30 Length-Weight Relationships of Alligator Gar

Length-Weight Relationships and Potential Biases

for Alligator Gar (Atractosteus spatula) from
Texoma Reservoir, Oklahoma
Richard A. Snow
Oklahoma Department of Wildlife Conservation, Oklahoma Fishery Research
Laboratory, Norman, OK 73072
Michael J. Porta
Oklahoma Department of Wildlife Conservation, Oklahoma Fishery Research
Laboratory, Norman, OK 73072
Clifton R. Sager
Oklahoma Department of Wildlife Conservation, Durant State Fish Hatchery,
Caddo, OK 74729

Abstract: Alligator Gar (Atractosteus spatula) is the largest fish species found in Oklahoma, which
makes obtaining field measurements (particularly weight) by fisheries managers or recreational
anglers challenging. A method to overcome the logistical hurdles associated with handling and
measuring large fish in the field would be beneficial to managers and anglers. Therefore, the
objective of this study is to develop length-weight and length-length relationships for Alligator Gar
collected from Texoma Reservoir, Oklahoma using total length (TL), standard length (SL), and
girth (G) measurements to predict weights and length measurements of Alligator Gar. A total of 339
Alligator Gars averaging 1,666 mm TL (range = 590 to 2,357 mm TL) and weighing 38.2 kg (range
= 0.77 to 105.3 kg) were used in this evaluation. No significant differences in weight of Alligator
Gar were detected between sex or season (winter-spring and summer-fall), so all fish were pooled
for the remaining analyses. All simple linear regression and multiple regression models produced
significant relationships (P ≤ 0.05), were highly correlated (r2 = 0.88 to 0.99), and mean predicted
error values were < 13% for all models. However, the simple linear regression model using TL
and the multiple regression model using TL and G were the best predictive models to estimate
weight in this evaluation. All weight and length estimates predicted with regression analysis were
not statistically different than actual Alligator Gar weights and lengths. Visual inspection of weight
bias plots suggests little bias between predicted and actual weights for Alligator Gar ≤ 60 kg, but
became more variable for fish > 60 kg. However, the simple linear regression model using TL and
the multiple regression model using TL and G produced less biased weight estimates for fish up to
80 kg, but appeared to underestimate weights of Alligator Gar ≥ 100 kg. All of the models evaluated
in this study provide managers and anglers with a relatively accurate estimate of Alligator Gar
weights from Texoma Reservoir and provides fisheries managers with weight-length information to
which other Alligator Gar populations can be compared.

anthropogenic influences such as dams,

dredging, habitat change, unlimited recreational
Introduction harvest and commercial fishing (Brinkman 2008,
Inebnit 2009, Snow et al 2018). The Red River
Alligator Gar (Atractosteus spatula) has basin remains the stronghold for Alligator Gar in
a limited distribution in Oklahoma due to Oklahoma, although anglers occasionally report
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
 R.A. Snow, M.J. Porta, and C.R. Sager 31

