<%BANNER%>

Direct and Indirect Estimates of Black Crappie Size Selectivity to Otter Trawls

Permanent Link: http://ufdc.ufl.edu/UFE0021818/00001

Material Information

Title: Direct and Indirect Estimates of Black Crappie Size Selectivity to Otter Trawls
Physical Description: 1 online resource (60 p.)
Language: english
Creator: Binion, Gregory R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crappie, fisheries, selectivity
Fisheries and Aquatic Sciences -- Dissertations, Academic -- UF
Genre: Fisheries and Aquatic Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: I estimated size selectivity of bottom trawl sampling for black crappie Pomoxis nigromaculatus utilizing direct and indirect approaches. I used capture-recapture methods to directly measure the effects of fish size on catchability (q, the fraction of a fish stock collected with a given unit of fishing effort) at Lake Jeffords, Florida. Estimates of q were made for different length-groups roughly resembling age-classes 0 to 2 and fish 3+ (90-119, 120-149, 150-179, 180+ mm) by marking a subpopulation collected using three gear types (bottom trawls, hoopnets, and electrofishing). Recapture sampling with otter trawls occurred two weeks after marking events ended, allowing a direct estimate of q from recaptures of tagged fish from each gear and size class. Indirect estimates of selectivity were obtained with a population model applied to long-term data at four Florida lakes. I constructed age-structured models for each lake that predicted annual catches-at-age as a function of measured growth rates, a time series of recruitment anomalies, assumed survival rates, and unknown age/size selectivities. Selectivity parameters were estimated by fitting model predicted catches-at-age to a time series of bottom trawl catch-at-age using maximum likelihood. Direct measures of selectivity indicated catchability was highest for the 90-119 length-group and lowest for fish greater than or equal to 180 mm, with q declining by a factor of 2 or 3 for large fish relative to small fish. Model simulations from the age-structured indirect approach revealed dome-shaped selectivity patterns with relative selectivities peaking at age-1 for three of four lakes. Lake Johns was the only exception where age-0 fish was the most efficiently captured age-group when survival was low. Overall model trends indicated greater selectivity of younger fish (age-0 and age-1) to the gear followed by decreasing relative selectivity to older age-classes (age-2+). Trawl selectivity patterns suggested that otter trawls would be best for monitoring the abundance of small black crappie. My results indicate that adult black crappie will likely be underrepresented in bottom trawl samples, which would influence age structure and growth rate estimates and the effectiveness of this gear as an assessment tool for tracking adult crappie populations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Gregory R Binion.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Allen, Micheal S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021818:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021818/00001

Material Information

Title: Direct and Indirect Estimates of Black Crappie Size Selectivity to Otter Trawls
Physical Description: 1 online resource (60 p.)
Language: english
Creator: Binion, Gregory R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crappie, fisheries, selectivity
Fisheries and Aquatic Sciences -- Dissertations, Academic -- UF
Genre: Fisheries and Aquatic Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: I estimated size selectivity of bottom trawl sampling for black crappie Pomoxis nigromaculatus utilizing direct and indirect approaches. I used capture-recapture methods to directly measure the effects of fish size on catchability (q, the fraction of a fish stock collected with a given unit of fishing effort) at Lake Jeffords, Florida. Estimates of q were made for different length-groups roughly resembling age-classes 0 to 2 and fish 3+ (90-119, 120-149, 150-179, 180+ mm) by marking a subpopulation collected using three gear types (bottom trawls, hoopnets, and electrofishing). Recapture sampling with otter trawls occurred two weeks after marking events ended, allowing a direct estimate of q from recaptures of tagged fish from each gear and size class. Indirect estimates of selectivity were obtained with a population model applied to long-term data at four Florida lakes. I constructed age-structured models for each lake that predicted annual catches-at-age as a function of measured growth rates, a time series of recruitment anomalies, assumed survival rates, and unknown age/size selectivities. Selectivity parameters were estimated by fitting model predicted catches-at-age to a time series of bottom trawl catch-at-age using maximum likelihood. Direct measures of selectivity indicated catchability was highest for the 90-119 length-group and lowest for fish greater than or equal to 180 mm, with q declining by a factor of 2 or 3 for large fish relative to small fish. Model simulations from the age-structured indirect approach revealed dome-shaped selectivity patterns with relative selectivities peaking at age-1 for three of four lakes. Lake Johns was the only exception where age-0 fish was the most efficiently captured age-group when survival was low. Overall model trends indicated greater selectivity of younger fish (age-0 and age-1) to the gear followed by decreasing relative selectivity to older age-classes (age-2+). Trawl selectivity patterns suggested that otter trawls would be best for monitoring the abundance of small black crappie. My results indicate that adult black crappie will likely be underrepresented in bottom trawl samples, which would influence age structure and growth rate estimates and the effectiveness of this gear as an assessment tool for tracking adult crappie populations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Gregory R Binion.
Thesis: Thesis (M.S.)--University of Florida, 2007.
Local: Adviser: Allen, Micheal S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021818:00001


This item has the following downloads:


Full Text





DIRECT AND INDIRECT ESTIMATES OF BLACK CRAPPIE SIZE SLECTIVITY TO
OTTER TRAWLS




















By

GREGORY R. BINION


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2007




























2007 GREGORY R. BINION




























To my grandfather "Ace" who instilled in me at a young age a passion for fish.









ACKNOWLEDGMENTS

I would like to thank and acknowledge my supervisory committee members Dr. Bill Pine,

Jim Estes, and Marty Hale for their encouragement and support, and especially my committee

chair, Dr. Mike Allen, for his mentorship and direction throughout my research.

I would also like to thank various members of the Allen lab for their help in field collection

and processing; Galen Kaufman, Jason Dotson, Kevin Johnson, Christian Barrientos, Erika

Thompson, Aaron Bunch, Melissa Woods-Jackson and Drew Dutterer. I would like to

acknowledge various members of the Florida Fish and Wildlife Conservation Commission (Eric

Nagid, Travis Tuten, Will Strong, Bill Johnson, and Janice Kerns) for collaboration in research.

A special thanks goes to office mates Matt Catalano and Mark Rogers whom encouraged and

supported my development as a student, mentoring and answering questions when I encountered

problems. Finally, I want to thank my loving wife, parents and grandparents for their endless

encouragement, love, and support.









TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ..............................................................................................................4

LIST OF TABLES ......... ..... .... ....................................................6

LIST OF FIGURES .................................. .. ..... ..... ................. .7

ABSTRAC T .........................................................................................

CHAPTER

1 INTRODUCTION ............... .............................. ............................. 10

2 M E T H O D S .......................................................................................................16

D direct M measure of Selectivity ................................................................. ........................16
Indirect M measure of Selectivity .............................................................................. ...... 19

3 R E SU L T S .............. ... ................................................................25

D direct M measure of Selectivity ........................................................................ ...................25
Indirect M measure of Selectivity............................................................ ...............26

4 D IS C U S S IO N ........................................................................................................4 6

L IST O F R E F E R E N C E S ..................................................................................... ....................54

B IO G R A PH IC A L SK E T C H .............................................................................. .....................60









LIST OF TABLES


Table page

3-1 Summary of Lake characteristics for study locations................................... .................29

3-2 Summary of CPUE-at-age (fish/min) data showing the mean, minimum, and
maximum values for each age and lake and combined across lakes. ................................30









LIST OF FIGURES


Figure page

3-1 Lake Jeffords located in Alachua County, North Central Florida.............. ................ 31

3-2 Geographic location of study lakes used for indirect measure of trawl size selectivity ....32

3-3 Bootstrap estimates of mean tagging mortality. Error bars represent 95% confidence
in terv als ................... .......................................................... ................ 3 3

3-4 Likelihood profiles for q by length-group using mean tagging mortality rates.
Maximum likelihood estimates (MLE's) represented at peak of each curve ..................34

3-5 Maximum likelihood estimates of q by length-group and associated 95% confidence
intervals using m ean m ortality rates. ............................................................................ 35

3-6 Maximum likelihood estimates for q calculated from the lower, mean, and upper
tagging mortality rates. ................................ .......... .....................36

3-7 Predicted mean length at age for each lake. The least-squares equations are shown.......37

3-8 Observed time series of relative recruitment anomalies by lake. ................ ..............38

3-9 Relative selectivity by age for each study lake from base model......................................39

3-10 Average relative selectivity for lakes sampled with standard trawl (Lakes Griffin,
Johns, and L ochloosa)......... .................................................................. ......... .......40

3-11 Relative selectivity by age under varying assumptions of survival.................................41

3-12 Lake Griffin probability profiles with relative selectivity on the x-axis and relative
probability on the y-axis. ............................ ......... ...... ... ....... ..... 42

3-13 Lake Johns probability profiles with relative selectivity on the x-axis and relative
probability on the y-axis. ............................ ......... ...... ... ........ .... 43

3-14 Lake Lochloosa probability profiles with relative selectivity on the x-axis and
relative probability on the y-axis. ........................................................... .....................44

3-15 Lake Okeechobee probability profiles with relative selectivity on the x-axis and
relative probability on the y-axis. ........................................................... .....................45









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

DIRECT AND INDIRECT ESTIMATES OF BLACK CRAPPIE SIZE SLECECTIVITY TO
OTTER TRAWLS

By

GREGORY R. BINION

December 2007

Chair: Mike Allen
Major: Fisheries and Aquatic Sciences

I estimated size selectivity of bottom trawl sampling for black crappie Pomoxis

nigromaculatus utilizing direct and indirect approaches. I used capture-recapture methods to

directly measure the effects of fish size on catchability (q, the fraction of a fish stock collected

with a given unit of fishing effort) at Lake Jeffords, Florida. Estimates of q were made for

different length-groups roughly resembling age-classes 0 to 2 and fish 3+ (90-119, 120-149, 150-

179, 180+ mm) by marking a subpopulation collected using three gear types (bottom trawls,

hoopnets, and electrofishing). Recapture sampling with otter trawls occurred two weeks after

marking events ended, allowing a direct estimate of q from recaptures of tagged fish from each

gear and size class. Indirect estimates of selectivity were obtained with a population model

applied to long-term data at four Florida lakes. I constructed age-structured models for each lake

that predicted annual catches-at-age as a function of measured growth rates, a time series of

recruitment anomalies, assumed survival rates, and unknown age/size selectivities. Selectivity

parameters were estimated by fitting model predicted catches-at-age to a time series of bottom

trawl catch-at-age using maximum likelihood. Direct measures of selectivity indicated

catchability was highest for the 90-119 length-group and lowest for fish greater than or equal to

180 mm, with q declining by a factor of 2 or 3 for large fish relative to small fish. Model









simulations from the age-structured indirect approach revealed dome-shaped selectivity patterns

with relative selectivities peaking at age-1 for three of four lakes. Lake Johns was the only

exception where age-0 fish was the most efficiently captured age-group when survival was low.

Overall model trends indicated greater selectivity of younger fish (age-O and age-1) to the gear

followed by decreasing relative selectivity to older age-classes (age-2+). Trawl selectivity

patterns suggested that otter trawls would be best for monitoring the abundance of small black

crappie. My results indicate that adult black crappie will likely be underrepresented in bottom

trawl samples, which would influence age structure and growth rate estimates and the

effectiveness of this gear as an assessment tool for tracking adult crappie populations.









CHAPTER 1
INTRODUCTION

Effective management of fishery resources is dependent on the quality of information

available for decisions. To develop optimal management strategies, biologists must be confident

that sampling reliably tracks population metrics. These strategies are often reliant on precise

estimates of some metric of population size and its corresponding level of production

(biomass/numbers). Gear selectivity can influence precision and accuracy of these measures

(Hilbom and Walters 1992). Samples collected from many fish populations with a variety of

gears often don't accurately describe the true age and size structure of the target population.

Therefore, obtaining an abundance index that reflects the actual age/size composition of a

population allows managers to monitor population trends such as recruitment, growth, and

mortality and evaluate population responses to management policies (e.g., size limits) (Hilborn

and Walters 1992).

When evaluating gear types, an important distinction between gear selectivity and gear

efficiency must be delineated. Gear selectivity is defined as the composition of a sample relative

to the true population metric (e.g. size, age, growth rate), and relative selectivity refers to the

effectiveness of a sampling gear to capture a particular size or species of fish relative to its

efficiency at capture of other sizes or species (Hubert 1996). In contrast, gear efficiency

describes the magnitude of effort required to catch adequate sample sizes.

Gear selectivity patterns are commonly attributed to intrinsic factors like fish size

(Beamesderfer and Rieman 1988, Myers and Hoenig 1997, Wakefield et al. 2007), fish density

(Mclnerny and Cross 2000, Rogers et al. 2003), species (Laarman and Ryckman 1982, Sammons

et al. 2002), sex (Jagielo 1999), behavioral patterns (Reynolds 1996, Jagielo 1999), and habitat

preferences (Jacobson et al. 2001) as well as extrinsic factors such as seasonal variation (Pope









and Willis 1996), environmental conditions or characteristics (Hayes et al. 1996; McInerny and

Cross 2000), diel variation (Paragamian 1989, Dumont and Dennis 1997), gear construction

(O'Neil and Kynoch 1996, Lok et al. 1997, Farmer et al. 1998), gear type (Kraft and Johnson

1992, Jackson and Noble 1995, Otway et al. 1996), and sampling crew expertise.

Estimates of gear selectivity are important for fish stock assessment. Estimates of

selectivity allow managers to assess population composition based on samples which may not

represent the true population, and estimates of selectivity provide information on aspects of the

population which is not readily observable. Adjusting for selectivity allows managers to obtain a

more accurate abundance index for the age and size structure of a stock because many samples

do not adequately represent the true population age or size structure. This enhances the ability of

managers to draw inferences about stock trends like recruitment, growth, and mortality (Hilborn

and Walters 1992).

Gear selectivities are commonly used to determine the effects of fishing on the size and

age composition of a fishery and are commonly used in assessment models to link size/age

structure of catch data to the size/age structure of the fish population (Walters and Martell 2004;

Taylor et al. 2005). Such models are required to predict effects of different harvest rates,

calculate biological reference points like spawning potential ratio (SPR), and determining

appropriate levels of sustainable yield for a fishery (Maunder 2002). Thus, quantifying gear

selectivity allows biologists to adjust abundance indices to represent the true size/age

composition which guides future management actions.

Measurements of the selective properties of fishing gears are often made utilizing direct

and indirect methods (Pollock et al. 1990; Walters and Martell 2004). Direct methods involve

comparing catch composition against a known population structure. The most direct method for









estimating selectivity is a mark recapture experiment creating a known population, then

calculating the proportion of fish caught by the gear in a given length category from the marked

subpopulation (Hamley and Regier 1973; Myers and Hoenig 1997; McInerny and Cross 2006).

Accurate estimation of selectivity using capture-recapture methods requires several assumptions

including: (1) the population of interest is closed to additions and deletions. That is, recruitment,

natural mortality, immigration, and emigration must be minimal, (2) tags are not lost or go

undetected, and (3) equal capture probability (i.e. no capture heterogeneity and/or trap response)

(Pollock et al. 1990). Unlike direct methods, indirect measures of selectivity require no prior

knowledge about the age composition of a population. If catch-at-age data from the commercial

or recreational sectors are available, age structured population models like virtual population

analysis (VPA) can estimate the age/size selective properties of the fishing gear used. Other

approaches incorporate the catch rates of various sizes of fish from different gear types and/or

mesh size to compare relative gear selectivity between gears, but such studies do not identify the

true selectivity of either gear (e.g., Boxrucker and Plosky 1989; Miranda et al. 1992; Millar and

Holst 1997).

Indirect or relative measures of abundance such as catch per unit effort (CPUE) are

commonly used by managers to assess and analyze trends in fish population abundance. In order

for CPUE to directly index population abundance, the relationship between catch rate and

abundance must be:

C N
= q (1)
f A

where C = catch,f= fishing effort, q = catchability coefficient (the fraction of population

removed per unit of effort), N= fish abundance and A = area occupied by stock (Ricker 1975).









This equation infers a linear relationship between CPUE and abundance with a constant slope q,

which is often not the case.

Catch per effort is a function of two factors: catchability and fish density (Hilborn and

Walters 1992; Arreguin-Sanchez 1996). Therefore, variability in q causes variability in CPUE

that is not related to population size, so catch statistics should be adjusted to account for

variation in catchability. Because catchability is a function of selectivity, it is an important

parameter when using CPUE to index abundance. Furthermore, when catchability coefficients

are apportioned by age/size classes, the estimated coefficients are actual age/size gear selectivity

estimates.

Like selectivity, catchability differs with a wide range of factors including fish age (Pierce

and Tomcko 2003), fish size (Bayley and Austen 2002; McInemy and Cross 2006), species

(Bayley and Austen 2002; Schoenebeck and Hansen 2005), fish density (Peterman and Steer

1981; Rogers et al. 2003), sample gear type (Hansen et al. 2000; Pierce and Tomcko 2003),

environmental conditions during sampling (Bayley and Austen 2002), and sampling season

(Schoenebeck and Hansen 2005; McInerny and Cross 2006). Nielson (1983) found catchability

from otter trawls to be similar across age-classes of adult yellow perch. Pierce and Tomcko

(2003) showed q of northern pike to vary with age in gill-nets. McInemy and Cross (2006)

quantified the effects of size, season, and density of black crappie on trap-net catchability. They

found q increased with fish size, and catchability was higher in spring than fall. Catchability also

varied with density as both increased and decreased values of catchability were observed for

different length-groups and sampling periods. Bayley and Austen (2002) provided a

comprehensive evaluation of electrofishing q estimates for different fish species, sizes, and

seasons under variable environmental conditions. Overall, knowledge of q can allow managers









to adjust abundance indices and estimate absolute abundances, estimate gear selectivity patterns,

identify seasonal and/or environmental biases associated with sample gears, and aid in the

selection of appropriate gears to maximize management objectives.

