The effects of logging on bird life were evaluated in the context of natural forms of disturbance and in the context of regional variation in the avifauna. Eleven sampling plots were established in a variety of logged and unlogged vegetation types in the Trans-Gogol timber area (near Madang, P.N.G.) where both clearfelling and less intensive logging had been underway for as long as five years before data were collected. An additional four plots were sampled in another region of lowland rainforest at Buso, near Lae. Seven plots in the Trans-Gogol were sampled at approximately six-weekly intervals for ten months , while on two of these plots more regular sampling continued for a further eight months . The vegetation types investigated included well-drained terrace forest, swamp forest, hill forest, and regrowth vegetation resulting from logging in each of the main forest types.
Bird data were collected using mist nets at a plot for two days at a time (a netting episode), and by taking bird counts from a canopy platform erected at each of the two main plots on river terrace where mist netting had been most intensive. A total of 20,930 net hours were accumulated to yield 2,171 captures of 1,534 individuals, and 168 hours were spent counting birds from the canopy platforms. A thorough assessment was made of the possible biases in the results associated with each of the two techniques of data collection.
Major communities were identified on the basis of species composition using pat tern analyses (classification and principal coordinate analysis) of netting episodes, and daily bird counts from the canopy platforms. At 105 net sites at the seven main plots in the Trans-Gogol, variation in habitat structure within and between plots was assessed using 70 habitat parameters in a principal coordinate analysis. The same sites were ordinated on the basis of bird captures. A canonical correlation analysis was then used to relate habitat structure to the microspatial distribution patterns of bird species. Seasonal and other temporal changes in bird communities were assessed in relation to logging, as were the vertical stratification of bird species, and diurnal activity patterns.
Regional differences in avifaunas between Buso and the Trans-Gogol included a slightly lower diversity and density of understorey birds at Buso, and considerable variation in species composition. Buso lacked several lowland species that were present in the Trans-Gogol, while other species, more typical of higher altitudes, were recorded near sea level at Buso but nowhere in the Trans-Gogol. Isolation of small tracts of lowland rainforest, coupled with competitive interaction between bird species contributes to regional variation in the lowland avifauna. The patchy distribution of lowland rainforest birds is an important consideration in assessing the consequences of timber projects of the scale conducted in the Trans -Gogol.
For both understorey species and those of the upper layers of vegetation, the most prevalent pattern of diurnal activity was one of highest activity a few hours after dawn, followed by a decrease until early to mid afternoon, and a subsequent increase before sunset. There were notable exceptions to this pat tern involving several species, particularly mixed feeders and insectivores in mixed species feeding flocks. At the regrowth plot, nomadic movements of some species were also apparent throughout the day.
Vertical stratification of species occurred for all feeding classes, but it was les s pronounced for mixed feeders. Frugivores, and some insectivores, exhibited foraging height preferences between the canopy and emergent vegetation at the forest plot. In general, species present at both the regrowth and forest plots did not forage over a greater vertical range a t the regrowth plot. In fact, each feeding class at the regrowth plot, especially the frugivores, responded to the different vegetation structure by generally compressing the vertical range it occupied. Activity became concentrated in the crowns of emergent remnant trees left after the logging, although there were exceptions to this trend. Trees left standing after logging at the regrowth not only affected the vertical stratification of birds but became foci of activity which determined the local horizontal distribution of birds.
Of the 57 species frequently counted in the upper layers of vegetation in the Trans-Gogol, about 22 exhibited little difference in abundance between the forest plot and regrowth plot. A minimum of 22 canopy species were more abundant at the forest plot and a maximum of 13 were more abundant at the regrowth plot. Among the latter group of species there were about equal numbers of insectivores and obligate herbivores, and only two mixed feeders. Of the 22 species with higher abundance in the canopy at the forest plot, about two-thirds were obligate herbivores, five were mixed feeders and three were insectivores. Many of the obligate herbivores at the forest plot were large birds in contras t to the generally smaller species typical of the regrowth. In general, species richness was about 10% lower in the canopy at the regrowth plot, and the evenness of species abundance was also lower.
There was more activity in canopy vegetation during the dry season than during the wet season, due mainly to a greater number of frugivores at the forest plot and the presence of migratory insectivores at the regrowth plot. The number of species regularly frequenting the canopy was slightly higher during the dry season, especially at the forest plot. The activity of understorey birds increased on level terrain during the wet season, when insect abundances were probably highest. There was evidence of local movements of birds in response to seasonally induced heterogeneity of the environment.
The forest type most productive of timber (well-drained terrace forest) maintained the highest density of understorey birds, including a distinctive group of about 20 species. This group of species was also characteristic of hill forest but were present there at a lower density. With the exception of plantation forest, the lowest overall density of understorey birds occurred in regrowth and in swamp forest (approximately 40% lower than in terrace forest). Two small flycatchers were especially abundant in regrowth; one of which was also common in swamp forest. The similarity between understorey bird communities in swamp forest and regrowth was further strengthened by the occurrence of three of the same species of kingfisher, and the mutual lack of a fourth species of kingfisher. The evenness of species abundance was a little lower in regrowth than elsewhere, whereas species richness was least in swamp forest. The swamp forest understorey lacked species found in other forest. type s and also lacked a large contingent of transitory species , which was a feature of regrowth vegetation . A single net ting episode conducted in Eucalyptus deglupta plantation indicated a particularly low diversity and density of birds.
There were relatively few species primarily adapted to large expanses of secondary growth, although many species were capable of utilising these areas. The second growth faunas of Africa and New Guinea appear to be a subset of species found in forest habitats, whereas in the neotropics the second growth species are far more distinctive. New Guinea birds are often prematurely categorised as "second growth" species when in fact they may be locally distributed according to habitat structure evident in both virgin forest and grossly disturbed environments.
Birds were responding to microhabitat mosaics. Six basic microenvironments were depicted among net sites in the Trans-Gogol on the basis of habitat structure and the local occurrences of understorey birds. They included typical "logging regrowth", gap phase vegetation, building phase vegetation and three types of mature phase forest. Gap phase, building phase and mature phase forest occur in a temporal sequence in the natural regenerative cycle following small scale disturbances. The mature phase forest types were well-drained forest, poorly developed forest and swamp forest. They resulted from differences in soil fertility, soil drainage, local rainfall and topography. Logging resulted in a mixture of microenvironments that included "logging regrowth", gap phase, and building phase vegetation. Where logging was leas t intense small enclaves of mature phase forest persisted with appropriate elements of the bird life. That is, in terms of both bird life and habitat structure, there was a certain interrelationship between the natural dynamics of rainforest regeneration following small scale disturbances, and the conditions created by logging. Nevertheless, the logging caused certain major changes in both habitat structure and bird life that were not evident under natural conditions, i.e. the microenvironment of "logging regrowth". This microenvironment was especially different from that of mature phase, well-drained forest where the bird community was most distinctive and bird density was highest. There was less contrast between "logging regrowth" and other types of mature phase forest. Neither habitat structure nor birdlife were indicative of the original forest type once an area was logged.
Regeneration of the vegetation is least likely to restore well-drained mature phase forest. Over the long term, it is predicted there will be large scale replacement of the most favourable habitat conditions for birds with alternatives that cater for lower bird densities and fewer species.