EFFECTS OF OLFACTORY AND AUDITORY STIMULI ON LOCOMOTION OF PROCAMBARUS CLARKII By Douglas Clements A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Biosystems Engineering - Master of S cience 2020 ABSTRACT EFFECTS OF OLFACTORY AND AUDITORY STIMULI ON LOCOMOTION OF PROCAMBARUS CLARKII By Douglas Clements This study addressed solutions for the population control of invasive Procambarus clarkii (Red Swamp Crayfish) in Michigan. The infes tation was reported to the Michigan Department of Natural Resources in 2015. P. clarkii outcompetes native species of crayf ish, is highly fecund and causes bank erosion through burrowing. Intensive trapping allows slow ing the spread of the population and d etect ing new spread, but the practice is costly. In this study, auditory stimuli were tested as a means of affecting a loco motive response in P. clarkii . These trials tested various pure tone sounds and pink and white noise rang es in artificial habitats. F ollowing those results, a white noise frequency band was played underwater during the trapping season. The results indicated that a high frequency range of white noise (10 - 15 kHz) was most effective at eliciting a locomotory response. When used during the trapping season, a combination of sound and food bait performed at the high est capture per unit effort (CPUE) , 0.820 . Traps with only food and only sound performed at 0.644 and 0.675, respectively . Moreover, traps without sound or food bait performed at a high baseline CPUE of 0.487 . Artificial refuge traps performed at a higher CPUE than other trap types, despite the lack of food bait. The results suggest that the benefits of refuge, sound, and food bait are additive. A novel trap design was created using the advantages of artificial refuge, food bait, and acoustic stimuli. The i mplications of this study span from the control of invasive species in the Great Lakes region to increasing profits of crayfish farming in the southern United States. iii ACKNOWLEDGM ENTS This thesis was made possible because of the dedication of my advisor, Dr. Wei Liao, whose open - door policy and encouraging demeanor pushed me to work harder than I could have imagined ; h e was never too busy to hear about crayfish. A thanks is also in order for Dr. Brian Roth . W hile he is one of my committee members, he more importantly exudes his passion for fisheries work , inspiring others . I would also like to extend a thanks to my committee member Dr. Yan Liu, who made me feel appreciated for all my repair work around the laboratory ; she too wears her enthusiasm for her work on her sleeve. I would also like to acknowledge my graduate secretary, Barbara DeLong. Without her assistance in graduate school application procedures, and hoop - jumping, I wo uld simply be an engineer in a well - pa ying position, instead of a graduate student struggling on his thesis. While she retired at the end of my graduate school career, she has been and will be missed . Sarah Eubanks also deserves my thanks, for helping me f igure out graduation requirements thro ugh a pandemic - laden academic year. My fellow students deserve thanks for their fellowship, humor, and support; none more so than Henry Frost, who has been right beside me since our senior design project. Megan Beaver, and Xiaojing Ma also supported the se nior design project and helped me to realize that one student does not need to do all the work to succeed. The fisheries and wildlife crayfish crew, Aaron Sullivan, Samantha Strandmark , Cole Hazeltine, Kelley Smith, Ma rk Hamlyn, Megan Frick, and Greg Byfor d, who helped me overcome my fears of small red pinching crustaceans deserve my thanks as well. Without you, I would still be fearfully and slowly pulling crayfish out of last traps. iv The funding agencies who he lped make this possible, were a combination of Michigan Department of Natural Resources and U.S. Fish and Wildlife Service. Dr. Seth Herbst and Dr. Lucas Nathan were great contacts and remained in touch when I needed them, which is more than I can say for any other government agency. My family has always supported m y leap into higher education and deserves my unending gratitude. My eldest brother Jeromy, thank you for googling an academic program called biosystems engineering ; your drive in following your o wn education still inspires me t o no end. Christian Smith helped me when the fieldwork became overwhelming, not for the pay, but for the company; thank you, brother. My partner, Celine deserves the credit for convincing me that returning to school to pursu e what I am passionate about, wa s worthwhile and entirely possible . v TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ........................ vii LIST OF FIGURES ................................ ................................ ................................ ..................... viii CHAPTER 1 : INTRODUCTION ................................ ................................ ................................ 1 RESEARCH GAPS ................................ ................................ ................................ ........................ 1 LITERATURE REVIEW ................................ ................................ ................................ ............... 2 Biology and Ecology ................................ ................................ ................................ ............... 2 Sensory Stimuli ................................ ................................ ................................ ....................... 4 Control Efforts ................................ ................................ ................................ ........................ 7 RESEARCH OBJECTIVES ................................ ................................ ................................ ........... 9 CHAPTER 2: ACOUSTIC FREQUENCY RESPONSE TRIALS ................................ ........ 10 INTRODUCTION ................................ ................................ ................................ ........................ 10 MATERIALS AND METHODS ................................ ................................ ................................ .. 11 Experimental Design ................................ ................................ ................................ ............. 11 Analysis Methods ................................ ................................ ................................ .................. 13 RE SULTS ................................ ................................ ................................ ................................ ..... 14 DISCUSSION ................................ ................................ ................................ ............................... 16 CONCLUSION ................................ ................................ ................................ ............................. 18 CHAPTER 3: ACOUSTIC LOCOMOTORY RESPONSE TRIALS ................................ ... 19 INTRODUCTION ................................ ................................ ................................ ........................ 19 MATERIALS AND METHODS ................................ ................................ ................................ .. 20 Experiment Design ................................ ................................ ................................ ................ 20 Analysis Methods ................................ ................................ ................................ .................. 24 Data Recording and Manipulation ................................ ................................ ........................ 26 RESULTS ................................ ................................ ................................ ................................ ..... 27 DISCUSSION ................................ ................................ ................................ ............................... 28 CONCLUSION ................................ ................................ ................................ ............................. 29 CHAPTER 4: ENGINEERING DESIGN OF SOFTVALVE TRAP ................................ .... 30 DATA DRIVEN DESIGN ................................ ................................ ................................ ............ 30 INITIAL DESIGN ................................ ................................ ................................ ........................ 31 VISION FOR FINAL DESIGN ................................ ................................ ................................ .... 32 CHAPTER 5: CONCLUSIONS ................................ ................................ ................................ 34 CHAPTER 6: CURRENT AND FUTURE WORK ................................ ................................ . 35 CURRENT WORK ................................ ................................ ................................ ....................... 35 FUTURE WORK ................................ ................................ ................................ .......................... 36 APPENDICES ................................ ................................ ................................ ............................. 37 vi APPENDIX A: CHAPTER 2 SUPPLEMENTAL MATERIALS ................................ .......... 38 APPENDIX B: CHAPTER 3 SUPPLEMENTAL MATERIALS ................................ .......... 67 APPE NDIX C: CHAPTER 4 SUPPLEMENTAL MATERIALS ................................ .......... 76 REFERENCES ................................ ................................ ................................ ............................ 83 vii LIST OF TABLES Table 2.1: Conover Test Results for Zones 1 and 2 ................................ ................................ ...... 14 Table 3.1 ................................ ..... 27 Table 3.2: Statistical Results of Speaker Proximity as a Factor of Daily Catch ........................... 27 Table 3.3: CPUE D uring Conventional Use ................................ ................................ ................. 28 Table A.1: Number of Trials ................................ ................................ ................................ ......... 38 viii LIST OF FIGURES Figure 1.1 : Adult male crayfish. The red raised spots on the sickle shaped chela (claws) and b lack stripe on the underside of the tail are characteristic of Procambarus clarkii. Photography by Douglas Clements. ................................ ................................ ................................ .......................... 3 Figure 1.2 : Mechanoreceptors can be seen on an antenna under a Keyence microscope at 100x magnification. Ph otograph by Amy Albin. ................................ ................................ ..................... 6 Figure 2.1 : The cattle bunk feeders provided a long and narrow habitat shape for conducting the lab - scale research. This picture was taken during cleaning of the sponge filter in zone 2, which explains its absence. A layer of door sealing ins ulation was used to create a seal around the uneven top of the bunk feeders, to prevent escape. ................................ ................................ ...... 12 Figure 2.2 : Flowchart of the process used to play sounds in habitats. ................................ .......... 13 Figure 2.3 : The populati on distribution for sound treatment is represented for each zone. The zero denotes the silent condition (contro l). ................................ ................................ ................... 15 Figure 3.1 : Trap locations spaced in a linear density of 5 meters. Distance from shore was dependent on wate r depth; traps were completely submerged. ................................ .................... 21 Figure 3.2 : Trap types include APART (top left), ART (top right) (Green et al., 2018), pyramid (Stanc liffe - Vaughan, 2015) shown in bottom midd le, and a deconstructed luffa (bottom right). ................................ ................................ ...................... 22 Figure 3.3 : Process flow for sound generation using batteries as a remote power source. .......... 23 Figure 3.4 : Map of trap locations, courtesy of Google Ma ps ................................ ....................... 23 Figure 3.5 : The particle velocity is asymptotic about zero as radius from the speaker increases. 25 Figure 4.1 : Rice farmer showing pyramid style tra p (Boyd, n.d.). ................................ ............... 30 Figure 4.2 : The top view shows two RSC that moved through the SoftValve to the bristle side of the aquarium (left). The profile view (m iddle) show the directionality of the Softvalve. A 3D rendering was crea ................................ ........................... 31 Figure 4.3 : An AutoCAD rendering of the initial tra p design. SoftValves are placed low enough on the frame for crayfish to have access. The galvanized steel mesh is shown as a clear sur face to better view the SoftValves. ................................ ................................ ................................ ........... 32 Figure 4.4 : A floating mat is attached to top of the trap. Modular systems such as battery bank, heating elements, speakers and anchors for drone retrieval are attached. ................................ .... 33 ix Figure 6.1 : A canula is in stalled using a gel glue (left) and the crayfish is placed in a 50 mL tube before anesthetizing. ................................ ................................ ................................ ..................... 35 Figure 6.2 now Traps (right). ................................ ................................ ................................ ................................ . 36 Figure B.1 : Recorded schedule used for tracking the treatment being used. ................................ 67 Figure C.1 : PVC pipes are tipped upwards using electromagnets to empty crayfi sh into a cage. 76 Figure C.2 : Crayfish enter the refuge style traps before a pump intermittently empties traps into a cage through a flapper valve. ................................ ................................ ................................ ........ 77 Figure C.3 : The third design uses a combination of semi - cylindri cal trap and refuge trap to direct crayfi sh toward a central floating cage. ................................ ................................ ........................ 78 Figure C.4 : This design uses refuge tubes with an angle at one end to act as a one - way valve for crayfish entry into a cage. An entrance similar to each en d. 79 Figure C.5 : The Dragon design is made of a corrugated plastic tube that coils up for storage and is uncoiled during deployment. Refuge tubes with flapper valves allow crayfish to enter into the corrugated tube. ................................ ................................ ................................ ............................. 80 Figure C.6 : Later termed the SoftValve, the concept is to use flexible bristles or fibers to allow a crayfish in one - way but not the opposing direction. ................................ ................................ ..... 81 Figur e C.7 : Blue prints of the current design were created using Autodesk Inventor. ................. 82 1 CHAPTER 1 : INTRODUCTION The invasion of Procambarus clarkii (Girard, 1852) , better known as Red Swamp Crayfish (RSC) has been witnessed world wide. The average observer might notice a small lobster crawling across a pond - side lawn , while the ecologist might see a destructive aquatic invader come to change the ecological landsca pe forever. Although extreme, the effects of an invasive species on an ecosystem can be deleterious. The RSC is no exception ; they are typically more aggressive than native species , fecund, and mobile. The introduction of the invasive crayfish results in t he loss of biodiversity and reductions in populations of finfish an d crabs (Moonga & Musuka, 2014) . Additionally, RSC can burrow under civil infrastructure causing dama ge to dams, reservoirs, and levees (Booy, Cor nwell, Parrott, Sutton - Croft, & Williams, 2017) . The RSC was reported to be present in Michigan in 2015 and was listed as an invasive species by the Michigan Department of Natural Resources (MDNR). As of October 29 th , 2019, over 20 water bodies were infes ted. The MDNR previously de veloped a Red Swamp Crayfish Response Plan which helped to inform them on response efforts should P. clarkii invade Michigan. The goal of this research is to investigate and conclude engineering design s to assist the MDNR in cont rol and eradication of the Red Swamp Cray fish (RSC). Knowledge gaps were found during a literature review covering the understanding of the biology, ecology, invasive habits, and current trapping techniques. RESEARCH GAPS It was critical to understand th e biology and ecology of RS C, so as to ut ilize these properties to conclude an engineering solution. A method of luring RSC would be useful and would require an understanding of their sensory organs. No studies were found relat ed to the use of sensory stim ul us as a lure for RSC or a ny other crayf ish , aside from different food baits . Sensilla which could be tested included olfactory, optical, thermal, and auditory. In addition, 2 predator prey studies are lacking information on Michigan native species . The Fis heries and Wildlife department at MSU is already conducting trials to determine which native fish species will consume RSC. LITERATURE REVIEW A literature review was completed to better understand the context of the engineering problem of controlling inv asive populations and provide some inform ation to guide solutions. The topics studied were biology and ecology, including sensory stimuli, and p ast control efforts relating to RSC. From the literature review, the knowledge gaps were identified, and researc h objectives were created. Biology and E cology RSC are invertebrates under the order Decapoda , family Cambaridae , genus Procambarus and specie s clarkii . The RSC typically has a unique phenotype from other crayfish species. They have a dark red carapace, with claws (chelae) reaching out in front of them as seen in F igure 1 . 1 . Red spots are common on the chelae, but the color is not a reliable predictor of species for this crayfish as many juvenile RSC are not red (Boets, Lo ck, Cammaerts, Plu, & Goethals, 2009) . The adult length ranges from 5.5 to 12 centimeters. 3 Figure 1 .1 : Adult male crayfish. The red raised spots on the sickle shaped chela (claws) and black stripe on the underside of th e tail are characteristic of P rocambarus clarkii. Photograph y by Douglas Clements. Native to the United States, RSC have habitat along the Gulf of Mexico between Mexico and Florida. However, as of 2019, infestations have begun along the west coast, east co ast, and in the northern mid - west (Nagy, Fusaro, Conard, & Morningstar, 2019) . Worldwide distribution of c rayfish has led to infestations in Europ e, Africa, South America, and Asia (Hobbs III, 1993; Holdich, Gydemo, & Rog ers, 2017) . Many means of introduction exist, including live fish bait, aquarium trade, biological supply to laboratories and classrooms, and live seafood markets (Kilian et al., 2012) . RSC have an omnivorous diet and consume plants, snail s, macrophytes, insects, and detritus (Gherardi & Barbaresi, 2007; Hobbs III, 1993) . Studies have found that adult RSC preferentially feed on p lants and detritus, by volume. Conversely, juveniles consume mostly animal matter in the form of insects, gastropods (snails), and fi sh (Correia, 2003) . Due to the aggressive nature of the species, RSC out co mpeted native crayfish for territory and food (Gherardi & Cioni, 2004) . 