Nell: The Dog Who Guarded in Kennels and Not at Home

Nell came in as a stray. Three-year-old, collie cross, no history. Within a week she had a note on her kennel card that said food aggressive, do not approach while eating.

What that note was based on was this: a volunteer went in to top up her water while she had a chew, and Nell froze, took the chew to the back of the kennel, and gave a low grumble when the volunteer stepped forward. Nobody was hurt. Nothing escalated. But it went on the card, and after that every person who walked past her kennel had already been told what she was.

two months later she was rehomed. Her adopter has never seen her guard anything.

That gap is the whole point of this post.

The first thing we did was nothing

Not literally nothing, but nothing that looked like training.

Nell had been in the building 6 days when the note went up. She was undersocialised, underweight, and had been sleeping badly. Almost everything about her behaviour at that point was telling us about the environment rather than about her.

So the first week was management and safety:

• Chews and high-value food given at the end of the day, in her kennel, undisturbed
• Nothing removed from her while she had it. If she had something she shouldn’t, we waited or we used a door management routine
• Water topped up when she was out of the kennel, not while she was eating
• The kennel card rewritten in plain language: what to give her, when, and what to do instead of reaching in

We also stopped talking about her as a food aggressive dog and started writing down what she actually did. Freezes and repositions the item when someone approaches within two metres while she has a chew. That’s a description you can build a plan from. The label isn’t.

Every rehearsal of guarding makes guarding stronger. Before you teach a dog anything, we want to stop giving them reasons to practise.

Safety before sharing

There’s an order to this that’s easy to get wrong. A dog who doesn’t feel safe cannot learn that people are safe. If the environment is still producing daily low-level threat, your training sessions are three good minutes swimming against twenty-three bad hours.

What safety looked like for Nell: predictable routines, the same two handlers for the first month, somewhere in the kennel she couldn’t be seen from the corridor, and enrichment that arrived reliably rather than as a treat when someone had time.

Her guarding softened noticeably before we ran a single training session. That happens more often than people expect, and it’s worth knowing about, because it tells you how much of what you’re looking at is the building.

Changing the prediction: DS and CC

Desensitisation and counter-conditioning is doing two jobs at once. Desensitisation keeps the trigger at an intensity the dog can cope with. Counter-conditioning changes what that trigger predicts.

For Nell, the trigger was a person approaching while she had something. So we worked at the distance where she stayed loose, which at the start was about four metres, and we did this:

• Person appears at four metres. Person tosses something excellent. Person leaves.
• That’s it. Nothing asked for. Nothing taken. Nell keeps her item and gains a second one.

Two details do most of the work here.

The order. The person has to appear before the food arrives, every time. If you get the food out first and then walk over, you’ve conditioned the treat bag rather than the person. The approach has to be the thing that predicts the good outcome.

The distance. You set it by watching the dog, and you set it lower than you think. The signals that you’ve come in too close arrive well before anything obvious: eating faster, going still over the item, a quick hard glance sideways, standing over it, picking it up and moving it. If you’re seeing those, that rep is teaching the wrong thing.

Over about two weeks Nell’s working distance came down from four metres to arm’s reach. We didn’t push it. The distance closed because she started moving toward us, which is the only version of that progress worth having.

Rucksack walks

This is where the work stopped being a training exercise and started being something Nell enjoyed.

The idea is simple. You put a few items in a rucksack, take the dog somewhere quiet and neutral, and sit down. Not the yard, not the paddock she associates with being taken back to kennels. Somewhere that doesn’t already have history attached.

Then you let her have an item, and you sit there being extremely boring.

Neutral ground matters more than people give it credit for. In the kennel, Nell had months of associations about who comes past and what they want. In a field, at the base of a tree, with a person sitting on the ground reading their phone, none of that context is present. Dogs will often show you a version of themselves outside that you simply cannot access indoors.

We took a stuffed Kong, a cardboard box with kibble in it, a scrappy old towel, and a chew. Low on her value ladder first, working up over sessions.

The long line and the boring human

The mechanism underneath the rucksack walk is choice.

