Hydrobiologia 378: 105–114, 1998. R. M. O’Riordan, G. M. Burnell, M. S. Davies & N. F. Ramsay (eds), Aspects of Littorinid Biology. © 1998 Kluwer Academic Publishers. Printed in Belgium. 105 Increasing precision in randomised field experiments: barnacle microtopography as a predictor of Littorina abundance E. G. Boulding ∗ & F. M. Harper 1 Department of Zoology, University of Guelph, Guelph, Ontario, N1G 2W1, Canada ( ∗ author for correspondence) 1 Present address: Department of Biology, Memorial University of Newfoundland, St. John’s, Newfoundland, A1B 3V6, Canada Key words: barnacle, density-dependent, Littorina, microtopography, population density, refuges Abstract We investigated whether measurements of barnacle microtopography could be used as covariates to increase preci- sion in randomised field experiments with intertidal Littorina populations. We used image analysis to quantify the microtopography of the barnacles within 100 cm 2 quadrats in July 1994 using as variables: number of barnacles, number of dead empty barnacles, mean distance to nearest neighbouring barnacle, total area covered by barnacles, and number of pockets of different size-classes [extra-small (< 0.15 cm 2 ), small (≥ 0.15 and < 0.30 cm 2 ), medium (≥ 0.30 and < 0.45 cm 2 ), large (≥ 0.45 and < 0.60 cm 2 ), and extra-large (≥ 0.60 cm 2 )] at least 75% surrounded by barnacles. We then used these variables to predict the abundance of Littorina spp. in the same quadrat. Two variables: the number of extra-small pockets and the number of small pockets accounted for 50% of the variation in total littorinid snail abundance among quadrats for January 1995 and for 47% of the variation for August 1994 but for only 16% of the variation for June 1994. However the single variable, the number of barnacles, accounted for 29–36% of the variation in snail abundance on these dates and can be measured without special equipment. The best combination of covariates was the number of extra-small pockets alone. The number of extra-small pockets between barnacles may be influencing the snail abundance by providing a refuge from wave shock and heat stress/desiccation. We suggest that the availability of these refuges among the barnacles could act as a density-dependent regulator of the total population size of a particular species of Littorina. Introduction Rocky intertidal shores are well known as tractable systems for ecological field experiments (reviewed in Paine, 1994). However rocky shores have steep verti- cal environmental gradients that are often combined with complex topography. These factors cause high spatial heterogeneity in the distribution and abundance of intertidal organisms. Consequently a great number of replicates are usually required before even large dif- ferences among treatments of field experiments can be detected statistically. Unfortunately increasing the sta- tistical power of experiments by increasing the number of replicates is often impossible because the area avail- able at a particular tidal level on steep rocky shores is usually extremely limited. Fisher (1932) suggests that an alternative method of increasing the statistical power of a field experiment is to correct for any pre-existing differences among quadrats with a covariate so that they all have simi- lar means before the treatments are applied. He used the yields of tea bushes before an experiment began as a covariate; this corrected for the fact that some treatments would have by chance a more productive set of tea bushes than other treatments and reduced the experimental error (= mean square error). This makes it more likely that a small difference among treatments will be statistically significant for a given number of replicates.