4 Crayfish are ecosystem engine ers characterized by their burrowing. The RSC can dig burrows extending up to 90 cm below the water table (Ingle, 1997) . Burrows a re most commonly found in areas with fine sediment and are less prevalent in areas with sand and harder soil substrates (Barbaresi, Tricarico, & Gherardi, 2004) . The RSC are nocturnal, mostly active in the nighttime immediately after the sun sets. Male crayfish exhibit dimorphism, expressing a form 1 sexually active physicality or a form 2 sexually inactive physicality. Form 1 males can be characterized by the stiffness a nd definition of their gonopods, and usually grow, through calcification, sharp hooks on their walking legs. Form 2 males do not have hooks on their walking legs and have le ss shapely definition in their gonopods. Males show different behaviors characteriz ed by their locomotion . One behavior type is that of form 1 males in a mate - seeking phase. This phase is characterized by bursts of h ighspeed movement. The second behavior t ype is characterized by an immobile stage during which the crayfish hides in its bu rrow only coming out at night to forage (Nagy et al., 2019) . Sensory Stimuli It is important t o understand what motivates RSC, in orde r to inform an engineering solution to the eradication and control of the species. The crayfish eat and mate of course, but even those motivating activities are sensed and communicated beforehand. Therefore, and unde rstanding of the sensory organs and proc esses should reveal testable solutions to affecting locomotory responses. Though it is known that marine crustaceans create and respond to sounds, little is known about the effects of sound on crayfish locomotion (Edmonds, 7AD) . RSC do not have hollow, air - filled organs to hear with, but instead perceive sound through hair - like sensory structures called mechano - receptors shown in F igure 1 .2 (Popper, Salmon, & Horch, 2001) . Therefore, the pres sure of sound are not the direct cause of stimuli to the sensilla, but instead the particle 5 velocity is responsible (Goodall, Chapman, & Neil, 1990) . RSC emit sounds composed of wide - band frequency pulses lasting 0.4 millisecond with a 20 kHz RMS bandwidth, peaking at 28 kHz. Maximum SPL PK (intensity) of the signal is 146 dB relative to 1 µPa. Such sounds occur during tail flips, fighting, and encountering even t s. Since the sound carries efficiently through water, it is thought that dominance and territory may be communicated at distance through acoustics (Buscaino et al., 2012) . This is of importance, since peak sensitivity to hydroacoustic stimuli was determined to be at frequencies below 150 Hz (Breithaupt & Tautz, 1990) . An experiment was conducted by an undergraduate engin eering group at Michigan State University, to determine the effect sound stimuli had on RSC locomotion . V arious pure tone frequencies between 20 Hz and 500 Hz were tested using a speaker modified for und erwater use and it was found that the sound had a sig nificant attraction effect on crayfish locomotion , especially in the 500 Hz trials. Since the highest locomotory response came from the maximum frequency tested, the group recommended further testing wit h a higher range of frequencies (Ausmus, L, Kontorousis, A, Li, B, Tang, 2018) . 6 Figure 1 .2 : Mechanoreceptors can be seen on an antenna under a Keyence microscope at 100x magnification. Photograph by Amy Albin. The age of a crayfish and light expo sure affect which parts of the spectrum crayfish most readily see. An electroretinogram study showed that juvenile crayfish had a highe r response (voltage presence) to ultraviolet and blue light than do the adults. Adult crayfish have a higher response to red and green light. Spectral sensitivity is dependent on whether crayfish have been exposed to dark or light; adults only respond to u ltraviolet light when dark - adapted and shifts to red sensitivity when light - adapted . The study suggests that short wavele ngth and long wavelength receptor cells change proportion as juveniles become adults (Fanjul - Moles & Fuentes - Pardo, 1988) . The RSC primarily communicate dominan ce hierarchies through use of olfactory sensory organs. Crayfish emit olfactory stimuli through urination. Olfactory sensors on the ant ennules detect the excreted chemicals. Crayfish with the sensory organs removed fight each other more often, while crayfi sh with the sensory organs avoided fights, thus proving that olfactory stimuli control aggressive behavior (Horner, Schmidt, Edwards, & Derby, 2008) . Olfactory stimuli are 7 en hanced through the creation of jets of water which draw od ors in toward the antennules , which have a high concentration of sensilla . The odors would otherwise be limited to natura lly occurring currents and molecular diffusion (Denissenko, Lukaschuk, & Breithaupt, 2007) . The effects of heat stimuli on the heartrate of crayfish ( Cherax destructor ) w ere studied in cooling and warming environments. The heartrate drops significantly faster than body temperature du ring cooling events and increased slower than temperature during heating events (Goudkamp, Seebacher, Ahern, & Franklin, 2004) . No litera ture was found considering locomotory response of crayfish with respect to heat or infrared light. Control Efforts Current cont rol strategies vary due to the nature of the infested water bodies. Rivers and streams with continuous flow are not necessarily a good fit for chemical treatments as the residuals would wash down stream. Likewise, a large lake may be costly to trap intensi vely. Each water body will bring a different set of challenges to control strategies. The MDNR response efforts include the ulti mate goal of developing a framework to classify waters for different treatment types. Intensive trapping is the practice of pla cing traps in a high - density arrangement to trap a large quantity of the population. A common type of trap used for trapping is a semi - cylindrical minnow trap, although many geometries exist. Between dip - netting, Fyke - netting, cylindrical traps, and semi - c ylindrical traps, the semi - cylindrical wire mesh traps have the highest capture rates per unit effort (CPUE). However, variation s in habitat also influenced CPUE. Sex selectivity can