Rebecca Michaels-Walker

Rebecca Michaels-Walker is the Content Marketing Specialist at Canadian Science Publishing.

Bridging data and conservation: Dan Struthers on the future of stream temperature modelling

November 21, 2024 | 9 minute read

Water temperature might not be the first thing that comes to mind when you think of conservation, but for Dan Struthers and his team at Banff National Park, it’s a critical piece of the puzzle in safeguarding aquatic species. Water temperature plays a key role in the life cycle and survival of fishes, and as climate change and habitat disruption intensify, understanding and predicting temperature patterns is crucial to protect Canada’s native species. But predicting these patterns across vast, interconnected stream networks is no easy task. How can conservationists develop practical, effective models that are accurate enough to guide on-the-ground efforts?

Dan with westslope cutthroat from Marvel Creek. Photo Credit: Gunnar Rollheizer.

That’s the challenge Struthers and his team set out to tackle in their recent paper, “Statistical Stream Temperature Modelling with SSN and INLA: An Introduction for Conservation Practitioners,” published in the Canadian Journal of Fisheries and Aquatic Sciences. From deploying tiny temperature sensors across Banff National Park’s streams to collaborating with international experts, Struthers’ team worked to develop models that balance scientific rigour with practicality for conservation work. In this interview, Struthers sheds light on the study, the international collaboration that made it possible, and the implications of their findings for the future of aquatic conservation.

First of all, congratulations on your publication! It’s a fascinating study. It’s also a complex one, integrating topics like statistical modelling, spatial autocorrelation, and dendritic networks. It’s tough stuff! If you were explaining your paper to a friend over coffee, how would you break it down for them?

With this paper, we were trying to help break down the “knowledge-action” gap. We all know data can speak a thousand words, but it’s often difficult for practitioners to find accessible options to analyze these data and get their story out there. Modelling is getting more complicated—and for good reasons—but conservation practitioners don’t have a lot of time to keep up with the latest techniques. Doing a good job of modelling ecological data is challenging because spatial correlation complicates the approach to these data, which is further exacerbated when working in complex systems like stream networks. We need accessible solutions for practitioners to work these data in a manner that doesn’t neglect assumptions like spatial independence. That’s why our focus for this paper was to try to make complex modelling tools more approachable for fellow practitioners.

Dan installing stream temperture logger in the Cascade River valley. Photo Credit: Conrad Jaeger.

Can you tell us a little more about why it’s so important for conservation practitioners to have tools like this to accurately predict water temperature?

Water temperature is a major driver for the life histories and movement patterns of freshwater fishes. This parameter often influences multiple aspects of a fish’s life, like the timing of spawning, how long their eggs incubate, their metabolic demands, and when it’s time to move into overwintering habitat. By quantifying stream temperatures across a large landscape like Banff National Park—in other words, creating a stream “thermalscape”—we can determine which populations of native fishes are most resilient to climate change, where there might be candidate locations for restoration efforts, and where non-native fishes or other aquatic invaders might outcompete or negatively impact our native fishes. Over time, a stream temperature model can also explain local extinctions. In Banff National Park, we are using our thermalscape to explain the distribution and abundance of native trout, like the threatened westslope cutthroat trout and bull trout. We have a problem with the legacy of non-native trout stocking in the park. So, we’re also using it to identify areas where native trout could thrive if we were to eradicate non-native trout versus areas that are already on the warm edge boundary for native trout and where eradication of non-native trout wouldn’t be as effective for their recovery.

Stream temperature logger installations in the Cascade River. Bolted-style install (left) and glue-style install (right). Photo Credit: Dan Struthers.

The authors of this paper span provinces and continents! How did this international collaboration come to be?

The only way this paper was going to reach completion was through teamwork. We began this paper several years ago through a research collaboration with Dr. Neil Mochnacz. He was equally interested in stream temperature for his PhD thesis work. From that collaboration, we dreamt up the idea of exploring both a well-established modelling tool (SSN) alongside a novel modelling approach (R-INLA). Early on, we learned that nobody had modelled stream network data with R-INLA even though it held promise as an approachable tool for practitioners. After hitting some roadblocks with the model validation process for R-INLA, we realized that we needed higher-level INLA expertise. Through networking with the Cooke Lab at Carleton University, we connected with Dr. Lee Gutowsky to help with the project. Lee had been out to Banff for an INLA course taught by Dr. Alain Zuur that we hosted years earlier. As we dug into the online world of R-INLA we came across Dr. Tim Lucas’s work applying cross-validation techniques for R-INLA. That was the piece of the puzzle we were missing and an important step for model validation. So, we approached Tim via email, followed by virtual calls to explain where we were at, and he was keen to join our research project. What we learned from this project is that you really need to collaborate with the right people to make projects like these come to life, especially when you’re a conservation practitioner and you don’t have the luxury to spend all day modelling. Although we were all working in different time zones, virtual meeting platforms made it possible to brainstorm ideas and work modelling kinks out together as a team.

Dan and Gunnar sampling Bryant Creek with backpack electrofishing. Photo Credit: Nicole Sulewski.

Anyone who has conducted fieldwork (or written code in R!) knows it can be frustrating. What were some challenges you faced while conducting this research?

To be honest, the fieldwork for this research was the enjoyable part. On the other hand, the modelling was very challenging, particularly sorting out the R-INLA procedures. We are in a unique position working in a national park like Banff because we must have a variety of skills to travel in the backcountry to collect data. Furthermore, our largest priority is active management work, like eradicating non-native trout and stream restoration. We don’t have a lot of time for modelling. Luckily, with excellent resources available from folks like Dr. Alain Zuur and Dr. Dan Isaak, who have published papers and textbooks on INLA and SSN, we were able to get a decent start ourselves. A lot of our modelling challenges were alleviated after we reached out to Lee and Tim for their mentorship and expertise with R-INLA. They provided that critical missing piece, making it possible to wrap up our modelling issues and move forward with the writing.