Black crappie support one of the most popular sport fisheries in North America often

ranking first or second among angler preference, but can be difficult to manage. Sampling

crappies to accurately describe rate functions (such as growth and mortality), abundance and size

structure is often demanding requiring much effort. Indexing black crappie abundance and size

structure is challenging due to differences in gear performance and selectivity patterns. In the

Midwest, trap nets have been useful in collecting large samples of crappie of all sizes

(Gablehouse 1984; Colvin and Vasey 1986; Boxrucker and Ploskey 1989), but true gear

selectivity has rarely been measured (but see McInerny and Cross 2006). Conversely, trap nets

in some southeastern systems collected young fish but few adults (Sammons and Bettoli 1998;

Maceina et al. 1998). Miranda and Dorr (2000) quantified the size selective effects of crappie

angling in five southeastern systems and reported dome-shaped selectivity for crappie vulnerable

to angling (size range: 20.0 to 39.8 cm) with smaller and larger sized crappie less susceptible

than intermediate sizes indicating differences in catchability and thus, exploitation.

Trap net efficiency and relative selectivity has been evaluated and compared to other gears

in numerous studies to determine which methods of capture are most effective (McInerny 1989;

Boxrucker and Ploskey 1989; Miranda et al. 1992; St. John and Black 2004). McInerny (1989)

found trap nets were the most effective and cost-efficient gear deployed for sampling black

crappie populations at Lake Wylie, North Carolina. Miranda et al. (1992) reported a higher catch

per effort with trap nets compared to electrofishing in the spring and rotenone sampling in the

summer for four Mississippi waters. Boxrucker and Ploskey (1989) revealed greater sampling









efficiency and less variation in catch per effort with trap nets when compared to electrofishing

and gillnetting. Thus, trap nets have provided useful data in some cases, but the selectivity of

trap nets relative to the population size/age has seldom been measured.

Otter trawls have received far less attention than other capture methods to index crappie

abundance. In Florida waters, otter trawls have proven to be successful at capturing black

crappies (Schramm et al. 1985; Allen et al. 1999; Pine 2000). Allen et al. (1999) compared the

relative efficiency of trap nets versus otter trawls for sampling black crappie in two Florida lakes

and reported that trawl sampling was superior to trap nets based on the size range of fish

collected, accuracy of abundance estimates, required sampling effort, and expenditures

associated with gear. Pine (2000) compared the relative selectivity of two different sized bottom

trawls and found a smaller trawl was more effective at collecting juvenile black crappie than a

larger trawl. Despite the importance of black crappie in Florida and the popularity of bottom

trawls for sampling crappie populations, trawl selectivity of black crappie relative to the

population has not been measured. Thus, in order for biologists to utilize trawl catch data for

management it is important to understand the selective properties of the gear relative to the

population. This will enhance the ability of managers to use trawl CPUE as an index of

abundance as well as length and age frequency information to describe the size/age structure of

black crappie populations. My objectives were to (1) estimate size-specific catchability (q) of

black crappie collected with otter trawls (2) estimate relative age/size-specific selectivity of

bottom trawl gears and (3) use those selectivity patterns to evaluate the utility of otter trawls as

an assessment gear for black crappie for Florida lakes.









CHAPTER 2
METHODS

Direct Measure of Selectivity

Capture-recapture sampling took place at Lake Jeffords, Florida during January 2007.

Lake Jeffords is a 65 hectare, mesotrophic (Pine 2000) system located in Alachua County, North

Central Florida (Figure 1). I selected Jeffords because I felt I could adequately sample the entire

lake (i.e. sample all available habitat types) and create a large enough marked subpopulation to

obtain reliable catchability estimates.

Mark-recapture methods were used to create a tagged population using three gear types.

Marking took place over a 10 day period in January 2007, with electrofishing gear sampled on

day 1, otter trawls sampled on days 1, 2, and 3, and hoopnets sampled on days 7 10. I sampled

with three gears during the marking event to ensure all available habitat types of the lake were

sampled. The recapture event took place over a two day period with bottom trawls two weeks

after the first marking day. I only used bottom trawls during the recapture period, which allowed

estimation of trawl size selectivity based on my known tagged population. The perimeter of the

lake was electrofished at both events to ensure fish had not moved into the shallow littoral zone

where it is not possible to effectively trawl. Captured fish from all trawls were divided into

subgroups by length. This division allowed estimation of q by size, providing a measure of

actual trawl size selectivity. The length-groups (mm) roughly resembled ages 0 (90-119), 1

(120-149), 2 (150-179) and adult fish three or older (180+). Abundance estimates were obtained

using a Lincoln-Peterson estimator (due to closed system and 2-stage mark-capture sampling

event) and the proportions of marked fish were calculated as the number of fish caught in the

recapture divided by the abundance estimate. All black crappie captured in the field during

marking were measured for total length (TL) to nearest (mm) and pelvic fin clipped. Since only









two weeks passed from mark to recap events and fish were fin clipped instead of using

conventional tag types like a T-bar tag, I assumed tag loss to be negligible. All black crappie

captured during the recapture where measured for total length to nearest (mm) and checked for

fin clips.

Bottom trawls were pulled from a 7-m boat powered with a 70 hp outboard in all areas of

the lake except in the shallow littoral zone to avoid fouling by vegetation. Effort was constant

throughout the study at three minutes per trawl. The trawl net consisted of a 4.88-m long body

and 4.6-m mouth and the body is constructed with 38.1 mm stretch mesh and 31.8 mm stretch

mesh in the cod end (Allen et al. 1999). Under tow, the mouth of the trawl is spread open with

floats (25 x 50 mm) that are secured to the headrope of the trawl mouth. The sweep, or chain

line, was attached to the footrope of the net. Wooden doors (38.1 x 76.2 cm) were secured to

146 cm leglines and a 15.3-m trawl bridle. The weighted doors served to open the trawl mouth

and allowed the net to sample near the bottom.

Modified hoop nets were deployed in the middle of the lake at various sites. Hoop nets

consisted of four similar-sized fiber-glass hoops either 0.9, 1.2 or 1.5m in diameter and covered

with 5.1cm stretch nylon mesh webbing. A 23-m lead was used to connect two nets, which

would direct fish toward a hoop net as they traveled along the lead. All hoop nets were set

during the day, fished for 48 hours, and retrieved. Hoop nets were only used for capture

sampling event.

Electrofishing was conducted with a Smith-Root model SR18 electrofisher, equipped

with a Smith-Root 9.0 GPP pulsator powered by a 9,000 Watt generator. Approximately 7 amps

of DC current were produced at 120 pulses per second. The entire shoreline perimeter was

sampled (as described above) with an experienced crew of one netter and one boat operator.









I estimated tagging mortality, defined as mortality from capture, handling, and tagging for

each size-group to adjust the size of our marked population available for recapture. A sub-

sample of marked fish were held in aerated bait tanks and placed in holding pens as replicates (n

= 8) for 24 hours to estimate associated tagging mortality for different length-groups. Holding

pens were constructed out of pvc pipe which consisted of a rectangular frame that measured 3.0

m length by 1.25 m width. The body of the holding nets consisted of 19.3 mm stretch mesh

webbing that extended to a depth of 1.1 m. The observed mortality rates for each size-group and

holding pen were randomly re-sampled with replacement using a bootstrap to create 1,000 Monte

Carlo estimates (Haddon 2001). The 95% confidence intervals were calculated at the 2.5 and

97.5 percentiles using the means of the resample from the bootstrap.

I used maximum likelihood methods to estimate how q varied with fish size. The Poisson

log likelihood function was appropriate and indicated as

In L(O, I q) = -( (P,) +, (O,) In(P,)) (2)

where P, = (number available for recapture in size-group i q effort) and 0, = (number

of observed recaptures in size-group i). Catchability for each length-group was estimated by

maximizing the negative likelihood function (i.e. minimize the differences between the observed

and expected recaptures). Parameters for the model included survival from marking (S) = (1 -

observed tagging mortality), number available for recapture = (number marked S), and q, the

fraction of population caught per unit of effort.

To describe the uncertainty in q estimates, I constructed likelihood profiles for each

length-group. These profiles were probability distributions (i.e. p-values) for the parameters.

The 95% confidence intervals for each parameter were calculated using Wilk's likelihood ratio

test statistic (Pawitan 2001) equal to










W = 2*lnL(hat)(') (3)


where W= Wilk's statistic, L(yphat) = log likelihood at MLE, and L((p) = log likelihood at

some p value less than the maximum likelihood estimate (MLE). Wilk's statistic conforms to a

Chi-square distribution with one degree of freedom (Pawitan 2001).

Indirect Measure of Selectivity

Black crappie populations in Florida lakes have been sampled with otter trawls over the

last two to three decades. I used a long-term database from four lakes to estimate the size/age

selectivity of bottom trawls using an age-structured population modeling approach. In this

context, selectivity was defined as differences in relative fish susceptibility due to size/age. The

model was used to estimate selectivity at age by comparing observed and model-predicted

catches-at-age using maximum likelihood estimation.

Annual bottom trawl sample data were obtained from lakes Griffin, Johns, Lochloosa, and

Okeechobee (Figure 2). The length of the data time series varied among lakes and ranged from

five years for Lake Johns (2002-2006) to 20 years at Lake Okeechobee (1987-2006). All lakes

exceed 1,500 ha with mean depths ranging from 1.8 m to 2.8 m and are classified as eutrophic or

hypereutrophic (Florida Lakewatch 1999, 2001; Forsberg and Ryding 1980) (Table 1).

Black crappie populations were sampled using a combination of fixed and/or random sites.

Lakes Lochloosa and Johns were divided into 250m2 grids using ARC GIS software with buffers

built in to avoid sampling in the vegetated littoral zone. Vegetation fouls the gear, reduces gear

performance, and does not allow an accurate assessment of gear utility. At Lake Lochloosa,

fixed and random sites were used throughout, whereas Lake Johns data were obtained from

randomly generated fixed sites until years 2005 and 2006 where a combination of six fixed and

six random sites were used. At Lake Griffin, 25 to 68 trawls were pulled based on a similar but









slightly different SRS design (fixed sites to 2002 and random sites from 2003). One to 20 trawls

were pulled at Lake Okeechobee on the north end of the lake from 0.8 to 2.4 km offshore

between Taylor Creek Lock (S-193) and Nubbin Slough Spillway (S-191) until 500 black

crappies were captured. In 2005, due to a large drop in numbers likely attributed to the 2004

hurricane effects on the lake, the sampling focus changed from minimum numbers to minimum

effort (-150 minutes). In lakes where a combination of fixed/random sites were used in a given

year (Johns, Lochloosa) I tested for differences in mean CPUE and size structure to determine if

fixed and random site samples could be pooled. Analysis of variance (ANOVA) indicated no

significant site type or year*site type effects on mean CPUE for both lakes: Johns (P = 0.48, P =

0.44), Lochloosa (P = 0.86, P = 0.08). Differences in size structure by site type were evaluated

using a Chi-square test and results indicated no significant differences in size structures between

fixed or random sites in any year. Based on these findings, I combined fixed and random sites

for the analysis.

There were some method differences among the lakes. Trawls were towed at a speed 2.0 -

2.5 m/s (1800-2000 RPM) with the exception of Lake Okeechobee where trawls were pulled at

1.0 m/s. Effort was constant throughout the study at three minutes per trawl with the exception

of Lake Lochloosa where some trawls were pulled for five minutes, and Lake Okeechobee where

trawls were pulled for 15 or 30 minute intervals (see Miller et al. 1990). Samples were collected

during daylight hours from October December at Lakes Griffin, Johns and Lochloosa, and in

January at Lake Okeechobee. The trawl net used at Lakes Griffin, Johns, and Lochloosa was the

same as used for Lake Jeffords capture-recapture sampling (described above) which I will refer

to as the standard trawl net. Lake Okeechobee trawl net was similar in design and application









but was larger, having a 10.7-m headrope, 32-mm square body mesh, and 25-mm square cod end

mesh.

All black crappie captured in the field were measured for total length (TL) to nearest (mm).

Subsamples of five fish per one cm group were brought back to the laboratory for further

analysis. However, age data were collected only every other year from 2001 to 2006 at Lake

Lochloosa. At the laboratory, gender, total length (TL) to nearest (mm), total body weight (TW,

g) were determined and sagittal otoliths were removed. Ages were determined in whole view by

two independent readers as the number of opaque bands on sagittal otoliths. Otoliths from fish

older than two years as well as any discrepancies on whole reads were sectioned along the

dorsoventral plane before aging. Use of black crappie otoliths for aging in Florida has been

validated by Schramm and Doerzbacher (1982).

I used an age-structured population model to estimate the relative age/size-specific

selectivity of black crappie to bottom trawls at each lake. The model predicted catch-at-age as a

function of von Bertalanffy growth parameters, annual relative recruitment anomalies, an

arbitrary number of intital recruits, assumed instantaneous rates of total mortality (Zo = e-zo

survival (So) to age-1, Z= e-z = survival (S) past age-1), and unknown age/size-specific

selectivities (to be estimated). The numbers at age (Na,) in a given year were estimated as

No, = R, Ro (4)


Ni, = Na1,--1 So (5)

N ,t =N ,tl S (6)

where Rt are annual recruitment anomalies, Ro is the average annual recruitment numbers

(arbitrary value used to scale model), So is survival from age 0 to age 1, and S is annual survival

for age-l+ fish. Different survival rates for age-0 relative to older fish were used because of









expected lower survival for age-0 fish. Annual recruitment anomalies (Rt) for each lake and year

were estimated by dividing mean catch per effort (CPUE) of age-0 crappie in year t by the

median age-0 CPUE across all years. This provided an index of strong and weak year classes in

the population model and was used as a basis for predicting future catches-at-age with the trawls.

This model allowed prediction of the relative numbers of fish at each lake, age, and year based

on the input parameters. From the numbers at age matrix, I predicted a catch-at-age matrix from

a hypothesized selectivity schedule. Expected catch at age was calculated as

Ca,t = N,t Sa (7)

where Ca,t is the catch at age at time t, and Sa are unknown selectivity at age parameters.

Growth parameters were estimated using the von Bertalanffy growth equation fitted to

weighted mean length at age data obtained from age-length keys using the technique described

by Devries and Frie (1996). The von Bertalanffy equation is

La = L *(1- e(K*(a-to))) (8)

where La is length at age, Lo represents the average asymptotic length, K is the metabolic

growth coefficient, a is fish age, and tois the age at zero length. A growth model was

constructed for each lake by pooling annual length-age samples after determining that growth did

not differ widely among years. The von Bertalanffy growth model was also used to link my age-

based selectivities to length-based selectivities for each cohort.

Because fish suffer higher rates of mortality early in life than adults (Hjort 1914; Cushing

1975), my base model assumed different instantaneous rates of total mortality for these two life

stanzas: Z = 1.2 (So = 0.30) for survival to age-1, and Z = 0.6 (S = 0.54) for ages 1+. These rates

served as a base for comparison to other simulations under varying assumptions for So and S. I

evaluated the sensitivity of selectivity estimates to different survival rates by estimating









selectivity parameters under different values of So and S. Values of So ranged from 0.22 to 0.37

and S ranged from 0.45 to 0.67.

Observed catch-at-age for each lake and year was estimated using age-length keys.

Because age data were only collected at Lake Lochloosa every other year, we used the previous

and post years age data as the basis for an age length key (e.g., 2002 age structure was estimated

using 2001 and 2003 age-length subsamples), and apportioned fish to ages based on the existing

length data. Observed catch-at-age for each year was standardized for sampling effort by

dividing the catch-at-age by the total lake effort (trawl minutes) for that year.

I used a multinomial log likelihood function to estimate selectivity at age by minimizing

the differences between observed and expected (i.e., model-predicted) proportions of catch-at-

age using the Solver function in Excel. The multinomial log likelihood equation was

lnL(O,, I S) = nZ 0,, ln(P,,) (9)
a t

where n is the number of years model fit to catch data, Oa,t represents the observed

proportion of catch at age a in year t, and Pa,t is the predicted proportion of catch at age a in year

t. I used a logit transformation on selectivity parameter estimates in the optimization routine to

constrain selectivities between zero and one. When working with a parameter such as a

probability that must be between zero and one, the logit transformation allows parameter

estimates to range from -oo to o. The logit transformed selectivities were calculated as


X'= ln (10)


where X' is the logit transformed selectivity at age. This logit transformation was used

when solving for the point estimates of selectivity, as well as for the likelihood profiles (below).









Parameter uncertainty was evaluated by calculating 95% likelihood profile confidence

intervals via a likelihood ratio test (Hilborn and Mangel 1997) using the likelihood profile

function in Poptools for Excel (www.cse.csiro.au/poptools/). The profile function allowed me

to test alternative parameter estimates for all age-specific selectivities by holding one selectivity

estimate constant, then iteratively solving for the maximum likelihood estimate by varying the

remaining parameters and repeating with different values until the profile was constructed. The

likelihood ratio test (LRT) was expressed in terms of the differences in the deviance or twice the

difference between the negative log-likelihoods (Hilbom and Mangel 1997). The deviance for

each simulation was found by

2 (LnL LnL ) (11)

where LnLres is the likelihood value for restricted or nested model and LnLmax is the

maximum likelihood value for full model. The likelihood ratio test is described by a Chi-square

distribution with r degrees of freedom. The degrees of freedom were determined by the

difference in the number of parameters estimated between the models (Hilbom and Mangel

1997). The LRT allowed comparisons of the probabilities for each selectivity estimate occurring

relative to alternative parameter values (hypotheses).









CHAPTER 3
RESULTS

Direct Measure of Selectivity

The number of marked fish and size range captured varied by gear type. I marked 1,250

fish with bottom trawls (size range: 80 365 mm), 23 fish with hoopnets (160 312 mm), and 9

with electrofishing gear (123 318 mm). Recapture with bottom trawls netted 788 fish (88 -

304 mm), 54 of which were previously marked individuals. Based on my recapture rates and

adjusting for differential tagging mortality, I marked approximately 0.065 % of the total black

crappie population at Lake Jeffords. I marked 0.098 % of fish in group 90-119, 0.064 % of fish

in 120-149, 0.073 % in 150-179, and 0.041 % offish 180+.

Tagging mortality for the smaller length groups (90-119, 120-149 mm) was much higher

compared to the larger groups (150-179, 180+ mm) (Figure 3). The 90-119 and 120-149 length-

groups experienced high tagging mortality at 68 and 36%, respectively. The 150-179 and 180+

groups experienced much lower tagging mortality rate averaging only 12 and 1%. Overall,

tagging mortality was much higher for the two smallest length-groups relative to larger fish.

