PART 2
*BY MAREN WELLENREUTHER, ANNA RUNEMARK AND MIKKEL BRYDEGAARD
Did you know that species can be measured as aerosols? Using lasers and dark field spectroscopy for remote classification of animals.
To improve the
temporal and spatial resolution of insect movement, we started to think about
novel ways to use species color information to track individual movement and
interactions. At one of our meetings, the Physicists suggested to use LIDAR
technologies and dark field spectroscopy to track the movement of individual
damselflies as they had discovered that they could be treated just like large
bio-aerosols. Laser radar or LIDAR
stands for Light Detection And Ranging and is typically used to remotely
measure the size distributions and composition of aerosols in the atmosphere.
Recently, fluorescence LIDARs were developed for the detection of biological aerosols in
relation to bio-warfare such as anthrax, but they are slowly getting discovered
by other scientists and are starting to
be applied outside this area. In this context, it should be noted that Radar technology is of limited in insect detection because the
small size of insects means that they are often too small to be detected by the
Radar-wavelengths and hence no discrimination can be provided. , LIDARs, on the
other hand, have the capability to capture fine scale temporal and spatial
insect movements, as well as capture multiple color/polarisation details. Due
to these characteristics, LIDARs can capture multiple movements in time and
space and classify flying aerosol particles (e.g. insects) into discrete groups
(e.g. species and sexes). We have shown these applications of LIDAR
technologies in field feasibility studies here (Brydegaard, Guan et al. 2009), here (Guan, Brydegaard et al. 2010) and here (Brydegaard, Lundin et al. 2010).
Figure from (Runemark, Wellenreuther et al. 2012). Setup of the experiment. T: Newtonian telescope. C: Calibration site, B:
Dark box termination. Light is collected by a 1mm UV fiber in the focal plane
of the telescope and fed to a compact spectrometer. Distance from telescope to
termination was 95 m of which the first 55 m cover grassland and the remaining
40 m cover the river Klingavälsån in Skåne, Sweden.
In addition to the use
of LIDARs, it is also possible to use sunlight-based passive sensing methods to
study flying insects. One passive sensing method, which is more economic and
thus might be of interest to research groups with little funding, makes the use
of sunlight and uses the back-scattering of animals (reflexes) that intersect
the field of view. This method is called dark
field spectroscopy (see e.g.
http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=06134630). By analyzing the
spectral signature (color) of the reflex, one can classify insects to species
level and even separate sexes (if they differ in color). By applying this
method, we could determine the sex of each damselfly individual that
intersected the field of view, and this allowed us to study, for instance, sex
specific activity patterns in relation to temperature and wind speed (Runemark, Wellenreuther et al. 2012). This setup also has the advantage of
capturing interactions that occur at fast temporal scales, which, in contrast,
are impossible to capture with the human eye. Just think of individual
damselflies chasing each other above the river surface. In such a situation,
where many damselflies belonging to different species and sexes interact in a
dynamic interplay, the human eyes are not capable to capture the temporal and
spatial resolution of such species interactions. The passive sensing setup,
however, allowed us to study both pre-copula mating interactions between a male
and female crossing the river (a female crosses the telescopes field of view
approximately 0.1 seconds after a male) and males chasing after females (a male
crosses the telescopes field of view approximately 1 second after a female).
Another interesting
application of electro optical systems is the study of nocturnal migration in birds.
Traditionally, nocturnal bird migration has been studied using RADARs, thermal
cameras or moon watching. LIDAR systems, however, can significantly improve on
those methods because LIDARs capture information about the color and plumage
microstructure of migrating species. With this color information, it is then
possible to classify some of the birds to the species level, as it is shown in
the Figure below.
