Remote sensing as a tool to determine population genetics

Blonder et al. just had their paper entitled “Remote sensing of ploidy level in quaking aspen (Populus tremuloides Michx.)” published in Journal of Ecology. In this blog post, Benjamin Blonder explains how remote sensing can be used to determine population genetics of aspen in the Rocky Mountains (Colorado, USA).


Can you tell much about the genetics of a population just by looking at a landscape? Our new study suggests the answer is ‘yes’, based on measurements of the iconic tree of western North America, quaking aspen (Populus tremuloides). 

Monodominant quaking aspen forests near Marcellina Mountain, western Colorado.

Aspen plays a keystone role in many ecosystems, providing nesting cavities for thick-billed parrots in Mexico and forage for deer in Alaska. In the Rocky Mountains, aspen is the only significant broad-leaf tree species, and occurs in large monodominant stands that can cover whole valleys and mountainsides. However, the species is also experiencing high mortality in many parts of its range in recent decades, driven by extremely hot and dry conditions. Identifying ‘winner’ and ‘loser’ genotypes is a priority for this species, and may also become a priority for many others.

Fall phenology is largely determined at clone level, making clone boundaries easy to
visualize in autumn.

The genetics of aspen are complex and may help explain its large range size and the origin of contemporary mortality events. First, aspen can grow in large spatially extensive genetic clones, e.g the Pando clone in Utah that covers more than 100 acres. Second, aspen can occur as either diploid or triploid, meaning that clones may have either two or three copies of each chromosome. Both the clonality and the ploidy level variation in this species may be important in determining its response to changing climates. But traditionally, assessing both at landscape-scale is time-consuming and expensive – for example requiring DNA microsatellite analysis, or flow cytometry, or next-gen sequencing. 

In this study we examined the potential of remote sensing to detect variation in ploidy level across individuals and landscapes. The premise of the method is that genetic variation could drive phenotypic variation, which could then influence the spectral reflectance of different plant tissues (like bark and leaves). This reflectance variation could then be leveraged by a classifier algorithm. Previous studies have been able to use spectral data to successfully discriminate among species – the ambition here was simply to do discrimination of genetic variation within species.

To find out, we set up a range of sites in aspen forests in southwestern Colorado, spanning from high-desert habitats to near alpine treelines. At each, we collected canopy leaves using slingshots, and later obtained ground-truth ploidy level measurements for each using DNA microsatellite or flow cytometry methods. And at each, we also tried three different methods for assessing spectral variation. 

We first measured the bark and leaves at each site using a handheld spectrometer. Bark color variation among stems is often visually evident, and leaf greenness also seems to change as one walks through a forest. We also overflew each site using a multispectral camera on a drone, and captured high-resolution imagery of the canopies at each site. 

Henry Pai pilots an unoccupied aerial system to capture multispectral imagery in aspen
forest.

This project involved a number of undergraduate and graduate students in education programs based at the Rocky Mountain Biological Laboratory, who had a fun time exploring these diverse landscapes and learning a new set of tools. We all had a lot of fun bringing batteries and computers and other equipment out to some improbable mountain slopes. 

We started the project in a fit of curiosity, and as a small gamble. We weren’t expecting to find anything. But the results were exciting. We found that the leaf-level and the canopy-level data had enough spectral variation to predict ploidy level at reasonable levels of accuracy. And we found that the visually-evident variation in bark color (aspen bark is photosynthetic!) was not associated with ploidy level, leaving a mystery as to what factors determine its spectral properties. 

Left: false-color multispectral image of forest canopies; right: inferred ploidy levels.

The results of course should be taken as more thought-provoking than anything else. We had a fairly limited sample size (a few hundred points) and forests comprising far more triploids than diploids. We don’t know if they generalize elsewhere, but we hope and expect that they will.

The results on the whole suggest that some genetic information is indeed visible from remote sensing, opening the potential for larger landscape-scale studies of ploidy level variation. Many other species besides aspen vary in ploidy level – sagebrush, creosote, many range grasses, and more than a few tree species. I’m excited to see where else this approach may become useful.

Benjamin Blonder, School of Life Sciences, Arizona State University, Arizona, USA

One thought on “Remote sensing as a tool to determine population genetics

  1. Pingback: New paper: remote sensing of ploidy level – Macrosystems Ecology Laboratory

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s