The likelihood profiles for each group indicated an overall decreasing trend in q with

increasing fish size (Figure 4). Results showed that the maximum likelihood estimates (MLE's)

of q for the length-group 90-119 were 2 to 3 times higher than q estimates at larger sizes,

assuming the calculated mean tagging mortality. The likelihood profiles revealed high

uncertainty in all the estimates of q, but higher uncertainty for the 90-119 mm size-group when

compared to other groups. Figure 5 shows the lower and upper 95% confidence bounds for the

MLE's using mean mortality estimates.

I evaluated how uncertainty from tagging mortality estimates would influence estimates of

q. The maximum likelihood estimates for q based on the lower, mean and upper tagging









mortality rates are presented in Figure 6. Based on mean tagging mortality rates, maximum

likelihood estimates for length-group 90-119 were approximately twice the MLE values of

groups 120-149, 150-179 and three times as high as length-group 180+. Differences in the MLE

estimates based on low tagging mortality among size-groups were reduced, except for groups 90-

119 and 180+ which still varied by a factor of two. In contrast, MLE estimates applying high

tagging mortality exhibited large variation among length-groups where q varied by a factor

greater than 2 when comparing the 90-119 group to groups 120-149, 150-179 and a factor of 4 to

group 180+. Overall model trends indicated decreasing catchability to trawl gear as fish length

increased, with a greater amount of uncertainty in the estimates for the smallest length group (90-

119).

Indirect Measure of Selectivity

Black crappie growth varied among lakes (Figure 7). Average asymptotic length (L,)

among lakes varied from 335 to 398, whereas metabolic growth coefficient (K) ranged from 0.27

to 0.42, and to from -0.79 to -1.17. As expected, lakes with higher K values had lower L. values,

and vice versa (Figure 7).

The general patterns observed in the catch-at-age data included higher catches of age-0 and

age-1 fish relative to older age classes, as would be expected for any population. Age-0 catch

rates among the lakes varied from 0.018 to 16.84 with an average of 3.24 (fish/min). Age-1

catch at age ranged from 0.016 to 10.83 with a mean of 2.12, whereas age-2 CPUE varied from

0.007 to 7.45 and averaged of 0.94 (fish/min). Catch rates for crappie 3 and older were

considerably less relative to younger age-classes and ranged from 0 to 3.70 with average catch

rates to 0.37 (Table 2).

Relative recruitment anomalies varied by lake and some lakes exhibited large fluctuation in

recruitment while others showed little variability in recruitment strength (Figure 8). The









recruitment anomalies for Lake Griffin varied from 0.24 to 3.57 with an average value of 1.13.

Lake Johns recruitment values varied from 0.85 to 1.12 with a mean relative recruitment of 0.99

indicating little variability in recruitment. Lake Lochloosa anomalies ranged from 0.14 to 2.35

with an average recruitment value of 1.04. Lake Okeechobee exhibited large fluctuations in

year-class strength with recruitment values ranging from 0.01 to 5.01 with a mean of 1.50. Thus,

the recruitment trends as indexed with age-0 fish catch rates suggested substantial variation in

recruitment among years at each lake.

My age structured model estimated dome-shaped selectivity with peak values for black

crappie in bottom trawls at age-1. In general, age-0 and age-1 fish were more susceptible to

trawl gears than older age-classes (ages-2+). Model simulations for the base model (Zo = 1.2 =

So = 0.30, Z = 0.6 = S = 0.55) revealed peak selectivity at age-1 for all lakes (Figure 9). The

average selectivity schedule for Lakes Griffin, John, and Lochloosa sampled with the standard

bottom trawl gear also indicated peak selectivity at age-1 (Figure 10).

Varying assumptions for instantaneous rates of mortality (i.e. survival) influenced the

selectivity parameter estimates. When survival to age-1 increased (lower Zo) greater numbers of

older fish were available for capture decreasing the corresponding proportion of age-0 fish in the

catch, thus increasing selectivity estimates for age-0 fish (Figure 11). Under this scenario, all

selectivity schedules peaked at age-1 as before, but selectivity estimates for age-0 fish increased.

Lake Johns was the only exception which exhibited peak selectivity at age-0 declining with age

if survival to age-1 increased (Figure 11). Conversely, when survival to age-1 decreased (higher

Zo) fewer numbers of older age-classes were available for capture, decreasing their proportion in

the catch. The corresponding proportion of age-0 fish in the catch increased resulting in lower

age-0 selectivity estimates. When survival past age-1 increased (lower Z), greater numbers of 2+









age-class fish were available for capture which resulted in increased proportions of older fish in

the catch. These increased proportions of older age-classes represented in the catch resulted in

decreased selectivity estimates for those ages. If survival past age-1 decreased (higher Z) fewer

numbers of 2+ age-class fish became available for capture. Under this scenario, decreased

proportions of older fish resulted in increased selectivity estimates. Overall, the selectivity

estimates for the older age-classes were more sensitive to changes in survival compared with

age-0 and age-1 estimates (Figure 11). For example, Lake Okeechobee results indicated the

selectivity estimates for the older age-classes could vary by a factor of 2 or 3 from the base

model estimates. Nevertheless, changes in assumed survival rates did not change the overall

pattern of dome-shaped selectivity for bottom trawls (Figure 11).

The uncertainty in the age-specific selectivity estimates are described from probability

profiles (similar to p-values) for each lake in Figures 12 15. Most age-specific selectivity

profiles indicated wide probability bands with potential selectivity estimates ranging from 0 to 1

for most lakes. Lake Okeechobee estimates exhibited tighter intervals relative to other lake

selectivity estimates. This is likely attributed to a longer time-series of data (model fit to 10

years), whereas other lakes had shorter data time-series resulting in wider probability bands.

Age-1 selectivity exhibited tighter intervals (on average from 0.70 to 1) than all other age-

groups.










Table 3-1. Summary of Lake characteristics for study locations, including county of location, surface area in hectares (ha),
Chlorophyll-a, concentration measured in (mg/L), trophic status, and years sampled. Tropic state based on Forsberg and
Ryding (1980), Florida Lakewatch Data (1999, 2001), Bachman et al. (1996)

Lake County Surface Area Chlorophyll-a Trophic State Sample years
(ha) (pg/L)
Griffin Lake 6,679 159 Hypereutrophic 1999-2006
Johns Orange 1,676 13 Eutrophic 2002-2006
Lochloosa Alachua 2,631 101 Hypereutrophic 2000-2006
Okeechobee Glades 173,000 30 Eutrophic 1987-2006












Table 3-2. Summary of CPUE-at-age (fish/min) data showing the mean, minimum, and maximum values for each age and lake and
combined across lakes.
Lake Age-0 Age-1 Age-2 Age-3 Age-4 Age-5
Mean 5.32 2.04 0.43 0.09 0.03 0.01
Min 1.15 0.17 0.01 0.01 0.00 0.00
Griffin
Max 16.84 4.41 0.66 0.19 0.08 0.04

Mean 3.78 1.29 0.23 0.06 0.01
Min 3.25 0.26 0.04 0.02 0.00
Johns
Max 4.22 2.14 0.47 0.14 0.04

Mean 3.85 1.63 0.46 0.06 0.01 0.01
Min 1.23 0.43 0.19 0.02 0.00 0.00
Lochloosa
Max 9.31 4.46 0.84 0.12 0.04 0.02

Mean 2.06 2.53 1.48 0.67 0.22 0.13
Min 0.02 0.02 0.06 0.02 0.01 0.00
Okeechobee
Max 7.65 10.83 7.45 3.70 1.31 1.41

Mean 3.24 2.12 0.94 0.37 0.12 0.08
Combined Min 0.02 0.02 0.01 0.00 0.00 0.00
Max 16.84 10.83 7.45 3.70 1.31 1.41











Lake Jeffords, FL


Figure 3-1. Lake Jeffords located in Alachua County, North Central Florida









































Figure 3-2. Geographic location of study lakes used for indirect measure of trawl size selectivity




















1.0 -


0.8


-a.

a 0.6 -
o



g 0.4
(CU
I-


0.2




0.0


90-119


120-149


150-179


180+


Length Group (TL, mm)



Figure 3-3. Bootstrap estimates of mean tagging mortality. Error bars represent 95% confidence
intervals.


S, w
















90-119 TL (mm) 120-149 TL (mm)
1.0

0.8

0.6

0.4

0.2

0.0
S0.000 0.001 0.002 0.003 0.004 0.005 0.000 0.001 0.002 0.003 0.004 0.005

0
0L
nL


-0
0
0
-c

1.2
150-179 TL (mm)

1.0

0.8

0.6

0.4

0.2

0.0 ,
0.000 0.001 0.002 0.003 0.004 0.005


180+ TL (mm)


0.000 0.001 0.002 0.003 0.004 0.005


Catchability (q)


Figure 3-4. Likelihood profiles for q by length-group using mean tagging mortality rates.
Maximum likelihood estimates (MLE's) represented at peak of each curve.












0.005



0.004



0.003
4-

-c
0.002
0


0.001 I



0.000
90-119 120-149 150-179 180+

Length-Group (TL, mm)

Figure 3-5. Maximum likelihood estimates of q by length-group and associated 95% confidence
intervals using mean mortality rates.




















-*- Low Tagging Mortality
o Mean Tagging Mortality
-T- High Tagging Mortality


O .
\


S \


90-119


120-149


150-179


180+


Length Group (mm)



Figure 3-6. Maximum likelihood estimates for q calculated from the lower, mean, and upper
tagging mortality rates.


0.0030



0.0025



0.0020


0.0015 -


0.0010



0.0005



0.0000



















350



300


E
250
-C
C

Oj
0-
O


0 1 2 3 4 5 6 7 8 9 10

Age




Figure 3-7. Predicted mean length at age for each lake. The least-squares equations are shown.


-- La, Gfin = 335*(1-e-0 42+0 79))
SLa, John = 361 (1-e(-0 28(+1 08)))
L, Lochtoose = 398*(-e -0 26*(a+ 95))
L, Okhotee= 347*(-e(-0 29*(+1 171))

















Lake Okeechobee
Lake Griffin
Lake Johns
........ Lake Lochloosa


/*'*
/*


1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Year


Figure 3-8. Observed time series of relative recruitment anomalies by lake.

















1.0 Lake Griffin Lak


0.8


0.6


0.4


0.2


0.0 -




77,

U)
._>


100




Lake Lochloosa Lak(


0.8


0.6


0.4


0.2


0.0 -
0 1 2 3 4 5 0 1 2 3





Age




Figure 3-9. Relative selectivity by age for each study lake from base model


e Johns


4 5






















0.8




S0.6
a)
0)
U)
0)

a)
0.4




0.2 -




0.0
0.0 I I I I
0 1 2 3 4 5

Age



Figure 3-10. Average relative selectivity for lakes sampled with standard trawl (Lakes Griffin,
Johns, and Lochloosa).













Lake Griffin


/ Zo = 1.2, Z = 0.6
. Zo = 1.5, Z = 0.6
Zo = 1.0, Z = 0.6
Zo = 1.2, Z = 0.4
-- Zo=1.2,Z=0.8



L.-


Lake Lochloosa Lake Okeechobee
..- / \\


0.4 / \\ \ /
/ \\\ \ /
0.6 -

/ \\ I \V /
0.4 -


0.2 \ -


0.0 ,
0 1 2 3 4 5 0 1 2 3 4 5

Age







Figure 3-11. Relative selectivity by age under varying assumptions of survival.










Age-0


0.0 0.2 0.4 0.6 0.8 1.0 0.970 0.975 0.980 0.985 0.990 0.995 1.000

Age-2 Age-3


0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0


Age-4 Aae-5


0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Relative Selectivity



Figure 3-12. Lake Griffin probability profiles with relative selectivity on the x-axis and relative
probability on the y-axis.


Age-1


-~


"\


u,
-r
-e
c~















Age-O


0.0 0.2 0.4 0.6 0.8 1.0 0.80


Age-2


0.85 0.90 0.95


1.00


Age-3


0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0


Age-4


0.2 0.4 0.6 0.8 1.0

Relative Selectivity


Figure 3-13. Lake Johns probability profiles with relative selectivity on the x-axis and relative
probability on the y-axis.


1.0


0.8


0.6


0.4


0.2


0.0
0.0


1.0


0.8


0.6


0.4


0.2


0.0
0.


0


Age-1


-2
J
f




































0.0 0.2 0.4 0.6 0.8 1.0 0.75 0.80 0.85 0.90 0.95 1.00


Age-2 Age-3


0.0 0.2 0.4 0.6 0.8 1.0 0.0


0.2 0.4 0.6 0.8 1.0


Age-5



* -.


0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Relative Selectivity


Figure 3-14. Lake Lochloosa probability profiles with relative selectivity on the x-axis and
relative probability on the y-axis.


1.C


0.8

o.e


0.4


0.2


0.0





1.1


.0.

0..
0.




S0.

n0 o.

0.I


0


8


6


4


2


0


Age-4


Age-1


Age-0


h
r' ~L\


4
r,
-c~





































0.0 0.2 0.4 0.6 0.8 1.0 0.70 0.75 0.80 0.85 0.90 0.95 1.00

Age-2 Age-3
0.6











0.4


0.2


0.0 .
0.0 0.2 0.4 0.6 0.8 1.0 0.70 0.75 0.80 0.85 0.90 0.95 1.00

Age-2 Age-3

1.0


S0.8


a

0.4





0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0


Age-4


Age-5


0.2 0.4 0.6 0.8 1.0 0.0 0.2


0.4 0.6 0.8 1.0


Relative Selectivity


Figure 3-15. Lake Okeechobee probability profiles with relative selectivity on the x-axis and
relative probability on the y-axis.


1.0


0.8


0.6


0.4


0.2


0.0
0.0


Age-0


Age-1









CHAPTER 4
DISCUSSION

Bottom trawls are an extremely efficient capture gear (reflected by large catches of various

aquatic organisms and use in many fisheries worldwide), but often select for small finfish and

crustaceans (Kjelson and Johnson 1978; Rulifson et al. 1992; Diamond et al. 1999). The size

selectivity exhibited by bottom trawls results in large catches of small aquatic organisms (as

evident by their use in commercial shrimp fisheries) and the incidental catch of commercially

and recreationally important juvenile fishes (Howell and Langan 1992; Gallaway and Cole 1999;

Diamond et al. 1999; Wakefield et al. 2007). My direct and indirect estimates of bottom trawl

selectivity corroborated one another and indicated decreasing size selectivity with increasing

length, suggesting that bottom trawls would be best for monitoring the abundance of small black

crappie but may be inadequate to characterize the adult population. Catchability of black crappie

has seldom been measured, but Miranda and Dorr (2000) found q with angling gear to vary with

fish size. Furthermore, McInerny and Cross (2006) found q with trap-nets varied with size,

season, and density and used these estimates to improve the interpretation of CPUE data to index

abundance and describe size structures. My estimates of q lend insight into the size selective

properties of bottom trawls for black crappie, and could be incorporated with catch data to index

the true population age/size composition (e.g., Lake Jeffords).

My age-structured model simulations indicated dome-shaped selectivity to bottom trawls

for black crappie and the highest selectivity estimates were for age-1 fish. Dome-shaped

selectivity patterns are common for many sample gears and species (Erzini and Castro 1998;

Jackson and Noble 1995; Miranda and Dorr 2000; Tanaka 2002), and evaluating differences in

relative selectivity by age/size allows managers to adjust indices of abundance (Quinn and

Deriso 1999) or determine gear efficiencies on the different age/size classes (Pierce et al. 1994).









Selectivity peaked at age-1 for all lakes under all scenarios except Lake Johns which exhibited

decreasing relative selectivity with age when assumed age-0 survival was high. The results

corroborate other selectivity studies (Jagielo 1999; Bayley and Austen 2002; McInerny and

Cross 2006) and indicate the trawl gear may be most useful for tracking small-sized black

crappie through time. Selectivity estimates exhibited high uncertainty except for age-1, but the

observed selectivity patterns were similar across lakes. Individual selectivity estimates from my

analyses should be viewed with caution, but the overall pattern of dome-shaped selectivity

probably describes the general pattern of bottom trawl selectivity for black crappie.

Gear selectivity determines the effect of fishing on size/age structures. As such,

assessment models can link size/age composition of catch data to size/age composition of the

fishery (Taylor et al. 2005) to predict the effects of different harvest rates, calculate biological

reference points, and identify maximum sustainable yields (Maunder 2002). Assuming constant

catchability among size/age classes is a dangerous assumption and can lead to bias in yield

models (Ricker 1975) because differences in age/size specific catchability to fishing gears can

alter the maximum sustainable yield (MSY) obtainable from a fishery (Maunder 2002). When

fishing mortality is restricted through effort controls, the catchability coefficient becomes a vital

parameter in yield models, due to the relationship between q and yield and abundance.

Gear selectivity patterns and the cumulative effects of size-selective fishing practices often

produce bias in length at age samples (Sinclair et al. 2002, Taylor et al. 2005). In exploited

populations, fast growing young fish and slow growing older fish are overrepresented in length

at age samples. In turn, this biases the parameter estimates of growth models used to describe

mean length at age. My model results indicated bottom trawl selectivity for younger size/age-

classes. Furthermore, all lakes in the study experience some level of angler exploitation. Thus,









since trawls appear more effective at capturing small fish and the older age-classes are exposed

to harvest, growth models from trawl data collected from exploited populations may severely

underestimate the asymptotic length (L,) and overestimate the metabolic growth coefficient (K).

These inaccuracies can lead to biased mean length at age which would influence estimates of

maximum yield and optimal harvest policies.

Age/size selectivity of sampling gears causes errors in the estimation of population

structure (Walters and Hilbom 1992) and may limit the ability to draw inferences about trends in

abundance. Therefore determining size selectivity of fishery independent assessment gears is

important when survey CPUE data are used to index abundance, especially if assessment gears

may not adequately sample the age and size range targeted by the fishery. For example, I found

that large adult black crappies were poorly represented in bottom trawl samples, but these fish

are targeted by recreational fisheries. Thus, bottom trawls may not detect changes in abundance

even if large fish suffer high fishing mortality. This could lead to appearances of hyperstability

(i.e., relatively constant CPUE over a large range of true fish abundance) because trawl catches

of large black crappie could potentially be low and not change widely with changes in

abundance.

