Left: Emma setting up the 3D Anolis replicas (excuse the yoga clothes–it was hot!), Right: 3D replica in-situ.

A lot of us have been there…. setting up what seems like endless 3D anole replicas, often in the tropical heat, messing around with countless iButtons (which are a nightmare to get out of the replicas), to measure operative temperature (Te)–the temperature of the animal at equilibrium with its environment….

As frustrating as this can sometimes be, it is an integral part of measuring thermal habitat quality and availability, which as we all know, is important for such things as ectotherm energetics, abundance and predicting species responses to climate and land cover change.

However, using these 3D replicas, we only get point-based measures of Te at randomly selected points within the survey area. These points are sampling only a very small extent of thermal habitat, and therefore may not represent the conditions mere metres away.  This method therefore does not allow us to measure Te across the whole of the survey area at spatial resolutions relevant to the individual animal. This method is also, costly in terms of both time and money. Therefore, is there another way?

Well, we do have microclimate–biophysical modelling, which generally relies on mechanistic models that downscale broad scale (usually monthly) macro-climate (≥ 1km grid) data to estimate microclimate in specific habitats, e.g.  NicheMapR, Microclima and Microclimc (Kearney and Porter, 2017; Maclean et al., 2018; Maclean and Klinges, 2021). These estimates of microclimate must then be combined with biophysical heat exchange models to estimate animal operative temperature (Te), e.g. the ectotherm model in NicheMapR (Kearney and Porter, 2020).

These models have revolutionized our ability to model thermal environments across broad spatial extents, especially for species distribution modelling, and new developments have the potential to model much finer variation (e.g. Microclimc), but applications at scales of individual organismal movement (e.g. cms to m) are still rare.

These limitations of existing methods are particularly pertinent given the established importance of spatial heterogeneity of thermal environment for species, particularly ectotherms, and by extension our beloved anoles (Huey, 1974; Sears and Angiletta, 2015; Sears et al., 2016).

Luckily, we as a team had already pondered, if the canopy is key for regulating ectotherm operative temperatures (Te), then, can we predict Te using biophysical equations relating to canopy characteristics?

Part of the field team, helping process what is certainly not an anole, whilst setting up survey plots (photo credit Adam Algar).

This was the basis of this paper, “Unoccupied Aerial Vehicles as a Tool to Map Lizard Operative Temperature in Tropical Environments.”

So, to test this, we first needed to collect canopy data – which, for anyone who has done this type of work will agree, is not so easy! This is where Unoccupied Aerial Vehicles (UAVs) come in.

UAVs have transformed how we collect data and allow us to collect canopy metrics across the whole of a survey area, rather than point based measurements using devices like Ceptometers or densiometers, from the ground.

Left: Emma flying the UAV; Right: Example RGB (True-colour) UAV orthomosaic for a plot.

We flew each plot using a standard true colour (red, green, blue) camera, on the DJI Phantom 4, and extracted percent greenness (equation below) and texture indices for the images, representing canopy presence, and heterogeneity.

We then coupled these high-resolution (<10cm) UAV-based canopy data (greenness and texture) with ground-based data (air temperature of each plot from iButtons) to predict lizard operative temperature (from 3D printed anole replicas) for an endemic anole (Anolis bicaorum), on the island of Utila, Honduras. See the workflow below.

Above: Workflow of Methods. Ta = Air Temperature, LAI = Leaf Area Index.

And we found that it works….for this anole in this system at least.

By using UAV derived canopy data and coupling it with air temperature and random forest models. We can predict the operative temperature of Anolis bicaorum at solar noon across the whole of the survey plot. Not just for 20 3D replica points!

Left: RGB UAV orthomosaic of forest survey plot, right: Random forest model output indicating lizard operative Temperature at solar noon for the same plot.