be an issue with trap geometries; however, the semi - cylindrical trap does not appear to select one way or the other (Paillisson, Soudieux, & Damien, 2011) . One research team tested collapsible mesh netting traps, among many 10 other trap types, and found that each trap captures distinct size categories of crayfish, sex ratios, and quantity of bycatch. The team 8 conclude d that a combination of trap types might be useful for intensive trapping to cover all size ranges (De Palma - Dow, Curti, & Emi Fergus, 2020) . Artificial refuge traps were tested based on an indigenous method of crayfish capture using brush piles; the design used PVC tubes to mimic a cra yfish burrow. The benefit s are that artificial refuge traps do not require bait, and can catch egg - bearing females when they seek shelter for protection (Green, Bentley, Stebbing, Andreou, . Intensive trapping requires a large amount of human effort to reduce populations and must be maint ained or populations will retu rn to previous levels within a couple of breeding cycles (Holdich et al., 2017) . Sterile Male Release Technique (SMRT) was tested as a means to control populations. In SMRT, m ale crayfish are exposed to X - rays , which reduce the size of their testes and alter spermatogenesis. The result of SMRT is a 43% reduction of offspring from females that mated with irradiated males (Aquiloni et al., 2009) . The efficacy of such a treatment is yet to be field tested, but the expected impact is low compared to other control strategies (Holdich et al., 2017) . Pesticides have been used with some success. Biocide is the term used to describe pesticides targeting invasive organisms. Biocides work best for smaller bodies of water were biocide quantity does not have to be costly to achieve lethal doses for invasive crayfish. Biocides are not specific to a species and can harm native crayfish and other organisms. In addition, accumulation and magnification of toxins can cause undesirable results. Some biocides include organophosphate, rotenone, surfactants, a nd pyreth roids insecticides. Trends in chemical treatment lean toward low environmental persistence, since selectivity of biocides does not yet exist for crayfish (Holdich et al., 2017) . One promising biocide i s emamect in benzoate which has been used to force molting of egg - bearing American lobster, thus aborting the eggs in the 9 process. However, experimental trials have not been concluded (Freema n, Turnbull, Yeomans, & Bean, 2010) . Predator prey studies with various organisms have been documented. A potentially effective predator is the European eel ( Anguilla anguilla ). European eels tend to consume crayfish under 45 mm in length, of which nor mally tend to be trap - shy (Aquiloni et al., 2010) . While they may be effective in con trolling the population in combination with traps, biological controls come with a caution. The eels themselves may create another ecological problem if , with preferenc e given to native species . RESEARCH OBJECTIVES In order to assist the MDNR with their goals of RSC population control and eradication, this study seeks to develop a better understanding of locomotory responses to auditory, olfactory. and thermal stimuli. Specifically, the objectives for this study are to: i) determine the locomotory response of various sound frequencies on RSC; ii) test the ability of sound to enhance intensive trapping of RSC; iii) test the ability of a heat source to attract RS C in lower temperature water; and iv) design a solution to enhance the control of RSC. The results of this study will provide options for an engineering solution to controlling RSC as an invasive species. Additionally, the benefits of this research could r each beyond control measures for invasive populations; a n ovel lure could provide innovation to crayfish farmers worldwide. 10 CHAPTER 2 : ACOUSTIC FREQUENCY RESPONSE TRIALS INTRODUCTION It was clear from the literature review that hydroacoustics, or underwa ter sound, would be a good starting point for experimentat ion. RSC create and sense a variety of different frequency sounds; a locomotory response using different frequencies was needed to consider hydroacoustics as a possible solution to RSC population co ntrol. Little research on the subject of sound stimulus on RSC exists , but what does exist seems contradictory. Specifically, the high sensitivity of RCS to detect sound in a lower band of frequency (<150 Hz) is at odds with what the Crayfish Will group fo und to be the best frequency to attract a locomotory respo nse (500 Hz) (Breithaupt & Tautz, 1990) . However, if one considers the physics of sound propagation , an explanation may exist. Higher sensitivity at lower frequencies may be required to sense the lower energy sounds. As frequency increases, the energy required to produce the soun d increases. Th e speaker power drives the wave intensity, which drives the pressure level of the waves, which in - turn drives the particle velocity of the water. This particle velocity is the measureand of the crayfish mechanoreceptors, unlike the pressure level, which hu mans perceive. So, at higher frequencies, higher energy levels are required, and therefore, the crayfish may not need a high sensitivity at such high particle velocities. Since the attraction trial results showed the highest level of attra ction at the ma ximum frequency tested, further testing was completed to expand the range of frequencies. An experiment was designed to verify the results of Crayfish Will , and to expand the range of frequencies tested. Additionally, many naturally made noi ses are in a cl assification called colored noises. Two such noises were studied: white noise, and pink noise. Pink noise is characteristic of waterfalls and other sounds which have equal energy per octave. White noises 11 are characteristic of a randomly gene rated noise wit hin a spectrum, and usually require added energy such as a fan or car tires driving across a road. Using a modified method to the Crayfish Will study, an experiment was designed to further test sound frequencies. In a long aquarium, sound w as played at on e end, to stimulate the crayfish. After a period of time, the population distribution was recorded and compared to the pre - stimulus distribution. The hypothesis was that at higher frequencies, a higher population distribution would occur nea rest the speake r, thus, showing a sort of attraction effect to the sound treatment. The null hypothesis w as thus a