• Nikki

Bears getting into trash cans: the data behind a mountain town rite of spring

In the arms race between human-engineered bearproof dumpsters and hunger-driven bear ingenuity, it's hard to say who's winning.

The first time I saw a black bear, I had been waiting months for the moment. I was a young reporter at the Steamboat Pilot & Today, and I'd been keeping my ear on the police scanner since my first day on the job waiting for someone to call the cops on a bear so I get get my first glimpse. My chance finally appeared when a coworker texted me. "Bears" and the address. A Colorado Parks and Wildlife officer was on the scene, and kindly told me I could take a look - and where to stand so I could keep my distance. She told me they were considering shooting off some blanks to scare them off. The first thing I saw was two tiny cubs stretching their paws up toward the top of the dumpster. Then their mother ambles around the corner, gently lifting the top. She was showing them how. The cubs deftly crawled up the side and tumbled inside with a clanging thud. Empty. On to the next dumpster.

You can't have a mountain town without nuisance bears. These crafty giant raccoons are thrilled with the growing popularity of rural living and increased mountain town tourism: it's a full-on buffet once they figure out this season's dumpster locking mechanisms. But as adorable as their little padded paws and furry rumps are, human-bear interactions are on the rise and pose a risk to both parties. Generations of bears who only know the deliciousness of trash are passing on this addiction and reliance to their young. Humans and their pets can be - albeit very rarely - attacked, injured or killed.

This is why knowing where, when and how these interactions are occurring can help officials manage this growing issue: i.e. we need data. This month, we look at when and where bears are getting into trash cans (and other trouble) via a new, rich dataset from Colorado Parks and Wildlife.


When do bears get into trash cans and other trouble? As you can probably imagine, summer. All winter they hibernate, and they wake up skinny and hungry and restless.

In this visualization, we see weekly reports of black bears to Colorado wildlife officials peak in July, with the majority of calls indicating property damage relating to a food source. Attacks or aggressive behavior are rare, but occur throughout the summer season.

But, ust like people, bears from different towns are not the same. Fort Collins bears and Aspen bears have their own unique patterns; they're not on the same schedule, have different habits and food sources and live in different climates. So let's break it down a little further to look at how bear reports vary across the state.

Here, we see how much these annual patterns vary by county. Larimer County (Fort Collins) bears are the biggest nuisance in August, while Routt and Pitkin county bear reports peak in June/July.

Something stuck out to me in these graphs: what's going on with bears in the fall, right before they hibernate? El Paso County (Colorado Springs) and Eagle County (Vail) bears show substantial spikes in activity (or reported activity, at least) before they go to sleep, but other locations have little to no fall bear reports.

Let's zoom in on two Colorado mountain ski town regions: Routt (Steamboat Springs) and Eagle (Vail). While there are some differences in elevation and vegetation cover as well as human population and patterns, these regions are relatively similar. What are some factors that might be contributing to this pattern? Why are Eagle County bears so much more visible in the fall?

And do differences between Routt and Eagle bears extend into their daily patterns? Here's a look at a 24-hour clock of bear complaints. It looks like Eagle bear reports are much more skewed toward late night, while Routt bears get into the most trouble in the morning. Still, bears all over Colorado have a little spike in activity just before midnight. As someone who regularly makes that 11:30 p.m. trip to the fridge, I can definitely relate to that.

Learn data science fundamentals by

following along with this post

DATASET: 2020 Colorado bear reports


LEARN: Dates and times, mutate(), polar plots



I made a map. Of course. With spatial data, it's hard for me not to make a map. I aggregated the data to a 1,000m grid, where the height and color of each column denote the number of bear complaints in 2021.

As you tilt and pan and zoom around, remember this: at their core, data are a representation of reality. Data aren't all-knowing or objective. This map is just one facet of a prism. Bear complaints do not necessarily tell us where the bears actually are, how many exist and how they behave. Instead, these data points represent the place where specific bear behavior overlaps with specific human behavior, muddied by individual differences in both species, as well as highly localized policies and enforcement (e.g. on bear-proofing and bird feeders) as well as culture (e.g. tourism, familiarity with bears). To me, this makes exploring data even more interesting - these data are nuanced because reality is just as convoluted.

The Details

Colorado Parks and Wildlife recently started keeping more detailed, centralized records of bear reports. While it doesn't have have a long history - which is why I focused on 2020 - I'm excited for what future years of these data might tell us about how bear reports shift with changes to human behavior post-COVID as well as climate- and drought-related factors.

I first came across these data in a press release from CPW. In the stats and maps they released, I immediately recognized the scent of a rich spatial dataset - much like a bear can smell a dumpster from five miles away. It contains 4,947 reports of bear complaints from 2020, where they occured, what the complaint was, exact descriptions of food sources and any human interaction, the bear's sex and tagging history....etc. There is so much more potential in this dataset, and I'm hoping to drive deeper into it someday.

While these data are public record, you can't find them online currently. You may have to make a public records request to view the full dataset. Sometimes, it's as easy as just asking. I have uploaded a slightly modified version of the dataset, however, to the Github repository associated with this blog. You can follow along with the provided code and make your own charts using this month's tutorial.

Now the question is: should we do mountain lion sightings or moose encounters next?

Check back for the next post on June 8*.

*I'm off by approximately a week from my planned schedule of posting the first Tuesday of every month. Consider these dates estimates while I'm finishing my dissertation.