Chicken's best friend? Livestock guardian dog bonding with free-ranging chickens

The study addresses predator control in free-range and pastured egg production systems, which are expanding globally as consumers demand higher-welfare products. Outdoor hens face increased predation risk (e.g., foxes, raptors). Producers are increasingly using livestock guardian dogs (LGDs) as a non-lethal predator control tool, yet key questions remain about how LGDs guard poultry, whether bonding with humans versus livestock affects performance, and the combined efficacy of LGDs with other deterrents. On a Western Australian pastured egg farm with two Maremma LGDs, the authors posed four questions: (1) Are LGDs bonded to their chickens when free to roam? (2) Does LGD presence increase paddock use by chickens? (3) Is fox activity influenced by free-roaming LGDs and/or motion-activated spotlights? (4) What factors influence LGD effectiveness, based on an online farmer survey.

The paper reviews predator control options for poultry including electric fencing and netting (effective barriers), and chemical, auditory, or visual deterrents whose efficacy can wane due to habituation. Motion-activated and multi-modal deterrents may improve effectiveness, but empirical tests—especially for poultry and combinations of methods—are limited. LGDs are widely used for predator control in various contexts and are valued as a cost-effective, non-lethal tool. Most LGD research has focused on sheep, showing LGDs spend most time with flocks, and that their direct presence deters predators rather than territorial exclusion alone. These findings may not translate to poultry due to differences in livestock behavior, flock size, spatial distribution, and higher human interaction in egg production, which may shift LGD bonding toward humans. Few studies have examined LGDs with poultry, creating a need to understand LGD bonding and behavior around free-range poultry.

Study site: Blue Tractor Farm, Glen Mervyn, Western Australia (8 July–23 August 2019). Three 50×50 m chicken paddocks with solar electric fencing and mobile “chicken caravans.” Paddock M: two 2-year-old sibling Maremma LGDs and ~450 9-month-old hens; allowed to free-roam the property 2–3 days/nights weekly as part of husbandry. Paddock A: two alpacas and ~500 6-month-old hens; alpacas confined to paddock. Paddock N: no guardian; older hens, numbers declining over study.

LGD bonding/movement: Two i-gotu GT-120 GPS loggers on collars recorded positions every 3 min (or 10 s if speed >10 km/h). Power banks enabled 5–8 days operation. Collars swapped weekly over 7 weeks. Day = 06:30–18:30; night = 18:30–06:30. Separation distance from current chicken paddock centroid computed. Comparisons: day vs night separation distances (Kruskal–Wallis), and nights with spotlights on vs off for speed and separation distance.

Chicken paddock use: Eight downward-facing Reconyx HC500 cameras mounted at ~1.8 m across paddock M (positions adjusted as paddock moved). Time-lapse: one image every 5 min for 20 days. For each image: count chickens and note shade presence (>50% of field of view shaded). Hourly chicken activity = sum of chickens per hour; shade availability = hourly proportion of shaded images. GLM (lme4) with Tweedie link: dependent variable chicken activity; predictors: LGD treatment (dogs in/dogs out) × distance from caravan (m), covariate shade availability. Residuals checked with DHARMa.

Fox activity and spotlights: Twelve passive infrared Reconyx cameras at the corners of all paddocks (3 m away, 0.6 m high) in Rapidfire mode; four additional motion-activated solar LED spotlights placed 50–150 m from paddocks near boundary, each paired with a camera; spotlights alternated weekly on/off and remained physically in place. Cameras near spotlights also ran 1-min time-lapse to detect light activations without a thermal trigger. Monitoring lasted 46 days, with checks every 3–4 days. Independent fox events counted if ≥5 min apart on the same camera. Pearson’s chi-squared tests compared fox events across LGD (in/out) × spotlight (on/off) combinations and pairwise contrasts.

Farmer survey: Online SurveyMonkey (mid-Sep to mid-Nov 2018) disseminated via LGD and poultry Facebook groups in Australia and North America; asked about predator issues, LGD use, costs, perceived effectiveness, and advice. Locations grouped as Australia/New Zealand vs North America; poultry numbers compared with unequal-variance t-test. Persisting predator issues coded as yes if respondents reported any of: yes, some, occasional, rarely. Binomial GLM predictors: number of LGDs, log(number of poultry), and whether dogs worked inside a fenced area; model selection with MuMIn dredge favored the full model; residuals supported Gaussian fit. Additional Pearson chi-squared tests (due to missing data) examined associations with bonding (livestock vs people), amount of human interaction, breed (Maremma vs other), working alone vs in groups, and whether respondents mentioned training in their advice. Another chi-squared test compared livestock injuries between LGDs that spent all time with livestock vs not.