population distribution similar to that in a silent condition, which was used as the control. The second hypothesis is that the noises will h ave a higher im pact on population distribution than the pure tone frequencies. Both hypotheses were tested using the variance from the population distributions recorded during the silent condition. The conclusion from a frequency response trial would addre ss Objective i) , determine the locomotory response of various sound frequencies on RSC. MATERIALS AND METHODS Experimental Design Laboratory trials were conducted in 3 artificial habitats made from cattle bunk feeders. Pea gravel was used to level out the slope in the h abitat. Each habitat contained 3 sponge filters with air sponges attached to clean the water and maintain suitable oxygen levels for the crayfish. Extruded polystyrene lids were used to prevent RSC from escape and to block out light. Each ha bitat was divid ed into 3 zones of equal area . Three zones were chosen for ease of recording population distribution, because of the difficulty of counting a large quantity of moving crayfish. Three 8 cm lengths of 3.8 cm diameter clear flexible tube were a dded to each zo ne for artificial habitat. Preliminary habitat set - up is shown in F igure 2 .1. 12 Figure 2 . 1 : The cattle bunk feeders provided a long and narrow habitat shape for conducting the lab - scale research. This picture wa s taken during cleaning of the sponge filter in zone 2, which explains its absence. A layer of door sealing insulation was used to crea te a seal around the uneven top of the bunk feeders, to prevent escape. Between 16 and 20 crayfish were placed in each ha bitat, and randomly exchanged between habitats after each trial. RSC were sourced from Carolina Biological Supply Company. RSC were uns exed and varied in size from juvenile to adult. A Lube l l Labs UW30 30 - Watt speaker was placed into Zone 1 in each tank. T h e speaker was powered by a Bogen CC4301 amplifier. A Dell PC sent audio signals to the amplifier using Windows Media Player. Sound file s were generated using Audacity (Figure 2.2) . Pure tones sound files used a sine waveform and were exported as an MP3 fi l e with 320 kbps quality. Noise files were generated using a built - in function and both high - pass and low - pass virtual 5 th order Butterworth filter s w ere added 13 to each. Sound treatments included the following pure tones: 500 Hz, 1 k Hz, 2 k Hz, 3 k Hz, 4 k Hz, 5 k Hz, 6 k Hz, 7 k Hz, 8 k Hz, 9 k Hz, 10 kHz, 11 kHz, 12 kHz, 13 kHz, 14 kHz, and 15 kHz. Frequency bands of white noise wer e tested in 4 different frequency ranges, 1 - 5 kHz, 5 - 10 kHz, 10 - 15 kHz, and 1 - 15 kHz. Each treatment was tested a number of times as s h own in Appendix A . The quantity of RSC in each zone was recorded before the treatment and immediately after the treatment . Two habitats were chosen to use the same treatment, while the third habitat was the control without a sound treatment. The control a l ternated habitats for each 24 - hour trial. Treatments lasted for 24 hours and crayfish were fed and left without sound for another 24 hours. Habitats were cleaned according to a strict schedule (Appendix A ). Figure 2.2 : Flowchart of the process used to p l ay sounds in habitats. Analysis Methods The response variable for the analysis was the percent of total population of RSC in each habitat (population distribution), for each zone. The independent variable was the sound treatment used during the trial. Th e population distribution for each zone was tested for a normal distribution using a histogram for visual reference, a Shapiro Wilk test for normality, and a Q - Q plot. Data in zones 2 and 3 were not normally distributed as seen in Appendix A. Therefore, a n on - parametric test (Kruskal Wallis Rank Sum) was used to determine if frequency affected population distr ibution. Significance was set to P<0.05. A Conover Test was used for zones 1 and 2 to determine which frequencies had the largest impact upon populati o n distribution as compare between treatments with and without sound. The statistical analysis was complet ed in the programming language R. 14 RESULTS Overall, sound t reatments had a statistically significant effect on Zones 1 and 2, but not on Zone 3. The r e spective p - values were 0.008784 , 0.01737 , and 0.1557 . F igure 2 . 3 shows a bar chart of the population distributions of individual sound treatments . The Conover Test values with P<0.05 are listed in Table 2.1. No te that adjustments were made to the p - values using the Conover Test. Adjustments using the Benjamini & Hochbe rg method (1995) changed some previously significant values to nearly significant (p~0.05) and some values as not significant. It is important to note that near significance should not be dis m issed in the case of false negatives in the analysis . However, t ype I error was introduced when a method for adjustment was not used . The full analysis was completed using R Markdown (Appendix A ). Table 2.1: Conover Test Results for Zones 1 and 2 . Treatment (Hz) P P - adjust Zone 200 0 - control 0.0053 0.0514 1 3000 - control 0.0248 0.1333 1 6000 - control 0.0188 0.1132 1 11000 - control 0.0115 0.0912 1 12000 - control 0.0040 0.0508 1 W N6 k10k - control 0.0046 0.0528 1 WN10k15k - control 0.0052 0.0531 1 2000 - control 0.0247 0.1098 2 6000 - control 0.0146 0.0820 2 8000 - control 0.0234 0.1078 2 14000 - control 0.0228 0.1069 2 PN1k15k - control 0.0178 0.0902 2 15 Figure 2 . 3 : The population distr ib ution for sound treatment is represented for each zone. The zero denotes the silent condition (control) . 