Dan performing a snorkel survey in Johnson Lake. Photo Credit: Conrad Jaeger. 

In your open-access paper, you’ve provided practical advice and an R-script for non-experts. The Canadian Journal of Fisheries and Aquatic Sciences has just launched a new policy supporting Open Data and Open Code. What are your thoughts on the accessibility of conservation methodologies?

This idea is very exciting to me. I think it’s great step forward for researchers to publish their data in an open-access repository, along with their R code so that practitioners like me don’t have to waste precious time trying to figure out the intricacies on our own. When you have methods clearly explained and readily available in supplementary materials or online repositories, it cuts down on having to “rewrite the book”, so to speak. I think this is the natural next step for the research community given the advent of new statistical methodologies and the fact that practitioners are so often trying to learn these modern techniques off the side of our desks between operational demands.

In support of open science, the Canadian Journal of Fisheries and Aquatic Sciences has launched a new data and code availability policy.

Your career has covered a range of topics, from the impacts of military activities on biodiversity to the spatial ecology of fish. How do you decide which areas to focus on in your research, and what has driven your interest in these diverse topics?

I think the common thread for my research interests is investigating anthropogenic impacts to freshwater biodiversity, and how we can mitigate these impacts. I feel a heightened need for this work given the rate of decline in freshwater biodiversity. I have a soft spot for wild rivers as they’ve captivated me from an early age. Using my knowledge and research skillset to protect these special places is very important to me.

What future directions do you see for your research? Are there any emerging challenges or opportunities in aquatic science that excite you?

I think the future direction of my research will revolve around investigating impacts on species at risk, like our native trout species here in Banff. These species are under threat from climate change and biological invasions, and much of our active management work is related to protecting westslope cutthroat trout and bull trout and monitoring their populations into the future. These threats are at the forefront of my mind as I imagine a future where native trout could continue to live in the waters of Banff National Park.

Our most recent work in Banff National Park focuses on mitigating hydropower impacts to native trout. We’ve been actively conducting habitat restoration and environmental flow analyses. There is a myriad of methodologies for environmental flows, and new ones emerging all the time. Like stream temperature modelling, it’s difficult to find the time to both do the work and learn about new methodologies. Our approach has been to lean on some of the wonderful expertise found in Canada. I will always look for those opportunities to publish conservation case-studies to help support practitioners managing freshwater ecosystems in other regions.

The Wild Salmon Policy was hailed as transformative in 2005, yet salmon populations continued to decline. What has the policy accomplished, and how can it evolve? SHARE YOUR INSIGHTS

What advice would you give to early-career aquatic scientists at the start of their research journeys—especially those interested in working at one of Canada’s national or provincial parks?

My advice to all early-career scientists is to get immersed in wild spaces as often as possible. I think this is the best education for someone coming out of a university institute, especially after undergraduate studies. When immersed in true wilderness, you begin to observe and understand how wild ecosystems function. For example, you can see the aquatic-terrestrial interface found along streams that have developed over millennia, having been allowed to flow naturally without being confined by human-built infrastructure. Spending time in a protected area, like a provincial or national park, provides a perspective of how natural systems form and function that’s difficult to find elsewhere given how altered our natural landscape is now. This perspective acts as a blueprint when writing up impact assessments and restoration plans, or looking to assist with sustainable development. It’s challenging to become qualified to work as a conservation practitioner for a provincial or national park because you must build wilderness travel skills while also developing skills to do quantitative analyses with the data you collect from the field. Luckily, we have our passion for wild spaces to keep us motivated.

Dan sampling fish with rod and reel at Marvel Lake. Photo Credit: Conrad Jaeger. 

Thank you so much for sharing your insights with us, Dan! If we could ask you one last question: What advice would you give to early-career researchers about the publication process?

  1. Don’t rush the process. If you feel like your writing or analyses need work, take the time to do it right, whether this means getting colleagues or peers to provide constructive feedback on the manuscript or just to verbally brainstorm ideas over a coffee. It also demonstrates respect to the publisher and peer-reviewers when you submit a paper that shows you’ve carefully considered all aspects of the process, including grammar and journal formatting requests. I understand that early-career researchers need to publish at a certain rate to meet their career goals, but the impact that a publication has on conservation outcomes on the ground will likely be higher when statistical analyses are done correctly and the story is told in a simple yet compelling way.
  2. Collaborate well. Find the right people to help with the project, but researchers need to approach collaboration respectfully. If you’re struggling to make progress on a research project, oftentimes more experienced researchers are happy to help. But their time is limited, and they will shy away if they can see that you haven’t done everything you can to solve the problem yourself. Before you reach out to someone for assistance, it’s best to have some demonstrable progress underway to build trust.
  3. Consider open access. If funding permits, think about publishing your manuscript in an open-access format. This can be helpful to conservation practitioners who are not affiliated with a university and who may not have access to journal subscriptions. Consider online repositories for publishing your data, methods, and code. It’s a breath of fresh air when authors provide the supporting data, code, and even paper as open access to the audience. This is a huge help to fellow researchers who may be struggling to work through similar analyses.
Rebecca Michaels-Walker

Rebecca Michaels-Walker is the Content Marketing Specialist at Canadian Science Publishing.