When properly designed and implemented, mark-recapture studies can provide managers

with information on growth, mortality, and reproduction, population size and structure, and gear

selectivity (Seber 1982). Data for closed capture-recapture methods must meet several criteria.

Foremost, the population of interest must be closed to additions and deletions. That is,

recruitment, natural mortality, immigration, and emigration must be minimal. The immigration

and emigration assumptions were probably not violated in my study due to the closed system and

short time interval over which the study occurred. The mortality assumption was violated due to









tagging mortality but I adjusted for this in my calculations. Another assumption for mark

recapture experiments is that marks are not lost or undetected by the recorder. Tag retention and

detectability were likely very high in my study because I marked fish with a pelvic fin clip and

sampled over a short interval before regeneration of fins could occur. A third requirement for

closed mark recapture experiments is equal capture probability (i.e. no capture heterogeneity

and/or trap response). Trap response is when capture probability is dependent on an animal's

capture history, and is difficult to directly estimate (Pollock et al 1990). I estimated

heterogeneity in capture probability as a function of fish size and account for potential trap

responses by using different gear types for mark and recap events.

One potential cause of bias and uncertainty in q estimates from the mark-recapture

experiment could be the higher observed sampling mortality for the smaller length-groups (90-

119, 120-149) relative to the larger groups (150-179, 180+). This size selective mortality

significantly reduced the number of fish available for recapture among those two groups. An

overestimation in tagging mortality for these groups would cause a positive bias in my q

estimates, resulting in more of an effect on trawl size selectivity. Alternately, underestimating

sampling mortality could have caused a negative bias in the catchability estimates. The larger

observed mortality on the smaller size groups could have resulted from high handling times and

greater sampling stress on these classes relative to the larger length groups. The likelihood

profiles also show there is more uncertainty in the likelihood estimate for the 90-119 size-group

when compared to other groups. The larger variance can likely be attributed to the higher

observed sampling mortality for this group. In turn, this influenced the number available for

recapture (smaller N for this group) leading to a greater difference between observed and

expected recaptures.









Differences in growth among the lakes could explain some of the differences in selectivity

at age between lakes. Some lakes appeared to exhibit faster growth, which could have resulted

in fewer numbers of faster growing individuals in the catch (as described above). This may have

resulted in biased mean length at age for some lakes which may explain some of the observed

differences in age-specific selectivities across systems. However, these differences appeared

minimal because the overall pattern of dome-shaped selectivity occurred at all lakes.

My approximation of age-length keys using previous and post year data at Lake Lochloosa

could pose problems in the estimated age structure and thus, the catch rate indices for those

years. Inaccuracies in observed catch-at-age could have affected my proportions of catch-at-age

and model fit having the potential to cause unknown biases in the selectivity estimates.

However, selectivity patterns for Lake Lochloosa were similar to those observed at the other

lakes where age data were collected every year. Thus, I believe the selectivity patterns at Lake

Lochloosa reflected real differences in across age-classes despite the gaps in the age structure

data.

Furthermore, variation in mortality at size/age can affect the numbers in each age class

available for capture leading to biased selectivity estimates. I accounted for this variability in

survival to age by apportioning mortality into two classes (survival to age-1 and survival past

age-1) based on life stanzas. I also simulated various levels of survival to show how different

rates of survival on the age-classes could influence the numbers available for capture and thus,

my selectivity estimates. I did not attempt to quantify exploitation rates on fish vulnerable to

angling (>= age-2) which may have decreased survival for some of the older age-classes relative

to younger age-classes. If exploitation is high on older age-classes, the corresponding

proportions of catch in those age-classes would decrease resulting in an increase in the younger









age-class proportions and an upward bias in the selectivity estimates for older/larger fish and a

downward bias in selectivity estimates of younger/smaller fish. I assumed constant survival

among years and did not evaluate the effects of changes in mortality between years. Changes in

yearly natural and/or fishing mortality could significantly increase or reduce the numbers

available for capture creating both positive and negative biases in selectivity estimates.

The indirect approach I used to estimate trawl selectivity is common data collected for

recreational and/or commercially important fisheries. This methodology is quite common in

marine assessments (Walters and Martell 2004), but examples are lacking for freshwater

fisheries applications. I found this approach quite useful for estimating gear selectivity and

recommend its application when catch-at-age time series data are available and when direct

estimates are infeasible (e.g., large, open systems).

Use of my selectivity estimates should be restricted to similar seasons and lake types as

used in this study. Size-selectivity of bottom trawls was evaluated during fall for the indirect

method and winter for the direct method. Thus, the estimates provided may not apply to other

seasons or times of the year. Seasonal effects on fish behavior are well documented (Pope and

Willis 1996, Hayes et al. 1996) and sampling black crappie with bottom trawls at other times in

the year may result in unknown bias in the CPUE index. Use of my selectivity estimates should

be restricted to similar seasons from which they were derived. Lake characteristics and

differences in geomorphology may have influenced my results. For example, gear efficiency and

thus, catchability may vary due to differences in amount of suitable habitat, percent area

coverage of macrophytes (PAC), depth, dissolved oxygen (DO) levels, substrate type or a variety

of other factors. I did not attempt to quantify all of these effects and managers should be aware

of these potential sources of bias and use the selectivity estimates with prudence. Furthermore,









systems selected for trawl sampling should be determined carefully to maximize efficiency and

prevent gear fouling. Bottom trawl sampling may be an inappropriate gear on some lakes due to

excessive bottom debris. Submerged debris, such as stumps or large amounts of aquatic

macrophytyes may prevent effective trawl sampling limiting sampling and/or gear efficiency.

The likely cause for size selectivity of smaller black crappie to otter trawls may be the

ability of larger fish to detect or avoid the gear. Net avoidance and escapement after initial

capture could result since swimming speed tends to increase with fish length (Helfman et al.

1997). Furthermore, gear avoidance by the larger fish may result due to the trawl pressure wake.

This pressure wake may be detected due to larger and more developed lateral lines in larger fish.

Spatial distributional patterns and habitat availability and use may also be a source for trawl

selectivity. The smaller fish may utilize the pelagic zones of a lake and school more relative to

large fish, where the larger fish may be more patchily distributed utilizing both open water and

littoral areas of lakes. However, we controlled for differences in spatial distribution and habitat

availability/use in the direct measure at Lake Jeffords, where crappie of all size ranges and the

majority of fish were captured in open water habitats. Differences in spatial distribution patterns

may have had more of an influence on the indirect selectivity estimates where there is likely a

greater difference in habitat availability and use due to lake size and habitat complexity.

However, the direct measure of selectivity suggests that the ability of larger fish to detect/avoid

the gear is the likely source for trawl selectivity of younger/smaller crappie.

Stock assessments are important for fisheries management and attempt to recreate past

stock trends to explain current stock trends and abundance by making quantitative predictions

about the reactions of fish populations to alternative management options (Hilbom and Walters

1992). Therefore, bottom trawl catch data may be important for stock assessment analysis and









used for indices of recruitment, as measures of recruitment are one of the important inputs used

for stock assessments. These data could be used in conjunction with another gear type that

indexes the adult population (such as a creel) along with an estimate of fishing mortality to

evaluate current stock trends facilitating the ability to make choices among policy options and

evaluate the trade-offs associated with those decisions.

Otter trawls have been shown to effectively capture young black crappies (Allen et al.

1999, Pine 2000), and my results suggest that bottom trawls provided adequate catches of age-0

and age-1 fish. As such, otter trawls are probably most effective at tracking age-0 and age-1

abundances and can be used for estimates of year-class strength. However, my results suggested

that the trawls may be inadequate to describe the adult population, fish growth rates, and age

structure estimates due to substantially lower selectivity values for large fish. The selectivity

estimates provided will allow managers to adjust abundance indices and correct age/size

structures for relatively shallow Florida lakes.









LIST OF REFERENCES


Allen, M. S., M. M. Hale, and W. E. Pine III. 1999. Comparison of trap nets and otter trawls for
sampling black crappie in two Florida Lakes. North American Journal of Fisheries
Management 19:977-983.

Arreguin-Sanchez, F. 1996. Catchability: a key parameter for fish stock assessment. Reviews
in Fish Biology and Fisheries 6: 221-242.

Bachman, R. W., B. L. Jones, D. D. Fox, M. Hoyer, L. A. Bull and D. E. Canfield. 1996.
Relations between trophic state indicators and fish in Florida lakes. Canadian Journal of
Fisheries and Aquatic Sciences 53:842-855.

Bayley, P. B. and D. J. Austen. 2002. Capture efficiency of a boat electrofisher. Transactions
of the American Fisheries Society 131:435-451.

Beamesderfer, R. C., and B. E. Rieman. 1988. Size selectivity and bias in estimates of
population statistics of smallmouth bass, walleye, and northern squawfish in a Columbia
River Reservoir. North American Journal of Fisheries Management 8:505-510.

Boxrucker, J., and G. Ploskey. 1989. Gear and seasonal biases associated with sampling crappie
in Oklahoma. Proceedings Annual Conference Southeast Association Fish and Wildlife
Agencies 42:89-97.

Colvin, M. A., and F. W. Vasey. 1986. A method of qualitatively assessing white crappie
populations in Missouri reservoirs. Pages 79-85 in G. E. Hall and M. J. Van Den Avyle,
editors. Reservoir fisheries management: strategies for the 80's. American Fisheries
Society, Southern Division Reservoir Committee, Bethesda, Maryland.

Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge University Press, Cambridge,
England.

Devries, D. R., and R. V. Frie. 1996. Determination of age and growth. Pages 483-512 in B.
R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition. American
Fisheries Society, Bethesda, Maryland.

Diamond, S. L., L. B. Crowder, and L. G. Cowell. 1999. Catch and bycatch: The qualitative
effects of fisheries on population vital rates of Atlantic croaker. Transactions of the
North American Fisheries Society 128:1085-1105.

Doerzbacher, J. F. and H. L. Schramm, Jr. 1982. Enlarger-produced photographs for the
measurement of black crappie otoliths. North American Journal of Fisheries
Management 4:547-551.

Dumont, S. C., and J. A. Dennis. 1997. Comparison of day and night electrofishing in Texas
Reservoirs. North American Journal of Fisheries Management 17:939-946.









Erzini, K. and M. Castro. 1998. An alternative methodology for fitting selectivity curves to pre-
defined distributions. Fisheries Research 34:307-313.

Farmer, M. J., D. T. Brewer, S. J. M. Blaber. 1998. Damage to selected fish species escaping
from prawn trawl codends: a comparison between square-mesh and diamond- mesh.
Fisheries Research 38:73-81.

Florida LAKEWATCH. 1999. Florida LAKEWATCH 1999 data report. University of Florida,
Department of Fisheries and Aquatic Sciences, Gainesville, FL.

Florida LAKEWATCH. 2001. Florida LAKEWATCH 2001 data report. University of Florida,
Department of Fisheries and Aquatic Sciences, Gainesville, FL.

Forsberg, C., and S. O. Ryding. 1980. Eutrophication parameters and trophic state indices in 30
Swedish water-receiving lakes. Arch. Hydrobiologia 89:189- 207.

Gablehouse, D. W., Jr. 1984. An assessment of crappie stocks in small midwestern private
impoundments. North American Journal of Fisheries Management 4: 371-384.

Gallaway, B. J., and J. G. Cole. 1999. Reduction of juvenile red snapper bycatch in the U.S.
Gulf of Mexico shrimp trawl fishery. North American Journal of Fisheries Management
19:342-355.

Haddon, M. 2001. Modeling and quantitative methods in fisheries. Chapman and Hall, New
York.

Hamley, J. M., and H. A. Regier. 1973. Direct estimates ofgillnet selectivity to walleye.
Journal of Fisheries Research Board of Canada 30:817-830.

Hansen, M. J., T. D. Beard, and S. W. Hewett. 2000. Catch rates and catchability of walleyes in
angling and spearing fisheries in Northern Wisconsin Lakes. North American Journal of
Fisheries Management 20:109-118.

Hayes, D. B., C. P. Ferreri, W. W. Taylor. 1996. Active fish capture methods. Pages 193-220
in B. R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition. American
Fisheries Society, Bethesda, Maryland.

Helfman, G. S., B. B. Collete, and D. E. Facey. 1997. The diversity of fishes. Blackwell
Science, Malden, Massachusetts.

Hilborn, R. and M. Mangel. 1997. The ecological detective. Princeton Press, Princeton, New
Jersey.

Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics,
and uncertainty. Chapman and Hall, New York.









Hjort, J. 1914. Fluctuations in the great fisheries of Northern Europe viewed in the light of
biological research. Rapports et Proces-Verbaux des Reunions du Conseil International
pour l'Exploration de la Mer 20.

Howell, W. H., and R. Langan. 1992. Discarding the commercial groundfish species in the Gulf
of Maine shrimp fishery. North American Journal of Fisheries Management 12:568-580.

Hubert, W. A. 1996. Passive capture techniques. Pages 157-181 in B. R. Murphy and D. W.
Willis, editors. Fisheries techniques, 2nd edition. American Fisheries Society, Bethesda,
Maryland.

Jackson, J. R., and R. L. Noble. 1995. Selectivity of sampling methods for juvenile largemouth
bass in assessments of recruitment processes. North American Journal of Fisheries
Management 15:408-418.

Jacobson, L. D., J. Brodziak, and J. Rogers. 2001. Depth distributions and time-varying bottom
trawl selectivities for Dover sole, sablefish, and thornyheads in a commercial fishery.
Fishery Bulletin 99:309-327.

Jagielo, T. H. 1999. Movement, mortality, and size selectivity of sport and trawl caught lingcod
off Washington. Transactions of the American Fisheries Society 128:31-48.

Kjelson, M. A., and G. N. Johnson. 1978. Catch efficiencies of a 6.1-meter otter trawl for
estuarine fish populations. Transactions of American Fisheries Society 107:246-254.

Kraft, C. E., and B. L. Johnson. 1992. Fyke-net and gill-net size selectivities for yellow perch
in Green Bay, Lake Michigan. North American Journal of Fisheries Management
12:230-236.

Laarman, P. W., and J. R. Ryckman. 1982. Relative size selectivity of trap nets for eight species
of fish. North American Journal of Fisheries Management 2:33-37.

Lok, A., A. Tokac, Z. Tosunoglu, C. Metin, and R. S. T. Ferro. 1997. The effects of different
cod-end design on bottom trawl selectivity in Turkish fisheries of the Aegean Sea.
Fisheries Research 32:149-156.

Maceina, M. J., O. Ozen, M. S. Allen, and S. M. Smith. 1998. Use of equilibrium yield models
to assess different size limits for crappie in Weiss Lake, Alabama. North American
Journal of Fisheries Management 18:854-863.

Maunder, M. M. 2002. The relationship between fishing methods, fisheries management, and
the estimation of maximum sustainable yield. Fish and Fisheries 3:251-260.

McInerny, M. C. 1989. Evaluation of trapnetting for sampling black crappie. Proceedings
Annual Conference Southeast Association Fish and Wildlife Agencies 42:98-106.









McInemy, M. C., and T. K. Cross. 2000. Effects of sampling time, intraspecific density, and
environmental variables on electrofishing catch per effort of largemouth bass in
Minnesota Lakes. North American Journal of Fisheries Management 20:328- 336.

McInemy, M. C. and T. K. Cross. 2006. Factors affecting trap-net catchability of black crappies
in natural Minnesota lakes. North American Journal of Fisheries Management 26:652-
664.

Millar, R. B. and R. Holst. 1997. Estimation of gillnet selectivity using log-linear models.
ICES Journal of Marine Science 54:471-477.

Miller, S. J., D. D. Fox, L. A. Bull, and T. D. McCall. 1990. Population dynamics of black
crappie in Lake Okeechobee, Florida, following suspension of commercial harvest.
North American Journal of Fisheries Management 10:98-105.

Miranda, L. E., and B. S. Dorr. 2000. Size selectivity of crappie angling. North American
Journal of Fisheries Management 20:706-710.

Miranda, L. E., J. C. Holder, and M. S. Schorr. 1992. Comparison of methods for estimating
relative abundance of white crappie. Proceedings Annual Conference Southeast
Association Fish and Wildlife Agencies 44:89-97.

Myers, R. A., and J. M. Hoenig. 1997. Direct estimates of gear selectivity from multiple
tagging experiments. Canadian Journal of Fisheries and Aquatic Science 54:1-9.

Nielsen, L. A. 1983. Variation in the catchability of yellow perch in an otter trawl.
Transactions of the American Fisheries Society 112:53-59.

O'Neil, F. G., and R. J. Kynoch. 1996. The effect of cover mesh size and cod-end catch size on
cod-end selectivity. Fisheries Research 28:291-303.

Otway, N. M., J. R. Craig, and J. M. Upston. 1996. Gear dependent size-selection of snapper,
Pargus auratus. Fisheries Research 28:119-132.

Paragamian, V. L. 1989. A comparison of day and night electrofishing: size structure and catch
per unit effort for smallmouth bass. North American Journal of Fisheries Management
9:500-503.

Patiwan, Yudi. 2001. In all likelihood: statistical modeling and inference using likelihood.
Clarendon Press, Oxford.

Peterman, R. M. and G. J. Steer. 1981. Relation between sport-fishing catchability coefficients
and salmon abundance. Transactions of the American Fisheries Society 110:585-593.









Pierce, R. B., C. M. Tomcko, and T. D. Kolander. 1994. Indirect and direct estimates of gill-net
size selectivity for northern pike. North American Journal of Fisheries Management
14:170-177.

Pierce, R. B. and C. M. Tomcko. 2003. Variation in gill-net and angling catchability with
changing density of northern pike in a small Minnesota lake. Transactions of the
American Fisheries Society 132:771-779.

Pine III, W. E. 2000. Comparison of two otter trawls of different sizes for sampling black
crappies. North American Journal of Fisheries Management 20:819-821.