This method also gives us the benefit of mapping across a continuous spatial area, at a spatial resolution which is relevant to individual organism movement. Which until now was not possible…

Above: Left: RGB UAV orthomosaic of urban survey plot, right: Random forest model output indicating lizard operative Temperature at solar noon for the same plot.

Looking at how the model performed across land covers on the island, it performed best within forested areas and less well in other land covers, including highly urban plots. Such failures are to be expected since, in such cases, the UAV imagery is capturing variation in the ground surface, rather than the influence of shade.

Observed versus predicted Te for each plot using the Te.Air.UAV random forest model with a Jackknifing approach. Labels (A) to (P) correspond to Plot number sequentially from 1 to 16. Point colours refer to land cover where purple = forested plots, orange = urban forest plots and black = urban plots. Blue lines indicate a line of best fit.

This highlights the need to train the model further across different land covers, and in the case of urban areas, consider that the modelling approach may not be suitable due to a general lack of canopy. However, in such land cover types, we could potentially link thermal UAV data as a predictor. There is also scope to expand the model using volumetric data, such as from structure from motion photogrammetry.

This workflow and model will allow us to map ecologically relevant measures of the thermal environment across larger areas at scales relevant to the individual animals and populations, something that until now was not feasible with standard ground-based methods or with mechanistic niche modelling. This opens new avenues to understanding the impact of anthropogenic and climate change on species, especially in forests, that are dependent on suitable thermal environments, like A. bicaorum.

 

Above: Male Anolis bicaorum, endemic to the island of Utila (photo credit Tom Brown).

We would love to test how this method performs in different systems and with different species, including how we can expand the method to cover other times in the day; if you would like to collaborate on this, or have any questions feel free to get in touch with me.

Dr Emma Higgins – emma.a.higgins@hotmail.com

You can read the work in full here!

I would also just like to thank everyone again who was involved in this project, it was a lot of hard work, but great fun and it couldn’t have been done without the team effort.

Paper in collaboration with: Adam Algar, Lakehead University, Doreen Boyd, Geertje Van der Heijden and Sarah Owen, University of Nottingham and Tom Brown, Kanahau Utila Research and Conservation Organisation.

New Paper: Unoccupied aerial vehicles as a tool to map lizard operative temperature in tropical environments

DOI: https://doi.org/10.1002/rse2.393

References 

Huey, R.B. (1974). Behavioral thermoregulation in lizards: importance of associated costs. Science, 184, 1001–1003.

Kearney, M.R., Porter, W.P., 2017. NicheMapR – an R package for biophysical modelling: the microclimate model. Ecography 40, 664–674. https://doi.org/10.1111/ecog.02360

Kearney, M.R. & Porter, W.P. (2020). NicheMapR – an R package for biophysical modelling: the ectotherm and Dynamic Energy Budget models. Ecography, 43, 85–96.

Maclean, IMD, Mosedale, JR, Bennie, JJ., 2018,  Microclima: An r package for modelling meso- and microclimate. Methods Ecol Evol. ; 10: 280– 290. https://doi.org/10.1111/2041-210X.13093

Maclean, I.M.D., Klinges, D.H., 2021. Microclimc: A mechanistic model of above, below and within-canopy microclimate. Ecol. Model. 451, 109567. https://doi.org/10.1016/j.ecolmodel.2021.109567

Sears, M.W., Angilletta, M.J., 2015. Costs and Benefits of Thermoregulation Revisited: Both the Heterogeneity and Spatial Structure of Temperature Drive Energetic Costs. Am. Nat. 185, E94–E102. https://doi.org/10.1086/680008

Sears, M.W., Angilletta, M.J., Schuler, M.S., Borchert, J., Dilliplane, K.F., Stegman, M., Rusch, T.W., Mitchell, W.A., 2016. Configuration of the thermal landscape determines thermoregulatory performance of ectotherms. Proc. Natl. Acad. Sci. 113, 10595–10600. https://doi.org/10.1073/pnas.1604824113

Emma Higgins
Latest posts by Emma Higgins (see all)