The mean proportions
of time spent in different behaviours were arcsin
square root transformed to approximate a normal
distribution in both species. We also tested the correlation
between either time spent foraging or lying
with time spent rutting using Pearson correlation
tests on transformed proportions. For the computation
of ratios, a constant (k = 1) was added to the
proportion of time spent foraging and lying in
mountain goats to avoid a division by 0 when an
individual was not observed to rest or a ratio of 0
when a male was not observe to forage. Consequently,
ratios could vary between 0.5 and 2. Ratios
were inverse-transformed in mountain goats and
log-transformed in bighorn sheep to meet normality
before testing for variations between periods. As for
absolute changes in the behaviours investigated, F ⁄ L
ratios were compared between periods when
accounting for year of observation and male age
using a LMM with ‘male identity’ fitted as a random
term. In all analyses, the significance of individual
effects was tested by comparing the log likelihood of
models with and without the random terms (Steele
& Hogg 2003). When the random term was not significant,
we present the results of generalised linear
models. Finally, in bighorn sheep, we also assessed
whether time spent foraging increased with the
overall proportion of time a ram was observed investigating
and ⁄ or (i.e. when not engaged in one of the
three mating tactics) using a Pearson correlation test,
as more time should be allocated to foraging in the
absence of opportunities for reproduction. Proportions
of time spent in different mating tactics (or
not) were also arcsin square root transformed prior
to analyses.
The mean proportions
of time spent in different behaviours were arcsin
square root transformed to approximate a normal
distribution in both species. We also tested the correlation
between either time spent foraging or lying
with time spent rutting using Pearson correlation
tests on transformed proportions. For the computation
of ratios, a constant (k = 1) was added to the
proportion of time spent foraging and lying in
mountain goats to avoid a division by 0 when an
individual was not observed to rest or a ratio of 0
when a male was not observe to forage. Consequently,
ratios could vary between 0.5 and 2. Ratios
were inverse-transformed in mountain goats and
log-transformed in bighorn sheep to meet normality
before testing for variations between periods. As for
absolute changes in the behaviours investigated, F ⁄ L
ratios were compared between periods when
accounting for year of observation and male age
using a LMM with ‘male identity’ fitted as a random
term. In all analyses, the significance of individual
effects was tested by comparing the log likelihood of
models with and without the random terms (Steele
& Hogg 2003). When the random term was not significant,
we present the results of generalised linear
models. Finally, in bighorn sheep, we also assessed
whether time spent foraging increased with the
overall proportion of time a ram was observed investigating
and ⁄ or (i.e. when not engaged in one of the
three mating tactics) using a Pearson correlation test,
as more time should be allocated to foraging in the
absence of opportunities for reproduction. Proportions
of time spent in different mating tactics (or
not) were also arcsin square root transformed prior
to analyses.
การแปล กรุณารอสักครู่..