Pollock, K. H., J. D. Nichols, J. E. Hines, and C. Brownie. 1990. Statistical inference for
capture-recapture experiments. Wildlife Monographs 107:1-97.

Pope, K. L., and D. W. Willis. 1996. Seasonal influences on freshwater fisheries sampling data.
Reviews in Fisheries Science 4:57-73.

Quinn, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics. Oxford University Press,
New York.

Reynolds, J. B. 1996. Electrofishing. Pages 221-253 in B. R. Murphy and D. W. Willis,
editors. Fisheries techniques, 2nd edition. American Fisheries Society, Bethesda,
Maryland.

Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish populations.
Fisheries Research Board of Canada Bulletin 191.

Rogers, M. W., M. J. Hansen, and T. D. Beard. 2003. Catchability of walleyes to fyke netting
and electrofishing in northern Wisconsin lakes. North American Journal of Fisheries
Management 23:1193-1206.

Rulifson, R. A., J. D. Murray, and J. J. Bahen. 1992. Finfish catch reduction in South Atlantic
shrimp trawls using three designs ofbycatch reduction devices. Fisheries 17:9-20.

Sammons, S. M. and P. W. Bettoli. 1998. Larval sampling as a fisheries management tool: early
detection of year-class strength. North American Journal of Fisheries Management
18:137-143.

Sammons, S. M., D. A. Isermann, and P. W. Bettoli. 2002. Variation in population
characteristics and gear selection between black and white crappies in Tennessee
Reservoirs: Potential effects on management decisions. North American Journal of
Fisheries Management 22:863-869.

Schoenebeck, C. W. and M. J. Hansen. 2005. Electrofishing catchability of walleyes,
largemouth bass, smallmouth bass, northern pike, and muskellunge in Wisconsin lakes.
North American Journal of Fisheries Management 25:1341-1352.










Schramm, H. L., J. V. Shireman, D. E. Hammond, and D. M. Powell. 1985. Effects of
commercial harvest of sport fish on the black crappie population in Lake Okeechobee,
Florida. North American Journal of Fisheries Management 5:217- 226.

Seber, G. A. F. 1982. The estimation of animal abundance and related parameters. 2nd ed.
London: Griffin.

Sinclair, A. F., D. P. Swain, and J. M. Hanson. 2002. Disentangling the effects of size selective
mortality, density, and temperature on length at age. Canadian Journal of Fisheries and
Aquatic Sciences 59:372-382.

St. John, T. R., and W. B. Black. 2004. Methods for predicting age-0 crappie year-class strength
in J. Percy Reservoir, Tennessee. North American Journal of Fisheries Management
24:1300-1308.

Tanaka, E. 2002. A method for calculating numerical estimates of gear selectivity curve.
Fisheries Science 68:1081-1087.

Taylor, N. G., C. J. Walters, and S. J. D. Martell. 2005. A new likelihood for simultaneously
estimating von Bertalanffy growth parameters, gear selectivity, and natural and fishing
mortality. Canadian Journal of Fisheries and Aquatic Sciences 62:215-223.

Wakefield, C. B., M. J. Moran, N. E. Tapp, and G. Jackson. 2007. Catchability and selectivity
of juvenile snapper and western butterfish from prawn trawling in a large marine
embayment in Western Australia. Fisheries Research 85:37-48.

Walters, C. J. and S. J. D. Martell. 2004. Fisheries ecology and management. Princeton Press,
Princeton, New Jersey.









BIOGRAPHICAL SKETCH

Gregory Robert Binion was born on September 18, 1978, at a U.S. air force base in the

United Kingdom, to Mike and Peggy Binion. Shortly after, he moved and was raised in San

Antonio, Texas, with his older brother Pete. At a young age, he developed a passion for the

outdoors and enjoyed much of his time exploring the wide open spaces and beautiful country of

South Texas. He graduated from the University of Kentucky with a B.A. in political science in

December 2002. Shortly after graduation, he relocated to Florida to pursue an interest in

fisheries biology and management. In October 2003, he began to work as a fisheries technician

on various projects at the University of Florida and began his graduate work at the Department of

Fisheries and Aquatic Sciences at the University of Florida in January 2006.He will graduate

with a Master of Science in December 2007. His future plans include traveling, passing time

fishing and hunting, spending time with his family, and pursuing a career in fisheries

management.





PAGE 1

DIRECT AND INDIRECT ESTIMATES OF BLACK CRAPPIE SIZE SLECTIVITY TO OTTER TRAWLS By GREGORY R. BINION A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007 1

PAGE 2

2007 GREGORY R. BINION 2

PAGE 3

To my grandfather Ace who instilled in me at a young age a passion for fish. 3

PAGE 4

ACKNOWLEDGMENTS I would like to thank and acknowledge my supervisory committee members Dr. Bill Pine, Jim Estes, and Marty Hale for their encourag ement and support, and especially my committee chair, Dr. Mike Allen, for his mentorship and direction throughout my research. I would also like to thank various members of the Allen lab for their help in field collection and processing; Galen Kaufman, Jason Dotson, Kevin Johnson, Christian Barrientos, Erika Thompson, Aaron Bunch, Melissa Woods-Jackson and Drew Dutterer. I would like to acknowledge various members of the Florida Fi sh and Wildlife Conservation Commission (Eric Nagid, Travis Tuten, Will Strong, Bill Johnson, and Ja nice Kerns) for collaboration in research. A special thanks goes to office mates Matt Cata lano and Mark Rogers whom encouraged and supported my development as a student, mentori ng and answering questions when I encountered problems. Finally, I want to thank my loving wife, parents and grandparents for their endless encouragement, love, and support. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES.........................................................................................................................7 ABSTRACT.....................................................................................................................................8 CHAPTER 1 INTRODUCTION................................................................................................................. .10 2 METHODS...................................................................................................................... .......16 Direct Measure of Selectivity.................................................................................................1 6 Indirect Measure of Selectivity...............................................................................................1 9 3 RESULTS...................................................................................................................... .........25 Direct Measure of Selectivity.................................................................................................2 5 Indirect Measure of Selectivity...............................................................................................2 6 4 DISCUSSION................................................................................................................... ......46 LIST OF REFERENCES...............................................................................................................54 BIOGRAPHICAL SKETCH.........................................................................................................60 5

PAGE 6

LIST OF TABLES Table page 3-1 Summary of Lake charact eristics for study locations........................................................29 3-2 Summary of CPUE-at-age (fish/min) data showing the mean, minimum, and maximum values for each age and lake and combined across lakes.................................30 6

PAGE 7

LIST OF FIGURES Figure page 3-1 Lake Jeffords located in Alachua County, North Central Florida.....................................31 3-2 Geographic location of study lakes used for indirect measure of tr awl size selectivity....32 3-3 Bootstrap estimates of m ean tagging mortality. Error bars represent 95% confidence intervals...................................................................................................................... ........33 3-4 Likelihood profiles for q by length-group using mean tagging mortality rates. Maximum likelihood estimates (MLEs) represented at peak of each curve....................34 3-5 Maximum likelihood estimates of q by length-group and associated 95% confidence intervals using mean mortality rates..................................................................................35 3-6 Maximum likelihood estimates for q calculated from the lower, mean, and upper tagging mortality rates.......................................................................................................36 3-7 Predicted mean length at age for each lake. The least-squares equations are shown........37 3-8 Observed time series of rela tive recruitment anomalies by lake.......................................38 3-9 Relative selectivity by age for each study lake from base model......................................39 3-10 Average relative selectivity for lakes sampled with standard trawl (Lakes Griffin, Johns, and Lochloosa)........................................................................................................40 3-11 Relative selectivity by age under va rying assumptions of survival...................................41 3-12 Lake Griffin probability profiles with re lative selectivity on the x-axis and relative probability on the y-axis....................................................................................................42 3-13 Lake Johns probability profiles with rela tive selectivity on the x-axis and relative probability on the y-axis....................................................................................................43 3-14 Lake Lochloosa probability profiles with relative selectivity on the x-axis and relative probability on the y-axis.......................................................................................44 3-15 Lake Okeechobee probability profiles w ith relative selectivity on the x-axis and relative probability on the y-axis.......................................................................................45 7

PAGE 8

Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DIRECT AND INDIRECT ESTIMATES OF BLACK CRAPPIE SIZE SLECECTIVITY TO OTTER TRAWLS By GREGORY R. BINION December 2007 Chair: Mike Allen Major: Fisheries a nd Aquatic Sciences I estimated size selectivity of bottom tr awl sampling for black crappie Pomoxis nigromaculatus utilizing direct a nd indirect approaches. I used capture-recapture methods to directly measure the effects of fish size on catchability (q, the fr action of a fish stock collected with a given unit of fishing effo rt) at Lake Jeffords, Florida. Estimates of q were made for different length-groups roughly resembling ageclasses 0 to 2 and fish 3+ (90-119, 120-149, 150179, 180+ mm) by marking a subpopulation collected using three gear types (bottom trawls, hoopnets, and electrofishing). Recapture sampling with otter trawls occu rred two weeks after marking events ended, allowing a direct estimate of q from recaptures of tagged fish from each gear and size class. Indirect estimates of selectivity were obtained with a population model applied to long-term data at four Florida lakes. I constructed age-structured models for each lake that predicted annual catches-at-a ge as a function of measured gr owth rates, a time series of recruitment anomalies, assumed survival rates, and unknown age/size selec tivities. Selectivity parameters were estimated by fitting model predicte d catches-at-age to a time series of bottom trawl catch-at-age using maximum likelihood. Direct measures of selectivity indicated catchability was highest for the 90119 length-group and lowest for fish greater than or equal to 180 mm, with q declining by a factor of 2 or 3 for large fish relative to small fish. Model 8

PAGE 9

9 simulations from the age-structured indirect appr oach revealed dome-shaped selectivity patterns with relative selectivities peaking at age-1 for three of four lakes. Lake Johns was the only exception where age-0 fish was the most efficien tly captured age-group when survival was low. Overall model trends indicated gr eater selectivity of younger fish (age-0 and age-1) to the gear followed by decreasing relative selectivity to ol der age-classes (age-2+). Trawl selectivity patterns suggested that otter traw ls would be best for monitori ng the abundance of small black crappie. My results indicate that adult black crappie will likely be unde rrepresented in bottom trawl samples, which would influence age st ructure and growth rate estimates and the effectiveness of this gear as an assessment t ool for tracking adult cra ppie populations.

PAGE 10

CHAPTER 1 INTRODUCTION Effective management of fishery resources is dependent on the quality of information available for decisions. To deve lop optimal management strategies, biologists must be confident that sampling reliably tracks popul ation metrics. These strategi es are often reliant on precise estimates of some metric of population si ze and its correspondi ng level of production (biomass/numbers). Gear selectivity can influe nce precision and accuracy of these measures (Hilborn and Walters 1992). Samp les collected from many fish populations with a variety of gears often dont accurate ly describe the true age and size structure of the target population. Therefore, obtaining an abundance index that reflects the actual age/size composition of a population allows managers to monitor population trends such as recruitment, growth, and mortality and evaluate populati on responses to management po licies (e.g., size limits) (Hilborn and Walters 1992). When evaluating gear types, an important di stinction between gear selectivity and gear efficiency must be delineated. Gear selectivity is defined as the composition of a sample relative to the true population metric (e.g. size, age, growth rate), and re lative selectivity refers to the effectiveness of a sampling gear to capture a part icular size or species of fish relative to its efficiency at capture of other sizes or species (Hubert 1996). In cont rast, gear efficiency describes the magnitude of effort requir ed to catch adequate sample sizes. Gear selectivity patterns are commonly attrib uted to intrinsic f actors like fish size (Beamesderfer and Rieman 1988, Myers and Hoenig 1997, Wakefield et al. 2007), fish density (McInerny and Cross 2000, Rogers et al. 2003), species (Laarman and Ryckman 1982, Sammons et al. 2002), sex (Jagielo 1999) behavioral patterns (Reynol ds 1996, Jagielo 1999), and habitat preferences (Jacobson et al. 2001) as well as extrinsic factors su ch as seasonal variation (Pope 10

PAGE 11

and Willis 1996), environmental conditions or ch aracteristics (Hayes et al. 1996; McInerny and Cross 2000), diel variation (Paragamian 1989, Dumont and Dennis 1997), gear construction (ONeil and Kynoch 1996, Lok et al. 1997, Farmer et al. 1998), gear type (Kraft and Johnson 1992, Jackson and Noble 1995, Otway et al. 1996), and sampling crew expertise. Estimates of gear selectivity are important for fish stock assessment. Estimates of selectivity allow managers to assess populati on composition based on samples which may not represent the true population, and estimates of selectivity provid e information on aspects of the population which is not readily observable. Adjusting for selectivity allows managers to obtain a more accurate abundance index for the age and size structure of a stock because many samples do not adequately represent the true population age or size structure. This enhances the ability of managers to draw inferences about stock trends like recruitment, growt h, and mortality (Hilborn and Walters 1992). Gear selectivities are commonly used to dete rmine the effects of fishing on the size and age composition of a fishery and are commonly us ed in assessment models to link size/age structure of catch data to the size/age structure of the fish pop ulation (Walters and Martell 2004; Taylor et al. 2005). Such models are required to predict effect s of different harvest rates, calculate biological reference points like spaw ning potential ratio ( SPR), and determining appropriate levels of sustainable yield for a fishery (Maunder 2002). Thus, quantifying gear selectivity allows biologists to adjust abundance indices to represent the true size/age composition which guides future management actions. Measurements of the selective properties of fishing gears are often made utilizing direct and indirect methods (Pollock et al. 1990; Walters and Martel l 2004). Direct methods involve comparing catch composition against a known populati on structure. The most direct method for 11

PAGE 12

estimating selectivity is a mark recapture experiment creating a known population, then calculating the proportion of fish caught by the g ear in a given length category from the marked subpopulation (Hamley and Regier 1973; Myers and Hoenig 1997; McInerny and Cross 2006). Accurate estimation of selectiv ity using capture-recapture methods requires several assumptions including: (1) the population of inte rest is closed to additions and deletions. That is, recruitment, natural mortality, immigration, and emigration must be minimal, (2) tags are not lost or go undetected, and (3) equal capture probability (i.e. no capture heterogeneity and/or trap response) (Pollock et al. 1990). Unlike dire ct methods, indirect measures of selectivity require no prior knowledge about the age composition of a population. If catch-at-age data from the commercial or recreational sectors are available, age stru ctured population models like virtual population analysis (VPA) can estimate the age/size selectiv e properties of the fish ing gear used. Other approaches incorporate the catch ra tes of various sizes of fish fr om different gear types and/or mesh size to compare relative gear selectivity be tween gears, but such studies do not identify the true selectivity of e ither gear (e.g., Boxrucker and Plosky 1989; Miranda et al. 1992; Millar and Holst 1997). Indirect or relative measures of abundance such as catch per unit effort (CPUE) are commonly used by managers to assess and analyze trends in fish population abundance. In order for CPUE to directly index population abundance, the relation ship between catch rate and abundance must be: A N q f C (1) where C = catch, f = fishing effort, q = catchability coefficient (the fraction of population removed per unit of effort), N = fish abundance and A = area occupied by stock (Ricker 1975). 12

PAGE 13

This equation infers a linear relationship be tween CPUE and abundance with a constant slope q, which is often not the case. Catch per effort is a function of two factors: catchability and fish density (Hilborn and Walters 1992; Arreguin-Sanchez 1996) Therefore, variability in q causes variability in CPUE that is not related to populati on size, so catch statistics should be adjusted to account for variation in catchability. Because catchability is a function of selectivity, it is an important parameter when using CPUE to index abundance. Furthermore, when catchability coefficients are apportioned by age/size classes, the estimated co efficients are actual ag e/size gear selectivity estimates. Like selectivity, catchability differs with a wi de range of factors including fish age (Pierce and Tomcko 2003), fish size (Bayley and Au sten 2002; McInerny and Cross 2006), species (Bayley and Austen 2002; Schoenebeck and Hans en 2005), fish density (Peterman and Steer 1981; Rogers et al. 2003), sample gear type (Hansen et al. 2000; Pierce and Tomcko 2003), environmental conditions during sampling (B ayley and Austen 2002), and sampling season (Schoenebeck and Hansen 2005; McInerny and Cross 2006). Nielson (1983) found catchability from otter trawls to be similar across age-cl asses of adult yellow perch. Pierce and Tomcko (2003) showed q of northern pike to vary with age in gill-nets. McInerny and Cross (2006) quantified the effects of size, se ason, and density of black crappie on trap-net catchability. They found q increased with fish size, and catchability was higher in spring than fall. Catchability also varied with density as both increased and decr eased values of catchability were observed for different length-groups and sampling periods Bayley and Austen (2002) provided a comprehensive evaluation of electrofishing q estimates for different fish species, sizes, and seasons under variable environmental conditions. Overall, knowledge of q can allow managers 13

PAGE 14

to adjust abundance indices and estimate absolu te abundances, estimate gear selectivity patterns, identify seasonal and/or environmental biases as sociated with sample gears, and aid in the selection of appropriate gears to maximize management objectives. Black crappie support one of the most popular sport fisheries in North America often ranking first or second among angler preference, but can be difficult to manage. Sampling crappies to accurately describe rate functions (such as growth and mortality), abundance and size structure is often demanding re quiring much effort. Indexing black crappie abundance and size structure is challenging due to differences in ge ar performance and selec tivity patterns. In the Midwest, trap nets have been useful in co llecting large samples of crappie of all sizes (Gablehouse 1984; Colvin and Vasey 1986; Boxr ucker and Ploskey 1989), but true gear selectivity has rarely been measured (but s ee McInerny and Cross 2006). Conversely, trap nets in some southeastern systems collected young fish but few adults (Sammons and Bettoli 1998; Maceina et al. 1998). Miranda and Dorr (2000) quantified the size selectiv e effects of crappie angling in five southeastern systems and reporte d dome-shaped selectivity for crappie vulnerable to angling (size range: 20.0 to 39.8 cm) with sma ller and larger sized cr appie less susceptible than intermediate sizes indicating differences in catchabili ty and thus, exploitation. Trap net efficiency and relative selectivity ha s been evaluated and compared to other gears in numerous studies to determine which methods of capture are most effective (McInerny 1989; Boxrucker and Ploskey 1989; Miranda et al. 1992; St. John and Black 2004). McInerny (1989) found trap nets were the most effective and cost-efficient gear deployed for sampling black crappie populations at Lake Wylie, North Carolina. Miranda et al. (1992) reported a higher catch per effort with trap nets compared to electr ofishing in the spring and rotenone sampling in the summer for four Mississippi wate rs. Boxrucker and Ploskey ( 1989) revealed greater sampling 14

