through 11-20 April. Recoveries of immatures remained 
high from 11-20 October through 11-20 November; about 
52 recoveries were made per 10-day period. They decreased 
to about 13 recoveries per 10-day period through 11—20 
April as did adults. After 20 April, no more than one adult 
and three immature recoveries were reported during any 
of the remaining 10-day periods. 
The earliest direct recovery in Mexico (Jalisco) occurred 
during the 21-31 August period; it was an adult bird and 
was the only Mexican recovery during August. There were 
only two adult and 15 immature recoveries before October. 
Detection of differential arrival dates between the two age 
cohorts was precluded due to small samples during early 
stages of migration. 
The arrival of CMU mourning doves in Mexico appeared 
to be well synchronized, The number of recoveries rose 
sharply during 11-20 October, at least 20% more for adults 
and 50% more for immatures than during any other 10-day 
period. These results agree closely with those of Blanken- 
ship and Reeves (1970). 
Composite Evaluation 
Doves from nearby States began entering Texas before 
September; the bulk of the population arrived later in Sep- 
tember. Doves from northern States began entering Texas 
by mid-September; their peak occurred during late Sep- 
tember and October. 
Immature doves from nearby States entered North Texas 
sooner than adults. However, immature and adult doves 
from northern States arrived at about the same time. The 
early arrival of immatures from nearby States may have 
been more the result of pre-migratory movements of imma- 
tures than migration. Both adult and immature doves began 
arriving in wintering areas of southern Mexico during Sep- 
tember, but peak arrival dates were during mid-October. 
Survival and Recovery Rates 
Kstimated Survival and Recovery Rates 
A common objective of banding studies is the estimation 
of parameters which reflect survival (Brownie et al. 1978). 
Anderson (1975), Anderson and Burnham (1976), and Burn- 
ham and Anderson (1979) described methods commonly 
used and accepted during earlier years to estimate such 
parameters and explained in detail their inadequacies. Early 
approaches were based on the deterministic life table con- 
cept and were biased and inefficient (Anderson 1975). The 
conceptualization and development of more efficient sta- 
tistical methods to estimate survival are well documented 
(Brownie et al. 1978), These newer methods are based on 
stochastic models which permit the computation of sam- 
pling variances and covariances of the various estimators 
by using the Maximum Likelihood approach. Use of these 
models allows the precision of individual estimates to be 
statistically examined. Precision estimates are of value only 
29 
when coupled with precision estimates from other cohorts, 
banding periods, geographical areas, etc. Stochastic models 
as described by Brownie et al. (1978) use such estimates in 
the development and testing of hypotheses which may lead 
to meaningful inferences. 
Banding and recovery matrices for adult female, adult 
male, all adult (sexes combined), and immature mourning 
doves were assembled for each banding reference area in 
the CMU. We estimated survival and recovery parameters 
for the various cohorts using models in Brownie et al. 
(1978). All estimates presented may be considered as aver- 
ages for selected years of banding. Absences of estimates 
indicate either insufficient or poorly distributed data. 
Estimates with low values in the goodness of fit tests 
(P < 0,05) indicate the data do not satisfy the assumptions 
of the model and are therefore subject to question. Recovery 
rates derived from these models are computed differently 
from traditional methods and should not be confused with 
the traditional definition of Recovery Rate (see Glossary). 
Adult females, — Although data on sexes within each ref- 
erence area were pooled (see Sex Related Differences), the 
data by sex class were also examined separately. Bandings 
and recoveries of adult females were generally insufficient 
for computation of survival rates; however, survival and 
recovery rates from 5 of 15 reference areas were estimated 
for adult female mourning doves by using program EsTI- 
MATE (Table A-70). Female doves from South Dakota had 
the highest survival rate (59.1%) whereas those from 
Missouri had the lowest (33.2%). Some estimates were 
imprecise as indicated by high standard errors. Recovery 
rates ranged from 0.9% in Colorado to 2.2% in Missouri, 
Differences among reference areas were obvious, especially 
in recovery rates, and prevented the data from being pooled 
for analysis in this section. For the entire CMU the un- 
weighted average survival rate was 51.0% , whereas the un- 
weighted average recovery rate was 1.6%. 
Adult males. — We calculated survival and recovery rates 
from 9 of 15 reference areas for adult male mourning doves 
using program ESTIMATE (Table A-71). Male doves from 
Nebraska had the highest survival rate (65.3%) and those 
from North Texas had the lowest (43.9%). Recovery rates 
ranged from 0.6% in Nebraska to 2.4% for doves banded 
in Missouri. For the entire CMU the unweighted average 
survival rate was 51.7% and the unweighted average re- 
covery rate was 1.6%. 
All adults. —Data from adult males and females were 
pooled within each reference area and subjected to program 
ESTIMATE. Survival and recovery rates were calculated for 
13 of 15 reference areas (Table A-72). Adult doves from 
New Mexico had the highest survival rate (65.1%) and those 
from Arkansas had the lowest (31.5%), Recovery rates 
ranged from 0.6% for birds banded in North Dakota, Ne- 
braska, and Iowa to 2.3% for those from Missouri. 
For the entire CMU the unweighted average survival and 
recovery rates were 52.1 and 1.4%, respectively. The un- 
weighted average survival rates for adult females, adult 