16 DISCUSSION From the pure tone frequencies, 12 kHz had the highest population distribution in Zone 1. B oth high and medium ranges of white noise had th e highest population distribution of all noise treatments. The high and medium frequency ranges of white noise showed a locomotory response that hints at an attraction, since a higher population distribution was closer to the speaker and a lower distribu ti on was in the farthest zone from the speaker. The high range showed a lower population distribution in Zone 2 than the medium range. Zone 3 showed high variability in the medium range, but a lower mean distri bution than in the high range. Aside from 2 kH z and 12 kHz, p ure tone frequencies and pink noises underperformed compared to the two highest white noise ranges. The attraction type effect of white noise may come from an interest in catching prey who also make such noises, such as turbulent movement th rough water. Further investigation is required to determine the cause. For the first hypothesis, the null hypothesis was not rejected, however, 12 kHz was near the higher end of frequencies tested. The resea rch once again showed that the highest populat io n distribution differences occurred near the top of the range of frequencies tested. The speakers used had a poor frequency response above 15 kHz. Thus, further testing should be done with speakers that have a higher frequency response range. For the sec on d hypothesis, the null hypothesis was not rejected. However, at medium and high frequency ranges, white noise had a higher population distribution than the pure tone frequencies within the range, except for t he 12 kHz pure tone. The locomotory response fi ndings bring question to their relationship with frequency sensitivity findings using methods such as - conditioning technique. Offutt found peak sensitivity at 75 Hz for the Americ an lobster ( Homarus americanus ) between 10 - 150 H z (Offutt, 1970) . Similarly, Breithaupt used a vibrat io n chamber to test sta tocyst 17 vibration sensitivity of Orcenectes limnosus and found optimal sensitivity at the lowest frequency tested 3 Hz (Breithaupt & Tautz, 1990) . Since the locomotory response was maximized at relatively high frequencies, the frequency may play less of a role than the particle velocity. A similar test with variety o f particle velocities could elude to the relationship. Indeed, Mark Plumber and Jürgen Tautz studied the effect of water vibrations on 9 interneurons and found that RSC were not sensitive to high frequency sounds (greater than 400 Hz). All low pass inter ne urons were inhibited by stimulus above 100 Hz. Broad band neurons were sensitive up to 80 Hz, but high pass neurons respond poorly above 60 Hz (Plummer, Tautz, & Wine, 1986) . Therefore, the high frequency white noise may be inhibiting t he ability to sense low er frequency noises nearby. Without the ability to sense predators or prey with their mechanosensors, RSC are limited to olfactory and optical sensory input. Since crayfish are more active in the darkness, even vision becomes unrelia bl e , since the experime nt was performed in complete darkness . An on - going study using manganese - enhanced magnetic resonance imaging to study the nervous responses of RSC to food and sound will shed light on mechanisms of different capture strategies. N o consideration was mad e to varying the amplitude (particle velocity of water) during the experiment. The habitat s used had very limiting boundary conditions for sound propagation. The plastic walls and pea gravel floor, as well as the shallow water surfac e allowed sound to spread in an unorganized fashion. A hydrophone was used to record the 1000 Hz signal at various distances from the speaker. Noise was present at relatively low levels outside of the 1000 Hz frequency . However, the sensitivity at which RS C will show a locomotory response is unknown and the unintended sound reflections could have been a factor in the response variable. 18 CONCLUSION The best frequenc ies for attracting crayfish w ere the medium and high range of white noise and the 12 kHz pure t one . Since the high range white noise had less variability than the medium range white noise and 12 kHz pure ton e, and nearly identical population distribution , it was chosen to be used for the acoustic locomotion response trials. Thus, research objectiv e i ) was completed. Because the high range of white noise performed the best, higher pure tone frequencies and ra nges of white noise should be tested to find an optimal frequency . However, testing may be limited by the frequency response range of the spea ke rs used. If similar speakers are used for further testing, the signal should first be recorded with a hydrophone and a spectral analysis completed across each pure tone to confirm the quality of the signal above the speaker response range. Additionally, pa rticle velocity should be varied and recorded to create a relationship between locomotory resp onse and signal amplitude. 19 CHAPTER 3 : ACOUSTIC LOCOMOTORY RESPONSE TRIALS INTRODUCTION After testing out sound frequencies in a laboratory setting, a more re al istic test was needed to understand how the population distribution in an infested pond will b e affected by underwater sounds. The laboratory habitats did not create a realistic spatial crayfish density ( crayfish per area) or boundary conditions for soun d propagation. In particular, the habitats were too small to view a gradient in the amplitude of the soundwave. Sound intensity drops as it travels away from the source or reflects off a boundary such as the surface of the water or bottom of the pond. In d es igning a solution that involved sound emission, it would be critical to know the effective amp litude needed in order to attract crayfish , as this information could help to determine power consumption and speaker specifications. Trap research was already b eing conducted since the beginning of the infestation by the The research focused on early detection, trap densities, and testing various trap types. The most commo n opening to allow for larger crayfish to enter. Th e ongoing research was an opportunity to test the ability of sound to attract crayfish outside of a l aboratory setting. T o address objective ii ), test the ability of sound to enhance intensive trapping of RSC, an experiment was designed to determine the effects of sound stimuli on intensive trapping of RSC . Originally, two infested pond s in Michigan w ere used as the testing si te . The response variable for the research was the mean capture per unit effort (CPUE). A unit of effort was defined as one crayfish trap, used for a duration of one day. Therefore, the CPUE was the daily amount of crayfish caught in one trap over one da y. The manipulated variable was the bait treatment used. Standard baited traps use dogfood as an inexpensive but effective bait. Four treatments were 20 used, including no bait, sound, sound and food bait, and only food bait. Other uncont rollable variables in cl ude d weather, water temperature, water depth, and the ponds p hysi ological characteristics . The hypothes e s are as follows: 1) Sound will have a statistically significant impact on mean CPUE for both treatments which include sound (p< 0.05). 