fish caught from the Arkansas River (Oklahoma Reservoir using experimental gillnets (net
Department of Wildlife Conservation [ODWC] dimensions described by Binion et al. 2015 and
unpublished data). Because Alligator Gar Schlechte et al. 2016). Nets were set in coves
have a limited distribution in Oklahoma, or on main-lake flats in depths ranging 2 - 9 m.
ODWC manages them as a species of greatest Site selection was aided by the use of visual
conservation need and regulates recreational observation of Alligator Gar surfacing (aerial
harvest with a one fish daily bag limit (which breathing) and side scan sonar. In deeper water
must be reported to ODWC). Numerous angling or when fish were observed with side scan sonar
methods are permitted to catch Alligator Gar, near the lake bottom, weights (9.1 kg keg- style
however bowfishing, rod and reel, and snagging anchors) were attached to the lead line in the
are the most common (ODWC 2019). Alligator middle of the net to ensure the net maintained
Gar anglers rarely have the means to weigh fish contact with the lake bottom. Nets were
due to their large size, even though fish weight monitored every 15 to 30 min to ensure quick
is often used as a measure of fishing success release of Alligator Gar and reduce mortalities.
by trophy anglers (Meerbeek and Crane 2017). Besides gillnetting, several angling methods
However, anglers often record total length (jug fishing, rod and reel, and snagging) were
or girth measurements to describe the size of used to collect Alligator Gar and harvested fish
the fish that they catch. Anglers and fisheries were donated by recreational anglers.
managers would benefit by having a method to
estimate the weight of an Alligator Gar using the Upon capture, each Alligator Gar was
measurements commonly collected in the field. measured using a fabric measuring tape for snout
length (snout tip to anterior start of eye orbit),
Length-weight relationships provide anal fin base length, girth (G; taken anterior of
fisheries managers with important basic biology the pelvic fins), standard length (SL; snout tip to
information about fish populations, including insertion of epichordal lobe of caudal fin) and
fish condition, growth rates, morphological total length (TL). These measurements were
differences, and reproductive potential (Santos used to identify sex of each gar using methods
et al. 2012, Torres et al. 2012, Meerbeek and outlined by McDonald et al. (2013). Due to their
Crane 2017, Maurya et al. 2018). Further, the large size, individual Alligator Gar were placed
mathematical relationship between various into a fish sock (25.4 mm #60 green-twine mesh
fish length measurements and weight allows x 3.05 m long x 457.2 mm diameter), which
either measurement to be predicted simply by were then hooked to a hanging scale (Intercomp
having the other (Sarkar et al. 2009, Nazir and CS200; Intercomp CO., Medina, Minnesota)
Khan 2017). Development of length-weight attached to a winch (Badland 2500 ATV/Utility
equations for Alligator Gar would provide Winch; Badland Winches, Camarillo, California)
fisheries managers and anglers with a means and lifted to attain the weight (W; kg) of each
to overcome physical and logistical issues fish (Figure 1).
associated with weighing large fish in the field,
which reduces handling stress or unnecessary To meet the assumptions of normality,
harvest just to obtain a weight. Therefore, the length and weight data were log transformed
objective of this study is to develop predictive prior to regression analyses. Differences in
equations to estimate Alligator Gar weights weight between seasons (winter/spring versus
using total length, standard length, and girth summer/fall) and sex of Alligator Gar were
measurements. Additionally, bias was evaluated tested using ANCOVA. Simple linear regression
between weights predicted from length-weight models were used to describe the relationships
relationships and actual Alligator Gar weights. between girth, length, and weight (TL:W, SL:W,
G:W, TL:SL, TL:G, and SL:G) of Alligator
Methods Gar. Multiple regression analysis was used
to determine the relationships between girth,
Alligator Gar were sampled in Texoma length, and weight (TL/G:W and SL/G:W) for
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
32 Length-Weight Relationships of Alligator Gar

Figure 1. Photographs depicting the winching apparatus used to weigh Alligator Gar and the
fish sock (25.4 mm #60 green-twine mesh x 3.05 m long x 457.2 mm diameter) used to support
Alligator Gar.
Alligator Gar. Differences between individual Results
weights and lengths predicted from the simple
linear and multiple regression equations were A total of 339 Alligator Gar averaging
tested against actual fish weights and lengths 1,666 mm TL (range = 590 to 2,357 mm TL)
using paired t-tests. Strength of each model and weighing 38.2 kg (range = 0.77 to 105.3
was evaluated by comparing percent error kg) were collected from Texoma Reservoir.
[(Observed - Predicted)/Predicted*100] by The sex ratio of Alligator Gar was 52.5% male
averaging the percent predicted error across and 47.5% female. More Alligator Gar were
all observations for each model (Wood 2005, collected during winter and spring (n=203) than
Scharf et al. 1998, Snow et al. 2017, Jeter et al. during summer and fall (n=136). No significant
2019). Additionally, bias plot were constructed difference in weight was detected between sex
to compare bias between actual and predicted (F1, 337 = 0.619, P = 0.57) or season (F1, 337 =
girths, lengths, and weights of Alligator Gar. A 0.116, P = 0.26), therefore all fish were pooled
bias plot was not constructed for TL:SL, as this for the remaining analyses.
relationship was highly correlated (r2 = 0.99).
All statistical analyses were conducted at a Simple linear regression models indicated
significance level of P ≤ 0.05. significant relationships between all length and
weight measurement combinations (P ≤ 0.01;
Table 1. Simple linear and multiple regression equations for predicting length and weight
measurements of Alligator Gar using total length (TL), standard length (SL), and girth (G)
measurements, with associated r² and P-values.