LGD bonding/movement: When free-roaming, both LGDs spent most time near the farmhouse (550–700 m from paddocks). Overall, 82% of all GPS locations were within 75 m of the farmhouse (night: 96%; day: 66%). Only 3.3% of all locations were within 75 m of their chicken paddock (night: 0.1%; day: 6.6%). Dogs were significantly farther from the chicken paddock at night than day (Coco: χ2=117.52, P<0.001; Hiro: χ2=10.38, P=0.001). On spotlight-on nights, dogs moved more slowly (Coco: χ2=12.93, P<0.001; Hiro: χ2=99.0, P<0.001) and stayed closer to the farmhouse (Coco: χ2=284.2, P<0.001; Hiro: χ2=6.73, P=0.009) than on spotlight-off nights.

Chicken paddock use: From 15,373 chicken sightings (0.27% at night), shade strongly increased activity (t=8.05, P<0.001), and activity was greater near the caravan (t=−3.31, P=0.001). After accounting for these factors, LGD presence had no effect on chicken activity (z=−0.14, P=0.999) and no LGD×distance interaction (t=−0.25, P=0.803).

Fox activity and spotlights: No predation events occurred during monitoring. Forty fox events were recorded over 46 days across all paddocks and spotlight cameras. Overall differences across the four LGD×spotlight treatments were not significant (χ2=6.07, df=3, P=0.108). Pairwise: nights with dogs out + spotlights had fewer fox events than spotlights alone (χ2=3.90, P=0.048): 2 events over 8 nights vs 13 events over 13 nights, respectively. LGD treatment alone (on spotlight-off nights) was not significant (χ2=1.56, P=0.211). Time-lapse indicated 19 spotlight activations without detected fox images; given few other nocturnal species, these likely represented additional fox activity not captured by triggers.

Farmer survey (n=59): Respondents averaged 2.39±2.42 LGDs and 743±1,596 poultry; Maremmas comprised 56% of LGDs (83% in Australia). Main predators: foxes (66% overall; 85% Australia), coyotes (77% North America), raptors (64% overall). About half (46% of 54 respondents) reported persisting predator issues; most believed LGDs effective (49/51) and would recommend them (49/50). In GLM, more poultry predicted persisting issues (z=2.13, P=0.034); number of LGDs (z=0.04, P=0.964) and fencing status (z=1.40, P=0.162) were not significant. No associations with reported bonding (livestock vs people), human interaction time, breed, or working alone vs in groups. Respondents who mentioned training in their advice were less likely to report persisting issues (χ2=5.85, P=0.016). LGDs that spent all their time with livestock were more likely to injure/kill livestock than those that did not (χ2=7.15, P=0.007).

The study’s four questions are addressed as follows. (1) Bonding: When allowed to roam, the two Maremma LGDs predominantly stayed near the farmhouse rather than with chickens, especially at night when foxes were active, indicating stronger bonding to people than to poultry. While survey respondents generally believed LGDs would protect poultry even when not confined with them, the case study suggests that human bonding can draw dogs away from livestock, potentially increasing risk when protection relies on the dogs’ direct presence. (2) Chicken behavior: LGD presence did not expand hens’ paddock use; shade and proximity to the caravan were the primary drivers of hen distribution. This aligns with most survey responses indicating LGDs do not alter hen spatial behavior. (3) Predator activity: The combination of free-roaming LGDs and motion-activated spotlights yielded fewer fox detections than spotlights alone, suggesting complementary or reinforcing effects—spotlights may alert dogs to predators or increase deterrence when combined with canine presence. (4) Effectiveness factors: Producers with ≥100 birds were more likely to report ongoing predator issues, implying that larger flocks may require more LGDs and/or complementary methods (e.g., fencing). Reported human bonding and contact levels did not predict producer-reported predation issues, but training emphasis was associated with fewer reported problems. Overall, findings support integrating LGDs with other tools (especially electric fencing and targeted deterrents), recognizing that dog–human–livestock bonding dynamics and property layout influence how and where LGDs operate.

LGD-to-livestock ratio remains a key predictor of protection success, but individual dog behavior and bonding patterns also matter. In this pilot case, two LGDs strongly bonded to people spent little time near chickens when free to roam, yet the combined system of electric fencing, LGDs, and motion-activated spotlights coincided with no fox predation during the study and fewer fox detections when dogs and spotlights were used together. Effective management should balance LGD welfare (opportunities for movement and stimulation) with the need for dogs to remain near livestock, and should include training and complementary deterrents. Future work should standardize measures of human vs livestock bonding in LGDs, examine larger samples across diverse poultry operations, and test integrated strategies (LGD numbers, fencing types, and deterrent combinations) to optimize protection for larger flocks.

This was a single-site pilot focused on two sibling Maremma LGDs; individual dog differences and farm-specific contexts limit generalizability. The study design followed commercial husbandry, constraining experimental manipulation (e.g., limited replication for alpaca paddock, alternating spotlight schedules). GPS proximity metrics defined “near” using distance bins, which may not fully capture protective effectiveness. Camera traps can miss events; some spotlight activations likely represented undetected fox activity. The farmer survey relied on self-selection and self-report, with missing data for some variables and limited before/after predation counts.