Nell went out on a long line for these sessions. The line was never used to move her, control her, or bring her in. It was there for safety and nothing else. What it bought her was the ability to decide how close to be.

We’d put the item down, walk away to the end of the line, and sit. Nell would go to it, take it, and then choose her own distance. Early sessions she took things twenty feet away and ate with her back to us. That’s fine. That’s the dog telling you exactly where she is with all this, and it costs you nothing to let her.

The measure of progress isn’t whether you can approach her. It’s whether she brings the item closer to you.

By week three Nell was lying down to work on a chew about six feet away, checking in between bites. Nobody made that happen. We just stopped being a reason to move away.

Some practical notes if you’re going to try rucksack walks with a dog known for guarding:

• Handle the long line properly. No loops around hands or wrists, gloves if it’s biothane, and a clear route back that doesn’t involve stepping over the line
• One dog, one handler, quiet location, no other dogs in sight
• Start with items low on the dog’s value ladder and let the dog’s response set the pace
• This is for lower-level guarding. Dogs with a bite history, dogs who guard across many contexts, or dogs who escalate quickly go to your behaviour lead with a proper risk assessment. A long line does not make a session safe on its own

The part where she got adopted and nothing happened

Nell’s adopter filled in her six-week follow-up and answered “no” to every question about guarding.

This surprises people. It shouldn’t, because it is one of the better-evidenced findings in shelter behaviour work.

A few things worth knowing about:

• Mohan-Gibbons, Weiss and Slater (2012) followed shelter dogs identified as food guarders into their adoptive homes. The majority were not guarding at home, and among the adopters who did see something, most didn’t regard it as a problem.
• Marder and colleagues (2013) compared in-shelter food aggression assessments against owner reports after adoption. The assessment predicted home behaviour poorly in both directions. Plenty of dogs who guarded on the test didn’t guard at home, and some dogs who passed cleanly did.
• A later study looking at shelters that removed the food guarding assessment from their evaluation protocol found no rise in adopter-reported problems or returns, alongside improved adoption outcomes.
• More broadly, Patronek and Bradley (2016) argued that shelter behaviour evaluations have limited predictive validity for home behaviour generally, and that the base rates involved mean a positive result on a rare-behaviour test tells you much less than it feels like it does.

The mechanism isn’t mysterious. Guarding is a response to resources being scarce, unpredictable and contestable. A kennel is scarcity, unpredictability and contest made architectural. There are dogs on both sides of you. Items arrive rarely and get taken away. Strangers pass constantly and some of them want things from you. Arousal is high and control is nil.

Then the dog goes to a house where food appears twice a day without fail, nobody takes anything, and the nearest other dog is a hundred metres away. The conditions that produced the behaviour aren’t there, so neither is the behaviour.

What this doesn’t mean

It doesn’t mean ignore guarding. It doesn’t mean stop recording it, and it certainly doesn’t mean skip the adopter conversation.

What it means is that in-shelter guarding is weak evidence about the dog and strong evidence about the situation, and it should be written up accordingly. A behaviour observed under those conditions is not a diagnosis and shouldn’t be treated as a permanent property of the animal.

Think about:

• Describe the behaviour, don’t label the dog
• Note the conditions it happened under, including how long the dog had been in
• Reassess in foster wherever you can, because foster data is worth more than kennel data
• Tell adopters what to do rather than what to fear. “Feed her somewhere quiet and don’t take things off her” is more useful than “she has a history of food aggression”
• Never let a single in-kennel observation carry a decision on its own

Nell’s original kennel card said food aggressive, do not approach while eating. two months of work went into a dog who, on the evidence, may not have needed most of it once she was out of the building.

The work wasn’t wasted. It made her safe to handle while she was with us, and it gave her adopter a real plan instead of a warning. But the note came first, and the note shaped everything that happened afterwards.

Write the note carefully.

Try this with one dog

1. Rewrite one kennel card so it describes behaviour rather than assigning a label
2. Run one rucksack session on neutral ground. Sit down, be boring, let the dog choose the distance (practice with a non-guarder first)
3. Record the distance the dog chooses each session and watch whether it closes on its own