PAGE 15

efficiency and less variation in catch per effort with trap nets when co mpared to electrofishing and gillnetting. Thus, trap nets have provided us eful data in some cases, but the selectivity of trap nets relative to the population size/age has seldom been measured. Otter trawls have received far less attention than other capture methods to index crappie abundance. In Florida waters, otter trawls ha ve proven to be successful at capturing black crappies (Schramm et al. 1985; Allen et al. 1999 ; Pine 2000). Allen et al. (1999) compared the relative efficiency of trap nets versus otter traw ls for sampling black crappie in two Florida lakes and reported that trawl sampling was superior to trap nets based on the size range of fish collected, accuracy of abundance estimates, required sampling effort, and expenditures associated with gear. Pine (2000) compared the relative selectivity of two different sized bottom trawls and found a smaller trawl was more effectiv e at collecting juvenile black crappie than a larger trawl. Despite the importance of black crappie in Florida and the popularity of bottom trawls for sampling crappie populations, trawl se lectivity of black cr appie relative to the population has not been measured. Thus, in order for biologists to utilize trawl catch data for management it is important to understand the sele ctive properties of the gear relative to the population. This will enhance the ability of managers to use trawl CPUE as an index of abundance as well as length and age frequency information to describe the size/age structure of black crappie populations. My objectives were to (1) estimate size-spec ific catchability (q) of black crappie collected with otte r trawls (2) estimate relative age/size-specific selectivity of bottom trawl gears and (3) use those selectivity patte rns to evaluate the utility of otter trawls as an assessment gear for black crappie for Florida lakes. 15

PAGE 16

CHAPTER 2 METHODS Direct Measure of Selectivity Capture-recapture sampling took place at La ke Jeffords, Florida during January 2007. Lake Jeffords is a 65 hectare, mesotrophic (Pin e 2000) system located in Alachua County, North Central Florida (Figure 1). I selected Jeffords b ecause I felt I could adequately sample the entire lake (i.e. sample all available habitat types) and create a la rge enough marked subpopulation to obtain reliable catchabi lity estimates. Mark-recapture methods were used to create a tagged population using three gear types. Marking took place over a 10 day period in Janu ary 2007, with electrofishing gear sampled on day 1, otter trawls sampled on days 1, 2, and 3, and hoopnets sampled on days 7 10. I sampled with three gears during the marking event to ensure all available habitat t ypes of the lake were sampled. The recapture event took place over a two day period with bottom trawls two weeks after the first marking day. I only used bottom trawls during the recaptu re period, which allowed estimation of trawl size selectivity based on my known tagged population. The perimeter of the lake was electrofished at both events to ensure fish had not moved into the shallow littoral zone where it is not possible to effectively trawl. Ca ptured fish from all trawls were divided into subgroups by length. This division allowed estimation of q by size, providing a measure of actual trawl size se lectivity. The length-groups (mm) roughly resembled ages 0 (90-119), 1 (120-149), 2 (150-179) and adult fish three or ol der (180+). Abundance estimates were obtained using a Lincoln-Peterson estimator (due to closed system and 2-stage mark-capture sampling event) and the proportions of marked fish were calculated as the number of fish caught in the recapture divided by the abundance estimate. Al l black crappie captured in the field during marking were measured for total length (TL) to nearest (mm) and pelvic fin clipped. Since only 16

PAGE 17

two weeks passed from mark to recap events and fish were fin clipped instead of using conventional tag types like a T-bar tag, I assumed tag loss to be negligible. All black crappie captured during the recapture wher e measured for total length to nearest (mm) and checked for fin clips. Bottom trawls were pulled from a 7-m boat pow ered with a 70 hp outboard in all areas of the lake except in the shallow littoral zone to avoid fouling by vegetation. Effort was constant throughout the study at three minutes per trawl. The trawl net consisted of a 4.88-m long body and 4.6-m mouth and the body is constructed with 38.1 mm stretch mesh and 31.8 mm stretch mesh in the cod end (Allen et al. 1999). Under to w, the mouth of the trawl is spread open with floats (25 x 50 mm) that are secured to the head rope of the trawl mout h. The sweep, or chain line, was attached to the footrope of the ne t. Wooden doors (38.1 x 76.2 cm) were secured to 146 cm leglines and a 15.3-m trawl bridle. The we ighted doors served to open the trawl mouth and allowed the net to sample near the bottom. Modified hoop nets were deployed in the middl e of the lake at various sites. Hoop nets consisted of four similar-sized fiber-glass hoops either 0.9, 1.2 or 1.5m in diameter and covered with 5.1cm stretch nylon mesh webbing. A 23-m lead was used to connect two nets, which would direct fish toward a hoop net as they tr aveled along the lead. All hoop nets were set during the day, fished for 48 hours, and retrieved. Hoop nets were only used for capture sampling event. Electrofishing was conducted with a Smith-R oot model SR18 electr ofisher, equipped with a Smith-Root 9.0 GPP pulsator powered by a 9,000 Watt generator. Approximately 7 amps of DC current were produced at 120 pulses pe r second. The entire shoreline perimeter was sampled (as described above) with an experienced crew of one nette r and one boat operator. 17

PAGE 18

I estimated tagging mortality, defined as morta lity from capture, handling, and tagging for each size-group to adjust the size of our mark ed population available for recapture. A subsample of marked fish were held in aerated bait tanks and placed in holding pens as replicates (n = 8) for 24 hours to estimate associated taggi ng mortality for different length-groups. Holding pens were constructed out of pvc pipe which co nsisted of a rectangular frame that measured 3.0 m length by 1.25 m width. The body of the holding nets consisted of 19.3 mm stretch mesh webbing that extended to a depth of 1.1 m. The observed mortality rates for each size-group and holding pen were randomly re-sampled with re placement using a bootstrap to create 1,000 Monte Carlo estimates (Haddon 2001). The 95% confiden ce intervals were calculated at the 2.5 and 97.5 percentiles using the means of the resample from the bootstrap. I used maximum likelihood methods to estimate how q varied with fish size. The Poisson log likelihood function was appropriate and indicated as ))ln()()()|(lni i i iPOPqOL (2) where Pi = (number available for recapture in size-group i q effort) and Oi = (number of observed recaptures in size-group i ). Catchability for each length-group was estimated by maximizing the negative likelihood function (i.e. minimize the differences between the observed and expected recaptures). Parameters for the model included survival from marking ( S) = (1 observed tagging mortality), number availa ble for recapture = (number marked S), and q, the fraction of population caught per unit of effort. To describe the uncertainty in q estimates, I constructed likelihood profiles for each length-group. These profiles were probability dist ributions (i.e. p-values) for the parameters. The 95% confidence intervals for each parameter were calculated using Wilks likelihood ratio test statistic (Pawitan 2001) equal to 18

PAGE 19

)( )( ln*2 L hatL W (3) where W = Wilks statistic, L ( hat ) = log likelihood at MLE, and L ( ) = log likelihood at some value less than the maximum likelihood estimat e (MLE). Wilks stat istic conforms to a Chi-square distribution with one degree of freedom (Pawitan 2001). Indirect Measure of Selectivity Black crappie populations in Florida lakes have been sampled with otter trawls over the last two to three decades. I used a long-term da tabase from four lakes to estimate the size/age selectivity of bottom trawls us ing an age-structured population modeling approach. In this context, selectivity was defined as differences in re lative fish susceptibility due to size/age. The model was used to estimate selectivity at ag e by comparing observed and model-predicted catches-at-age using maximum likelihood estimation. Annual bottom trawl sample data were obtaine d from lakes Griffin, Johns, Lochloosa, and Okeechobee (Figure 2). The length of the data tim e series varied among lakes and ranged from five years for Lake Johns (2002-2006) to 20 years at Lake Okeechobee (1987-2006). All lakes exceed 1,500 ha with mean depths ranging from 1.8 m to 2.8 m and are classified as eutrophic or hypereutrophic (Florida Lakewatch 1999, 2001; Forsberg and Ryding 1980) (Table 1). Black crappie populations were sampled using a combination of fixed and/or random sites. Lakes Lochloosa and Johns were divided into 250m2 grids using ARC GIS software with buffers built in to avoid sampling in the vegetated littoral zone. Vegetation fouls the gear, reduces gear performance, and does not allow an accurate assessment of gear utility. At Lake Lochloosa, fixed and random sites were used throughout, wh ereas Lake Johns data were obtained from randomly generated fixed sites until years 2005 and 2006 where a combination of six fixed and six random sites were used. At Lake Griffin, 25 to 68 trawls were pulled based on a similar but 19

PAGE 20

slightly different SRS design (fixed sites to 2002 and random sites from 2003). One to 20 trawls were pulled at Lake Okeechobee on the north en d of the lake from 0.8 to 2.4 km offshore between Taylor Creek Lock (S-193) and Nubbin Slough Spillway (S-191) until 500 black crappies were captured. In 2005, due to a large drop in numbers likely attributed to the 2004 hurricane effects on the lake, the sampling fo cus changed from minimum numbers to minimum effort (~150 minutes). In lakes where a combination of fixed/random sites were used in a given year (Johns, Lochloosa) I tested for differences in mean CPUE and size st ructure to determine if fixed and random site samples could be pooled. Analysis of variance (ANOVA) indicated no significant site type or year*sit e type effects on mean CPUE fo r both lakes: Johns (P = 0.48, P = 0.44), Lochloosa (P = 0.86, P = 0.08). Differences in size structure by site type were evaluated using a Chi-square test and results indicated no si gnificant differences in si ze structures between fixed or random sites in any year. Based on th ese findings, I combined fixed and random sites for the analysis. There were some method differences among the lakes. Trawls were towed at a speed 2.0 2.5 m/s (1800-2000 RPM) with the exception of La ke Okeechobee where trawls were pulled at 1.0 m/s. Effort was constant throughout the st udy at three minutes per trawl with the exception of Lake Lochloosa where some trawls were pul led for five minutes, and Lake Okeechobee where trawls were pulled for 15 or 30 minute intervals (see Miller et al. 1990). Samples were collected during daylight hours from October December at Lakes Griffin, Johns and Lochloosa, and in January at Lake Okeechobee. The trawl net used at Lakes Griffin, Johns, and Lochloosa was the same as used for Lake Jeffords capture-recapture sampling (described above) which I will refer to as the standard trawl net. Lake Okeechobee trawl net was si milar in design and application 20

PAGE 21

but was larger, having a 10.7-m headrope, 32-mm square body mes h, and 25-mm square cod end mesh. All black crappie captured in the field were meas ured for total length (TL) to nearest (mm). Subsamples of five fish per one cm group were brought back to the laboratory for further analysis. However, age data were collected only every other year fr om 2001 to 2006 at Lake Lochloosa. At the laboratory, gender, total le ngth (TL) to nearest (mm) total body weight (TW, g) were determined and sagittal otoliths were re moved. Ages were determined in whole view by two independent readers as the number of opaque bands on sagittal otoliths. Otoliths from fish older than two years as well as any discrepa ncies on whole reads were sectioned along the dorsoventral plane before aging. Use of black crappie otoliths for aging in Florida has been validated by Schramm and Doerzbacher (1982). I used an age-structured popul ation model to estimate th e relative age/size-specific selectivity of black crappie to bottom trawls at each lake. The model predicted catch-at-age as a function of von Bertalanffy growth parameters, annual relative recruitment anomalies, an arbitrary number of intital recruits, assumed instantaneous rates of total mortality ( Zo = e-Zo= survival ( So ) to age-1, Z = e-Z = survival ( S) past age-1), and unknown age/size-specific selectivities (to be estimated). The numbers at age (Na,t) in a given year were estimated as ot tRRN *, 0 (4) otatSNN *1,1 ,1 (5) SNNtat*1,1 ,2 (6) where Rt are annual recruitment anomalies, Ro is the average annua l recruitment numbers (arbitrary value used to scale model), So is survival from age 0 to age 1, and S is annual survival for age-1+ fish. Different survival rates for ag e-0 relative to older fish were used because of 21

PAGE 22

expected lower survival for age-0 fi sh. Annual recruitment anomalies ( Rt) for each lake and year were estimated by dividing mean catch per e ffort (CPUE) of age-0 crappie in year t by the median age-0 CPUE across all years. This provide d an index of strong and weak year classes in the population model and was used as a basis for pred icting future catches-at-age with the trawls. This model allowed prediction of the relative numb ers of fish at each lake, age, and year based on the input parameters. From the numbers at age matrix, I predicted a catch-at-age matrix from a hypothesized selectivity schedule. Expected catch at age was calculated as atataSNC *,, (7) where Ca,t is the catch at age at time t and Sa are unknown selectivity at age parameters. Growth parameters were estimated using th e von Bertalanffy growth equation fitted to weighted mean length at age data obtained from age-length keys using the technique described by Devries and Frie (1996). The von Bertalanffy equation is ) 1(*))*((0taK aeLL (8) where La is length at age, L represents the average asymptotic length, K is the metabolic growth coefficient, a is fish age, and t0 is the age at zero length. A growth model was constructed for each lake by pooling annual length-a ge samples after determining that growth did not differ widely among years. The von Bertalanffy growth model was also used to link my agebased selectivities to length-based selectivities for each cohort. Because fish suffer higher rates of mortality ear ly in life than adults (Hjort 1914; Cushing 1975), my base model assumed different instantane ous rates of total mort ality for these two life stanzas: Z = 1.2 ( So = 0.30) for survival to age-1, and Z = 0.6 ( S = 0.54) for ages 1+. These rates served as a base for comparison to othe r simulations under vary ing assumptions for So and S I evaluated the sensitivity of selectivity estimat es to different survival rates by estimating 22

PAGE 23

selectivity parameters under different values of So and S Values of So ranged from 0.22 to 0.37 and S ranged from 0.45 to 0.67. Observed catch-at-age for each lake and y ear was estimated using age-length keys. Because age data were only collected at Lake Lo chloosa every other year, we used the previous and post years age data as the basis for an ag e length key (e.g., 2002 age structure was estimated using 2001 and 2003 age-length subsamples), and a pportioned fish to ages based on the existing length data. Observed catch-at-age for each year was standardized for sampling effort by dividing the catch-at-age by the total lake effort (trawl minutes) for that year. I used a multinomial log likelihood function to estimate selectivity at age by minimizing the differences between observed and expected (i.e., model-predicted) proportions of catch-atage using the Solver function in Excel. The multinomial log likelihood equation was at tata ataPOnSOL,, ,ln )|( ln (9) where n is the number of years model fit to catch data, Oa,t represents the observed proportion of catch at age a in year t and Pa,t is the predicted propor tion of catch at age a in year t I used a logit transformation on selectivity pa rameter estimates in the optimization routine to constrain selectivities between zero and one. When working with a parameter such as a probability that must be between zero and one the logit transformation allows parameter estimates to range from to The logit transformed selectivities were calculated as a aS S X 1 ln' (10) where X is the logit transformed selectivity at age. This logit transformation was used when solving for the point estimates of selectivit y, as well as for the lik elihood profiles (below). 23

PAGE 24

24 Parameter uncertainty was evaluated by cal culating 95% likelihood profile confidence intervals via a likelihood ratio test (Hilbor n and Mangel 1997) usi ng the likelihood profile function in Poptools for Excel ( www.cse.csiro.au/poptools/). Th e profile function allowed me to test alternative parameter estimates for all age-specific selectivities by holding one selectivity estimate constant, then iteratively solving fo r the maximum likelihood estimate by varying the remaining parameters and repeati ng with different valu es until the profile was constructed. The likelihood ratio test (LRT) was expr essed in terms of the differences in the deviance or twice the difference between the negative log-likelihoods (Hilborn and Mangel 1997). The deviance for each simulation was found by (11) max*2 LnL LnLrest where LnLrest is the likelihood value for rest ricted or nested model and LnLmax is the maximum likelihood value for full model. The like lihood ratio test is desc ribed by a Chi-square distribution with r degrees of freedom. The degrees of freedom were determined by the difference in the number of parameters estim ated between the mode ls (Hilborn and Mangel 1997). The LRT allowed comparisons of the probabilities for each selectivity estimate occurring relative to alternative para meter values (hypotheses).

