3. Comparison of TubeTraps and Longworth Traps


During 2012 and 2013, I have received reports from four users who, between them carried out eight trials where TubeTraps were compared with Longworth Traps. In addition, I have carried out five trials myself on three sites.

In all cases, the trials involved matched pairs of traps – one of each type – set in either grids or lines. The number of pairs varied from nine to 20 and the number of sessions varied from two to six. All captures were recorded and no distinction has been made between new captures and recaptures.


The data were subject to a “quick-and-dirty” analysis. A 3-way ANOVA model was constructed, utilising a Poisson error term and log-link function in a generalised linear model-building process. Although the model was not a particularly good fit to the data, it did allow a number of simple hypotheses to be tested;

  • Is there any difference in the numbers of captures overall between the two traps?
  • Does this difference vary between species?
  • Does the difference vary between trials?


A total of 334 captures were made in approx 1520 trap-nights, belonging to seven species (Table 1). Not surprisingly the majority of these were wood mice, although a good number of bank voles and field voles were also caught.

The ANOVA showed a very highly significant difference between overall captures between trials and between species. However, there was no significant difference between captures in the two different trap types (Wald χ2(1) = 0.864, p > 0.1). Furthermore, the two interactions between trap type and species (Fig. 1), and trap type and trial (Fig. 2) were not significant, indicating there was no evidence that the efficacy of the trap types varied between species or trials.

Simon Poulton: 13th February 2014