BR. J. ENT. NAT. HIST., 7: 1994 



Results 



The total number of individuals taken in the yew woodland and the chalk grassland 

 habitats, for each species, over the two trapping periods combined, are given in 

 Table 2. The species are named and ordered following Skinner (1984). 



Because only one anemometer was available, windspeed could only be as- 

 in one habitat at a time, so the two habitats could not be monitored concurrently. 

 The data, while showing windspeed to be generally lower in the yew woodland than 

 outside it, are not strictly comparable. However, further work in Juniper Bottom, 

 in 1991 and 1992, using paired anemometers, has shown that the mean windspeed 

 is markedly lower in the yew woodland than outside it (Fraiers & Cox, pers. comm.). 

 In 1990, and subsequent years, the temperature inside the yew woodland was generally 

 the same as, or slightly lower than, that in the grassland when trapping began each 

 night, but declined less rapidly, so that by the end of trapping it was fairly consistently 

 0.5-1.5°C higher inside the yew woodland. 



The background light intensity, measured during the day, was substantially higher 

 in the grassland than in the yew, the mean ratio being 32.2:1. 



Analysis 



For species taken in reasonable numbers (more than 10) over the 2 years, the number 

 of moths taken inside and outside the woodland was compared using the chi-squared 

 test (a statistical test which estimates the probability of any bias in the actual result, 

 away from the expected result, being due to chance alone, i.e. sampling error). Initially 

 tests were performed using the simple expectation that there was an equal probability 

 of a moth being caught in either habitat: i.e. an expected ratio of 0.5:0.5 for a species 

 in the two habitats. The normal level of statistical significance, that there is less than 

 a 5% probability of any deviation in the observed data away from expectation being 

 due to chance sampling error alone, is used. The results of these analyses are given 

 in Table 3. 



Of course, these tests may be unreliable were traps in one habitat much more 

 efficient in attracting and catching macrolepidoptera than those in the other habitat. 

 The total number of moths of all species recorded in each habitat (1624 in the yew 

 wood, 883 in the grassland) show that this might be the case. One method of 

 circumventing this problem would be to bias the chi-squared expectations by the ratio 

 of total moths captured in each type of habitat. This would give an expectation 

 ratio of 0.648:0.352. The results of repeating the chi-squared tests with these 

 transformed expectations are also given in Table 3. 



This statistical recourse could also be criticized because the greater number of moths 

 taken in the yew woodland traps, compared to the grassland traps, is primarily a 

 consequence of four common species, Idaea aversata, Peribatodes rhomboidaria. 

 A. repandata and Campaea margahtata, which all show significant bias towards the 

 yew woodland, with either expectation ratio. If the data for these four species are 

 removed from the catch totals for each habitat, these are reduced to 699 for yew 

 woodland and 652 for chalk grassland. 



It is notable that there is general tendency for geometrid species to be taken in 

 larger numbers in the yew woodland and noctuid species to be taken in larger 

 numbers in the grassland. This may result from differences in windspeed in the two 

 habitats. The windspeed in woodland is generally lower than in open habitats. This 

 has been confirmed for the two habitats in question. The yew woodland may 

 thus provide sheltered conditions that would be more conducive to flight for 