PAGE 25

CHAPTER 3 RESULTS Direct Measure of Selectivity The number of marked fish and size range captu red varied by gear type. I marked 1,250 fish with bottom trawls (size range: 80 365 mm), 23 fish with hoopnets (160 312 mm), and 9 with electrofishing gear (123 318 mm). Recapture with bottom trawls netted 788 fish (88 304 mm), 54 of which were previously marked individuals. Based on my recapture rates and adjusting for differential tagging mortality, I ma rked approximately 0.065 % of the total black crappie population at Lake Jeffords. I marked 0.098 % of fish in group 90-119, 0.064 % of fish in 120-149, 0.073 % in 150-179, and 0.041 % of fish 180+. Tagging mortality for the smaller length groups (90-119, 120-149 mm) was much higher compared to the larger groups (150-179, 180+ mm) (Figure 3). The 90-119 and 120-149 lengthgroups experienced high tagging mortality at 68 and 36%, respectively. The 150-179 and 180+ groups experienced much lower tagging mortalit y rate averaging only 12 and 1%. Overall, tagging mortality was much higher for the two sma llest length-groups rela tive to larger fish. The likelihood profiles for each group indi cated an overall decreasing trend in q with increasing fish size (Figure 4). Results showed that the maximum likelihood estimates (MLEs) of q for the length-group 90-119 were 2 to 3 times higher than q estimates at larger sizes, assuming the calculated mean tagging mortality. The likelihood profiles revealed high uncertainty in all the estimates of q but higher uncertainty for the 90-119 mm size-group when compared to other groups. Figure 5 shows th e lower and upper 95% confidence bounds for the MLEs using mean mortality estimates. I evaluated how uncertainty from tagging mortality estimates would influence estimates of q The maximum likelihood estimates for q based on the lower, mean and upper tagging 25

PAGE 26

mortality rates are presented in Figure 6. Based on mean tagging mo rtality rates, maximum likelihood estimates for length-group 90-119 were approximately twice the MLE values of groups 120-149, 150-179 and three times as high as length-group 180+. Differences in the MLE estimates based on low tagging mortality among size-groups were reduced, except for groups 90119 and 180+ which still varied by a factor of two. In contra st, MLE estimates applying high tagging mortality exhibited large va riation among length-groups where q varied by a factor greater than 2 when comparing the 90-119 group to groups 120-149, 150-179 and a factor of 4 to group 180+. Overall model trends indicated decreasi ng catchability to trawl gear as fish length increased, with a greater amount of uncertainty in the estimates for the smallest length group (90119). Indirect Measure of Selectivity Black crappie growth varied among lakes (F igure 7). Average asymptotic length ( L) among lakes varied from 335 to 398, whereas metabolic growth coefficient (K ) ranged from 0.27 to 0.42, and t0 from -0.79 to -1.17. As expected lakes with higher K values had lower L values, and vice versa (Figure 7). The general patterns observed in the catch-at-age data include d higher catches of age-0 and age-1 fish relative to older age classes, as w ould be expected for any population. Age-0 catch rates among the lakes varied from 0.018 to 16.84 with an average of 3.24 (fish/min). Age-1 catch at age ranged from 0.016 to 10.83 with a m ean of 2.12, whereas age-2 CPUE varied from 0.007 to 7.45 and averaged of 0.94 (fish/min). Catch rates for crappie 3 and older were considerably less relative to younge r age-classes and ranged from 0 to 3.70 with average catch rates to 0.37 (Table 2). Relative recruitment anomalies varied by lake an d some lakes exhibited large fluctuation in recruitment while others showed little variabil ity in recruitment strength (Figure 8). The 26

PAGE 27

recruitment anomalies for Lake Griffin varied fro m 0.24 to 3.57 with an average value of 1.13. Lake Johns recruitment values varied from 0.85 to 1.12 with a mean rela tive recruitment of 0.99 indicating little variability in recruitment. Lake Lochloosa anomalies ranged from 0.14 to 2.35 with an average recruitment va lue of 1.04. Lake Okeechobee exhibited large fluctuations in year-class strength with recruitmen t values ranging from 0.01 to 5.01 with a mean of 1.50. Thus, the recruitment trends as indexed with age-0 fish catch rates suggested su bstantial variation in recruitment among years at each lake. My age structured model estimated dome-shape d selectivity with peak values for black crappie in bottom trawls at age-1. In general, age-0 and age-1 fish were more susceptible to trawl gears than older age-cl asses (ages-2+). Model simulations for the base model ( Zo = 1.2 = So = 0.30, Z = 0.6 = S = 0.55) revealed peak se lectivity at age-1 for all lakes (Figure 9). The average selectivity schedule for Lakes Griffin, John, and Lochl oosa sampled with the standard bottom trawl gear also i ndicated peak selectivity at age-1 (Figure 10). Varying assumptions for instantaneous rates of mortality (i.e. survival) influenced the selectivity parameter estimates. When survival to age-1 increased (lower Zo) greater numbers of older fish were available for capture decreasing the corresponding proportion of age-0 fish in the catch, thus increasing selectivity estimates for age-0 fish (Figure 11). Under this scenario, all selectivity schedules peaked at age-1 as before, but selectivity estimates for age-0 fish increased. Lake Johns was the only exception which exhibited peak selectivity at age-0 declining with age if survival to age-1 increased (Figure 11). Conve rsely, when survival to age-1 decreased (higher Zo) fewer numbers of older age-classes were availa ble for capture, decrea sing their proportion in the catch. The corresponding propo rtion of age-0 fish in the catch increased resulting in lower age-0 selectivity estimates. When su rvival past age-1 increased (lower Z ), greater numbers of 2+ 27

PAGE 28

28 age-class fish were available for capture which re sulted in increased proportions of older fish in the catch. These increased proportions of older age-classes represented in the catch resulted in decreased selectivity estimates for those ages. If survival past age-1 decreased (higher Z ) fewer numbers of 2+ age-class fish became available for capture. Under this scenario, decreased proportions of older fish resulted in increased selectiv ity estimates. Overall, the selectivity estimates for the older age-classes were more sensitive to changes in survival compared with age-0 and age-1 estimates (Figure 11). For example, Lake Okeechobee results indicated the selectivity estimates for the olde r age-classes could vary by a factor of 2 or 3 from the base model estimates. Nevertheless, changes in assu med survival rates did not change the overall pattern of dome-shaped selectivity for bo ttom trawls (Figure 11). The uncertainty in the age-specific selectivity estimates are described from probability profiles (similar to p-values) for each lake in Figures 12 15. Most age-specific selectivity profiles indicated wide probability bands with pot ential selectivity estimates ranging from 0 to 1 for most lakes. Lake Okeechobee estimates exhi bited tighter intervals relative to other lake selectivity estimates. This is likely attributed to a longer timeseries of data (model fit to 10 years), whereas other lakes had shorter data tim e-series resulting in wider probability bands. Age-1 selectivity exhibited tighter intervals ( on average from 0.70 to 1) than all other agegroups.

PAGE 29

Table 3-1. Summary of Lake characteristics for study loca tions, including county of location, surface area in hectares (ha), Chlorophyll-a concentration measured in (mg/L), trophic st atus, and years sampled. Tropic state based on Forsberg and Ryding (1980), Florida Lakewatch Data (1999, 2001), Bachman et al. (1996) Lake County Surface Area (ha) Chlorophyll-a (g/L) Trophic State Sample years Griffin Lake 6,679 159 Hypereutrophic 1999-2006 Johns Orange 1,676 13 Eutrophic 2002-2006 Lochloosa Alachua 2,631 101 Hypereutrophic 2000-2006 Okeechobee Glades 173,000 30 Eutrophic 1987-2006 29

PAGE 30

30 Table 3-2. Summary of CPUE-at-age (fish/ min) data showing the mean, minimum, and maximum values for each age and lake and combined across lakes. Lake Age-0 Age-1 Age-2 Age-3 Age-4 Age-5 Mean 5.32 2.04 0.43 0.09 0.03 0.01 Min 1.15 0.17 0.01 0.01 0.00 0.00 Max 16.84 4.41 0.66 0.19 0.08 0.04 Griffin Mean 3.78 1.29 0.23 0.06 0.01 Min 3.25 0.26 0.04 0.02 0.00 Max 4.22 2.14 0.47 0.14 0.04 Johns Mean 3.85 1.63 0.46 0.06 0.01 0.01 Min 1.23 0.43 0.19 0.02 0.00 0.00 Max 9.31 4.46 0.84 0.12 0.04 0.02 Lochloosa Mean 2.06 2.53 1.48 0.67 0.22 0.13 Min 0.02 0.02 0.06 0.02 0.01 0.00 Max 7.65 10.83 7.45 3.70 1.31 1.41 Okeechobee Mean 3.24 2.12 0.94 0.37 0.12 0.08 Min 0.02 0.02 0.01 0.00 0.00 0.00 Combined Max 16.84 10.83 7.45 3.70 1.31 1.41

PAGE 31

Figure 3-1. Lake Jeffords located in Alachua County, North Central Florida 31

PAGE 32

32 Figure 3-2. Geographic lo cation of study lakes used for indirect measure of traw l size selectivity

PAGE 33

Figure 3-3. Bootstrap estimates of mean tagging mortality. E rror bars represent 95% confidence intervals. 33

PAGE 34

Figure 3-4. Like lihood profiles for q by length-group using mean tagging mortality rates. Maximum likelihood estimates (MLEs) represented at peak of each curve. 34

PAGE 35

Figure 3-5. Maximum likelihood estimates of q by length-group and associated 95% confidence intervals using mean mortality rates. 35

PAGE 36

Figure 3-6. Maximum likelihood estimates for q calculated from the lower, mean, and upper tagging mortality rates. 36

PAGE 37

Figure 3-7. Predicted mean length at age for each lake. The least-squares equations are shown. 37

PAGE 38

Figure 3-8. Observed time series of relative recruitment anomalies by lake. 38

PAGE 39

Figure 3-9. Relative selectivity by age for each study lake from base model 39

PAGE 40

Figure 3-10. Average relative selectivity for lakes sampled with standard trawl (Lakes Griffin, Johns, and Lochloosa). 40

PAGE 41

Figure 3-11. Relative sel ectivity by age under varying assumptions of survival. 41

PAGE 42

Figure 3-12. Lake Griffin probabilit y profiles with relative selectiv ity on the x-axis and relative probability on the y-axis. 42

PAGE 43

Figure 3-13. Lake Johns probability profiles with relative selectivity on the x-axis and relative probability on the y-axis. 43

PAGE 44

Figure 3-14. Lake Lochloosa proba bility profiles with relative selectivity on the x-axis and relative probability on the y-axis. 44

PAGE 45

45 Figure 3-15. Lake Okeechobee probability profiles with relative selectivity on the x-axis and relative probability on the y-axis.

PAGE 46

CHAPTER 4 DISCUSSION Bottom trawls are an extremely efficient captur e gear (reflected by large catches of various aquatic organisms and use in many fisheries worl dwide), but often select for small finfish and crustaceans (Kjelson and Johnson 1978; Rulifson et al. 1992; Diamond et al. 1999). The size selectivity exhibited by bottom trawls results in large catches of small aquatic organisms (as evident by their use in commercial shrimp fisheries) and the inci dental catch of commercially and recreationally important j uvenile fishes (Howell and La ngan 1992; Gallaway and Cole 1999; Diamond et al. 1999; Wakefi eld et al. 2007). My direct and i ndirect estimates of bottom trawl selectivity corroborated one another and indicated decr easing size selectiv ity with increasing length, suggesting that bottom tr awls would be best for monitoring the abundance of small black crappie but may be inadequate to characterize the adult population. Catchability of black crappie has seldom been measured, but Miranda and Dorr (2000) found q with angling gear to vary with fish size. Furthermore, Mc Inerny and Cross (2006) found q with trap-nets varied with size, season, and density and used these estimates to im prove the interpretation of CPUE data to index abundance and describe size stru ctures. My estimates of q lend insight into the size selective properties of bottom trawls for black crappie, and could be incorporated w ith catch data to index the true population age/size compos ition (e.g., Lake Jeffords). My age-structured model simulations indicated dome-shaped selectivity to bottom trawls for black crappie and the highest selectivity estimates were for age-1 fish. Dome-shaped selectivity patterns are common for many sample gears and species (E rzini and Castro 1998; Jackson and Noble 1995; Miranda and Dorr 2000; Tanaka 2002), a nd evaluating differences in relative selectivity by age/size allows managers to adju st indices of abundance (Quinn and Deriso 1999) or determine gear efficiencies on th e different age/size classes (Pierce et al. 1994). 46

PAGE 47

Selectivity peaked at age-1 for all lakes under all scenarios except Lake Johns which exhibited decreasing relative selectivity with age when assumed age-0 survival was high. The results corroborate other selectivity st udies (Jagielo 1999; Bayley and Austen 2002; McInerny and Cross 2006) and indicate the trawl gear may be most useful for tracking small-sized black crappie through time. Selectivity estimates exhi bited high uncertainty ex cept for age-1, but the observed selectivity patterns were similar across la kes. Individual selectivity estimates from my analyses should be viewed with caution, but the overall pattern of dome-shaped selectivity probably describes the general pattern of bo ttom trawl selectivity for black crappie. Gear selectivity determines the effect of fishing on size/age structures. As such, assessment models can link size/age composition of catch data to size/age composition of the fishery (Taylor et al. 2005) to predict the effect s of different harvest ra tes, calculate biological reference points, and identify maximum sustaina ble yields (Maunder 2002). Assuming constant catchability among size/age classes is a dangerous assumption and can lead to bias in yield models (Ricker 1975) because differe nces in age/size specific catchability to fishing gears can alter the maximum sustainable yield (MSY) obt ainable from a fishery (Maunder 2002). When fishing mortality is restricted th rough effort controls, the catchab ility coefficient becomes a vital parameter in yield models, due to the relationship between q and yield and abundance. Gear selectivity patterns and the cumulative eff ects of size-selective fishing practices often produce bias in length at age sa mples (Sinclair et al. 2002, Taylor et al. 2005). In exploited populations, fast growing young fish and slow growing older fish ar e overrepresented in length at age samples. In turn, this biases the parame ter estimates of growth models used to describe mean length at age. My model results indi cated bottom trawl selectivity for younger size/ageclasses. Furthermore, all lakes in the study experience some level of angler exploitation. Thus, 47

PAGE 48

since trawls appear more effec tive at capturing small fish and th e older age-classes are exposed to harvest, growth models from trawl data co llected from exploited populations may severely underestimate the asymptotic length ( L) and overestimate the metabo lic growth coefficient ( K ). These inaccuracies can lead to biased mean length at age which would influence estimates of maximum yield and optimal harvest policies. Age/size selectivity of sampling gears cause s errors in the estimation of population structure (Walters and Hilborn 1992) and may limit the ability to dr aw inferences about trends in abundance. Therefore determini ng size selectivity of fishery i ndependent assessment gears is important when survey CPUE data are used to index abundance, especially if assessment gears may not adequately sample the ag e and size range targeted by the fishery. For example, I found that large adult black crappies we re poorly represented in bottom trawl samples, but these fish are targeted by recreational fisher ies. Thus, bottom trawls may not detect changes in abundance even if large fish suffer high fish ing mortality. This could lead to appearances of hyperstability (i.e., relatively constant CPUE over a large range of true fish abundance) because trawl catches of large black crappie could potentially be low and not change widely with changes in abundance. When properly designed and implemented, mark-recapture studies can provide managers with information on growth, mortal ity, and reproduction, population size and structure, and gear selectivity (Seber 1982). Data fo r closed capture-recapture methods must meet several criteria. Foremost, the population of interest must be cl osed to additions and deletions. That is, recruitment, natural mortality, immigration, a nd emigration must be minimal. The immigration and emigration assumptions were probably not violat ed in my study due to the closed system and short time interval over which the study occurred. The mortality assumption was violated due to 48

PAGE 49

tagging mortality but I adjusted for this in my calculations. Another assumption for mark recapture experiments is that marks are not lost or undetected by the recorder. Tag retention and detectability were likely very hi gh in my study because I marked fi sh with a pelvic fin clip and sampled over a short interval befo re regeneration of fins could o ccur. A third requirement for closed mark recapture experiments is equal cap ture probability (i.e. no capture heterogeneity and/or trap response). Trap response is when capture probability is dependent on an animals capture history, and is difficult to directly estimate (Pollock et al 1990). I estimated heterogeneity in capture probability as a func tion of fish size and account for potential trap responses by using different gear types for mark and recap events. One potential cause of bias and uncertainty in q estimates from the mark-recapture experiment could be the higher observed sampli ng mortality for the sm aller length-groups (90119, 120-149) relative to the larger groups (150179, 180+). This size selective mortality significantly reduced the number of fish availa ble for recapture among those two groups. An overestimation in tagging mort ality for these groups would cause a positive bias in my q estimates, resulting in more of an effect on trawl size selectiv ity. Alternately, underestimating sampling mortality could have caused a negative bi as in the catchability estimates. The larger observed mortality on the smaller size groups could have resulted from high handling times and greater sampling stress on these classes relative to the larger length groups. The likelihood profiles also show there is mo re uncertainty in the likeli hood estimate for the 90-119 size-group when compared to other groups. The larger variance can likely be attributed to the higher observed sampling mortality for this group. In tu rn, this influenced the number available for recapture (smaller N for this group) leading to a greater difference between observed and expected recaptures. 49

PAGE 50

Differences in growth among the lakes could expl ain some of the differences in selectivity at age between lakes. Some lakes appeared to exhibit faster growth, which could have resulted in fewer numbers of faster growi ng individuals in the catch (as de scribed above). This may have resulted in biased mean length at age for some lakes which may explain some of the observed differences in age-specific selectivities across systems. However, these differences appeared minimal because the overall pattern of dome-s haped selectivity occurred at all lakes. My approximation of age-length keys using prev ious and post year data at Lake Lochloosa could pose problems in the estimated age structur e and thus, the catch ra te indices for those years. Inaccuracies in observed catch-at-age could have aff ected my proportions of catch-at-age and model fit having the potential to cause unknown biases in the selectivity estimates. However, selectivity patterns for Lake Lochloos a were similar to those observed at the other lakes where age data were collected every year. Thus, I believe the selectivity patterns at Lake Lochloosa reflected real differen ces in across age-classes despite the gaps in the age structure data. Furthermore, variation in mort ality at size/age can affect the numbers in each age class available for capture leading to bi ased selectivity estimates. I a ccounted for this variability in survival to age by apportioning mortality into two classes (surviva l to age-1 and survival past age-1) based on life stanzas. I also simulated vari ous levels of survival to show how different rates of survival on the age-cla sses could influence the numbers available for capture and thus, my selectivity estimates. I did not attempt to quantify exploitation rates on fish vulnerable to angling (>= age-2) which may have decreased survival for some of the older age-classes relative to younger age-classes. If exploitation is high on older age-cla sses, the corresponding proportions of catch in those age-classes would decrease resulting in an increase in the younger 50

