98 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 1. March 2012 
with one typically charging the other with rapid 
aggressive movements. Idle was recorded when a 
male was not doing any of the other behaviors. 
We noted the male that performed any 
copulation attempt and the location. A copulation 
attempt was defined as anytime a male was able to 
put at least one foot on the female’s back. Most 
copulation attempts were interrupted by other 
males at varying times throughout the attempt. 
We classified copulation attempts as successful 
when females vigorously ruffled their feathers and 
departed the lek after copulating (Hagen and 
Giesen 2005). 
Statistical Analyses—We used .Skew Calcula¬ 
tor 2003 (Nonacs 2003) to analyze mating skew 
on leks. We used two indices of mating skew: X 
and R (Binomial Skew Index). Lambda values 
ranged from 0 to I with larger values indicating 
greater skew (Kokko and Lindstrom 1997). 
Positive values of R indicate some monopoliza¬ 
tion ot matings (skew), while 0 indicated random 
mating, and negative values indicated a more 
equal distribution (Nonacs 2000). /i-values, gen¬ 
erally. ranged from - 1 to I. although it is possible 
to obtain values >1. Confidence intervals and P 
values can be calculated for R (Nonacs 2000). Lek 
attendance rates were calculated as the number of 
days a male was observed on the lek divided by 
the number ol days the lek was observed and at 
least one male was present. 
We assessed two characteristics of territories: 
size and distance to center of lek activity. We 
calculated two measures of territory size, 95% 
kernel and 95% minimum convex polygon 
(MCP) estimates, using plotted male locations 
with >24 locations/behavioral observations. 
Both metrics were computed in the ADEHABI- 
TAT package (Calenge 2006) of Program R (R 
Development Core Team 2008). Lesser Prairie- 
Chickens appear to spend a disproportionate 
amount of time at territory boundaries and we 
suspect that kernel estimators overestimated 
territory size. MCPs only outline the outer points 
of a distribution and may be more accurate in 
assessing individual territory sizes of ickking 
prairie-chickens. Thus, only MCP estimates of 
territory size were used Tor modeling. We report 
ernel estimates of territory size for comparative 
puiposas with other studies of lekking grouse 
( g„ Nooker and Sandcrcock 2008). Kernel 
(r = a oo«T e a " ed With MCP est ' ma tes 
, •)' A ma,c s center of activity was 
computed as the centroid of all its locations Each 
male’s centroid was averaged to ascertain the 
center of activity for the lek. 
Discrete choice models (DCM) allow inference 
to be drawn about resource preferences based on 
the attributes of the resource (Cooper and Mill- 
spaugh 1999). These models predict the probabil¬ 
ity that an individual will select a certain resource 
as opposed to any of the other available resources 
and assume that individuals make choices that 
will maximize utility (Cooper and Millspaugh 
1999). DCMs are used more frequently in habitat 
selection studies (e.g., Lesmeister et al. 2008. 
Vanak and Gompper 2010). We followed the 
example of Nooker and Sandercock's (2008) 
Studies of Greater Prairie-Chickens and used 
DCMs (PROC MDC, SAS Version 9.1. Can', 
NC, USA) to assess correlates of male mating 
success for Lesser Prairie-Chickens. Each copu¬ 
lation attempt represents one sample in the DCM 
A female chooses one male to mate with among a 
group ol males, which is considered the choice 
set. DCMs allow the choice set to vary by sample, 
which is necessary when multiple leks are 
involved. The males (or sample of males) on 
one lek compose the choice set for each 
copulation attempt on that lek. We had to colled 
>24 location/behavior points on the male in¬ 
volved in the copulation attempt for it to be 
included in this analysis. Not every male on a Id 
was included in the analysis, but we believe our 
sample is representative of all males attending the 
lek. We trapped across the entire lek area and did 
not focus trapping efforts on central or peripheral 
males. 
Behavioral variables included the proportion ol 
observations recorded as each behavior category 
display, face off, fighting, idle, and moving 
Morphological variables included wing cord length 
(cm), tarsus length (cm), pinnae length (cm), and 
mass (g). Territorial variables included distant 
to lek center (m), and territory size (MCP. m 
We did not use all variables in discrete choice 
models due to small sample sizes. We selected the 
behavioral variables display and idle to use w 
models because they were uncorrelated (r = "9.1, 
P — 0.59) and represented what we hypothesized to 
be important in mate choice. We selected the 
morphological variables mass and pinnae length 
because they were uncorrelated (/• = 0.1 ,P~ 0.591 
and represent a size component (mass) and a 
secondary sexual characteristic (pinnae). We ah'" 
included territory size, distance to lek center, and 
age in models. 