P-
Regression type n r² Regression equation
value
Simple Linear 338 ≤ 0.01 0.91 logW = 3.074(logTL)-8.364
331 ≤ 0.01 0.88 logW = 2.839(logSL)-7.451
227 ≤ 0.01 0.91 logW = 2.501(logG)-5.536
331 ≤ 0.01 0.99 logSL = 1.011(logTL)-0.0878
235 ≤ 0.01 0.93 logG = 1.111(logTL)-0.7514
234 ≤ 0.01 0.93 logG = 1.095(logSL)-0.6431
Multiple 227 ≤ 0.01 0.92 logW = 3.0806(logTL)-0.0037(logG)-8.3806
227 ≤ 0.01 0.90 logW = 2.8521(logSL)-0.0021(logG)-7.4931
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
 R.A. Snow, M.J. Porta, and C.R. Sager 33
Table 2. Percent error of length and weight measurements predicted from simple linear
regression and multiple linear regression using total length (TL), standard length (SL), and
girth (G) measurements taken from Alligator Gar, including outcomes of paired t-tests.

Predicting
Regression type % error t-statistic df P-value
variables
Simple Linear TL:W 8.1 0.596 337 0.28
SL:W 12.4 1.54 330 0.06
G:W 9.3 0.607 226 0.27
TL:SL 1.1 0.108 330 0.46
TL:G -3.9 0.443 234 0.33
SL:G -8.8 0.436 233 0.33
Multiple TL/G:W 2.86 -0.829 226 0.20
SL/G:W 2.14 -0.191 226 0.42

Table 1), and these relationships were highly actual weights of the largest Alligator Gar (>
correlated (r2 = 0.88 to 0.99). The simple linear 100 kg; Figure 2). Little bias was observed
regression model using TL was the best predictor between girth measurements predicted using
of Alligator Gar weight (P < 0.01, r2 = 0.91, mean TL and SL and actual girth measurements for
% error = 8.1%; Table 2). Regardless of model, smaller fish, however the variability of predicted
all predicted length and weight measurements girth measurements increased with increasing
were not significant different than actual values Alligator Gar length (Figure 3).
(P > 0.05; Table 2).
Discussion
The multiple regression models using TL
and G and SL and G to predict weight were This study provides length-weight equations
significant (P ≤ 0.01) and highly correlated (r2 = for Alligator Gar from Texoma Reservoir,
0.90-0.92; Table 1). The mean percent error was Oklahoma. Both models (simple linear
low for both multiple regression models (< 3%; regression and multiple linear regression) that
Table 2). The weights predicted using the TL/G best predicted Alligator Gar weights used TL to
and SL/G models were not different than actual predict weight, which had the highest correlation
fish weights (P > 0.05; Table 2), suggesting both coefficients (r2 = 0.91-0.92). Although all
models predicted fish weights similarly. predictive models produced reliable estimates of
Alligator Gar weight for smaller fish (≤ 60 kg),
Bias plots suggest predicted weights were the weights predicted using simple linear (TL)
relatively unbiased compared to actual fish and multiple regression (TL and G) models were
weights for all models, particularly for Alligator less variable than weights predicted using other
Gar < 60 kg (Figure 2). As Alligator Gar length measurements (SL or G), particularly
weights exceeded 60 kg, increased variability for larger fish (≥ 80 kg). The simple linear
was observed between predicted and actual regression model relating TL to weight was the
weights for all linear regression and multiple best predictor of Alligator Gar weights for fish
regression models (Figure 2). The linear and up to 100 kg. The variability in predicted weights
multiple regression models using TL were the of larger fish was likely due to low sample size
best predictors of fish weight compared to actual of large fish, which comprised <1% of sample.
fish weights for large fish (80-100 kg). Visual Additionally, variability in predicted weights
inspection of bias plots suggests that predicted may have also been caused by intrinsic (gonadal
weights from all models underestimate the development, age, sex and genetic makeup) and
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
34 Length-Weight Relationships of Alligator Gar