Lessons learned: What we as Biologists have learned from working with physicists
Color is of utmost
importance in many animal systems, and it is therefore clear that many
biologists wish to study and quantify color. The first step when doing so is to
realize that this is not a trivial undertaking, and that we all need to first
take a step back and carefully think about what we really want to do. Think of
the photons as living beings, where are they born? Where do they go, what do
they do and how do they change before their death when your detect them? What
part of the animal do we want to study, what spatial and temporal resolution do
we want, which animal is the receiver of the color signal and so on.
Where from now? Future perspectives
Some of the
applications described in this blog post have implications for the study of
vector-borne diseases, the study of pollination, quantification of habitat
availability for conservation issues, presence/absence of agricultural pests,
and the study of predator prey interactions. We are collaborating with a
Pan-African spectroscopy network, and in September, we will hold a workshop on
realistic remote sensing methods for entomology. The aim is to remotely
determine the flight direction, blood meal sizes and individual age of insects.
We are looking for new biological questions that these spectroscopy groups,
which are dispersed over 6 African countries, could address using setups that
are similar to that described above (Runemark, Wellenreuther et al. 2012). We,
therefore, greatly welcome comments and ideas in response to this blog post.
The member countries are from east Kenya, Ghana, Burkina Faso, Mali, Ivory
Coast and Senegal. If you wish to collaborate with any of the research groups,
please contact Mikkel.Brydegaard@fysik.lth.se.
Some basic advice for biologists that would like to measure color
- Read up on the vision system of your study species to make sure that the spectral domains you study are actually what they – or for instance their predators use. This might mean that you need to adjust your measuring setup to the question posed (e.g. if you study sexually selected traits the vision system of the study species is of interest, whereas that of the predator is more relevant if you want to study crypsis).
- Standardize the setup, including the light environment whenever possible. It is also whortwhile to spend a little time on making a standardized photobox because this will save you time in the long-run, when it comes to the evaluation of your data. In addition, using standardized setups ensures that you obtain more accurate data.
- When standardized illumination cannot be used, use a ColorChecker reference board to characterize the spectral bands of your imaging system for the given situation.
- When correcting for the spectral flatness and the illumination profile, you should get diffuse reflectances between zero and one.
- To measure ‘color’ of background environments, measurements that are taken over an entire day cycle are ideal. If that is not possible to achieve, try to take your measurements at the same time of the day, facing the same cardinal direction. Considering the radiative transport in the atmosphere, vegetation, canopies and water environment are complex issues that the remote sensing community has had to deal with for decades. In fact, there are even journals that are dedicated to this particular topic.
- Use polarization filters in front of the flash and objective to avoid specular reflexes – the shiny white reflex on, for instance, a tomato. These reflexes do not contain any valid information about the color of the tomato, and we need to, therefore, avoid them. This principle is, however, not applicable to species which have structural colors, such as the Calopteryx damselfly species that we study. For an example of how color can be quantified for such animals, see (Brydegaard, Guan et al. 2009).
- If the character studied is represented in a biologically relevant way by the RGB bands in e.g. commercially available cameras, the best way to make use of the information is to use the information given in each pixel. This provides information on, for instance, the variance in addition to the mean, see (Brydegaard, Runemark et al. 2012) for an example.
- Be aware that a large number of commercially available multispectral and hyper spectral imagers are capable of capturing continuous spectra in every pixel, or images at all wavelength visible to your study species. Systems with a few spectral bands can easily be built at home using cameras and filters wheels of the shelf.
- It is important to take color vision into account when designing experiments. It is of limited use with, for instance, mate choice trials in glass or plastics boxes for a species with a sexually selected ornament in the UV domain, as glass and plastic might filter out the relevant part of the UV spectrum.
- Don’t be afraid to knock on the door of the Physics department at your facility, they might just have the equipment you are looking for. Starting a communication process with physicists can help with your ideas and experiments, but this communication can also help the physicists to think about new problems that are biologically relevant. The problems that biologists face when studying the natural world the are seldom trivial and the challenge that these questions bring to a physicist might be both stimulating and entertaining from a Physics and engineering perspective.
