STUDY METHODS 



Bird and small mammal study plots were established 

 on each of the experimental watersheds before logging. 

 Bird and small mammal populations were estimated an- 

 nually beginning in 1975 (2 years prelogging) and ending 

 in 1979 (3 years postlogging). Trees were harvested on 

 the logged watershed after completion of field work in the 

 autumn of 1976. 



Breeding Birds 



Two plots of 20 acres each were censused for breeding 

 birds using the Williams spot-map method (International 

 Bird Census Committee 1970). Methodological difficulties 

 and other special problems of the mapping method are 

 summarized by Oelke (1981). Plots were rectangular and 

 varied slightly in shape depending on topography and 

 cutting unit boundaries. Plots were surveyed and gridded 

 in a Cartesian coordinate system with points flagged and 

 numbered with stakes at 164-ft intervals. At least 10 

 census visits were made annually to each plot from mid- 

 May to late July. Most of the bird registrations were 

 recorded from sunrise to late morning when birds were 

 most active. To ensure complete coverage, the plot was 

 censused by walking within 82 ft of all points on the grid. 

 Observations and registrations extended well beyond plot 

 boundaries. Census routes were varied. 



At the end of the sampling period, concentrated groups 

 of registrations and coded activity patterns were circled as 

 indicating areas of activity or approximate territories. 

 Fractional parts of boundary territories were estimated to 

 the nearest one-quarter territory. Results were converted 

 to the number of pairs of breeding birds per 100 acres. 

 Estimates of bird species diversity followed Hill (1973). 



Small Mammals 



Small mammal populations were estimated on two 

 permanently marked 5.6-acre trapping grids. Grids were 

 located near the center of the larger bird census grids. 

 Each grid measured 495 by 495 ft and consisted of 100 

 trap positions regularly spaced at 55-fl intervals in 10 

 rows and 10 columns. One metal Tomahawk live trap, 

 3 by 3 by 10 inches, was placed near each position. Traps 

 were baited with a mixture of cracked corn, wheat, and 

 rolled oats. Surgical cotton was placed in each trap to 

 minimize death from exposure. Captured animals were 

 ear-tagged with Monel fingerling tags and released at 

 the point of capture. Trapping was conducted annually 

 for a 6-night period on each grid in August and early 

 September. 



Populations of the deer mouse, yellow pine chipmunk, 

 and boreal redback vole were estimated by means of the 



mark-recapture method and the Schnabel estimator (as 

 described in Overton and Davis 1969). The effective 

 trapping area was considered to be the dimensions of the 

 grid plus a strip whose width was equal to half the aver- 

 age range length added to each side. Range length was 

 determined by the "adjusted range length" method 

 described by Stickel (1954). 



As an additional indicator of abundance, the number 

 of individual animals caught in each 6-night trapping 

 period was used for deer mice, yellow pine chipmunks, 

 redback voles, shrews, western jumping mice, and the 

 sum of other small mammals that were trapped in small 

 numbers. 



Data Analysis 



The analysis of breeding bird populations was per- 

 formed on the differences between corresponding means 

 for the two study plots (that is, the mean of the logged 

 plot minus the mean of the unlogged plot). We felt that 

 the mean level of such differences should have neither an 

 upward nor downward slope during the period prior to 

 logging. Once logging had occurred, an upward slope 

 would indicate that logging had increased the breeding 

 bird density while a downward slope would indicate a 

 decrease. With this in mind, a grafted-polynomial model 

 was fitted to the data. The model assumed a slope of zero 

 for the prelogging period (1975 to 1976) and allows either 

 a negative or positive slope for the three postlogging years 

 (1977 to 1979). The probabilities presented in table 1 test 

 the hypothesis of a change in slope beginning after log- 

 ging. A small probability, say P<0.05, suggests a change 

 in slope. 



A contingency table analysis was performed on the 

 number of individual small mammals trapped on logged 

 and unlogged study plots. The chi-square test was de- 

 signed to detect differences in the pattern of numbers of 

 mammals trapped between the logged and unlogged plots. 

 Although the analysis was based on individual counts 

 rather than means of counts, the means are presented in 

 table 2 to make it easier to see the patterns. The confi- 

 dence limits presented in figure 1 use calculations for 

 estimates of small mammal population densities given by 

 Overton and Davis (1969). 



Because this study was not replicated, it was necessary 

 to make the assumption of independence among yearly 

 observations in order to perform the analyses. Even with 

 this assumption, the tests are not powerful. Therefore, 

 we may have failed to detect some changes that took place 

 (large Type II error), but it is unlikely that the differences 

 we discuss are by chance alone (small Type I error). The 

 assumption of independence from year to year, although 

 not strictly true, is probably inconsequential in view of 

 the rather dramatic postlogging changes in some of the 

 bird and small mammal populations. 



2 