Figure 2. Bias plots comparing actual Alligator Gar weights to those predicted using regression
anaysis: A) TL:W, B) SL:W, C) G:W, D) SL and G:W, and E)TL and G:W) of Alligator Gars.
The diagonal line represents a 1:1 relationship between predicted and actual weights.

extrinsic factors (gut content, habitat, season; dimorphism (McDonald et al. 2013), we
Neumann et al. 2012). Despite this variability, found no significant differences in weights
the best predictive models estimated the weight by sex or season. Similarly, García de León et
of the current Oklahoma state record Alligator al. (2001) found no difference in the weight-
Gar (not used in this study) to within 1.9 to 2.4% length relationships of Alligator Gar collected
(TL:W = 118.05, TL and G:W = 117.49 kg) of from Vincente Guerrero Reservoir, Mexico.
the actual weight (115.27 kg). Previous research produced similar results for
other large-bodied fishes. Meerbeek and Crane
Although Alligator Gars exhibit sexual (2017) found inclusion of sex and reproductive
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
 R.A. Snow, M.J. Porta, and C.R. Sager 35

commercial harvest and unregulated harvest

by anglers may have also reduced the number
of old, large female Alligator Gar in Texoma
Reservoir (Taylor et al. 2019).

The equations presented in this study to

predict weight of Alligator Gar provide anglers
and fisheries managers with an important tool
that allows them to overcome physical and
logistical issues associated with obtaining
weights of large fishes. Obtaining weights of
large-bodied fishes, like Alligator Gars, in the
field requires specialized equipment (large
capacity scales), and can increase handling time
leading to additional physiological stress to a
fish (Neumann et al 2012). For these reasons,
numerous studies have provided length-weight
relationships or predictive equations for large-
bodied fishes including, Atlantic Tarpon
(Megalops atlanticus; Ault and Luo 2013),
Lake Sturgeon (Commanda 2018), Muskellunge
(Meerbeek and Crane 2017), Pacific Goliath
Grouper (Epinephelus quinquefasciatus; Boas
et al. 2016), and White Sturgeon (Acipenser
transmontanus; DeVore et al. 1995). Alligator
Gar anglers are often not equipped to measure
fish weights in the field, even though fishing
Figure 3. Bias plots comparing actual success is measured by the number of fish
Alligator Gar girth measurements to those caught, and the lengths and weights of those fish
predicted using regression analysis: A) TL:G, (Ault and Luo 2013, Meerbeek and Crane 2017).
B=SL:G) of Alligator Gars. The diagonal Our results promote Alligator Gar conservation
line represents a 1:1 relationship between by providing a method to assign weight to
predicted and actual girth measurements. Alligator Gar, reducing the need to over-handle
or harvest fish simply to obtain a weight.
status did not significantly improve regression
models for estimating weights of Muskellunge The models presented in this study provide
(Esox masquinongy). Similarly, Craig et al. fisheries managers and anglers with a method to
(2005) found no differences in length-weight estimate measurements or weights of Alligator
relationships between sexes of Lake Sturgeon Gar that are typically captured in the field (G, SL,
(Acipenser fulvescens) from the St. Clair River, TL, and). However, readers should keep in mind
Michigan. However, they attributed this to a that these models include only Alligator Gar
small sample size of old, large females in their from Texoma Reservoir and application of these
study, which tend to not become larger than equations to other Alligator Gar populations
males until they have spawned several times should be done with caution, due to population
(Bruch 1999). It is possible that our results specific morphological differences. Further, use
are affected by a limited sample of old, large of these equations to predict measurements of
females, as Alligator Gars exhibit a periodic Alligator Gar outside of the size distribution
life history (Buckmeier et al. 2017), which may included in this study should be done cautiously.
affect the ultimate size of female gar, similar As future Alligator Gar conservation efforts
to species of Acipenseridae (Bruch 1999). Past increase in Oklahoma and across the range of the
Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)
36 Length-Weight Relationships of Alligator Gar

