A response to Hesford & MacLeod (2022): Rejection of a model estimating high densities of mountain hares in the Peak District, England

• Main Authors: Bedson, Carlos, Metcalf, Oliver, Symeonakis, Elias, Mallon, David, Reid, Neil
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2022
• Subjects: Density estimation; Encounter rate; Game and Wildlife Conservation Trust; GWCT; Lamping survey; Lepus timidus; linear model; spatial capture-recapture; Macleod; mountain hare
• Online Access: https://pure.qub.ac.uk/en/publications/7281f998-0e6b-4122-a090-4a7cbdc66d06; https://pureadmin.qub.ac.uk/ws/files/311732697/MC2203_Response_To_Hesford_And_MacLeod_1_.pdf; https://www.mammal.org.uk/science-research/mammal-communications/browse-papers/

A recent Game and Wildlife Conservation Trust (GWCT) report (Hesford & MacLeod 2022)suggested densities of mountain hares (Lepus timidus) may reach 52 - 125/km2 in parts of thePeak District, England. These are notably higher than previous and current estimates of 5 - 33hares/km2 (Matthews et al. 2018; Bedson et al. 2022). We review Hesford & MacLeod (2022) whobased their methods on those used in a review of mountain hare survey methods in Scotland(Newey et al. 2018). This review demonstrated a weak, non-significant relationship between hareencounter rates using spotlight surveys of walked transects at night and estimated densitiesderived from spatial capture-recapture (SCR) methods on managed heather moorland (p=0.08).Newey et al. (2018) recommended that this relationship should not be used to estimate haredensities. We reproduce the Newey et al. (2018) linear model and confirm its poor predictiveability and show that removal of one outlier reduces an already marginal relationship to a near flatline (p=0.80). Hesford & MacLeod (2022), nonetheless, used this relationship to estimate haredensities along non-randomly placed transects. We conclude that reportedly high mountain haredensities estimated by Hesford & MacLeod (2022) are biased and based on a model with littlepredictive power; more recent Distance Sampling estimates are from 37 - 96% lower (Bedson etal. 2022). It is important that wildlife monitoring methods robustly account for survey bias anderror, detection probability and variation between habitats, especially if results are to informpotential population management interventions.