the waviness, for every Dalea species with photos at iNaturalist, as well as in some of my many pix of these species.

Before I present those results, I show some typical pictures of the leaves of the two species in Fig. 1. The examples were chosen from my analysis that confidently identified the two species in these examples.

During the course of my measurements, I was amazed at the variation in color and hairiness of the leaves, and I ended up categorizing the extent of that variation to see if leaf color could aid in separating the species.

Fig. 1 shows most of that variation, with darker-colored leaves in the top row, gray-green leaves in the second row, and greenish leaves in the third row.

Dalea mollis Dalea mollissima

Tom Chester, West Borrego Mountain, 8 March 2011
Leaflet lobe depth 0.00 of leaflet length

Fred Melgert and Carla Hoegen, Borrego Badlands, 2 March 2018
Leaflet lobe depth 0.08 of leaflet length - see how this was measured

Fred Melgert and Carla Hoegen, Smoke Tree Canyon, 25 December 2018
Leaflet lobe depth 0.00 of leaflet length

Tom Chester, west Borrego Mountain, 3 April 2011
Leaflet lobe depth 0.06 of leaflet length

Fred Melgert and Carla Hoegen, June Wash, 16 December 2018
Leaflet lobe depth 0.02 of leaflet length

Matthew Salkiewicz, Glorietta Canyon, 11 March 2017
Leaflet lobe depth 0.08 of leaflet length
Fig. 1. Left: Dalea mollis. Right: Dalea mollissima. Click on the pictures for larger versions of my pictures, or for the iNaturalist link with more pictures for the rest of the pictures.

What a bewildering variation in the leaves! Clearly, the color of the leaf is utterly useless to discriminate the species.

However, note that the leaves of D. mollissima on the right are clearly lobed, obvious at a glance. In contrast, the upper two photographs of D. mollis show clearly entire leaves, very different at a glance from the lobed leaves of D. mollisima.

The last picture of D. mollis shows that its leaves can indeed be weakly lobed. There is no doubt about the determination of this plant, since this plant has flowers with a notched wing, which only occurs in D. mollis. But it seems difficult to reliably separate that plant from D. mollissima from the leaves alone.

Analysis of the Leaf Margins

I measured the maximum depth of lobes, and/or the maximum width of the waviness, for every Dalea species with photos at iNaturalist, as well as in some of my many pix of these species. To be considered for measurement, there had to be at least one good photograph showing a leaf at a large enough scale that I could reliably measure the lobe depth on my screen. In nearly all cases of acceptable observations, I could clearly see the edge of the leaf and could measure the lobe depth.

I measured the full depth of the most extreme lobe, from the maximum extent of the lobe to the edge of the leaf between lobes. Since there are no scales on the pictures, I normalized the measurement by dividing by the leaflet length. I used the leaflet length rather than the width since the length can always reliably measured in a photograph, whereas it is difficult to measure the full leaf width due to how folded the leaves are, or due to the angle of the photograph.

A very small number of photographs showed the leaf from directly on top, and hence I could not measure the depth of the lobe. In those cases, I measured the maximum waviness of the leaf edge, measuring the full extent from the edge of one wave on one side to the edge of the neighboring wave on the other side.

See example of what was measured using the D. mollissima leaf at the upper right in Fig. 1.

A total of 57 plants were measured, 40 iNaturalist observations and 17 of my observations.

Fig. 2 present the raw results of the measurements on all Dalea plants, paying no attention to their determination.

Fig. 2. Histogram of the depth of the leaflet lobes for all measured Dalea plants. See also a histogram in which neighboring bins are added in order to get better statistics in each bin, producing a smoother distribution.

These are just about the best results one could hope for to show that there clearly are two species here! Twenty plants had only entire leaves, and were clearly D. mollis. Eight of those plants also had flowers that had notched wings, confirming that determination.

There is a very smooth distribution for non-entire leaves, peaking at values of 0.05 to 0.10, and falling to zero at lower values and higher values. Most of the plants in that smooth distribution have to be D. mollissima.

Seven plants had flowers with wing petals that were not notched. Their leaflet lobe depths ranged from 0.05 to 0.10, confirming that the broad peak, and the larger values, are due to plants of D. mollissima. (The individual values were 0.05, 0.06, 0.06, 0.07, 0.07, 0.09, and 0.10.)

That leaves just nine plants with values from 0.02 to 0.05 that cannot be unambiguously assigned from this histogram alone, which is just 7% of all the plants I measured.

There were three confirmed D. mollis plants, from the notched wings, with weak leaflet lobes, at values of 0.02, 0.04, and 0.05. Those values are not very precise, since the lobe depth was just ~1 mm or smaller on my screen, and the true values are likely to be somewhat smaller since I couldn't accurately measure anything less than a mm from those photographs. Furthermore, in some of these cases, only some leaves were lobed (remember, I measured the maximum leaf lobe I could find on any leaf I could see). However, taking those measurements at face value, one can only conclude that the plants with ratios of 0.06 and higher are almost surely all D. mollissima; there are 28 plants with those values.

Conclusion: ~93% of this set of Dalea plants in the Borrego Desert can be unambiguously determined from good photographs of the leaves alone, leaving ~7% indeterminate from just using the leaves.

This isn't bad at all, and is much better than I had thought it would be. Single characteristics are rarely 100% reliable in distinguishing any species.

Furthermore, Barneby, in his 1977 monograph on Dalea, states that Dalea mollis and D. mollissima "can be separated only with difficulty by coarser methods". Most of the characteristics he gave to separate them are on the flowers; here's his leaflet separation (italics his):

4. ... margins of leaflets plane or very obscurely gland-undulate ... D. mollis
4'. ... margins of leaflets undulately gland-crenate ... D. mollissima

One could easily describe the bottom photograph of D. mollis in Fig. 1 as being obscurely gland-undulate. But without the further information from the notched wing petal, it doesn't seem possible to determine such weakly-lobed leaves to the species.

Some plants will need flowers to reliably determine them.

Go to:

Copyright © 2019 by Tom Chester.
Commercial rights reserved. Permission is granted to reproduce any or all of this page for individual or non-profit institutional internal use as long as credit is given to me at this source:
Comments and feedback: Tom Chester
Updated 6 January 2019