PAGE 51

age-class proportions and an upward bias in the selectivity estimate s for older/larger fish and a downward bias in selectivity es timates of younger/smaller fish. I assumed constant survival among years and did not evaluate the effects of ch anges in mortality between years. Changes in yearly natural and/or fishing mortality could significantly increase or reduce the numbers available for capture creating both positive and negative biases in selectivity estimates. The indirect approach I used to estimate tr awl selectivity is common data collected for recreational and/or commercially important fish eries. This methodol ogy is quite common in marine assessments (Walters and Martell 2004) but examples are lacking for freshwater fisheries applications. I found this approach quite useful for estimating gear selectivity and recommend its application when catch-at-age time series data are available and when direct estimates are infeasible (e .g., large, open systems). Use of my selectivity estimates should be rest ricted to similar seasons and lake types as used in this study. Size-selec tivity of bottom trawls was evalua ted during fall for the indirect method and winter for the direct method. Thus, the estimates provided may not apply to other seasons or times of the year. Seasonal effects on fish behavior are well documented (Pope and Willis 1996, Hayes et al. 1996) and sampling black crappie with bottom trawls at other times in the year may result in unknown bias in the CPUE index. Use of my sel ectivity estimates should be restricted to similar seasons from which they were derive d. Lake characteristics and differences in geomorphology may have influenced my results. For example, gear efficiency and thus, catchability may vary due to differences in amount of suitable habitat, percent area coverage of macrophytes (PAC), depth, dissolved oxygen (DO) levels, substrate type or a variety of other factors. I did not atte mpt to quantify all of these effect s and managers should be aware of these potential sources of bias and use the selectivity estimates with prudence. Furthermore, 51

PAGE 52

systems selected for trawl sampling should be de termined carefully to maximize efficiency and prevent gear fouling. Bottom trawl sampling may be an inappropriate gear on some lakes due to excessive bottom debris. Submerged debris, su ch as stumps or large amounts of aquatic macrophytyes may prevent effective trawl sampling limiti ng sampling and/or gear efficiency. The likely cause for size selectivity of smaller black crappie to otter trawls may be the ability of larger fish to detect or avoid the gear. Net avoidance and escapement after initial capture could result since swimmi ng speed tends to increase with fish length (Helfman et al. 1997). Furthermore, gear avoidance by the larger fi sh may result due to the trawl pressure wake. This pressure wake may be detected due to larger and more developed lateral lines in larger fish. Spatial distributional patterns and habitat availa bility and use may also be a source for trawl selectivity. The smaller fish may utilize the pelagic zones of a lake and school more relative to large fish, where the larger fish may be more patchily distributed utilizing both open water and littoral areas of lakes. However, we controlled for differences in spatia l distribution and habitat availability/use in the direct measure at Lake Jeffords, where crappie of all size ranges and the majority of fish were captured in open water habitats. Differences in spatial distribution patterns may have had more of an influence on the indire ct selectivity estimates where there is likely a greater difference in habitat av ailability and use due to lake size and habitat complexity. However, the direct measure of selectivity suggests that the ability of larger fish to detect/avoid the gear is the likely source for trawl selectivity of younger/smaller crappie. Stock assessments are important for fisherie s management and attempt to recreate past stock trends to explain current stock trends and abundance by making quantitative predictions about the reactions of fish popul ations to alte rnative management options (Hilborn and Walters 1992). Therefore, bottom trawl catch data may be important for stock assessment analysis and 52

PAGE 53

53 used for indices of recruitment, as measures of recruitment are one of the important inputs used for stock assessments. These data could be used in conjunction with another gear type that indexes the adult population (such as a creel) along with an es timate of fishing mortality to evaluate current stock trends facilitating the ability to make choices among policy options and evaluate the trade-offs associat ed with those decisions. Otter trawls have been show n to effectively capture young black crappies (Allen et al. 1999, Pine 2000), and my results suggest that botto m trawls provided adequa te catches of age-0 and age-1 fish. As such, otter trawls are probably most effec tive at tracking age-0 and age-1 abundances and can be used for estimates of yea r-class strength. However, my results suggested that the trawls may be inadequa te to describe the adult populati on, fish growth rates, and age structure estimates due to substantially lower sele ctivity values for large fish. The selectivity estimates provided will allow managers to adjust abundance indice s and correct age/size structures for relatively shallow Florida lakes.

PAGE 54

LIST OF REFERENCES Allen, M. S., M. M. Hale, and W. E. Pine III. 1999. Comparison of trap nets and otter trawls for sampling black crappie in two Florida Lake s. North American Journal of Fisheries Management 19:977-983. Arreguin-Sanchez, F. 1996. Catchability: a key parameter for fish stock assessment. Reviews in Fish Biology and Fisheries 6: 221-242. Bachman, R. W., B. L. Jones, D. D. Fox, M. Hoyer, L. A. Bull and D. E. Canfield. 1996. Relations between trophic state indicators and fish in Florida lakes. Canadian Journal of Fisheries and Aquatic Sciences 53:842-855. Bayley, P. B. and D. J. Austen. 2002. Capture e fficiency of a boat electro fisher. Transactions of the American Fisheries Society 131:435-451. Beamesderfer, R. C., and B. E. Rieman. 1988. Size selectivity and bias in estimates of population statistics of smallmouth bass, wa lleye, and northern squawfish in a Columbia River Reservoir. North American Jo urnal of Fisheries Management 8:505-510. Boxrucker, J., and G. Ploskey. 1989. Gear and seasonal biases associated with sampling crappie in Oklahoma. Proceedings Annual Conferen ce Southeast Association Fish and Wildlife Agencies 42:89-97. Colvin, M. A., and F. W. Vasey. 1986. A method of qualitatively assessing white crappie populations in Missouri reservoirs. Pages 79-85 in G. E. Hall and M. J. Van Den Avyle, editors. Reservoir fisheries management: st rategies for the 80s. American Fisheries Society, Southern Division Reservoir Committee, Bethesda, Maryland. Cushing, D. H. 1975. Marine ecology and fish eries. Cambridge University Press, Cambridge, England. Devries, D. R., and R. V. Frie. 1996. Dete rmination of age and growth. Pages 483-512 in B. R. Murphy and D. W. Willis, ed itors. Fisheries techniques, 2nd edition. American Fisheries Society, Bethesda, Maryland. Diamond, S. L., L. B. Crowder, and L. G. Cowe ll. 1999. Catch and bycatch: The qualitative effects of fisheries on populat ion vital rates of Atlantic croaker. Transactions of the North American Fisherie s Society 128:1085-1105. Doerzbacher, J. F. and H. L. Schramm, Jr. 1982. Enlarger-produced photographs for the measurement of black crappie otoliths. North American Journal of Fisheries Management 4:547-551. Dumont, S. C., and J. A. Denni s. 1997. Comparison of day a nd night electrofishing in Texas Reservoirs. North American Journal of Fisheries Management 17:939-946. 54

PAGE 55

Erzini, K. and M. Castro. 1998. An alternativ e methodology for fitting selectivity curves to predefined distributions. Fi sheries Research 34:307-313. Farmer, M. J., D. T. Brewer, S. J. M. Blaber. 1998. Damage to selected fish species escaping from prawn trawl codends: a comparison between square-mesh and diamondmesh. Fisheries Research 38:73-81. Florida LAKEWATCH. 1999. Florida LAKEWATCH 1999 data report. University of Florida, Department of Fisheries and Aqua tic Sciences, Gainesville, FL. Florida LAKEWATCH. 2001. Florida LAKEWATCH 2001 data report. University of Florida, Department of Fisheries and Aqua tic Sciences, Gainesville, FL. Forsberg, C., and S. O. Ryding. 1980. Eutrophi cation parameters and trophi c state indices in 30 Swedish water-receiving lakes. Arch. Hydrobiologia 89:189207. Gablehouse, D. W., Jr. 1984. An assessment of crappie stocks in sm all midwestern private impoundments. North American Journal of Fisheries Management 4: 371-384. Gallaway, B. J., and J. G. Cole. 1999. Reductio n of juvenile red snapper bycatch in the U.S. Gulf of Mexico shrimp trawl fishery. No rth American Journal of Fisheries Management 19:342-355. Haddon, M. 2001. Modeling and quantitative me thods in fisheries. Chapman and Hall, New York. Hamley, J. M., and H. A. Regier. 1973. Direct estimates of gillnet selectivity to walleye. Journal of Fisheries Resear ch Board of Canada 30:817-830. Hansen, M. J., T. D. Beard, and S. W. Hewett. 2000. Catch rates and catchability of walleyes in angling and spearing fisheries in Northern Wisconsin Lakes. North American Journal of Fisheries Management 20:109-118. Hayes, D. B., C. P. Ferreri, W. W. Taylor. 1996. Active fish capt ure methods. Pages 193-220 in B. R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition. American Fisheries Society, Bethesda, Maryland. Helfman, G. S., B. B. Collete, and D. E. F acey. 1997. The diversity of fishes. Blackwell Science, Malden, Massachusetts. Hilborn, R. and M. Mangel. 1997. The ecological detective. Princeton Press, Princeton, New Jersey. Hilborn, R., and C. J. Walters. 1992. Quantitati ve fisheries stock assessment: choice, dynamics, and uncertainty. Chapman and Hall, New York. 55

PAGE 56

Hjort, J. 1914. Fluctuations in the great fisheries of Northern Europe viewed in the light of biological research. Rappor ts et Proces-Verbaux des Reunions du Conseil International pour lExploration de la Mer 20. Howell, W. H., and R. Langan. 1992. Discarding the commercial groundfis h species in the Gulf of Maine shrimp fishery. North American Journal of Fisheries Management 12:568-580. Hubert, W. A. 1996. Passive ca pture techniques. Pages 157-181 in B. R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition. American Fisher ies Society, Bethesda, Maryland. Jackson, J. R., and R. L. Noble. 1995. Select ivity of sampling methods for juvenile largemouth bass in assessments of recruitment processe s. North American Journal of Fisheries Management 15:408-418. Jacobson, L. D., J. Brodziak, and J. Rogers. 2001. Depth distributions and time-varying bottom trawl selectivities for Dover sole, sablefish, and thornyheads in a commercial fishery. Fishery Bulletin 99:309-327. Jagielo, T. H. 1999. Movement, mortality, and si ze selectivity of sport and trawl caught lingcod off Washington. Transactions of th e American Fisheries Society 128:31-48. Kjelson, M. A., and G. N. Johnson. 1978. Catc h efficiencies of a 6.1-meter otter trawl for estuarine fish populations. Transactions of American Fisheries Society 107:246-254. Kraft, C. E., and B. L. Johnson. 1992. Fyke-net and gill-net size selec tivities for yellow perch in Green Bay, Lake Michigan. North Am erican Journal of Fisheries Management 12:230-236. Laarman, P. W., and J. R. Ryckman. 1982. Relati ve size selectivity of trap nets for eight species of fish. North American Journa l of Fisheries Management 2:33-37. Lok, A., A. Tokac, Z. Tosunoglu, C. Metin, and R. S. T. Ferro. 1997. The effects of different cod-end design on bottom trawl selectivity in Turkish fisheries of the Aegean Sea. Fisheries Research 32:149-156. Maceina, M. J., O. Ozen, M. S. Allen, and S. M. Smith. 1998. Use of equilibrium yield models to assess different size limits for crappie in Weiss Lake, Alabama. North American Journal of Fisheries Management 18:854-863. Maunder, M. M. 2002. The relationship betwee n fishing methods, fisheries management, and the estimation of maximum sustainable yield. Fish and Fisheries 3:251-260. McInerny, M. C. 1989. Evalua tion of trapnetting for sampling black crappie. Proceedings Annual Conference Southeast Association Fish and Wildlif e Agencies 42:98-106. 56

PAGE 57

McInerny, M. C., and T. K. Cross. 2000. Effects of sampling time, intraspecific density, and environmental variables on electrofishing catch per effort of largemouth bass in Minnesota Lakes. North American Journal of Fisheries Management 20:328336. McInerny, M. C. and T. K. Cross. 2006. Factors affecting trap-net catchab ility of black crappies in natural Minnesota lakes. North Amer ican Journal of Fisheries Management 26:652664. Millar, R. B. and R. Holst. 1997. Estimation of gillnet selectivity using log-linear models. ICES Journal of Marine Science 54:471-477. Miller, S. J., D. D. Fox, L. A. Bull, and T. D. McCall. 1990. Popul ation dynamics of black crappie in Lake Okeechobee, Florida, follo wing suspension of commercial harvest. North American Journal of Fisheries Management 10:98-105. Miranda, L. E., and B. S. Dorr. 2000. Size sel ectivity of crappie angli ng. North American Journal of Fisheries Management 20:706-710. Miranda, L. E., J. C. Holder, and M. S. Schorr. 1992. Comparison of methods for estimating relative abundance of white crappie. Proceedings Annual Conference Southeast Association Fish and Wildlife Agencies 44:89-97. Myers, R. A., and J. M. Hoenig. 1997. Direct estimates of gear selectivity from multiple tagging experiments. Canadian Journal of Fisheries and Aquatic Science 54:1-9. Nielsen, L. A. 1983. Variation in the catch ability of yellow perch in an otter trawl. Transactions of the American Fisheries Society 112:53-59. ONeil, F. G., and R. J. Kynoch. 1996. The e ffect of cover mesh size and cod-end catch size on cod-end selectivity. Fish eries Research 28:291-303. Otway, N. M., J. R. Craig, and J. M. Upston. 1996. Gear depe ndent size-selection of snapper, Pargus auratus Fisheries Research 28:119-132. Paragamian, V. L. 1989. A comparison of day a nd night electrofishing: size structure and catch per unit effort for smallmouth bass. North American Journal of Fisheries Management 9:500-503. Patiwan, Yudi. 2001. In all likelihood: statistical modeling and inference using likelihood. Clarendon Press, Oxford. Peterman, R. M. and G. J. Steer. 1981. Relatio n between sport-fishing catchability coefficients and salmon abundance. Transactions of the American Fisheries Society 110:585-593. 57

PAGE 58

Pierce, R. B., C. M. Tomcko, and T. D. Kolander. 1994. Indirect and direct estimates of gill-net size selectivity for northern pike. North American Journa l of Fisheries Management 14:170-177. Pierce, R. B. and C. M. Tomcko. 2003. Variat ion in gill-net and ang ling catchability with changing density of northern pike in a sm all Minnesota lake. Transactions of the American Fisheries Society 132:771-779. Pine III, W. E. 2000. Comparison of two otte r trawls of different sizes for sampling black crappies. North American J ournal of Fisheries Management 20:819-821. Pollock, K. H., J. D. Nichols, J. E. Hines, a nd C. Brownie. 1990. St atistical inference for capture-recapture experiments. Wildlife Monographs 107:1-97. Pope, K. L., and D. W. Willis. 1996. Seasonal infl uences on freshwater fisheries sampling data. Reviews in Fisheries Science 4:57-73. Quinn, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics. Oxford University Press, New York. Reynolds, J. B. 1996. Electrofishing. Pages 221-253 in B. R. Murphy and D. W. Willis, editors. Fisheries techniques, 2nd edition. American Fisher ies Society, Bethesda, Maryland. Ricker, W. E. 1975. Computation and interpretati on of biological statistics of fish populations. Fisheries Research Board of Canada Bulletin 191. Rogers, M. W., M. J. Hansen, and T. D. Beard. 2003. Catchability of walleyes to fyke netting and electrofishing in northern Wisconsin lake s. North American Journal of Fisheries Management 23:1193-1206. Rulifson, R. A., J. D. Murray, and J. J. Bahen. 1992. Finfish catch re duction in South Atlantic shrimp trawls using three designs of by catch reduction devices. Fisheries 17:9-20. Sammons, S. M. and P. W. Bettoli. 1998. Larval sampling as a fisheries management tool: early detection of year-class strength. Nort h American Journal of Fisheries Management 18:137-143. Sammons, S. M., D. A. Isermann, and P. W. Bettoli. 2002. Variation in population characteristics and gear selection between black and white crappies in Tennessee Reservoirs: Potential effects on management decisions. North American Journal of Fisheries Management 22:863-869. Schoenebeck, C. W. and M. J. Hansen. 2005. Electrofishing catch ability of walleyes, largemouth bass, smallmouth bass, northern pike and muskellunge in Wisconsin lakes. North American Journal of Fisheries Management 25:1341-1352. 58

PAGE 59

59 Schramm, H. L., J. V. Shireman, D. E. Ha mmond, and D. M. Powell. 1985. Effects of commercial harvest of sport fish on the black crappie population in Lake Okeechobee, Florida. North American Journal of Fisheries Management 5:217226. Seber, G. A. F. 1982. The estimation of animal abundance and related parameters. 2nd ed. London: Griffin. Sinclair, A. F., D. P. Swain, a nd J. M. Hanson. 2002. Disentangl ing the effects of size selective mortality, density, and temper ature on length at age. Cana dian Journal of Fisheries and Aquatic Sciences 59:372-382. St. John, T. R., and W. B. Black. 2004. Method s for predicting age-0 cra ppie year-class strength in J. Percy Reservoir, Tennessee. Nort h American Journal of Fisheries Management 24:1300-1308. Tanaka, E. 2002. A method for calculating numeric al estimates of gear selectivity curve. Fisheries Science 68:1081-1087. Taylor, N. G., C. J. Walters, and S. J. D. Ma rtell. 2005. A new likelihood for simultaneously estimating von Bertalanffy growth parameters gear selectivity, and natural and fishing mortality. Canadian Journal of Fi sheries and Aquatic Sciences 62:215-223. Wakefield, C. B., M. J. Moran, N. E. Tapp, and G. Jackson. 2007. Catc hability and selectivity of juvenile snapper and western butterfis h from prawn trawling in a large marine embayment in Western Australia. Fisheries Research 85:37-48. Walters, C. J. and S. J. D. Martell. 2004. Fi sheries ecology and manageme nt. Princeton Press, Princeton, New Jersey.

PAGE 60

BIOGRAPHICAL SKETCH Gregory Robert Binion was born on September 18, 1978, at a U.S. air force base in the United Kingdom, to Mike and Peggy Binion. Shortly after, he moved and was raised in San Antonio, Texas, with his older brother Pete. At a young age, he developed a passion for the outdoors and enjoyed much of his time exploring the wide open spaces and beautiful country of South Texas. He graduated from the University of Kentucky with a B.A. in political science in December 2002. Shortly after graduation, he re located to Florida to pursue an interest in fisheries biology and management. In October 2003, he began to work as a fisheries technician on various projects at the Universi ty of Florida and began his gradua te work at the Department of Fisheries and Aquatic Sciences at the University of Florida in January 2006.He will graduate with a Master of Science in December 2007. Hi s future plans include traveling, passing time fishing and hunting, spending time with his family, and pursuing a ca reer in fisheries management. 60