species, models in this study provide managers Buckmeier, D. L., N. G. Smith, D. J. Daugherty,
with a method to compare body condition and and D. L. Bennett. 2017. Reproductive
growth potential across populations. ecology of Alligator Gar: Identification of
environmental drivers of recruitment success.
Acknowledgments Journal of the Southeastern Association of
Fish and Wildlife Agencies 4:8-17.
The authors thank Micah Waters, Jory Commanda, N. 2018. Lake Sturgeon (Acipenser
Bartnicki, Shelby Jeter, Matthew Lyons, ODWC Fulvescens) management and status update
Southcentral Regional Management Staff, for the Lake Nipissing Watershed - Part of the
and the Tishomingo National Fish Hatchery Great Lakes Upper St. Lawrence Population,
personnel for assisting with field sampling. We Ontario, Canada. Master’s thesis. Nipissing
thank Kurt Kuklinski (ODWC) and Jim Long University, North Bay, Ontario.
(OSU) for reviewing an earlier draft of this Craig, J., M. Thomas, and S. Nichols. 2005.
manuscript. Additionally, comments provided by Length–weight relationship and a relative
Dr. Mostafa Elshahed and anonymous reviewers condition factor equation for Lake Sturgeon
greatly improved this manuscript. Financial (Acipenser fulvescens) from the St Clair River
support for this publication was provided system (Michigan, USA). Journal of Applied
by Sport Fish Restoration Program grants Ichthyology 21:81-85.
[F-50-R-25] and [F-86-D-1] to the Oklahoma DeVore, J. D., B. W. James, C. A. Tracy, and
Department of Wildlife Conservation. D. A. Hale. 1995. Dynamics and potential
production of White Sturgeon in the Columbia
References River downstream from Bonneville Dam.
Transactions of the American Fisheries
Ault, J. S., and J. Luo. 2013. A reliable game fish Society 124:845-856.
weight estimation model for Atlantic Tarpon Garcia de Leon, F. J., L. Gonzalez-Garcia, J.
(Megalops atlanticus). Fisheries Research M. Herrera-Castillo, K. O. Winemiller, and
139:110-117. A Banda-Valdes, A. 2001. Ecology of the
Binion, G. R., D. J. Daugherty, and K. A. Bodine. Alligator Gar, Atractosteus spatula, in the
2015. Population dynamics of Alligator Gar in Vincente Guerrero Reservoir, Tamaulipas,
Choke Canyon Reservoir, Texas: Implications Mexico. The Southwestern Naturalist 46:151-
for management. Journal of the Southeastern 157.
Association of Fish and Wildlife Agencies Inebnit III, T. E. 2009. Aspects of the
2:57–63. reproductive and juvenile ecology of Alligator
Boas, R., G. A. Castellanos-Galindo, C. Gar in the Fourche LaFave River, Arkansas.
Chong-Montenegro, P. Tompkins, and L. A. Master’s thesis. University of Central
Zapata. 2016. Length-weight relationship Arkansas, Conway, Arkansas.
of the Pacific Goliath Grouper, Epinephelus Jeter, S. E., M. J. Porta, and R. A. Snow. 2019.
quinquefasciatus (Bocourt, 1868). Journal of Body size estimation and identification of
Applied Ichthyology 32:727-728. twelve fish species using cleithrum bones.
Brinkman, E. L. 2008. Contributions to the Proceedings of the Oklahoma Academy of
life history of Alligator gar, Atractosteus Science 99:21-30.
spatula (Lacepede), in Oklahoma. Master’s Maurya, A., L. Prasad, and R. Gautam. 2018.
thesis. Oklahoma State University, Stillwater, Length weight relationship and condition
Oklahoma. factor of Anabas testudineus (Bloch, 1792)
Bruch, R. M. 1999. Management of Lake from Rudrasagar Lake (A Ramsar site),
Sturgeon on the Winnebago System - long Tripura. Journal of Pharmacognosy and
term impacts of harvest and regulations on Phytochemistry 7:1395-1398.
population structure. Journal of Applied
Ichthyology 15:142–152.

Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)

 R.A. Snow, M.J. Porta, and C.R. Sager 37

McDonald, D. L., J. D. Anderson, C. Hurley, B. Scharf, F. S., R. M. Yetter, A. P. Summers, and

W. Bumguardner, and C. R. Robertson. 2013. F. Juanes. 1998. Enhancing diet analyses of
Sexual dimorphism in Alligator Gar. North piscivorous fishes in the Northwest Atlantic
American Journal of Fisheries Management through identification and reconstruction of
33:811-816. original prey sizes from ingested remains.
Meerbeek, J. R., and D. P. Crane. 2017. A Fishery Bulletin 96:575-588.
comparison of Muskellunge weight estimation Schlechte, J. W., K. A. Bodine, D. J. Daugherty,
equations to a modified length-girth and G. R. Binion. 2016. Size selectivity of
technique. Muskellunge management fifty multifilament gill nets for sampling Alligator
years of cooperation among anglers, scientist, Gar: Modeling the effects on population
and fisheries biologists. American Fisheries metrics. North American Journal of Fisheries
Society Symposium 85:539-554. Management 36:630-638.
Nazir, A., and M. A. Khan. 2017. Length-weight Snow, R. A., M. J. Porta, R. W. Simmons Jr,
and length-length relationships of Cirrhinus and J. B. Bartnicki. 2018. Early life history
mrigala (Cyprinidae) and Xenentodon cancila characteristics and contribution of stocked
(Belonidae) from the River Ganga. Journal of juvenile Alligator Gar in Lake Texoma.
Applied Ichthyology 57:787-790. Proceedings of the Oklahoma Academy of
Neumann, R. M., C. S. Guy, and D. W. Willis. Science 98:60-68.
2012. Length, weight, and associated indices. Snow, R. A., M. J. Porta, and C. P. Porter. 2017.
Pages 637–676 in Zale, A. V., D. L. Parrish, Estimating fish length, weight, and age of
and T. M. Sutton. (Eds) Fisheries techniques, Central Stoneroller (Campostoma anomalum)
3rd edition. American Fisheries Society, using bone measurements. American Currents
Bethesda, Maryland. 42:5-10.
Oklahoma Department of Wildlife Conservation. Taylor, A. T., J. M. Long, R. A. Snow, and M.
2019. Oklahoma fishing regulations guide. J. Porta. 2019. Hybridization and population
Santos, M., J. Lemos, C. Pereira, A. Oliveira, M. genetics of Alligator Gar in Lake Texoma.
Tavares-Dias, and J. Marcon. 2012. Length- North American Journal of Fisheries
weight relationships of four ornamental fish Management. DOI: 10.1002/nafm.10346.
species from the Brazilian Negro River basin. Torres, M., F. Ramos, and I. Sobrino. 2012.
Journal of Applied Ichthyology 28:148-149. Length–weight relationships of 76 fish species
Sarkar, U. K., P. K. Deepak, and R. S. Negi. from the Gulf of Cadiz (SW Spain). Fisheries
2009. Length–weight relationship of Clown Research 128:171–175.
Knifefish, Chitala chitala (Hamilton 1822) Wood, A. D. 2005. Using bone measurements
from the River Ganga basin, India. Journal of to estimate the original sizes of bluefish
Applied Ichthyology 25:232-233. (Pomatomus saltatrix) from digested remains.
Fisheries Bulletin 103:461-466
Submitted August 7, 2020 Accepted November 22, 2020

Proc. Okla. Acad. Sci. 100: pp 30 - 37 (2020)