232
M.F. Willson et al.
For the broader comparison among regions , we divided both northern
and southern forests into two types (conifer and broadleaf), because
experience in Alaska has suggested important biological differences
between them and because other studies (e .g., James & Rathbun, 1981)
have found differences in avian community structure between the two
kinds of forest . We search for patterns, using a Model I two-way parametric ANOVA (with arcsin-transformation of percents) or region
(south, midwest , northwest) X forest type (broadleaf, conifer) . This
approach allows us to detect interactions between the chosen factors and
the effects of each factor. In some cases , directional differences are
apparent despite a significant interaction term, and these can be treated,
properly, along with the results of ANOVAs that lack significant interaction terms (Sokal & Rohlf, 1981). In addition, we can make predictions,
based on what is known or believed about the basic biology of each kind
of forest, about the direction of differences between communities of these
forests. Post-hoc tests following the ANOVA permit a ranking of all the
regional forest types , which is then examined for trends. The predictive
approach is stronger in some ways, because it permits the negation of
some possible explanations. However, it has the obvious frailty that the
prediction is only as good as the information on which it is based;
negation can occur because the background logic and information-and
hence the prediction-are faulty. Nevertheless, it can be a useful tool in
this preliminary phase as a means of examining some possible sources of
differences among the communities.
Comparison of Alaskan and Chilean Rainforests: Results
Diversity and Abundance
Site diversity (the number of regularly occurring species at each site)
differed with marginal significance among the three forest types (refer to
Table 11.1; one-way ANOVA, F = 3.64 , P = 0.065) . However, point
diversity (the average number of regularly occurring species per point)
differed significantly among forest types: point diversity was highest in
Alaskan deciduous forest and lowest in Alaskan conifer stands; Chiloe
was intermediate and not significantly different from either (see Table
11.1; one-way ANOVA, F = 4.0, P = 0.05; Tukey post-hoc test). On
average, point diversity in Chiloe accounted for a similar proportion of
site diversity (68%) as in the other sites (74%, 80%), probably indicating
similar spatial heterogeneity in avian distributions within sites in all three
kinds -of forest.
'
The greatest average number of birds per census point per day was
found in Chilean forests, and the lowest, in Alaskan coniferous forests;
Alaskan deciduous forests were intermediate (Table 11.1; one-way
M.F. Willson et al.
For the broader comparison among regions , we divided both northern
and southern forests into two types (conifer and broadleaf), because
experience in Alaska has suggested important biological differences
between them and because other studies (e .g., James & Rathbun, 1981)
have found differences in avian community structure between the two
kinds of forest . We search for patterns, using a Model I two-way parametric ANOVA (with arcsin-transformation of percents) or region
(south, midwest , northwest) X forest type (broadleaf, conifer) . This
approach allows us to detect interactions between the chosen factors and
the effects of each factor. In some cases , directional differences are
apparent despite a significant interaction term, and these can be treated,
properly, along with the results of ANOVAs that lack significant interaction terms (Sokal & Rohlf, 1981). In addition, we can make predictions,
based on what is known or believed about the basic biology of each kind
of forest, about the direction of differences between communities of these
forests. Post-hoc tests following the ANOVA permit a ranking of all the
regional forest types , which is then examined for trends. The predictive
approach is stronger in some ways, because it permits the negation of
some possible explanations. However, it has the obvious frailty that the
prediction is only as good as the information on which it is based;
negation can occur because the background logic and information-and
hence the prediction-are faulty. Nevertheless, it can be a useful tool in
this preliminary phase as a means of examining some possible sources of
differences among the communities.
Comparison of Alaskan and Chilean Rainforests: Results
Diversity and Abundance
Site diversity (the number of regularly occurring species at each site)
differed with marginal significance among the three forest types (refer to
Table 11.1; one-way ANOVA, F = 3.64 , P = 0.065) . However, point
diversity (the average number of regularly occurring species per point)
differed significantly among forest types: point diversity was highest in
Alaskan deciduous forest and lowest in Alaskan conifer stands; Chiloe
was intermediate and not significantly different from either (see Table
11.1; one-way ANOVA, F = 4.0, P = 0.05; Tukey post-hoc test). On
average, point diversity in Chiloe accounted for a similar proportion of
site diversity (68%) as in the other sites (74%, 80%), probably indicating
similar spatial heterogeneity in avian distributions within sites in all three
kinds -of forest.
'
The greatest average number of birds per census point per day was
found in Chilean forests, and the lowest, in Alaskan coniferous forests;
Alaskan deciduous forests were intermediate (Table 11.1; one-way