(n = 13), and C. hoffmanni initially marked as juveniles (n = 36)
and subadults (n = 47). We defined our sampling period from
February to May in 2012 for these analyses; shorter windows
resulted in low capture probabilities.
We considered two competing mark–recapture models {φage,
p.} and {φage(JS,A), p.} for both species. The first model hypothesized
that survival differed between juveniles, subadults and
adults. Adult survival was estimated from subadults marked in
2010 that transitioned to the adult stage in 2011 and was treated
as a nuisance parameter since it was estimated more robustly
using the known fates analyses described above. The second
model hypothesized that juveniles and subadults survived at the
same rate and that adult survival differed from juvenile and subadult
survival. We assumed constant recapture probabilities over
time and ages classes because only one instance of a ‘missed’
encounter (i.e. an encounter history of ‘101’) was observed in
C. hoffmanni and no missed encounters occurred for
B. variegatus.
ESTIMATING POPULATION GROWTH RATES
We used a three-stage-class matrix model parameterized with
survival and reproductive rates to estimate the asymptotic population
growth rate (k) for both B. variegatus and C. hoffmanni
(Table 1). The three age classes considered were juveniles, subadults
and adults, where only adults were able to produce offspring.
We assumed a birth-pulse model for B. variegatus as most young
are produced from February to April in this species, and a birthflow
model for C. hoffmanni since young are likely to be
produced year-round in this species. We followed the approach
described in Morris & Doak (2002) when parameterizing the
reproductive elements of the matrix models for both species. Age
of first breeding (AFB) is uncertain for both species but has been
estimated to be 2–3 years in maned sloths of both sexes (Lara-
Ruiz & Chiarello 2005). Our data support this estimate as one
B. variegatus weighed 31 kg when it was captured at 18 months
of age, and given that B. variegatus can breed at 37 kg, we suspect
that this species can reach adult size by 2–3 years of age.
Therefore, in separate analyses, we assumed that AFB was 20
and 30 years in B. variegatus. We set AFB to 25 and 35 years
in C. hoffmanni based on their c. 20% greater body mass (Pauli
& Peery 2012); doing so was facilitated by the fact that the
employed birth-flow model assumes that all young are produced
at the midpoint of the interval between sampling events (Morris
& Doak 2002).
When parameterized with estimates of true survival rates (i.e.
emigrants are not considered as losses from the population),
matrix models yield estimates of k that reflect the extent to which
sampled individuals are replacing themselves demographically
(i.e. whether birth rates balance death rates) regardless of dispersal
movements (Peery, Becker & Beissinger 2006). When
parameterized with estimates of local survival rates, matrix models
yield an estimate of k that reflects the extent to which individuals
lost via death and emigration are replaced by local
reproduction and immigration (i.e. the rate of change in population
size within study area boundaries). However, the latter estimate
of k will be biased low if immigration is not considered
and, as a result, matrix models can yield estimates of k < 1 in
stable populations (Peery, Becker & Beissinger 2006). Here, we
estimated k using local survival rates to estimate population
growth in the absence of immigration and then determined how
much immigration would be required to maintain a stable population.
We also estimated k using true survival rates for radiomarked
adults to determine whether sloths initially marked
within the study area were replacing themselves demographically.
We characterized the possible effects of immigration on estimates
of local k using a range of annual immigration rates from
0 to 010 (in increments of 001). The annual immigration rate
was defined as the number of immigrants in year t divided by the
total population size in year t1. We assumed that all immigration
occurred into the subadult stage class, since, with one exception
(see below), adults of both species retain defined home
ranges within study area boundaries (Pauli & Peery 2012; Peery
& Pauli 2012). We estimated k from the dominant right eigenvalue
of the matrix model for each factorial combination of species,
AFB and potential immigration rate (Morris & Doak 2002).
We estimated sampling variation associated with k using Monte
Carlo simulations (Alvarez-Buylla & Slatkin 1993).
CHARACTERISING THE POTENTIAL IMPACTS OF
IMMIGRATION
We assessed whether sloth immigration into our study area
occurred using a recently developed genetic kinship approach
(Palsbøll 1999; Peery et al. 2008). The number of parent–
offspring pairs in a sample of individuals should be relatively
low when immigration rates are high. Thus, we hypothesized
that few subadults should have a parent present in the popu