128 
Fishery Bulletin 117(3) 
Table 1 
Comparisons of weight-length relationships of red lionfish (Pterois volitans) for each ecological region (ecoregion) sampled in the 
northern Gulf of Mexico from 2012 through 2015: southeast (Florida Keys north to Anclote Keys, Florida), northeast (Anclote Keys 
north to Mobile Bay, Alabama), and central (west of Mobile Bay to Galveston Bay, Texas). (A) Regression equations for comparisons 
by sex within each ecoregion. (B) Pairwise comparisons by sex between ecoregions. Between-sex and between-ecoregion compari¬ 
sons of weight (W) were evaluated by using analysis of covariance with total length (TL) as the covariate. 
Ecoregion 
Male 
Female 
Comparison 
n 
Regression equation 
n 
Regression equation 
Southeast 
857 
W = 2.00 x 10- 6 (TL) 3 - 34 
671 
W= 1.00 x 10~ 6 (TL) 3 44 
F, ,„ S =12.68,P<0.001 
Northeast 
1181 
W = 3.00 x 10- 6 (TL) 3 - 26 
1239 
W = 3.00 X 10~ 6 (TL) 3 - 30 
F, 9 4 9 n =2.32, P=0.128 
Central 
368 
W = 3.00 X 10- 6 (7T,) 3 - 30 
354 
W = 2.00 X 10 - 6 (TL) 3Al 
F 4 722 =6.84, P=0.090 
Pooled 
2406 
W = 3.00 x KT 6 (TL) 3 - 29 
2264 
W = 2.00 X 10’ 6 (TL) 3 - 37 
F W6 7o=2:L.96,P<0.001 
B 
Ecoregion comparison 
Males 
Females 
Southeast vs. northeast 
Ft 9oos=8.16, P=0.004 
F, 1Qin =27.72, P<0.001 
Northeast vs. central 
Fj 1549 =1.42, P=0.223 
F1 1593=10.19, p=0.001 
Southeast vs. central 
F i j225=0.86, P=0.354 
F 1 102 5=0.61, P=0.436 
Pooled ecoregions 
F2.2406=4-17, P=0.015 
F 2 ,2264= 15 - 88 > P<0.001 
were read by 2 independent readers to assign age infor¬ 
mation. The average percent error (APE) between readers 
was estimated following Vanderkooy and Guindon-Tisdel 3 . 
Otolith sections had 2 distinct band types, with the opaque 
bands formed during slow growth periods and the trans¬ 
lucent bands formed during periods of faster growth (see 
fig. 7 in Fogg, 2017; Vanderkooy and Guindon-Tisdel 3 ). Any 
otoliths with discrepancies in ages between the 2 readers 
were reexamined, and if a consistent age could not be 
determined for an otolith, it was removed from analysis. 
Following agreement on an age, marginal increment anal¬ 
ysis was conducted by measuring marginal increments (to 
the nearest 0.001 mm) and comparing them to the width 
of the previous complete annuli by using the following 
equation (Tanaka et al., 1981): C=W n /W n _ 1 , where C is 
the index of completion, W n is the width of the marginal 
increment, and W n _ 1 is the width of the previous complete 
annulus. This method was used to confirm the periodic¬ 
ity of annuli formation. To determine ages when the most 
recent band was incomplete (indicating only a partial year 
of growth), we assigned one of 4 otolith margin codes to 
the incomplete outer ring, as viewed under transmitted 
light (Vanderkooy and Guindon-Tisdel 3 ): 1 (opaque), 2 (1/3 
translucent), 3 (1/2 translucent), and 4 (2/3 translucent). 
Each marginal code equals 0.25 years. 
Because our sample collections lacked smaller fish 
(<100 mm TL), a truncated normal distribution was used 
to fit a 3-parameter von Bertalanffy growth curve to the 
3 VanderKooy, S., and K. Guindon-Tisdel. 2003. A practical hand¬ 
book for determining the age of Gulf of Mexico fishes. Gulf States 
Mar. Fish. Commission Publ. Ill, 109 p. [Available from website.] 
age data, and separate model parameters were determined 
for each sex and ecoregion for comparison (Diaz et al. 4 ). 
Because the von Bertalanffy growth curve is nonlinear, a 
sum of squares reduction test (Schabenberger and Pierce, 
2002) was used with statistical software SAS (vers. 9.4; 
SAS Institute, Inc., Cary, NC), instead of a traditional anal¬ 
ysis of variance to determine if there were differences in 
growth between ecoregions and sexes by comparing non¬ 
linear trends between groups (a=0.05). The sum of squares 
reduction test was conducted by fitting a full and reduced 
model to the data. The test statistic (F) was calculated fol¬ 
lowing this equation: F=[(SSRR-SSRF)/(DFRR-DFRF)]/ 
MSRF, where SSRR and SSRF are the residual sums 
of squares from the reduced and full model, respectively, 
DFRR and DFRF are the residual degrees of freedom for 
the reduced and full model, respectively, and MSRF is the 
mean sum of squares from the full model. This test evalu¬ 
ates the SSRF when the SSRR is removed from the model. 
Results 
The slopes of the weight-length relationships for male 
and female red lionfish were significantly different within 
the southeast ecoregion and did not differ in the 2 other 
ecoregions (ANCOVA: F x 1528 =12.68, P<0.001) (Table 1A, 
Fig. 2). The weight-length relationships from data pooled 
4 Diaz, G. A., C. E. Porch, and M. Ortiz. 2004. Growth models for red 
snapper in U.S. Gulf of Mexico waters estimated from landings 
with minimum size limit restrictions. Southeast Data Assess¬ 
ment and Review SEDAR7-AW-01,13 p. [Available from website.] 