2) Traps close r to the underwater speaker will have a higher CPUE than traps further from the speaker. Because different trap types will be tested, another hypothesis can be made about how trap types are a ffected by sound. Three trap types are not baited while two trap ty pes are baited with dog food. The trap types which do not use bait are refuge style traps , which create safe places for crayfish to hide. The third hypothesis is as follow: 3) Refuge traps will increase in CPUE nearer to the speaker. While the number of ea ch trap types are uneven (the vast majority MATERIALS AND METHODS Experiment Design Field trials to ok place in an infested stormwater retention pond in Novi, Michig an (42.442431, - 83.434977). The pond was characterized by a soft, silt bottom and the bank was a spatial mixture of clay and cobble. The pond was shallow, estimated at less than 1.5 m depth a t the center. Heavy microbial matting suggested nutrient - rich wat er . A population of Pimephales promelas (fathead minnows) was present in the pond, in addition to the RSC. The shape of the pond was a long oval with an approximate major diameter of 118 m and minor diameter of 32 m. ra ps was performed for the previous 3 trapping seasons. T he RSC present had sizes ranging from juvenile to large adult (> 50 mm carapace length). The perimeter of the pond was lined with 57 crayfish traps, spaced 5 m apart (Figure 3. 1 ) . 21 Figure 3. 1 : Trap l ocations spaced in a linear density of 5 meters. Distance from shore was dependent on water depth; traps were completely submerged. with expanded openings , a pyramid shaped trap with a sing le opening at the top , and three custom designed artificial refuge type traps . Traps , which filled in the remaining 37 trap locations in the pond . Two of the artificial refuge trap types were co nstructed from varying lengths and diameters of PVC pipe, attached together with adhesive and a fine mesh screen to allow water to flow out of one end. One configuration of arran gement was d en oted artificial refuge trap (ART ). The second configuration was made of two horizontal configurations stacked on top of one another and was referred to as APART (short for artificial apartment). The third trap was made from a deconstructed lu ffa and a st ai nless - steel nut , with the intent to capture juvenile crayfish, by providing a safe place for them to hide from predation . These were called juvenile traps. All three custom traps were designed to trap crayfish by means of providing an artific ial protecti ve space; these traps were not baited with food during any treatment. Trap types are shown in Figure 3. 2 . 22 Figure 3 . 2 : Trap types include APART ( top left) , ART (top right) (Green et al., 2018) , pyramid trap (bottom lef t) (Stancliffe - Vaug han, 2015) shown in bottom middle, and a deconstructed luffa (b ottom right) . Stakes were labeled by trap location number 1 - 57 beginning at the northeastern most point and numbered in a clockwise manner. Figure 3. 3 shows the process for sound generation. An LL916C - 100 UW underwater speaker was placed 1 meter from a d e signated trap location (Lubell Labs, Inc). The speaker was powered by 3 Duracell SLI31MDC 12 - volt deep cycle batteries through a CA - 160R TOA 60 - Watt am plifier (TOA Electronics, Inc). A SanDisk Clip Jam MP3 Player was connected to the amplifier and contain e d a 10 - minute sound file with a 10 - 15 kHz band of white noise. The white noise sound file was generated in Audacity software using a high - pass and low - pass 5 th order Butterworth filter for 10 and 15 kHz, respectively , with a sound quality of 320 kbps . The sound file was looped for the duration of the sound treatments. The speaker location was moved 3 times throughout the trapping season. 23 Figure 3. 3 : P rocess flow for sound generation using batteries as a remote power source. Distance to each trap locatio n was measured from the speaker location using Google Maps in combination with Autodesk AutoCAD 2020; measuring accuracy was confirmed using a meter sti ck to measure objects on either end of the major diameter with overall readings ±10 cm (Figure 3. 4 ) . Sin c e traps moved around freely on 0.5 - meter twine, the accuracy of trap location was accurate within 1 m. Figure 3. 4 : Map of trap locations, courtesy of Google Maps . 24 Four treatments were tested: 1) Sound and Food bait, 2) Food bait only, 3) Sound only, an d 4) No treatment. Dog food was used as a food bait in treatments 1 and 2. Each treatment was performed for a 24 - hour period, starting in the m orning, and ending the next morning. Traps were emptied each day and the trap type, number of crayfish caught, an d trap location were recorded. Methods were approved by the MDNR before experimentation began. Analysis Methods Due to differences in trap typ es (baited versus not baited), treatment analysis was only in now Traps. The first step of the analysis was to determine which metric to use to analyze efficacy of each treatment . Since the initial population of RSC in the pond was unknown, CPUE was used as the response variable . The second step was to determine if t he da ily catch data were normally distributed. A histogram was used to visualize the CPUE and a Shapiro Wilk test was used to te st normality of the data. Since the number of samples was suitably large (n>20), the Central Limit Theorem justified the use o f parametric tests such as the analysis of variance ( ANOVA ) . However, non - parametric tests were also used to confirm results. The Kruskal Wallis Rank Sum Test was used to verify the ANOVA. Significance was defined as p<0.05 . Given significance, a post hoc te st (pairwise Wilcoxon Test) was used to c ompare treatments. The third step was to determine role of the variables: treatment , di stance to the speaker, and t rap type as a factor of CPUE . In addition to treatment, d istance from the speaker was also impor ta nt to analyze, as this would help to determine the minimum particle velocities that RSC could perceive acoustic signals with . Calculations were performed to see if a zoned analysis was necessary, due to any abrupt c hanges in particle velocity over the le ng th of the pond. Sound intensity , J was calculated using an inverse square law relationship (Equation 1) . P is the power of the speaker in Watts, while r is the radius from the speaker. 25 E q. 1 Then, acoustic impedance , Z was calculated using Equation 2, where i s the density of water at 20ºC and c is the speed of sound in fresh water at 20ºC. 1482.66 m/s was used for c (Greenspan & Tschiegg, 1 0A D) and 998 kg/m 3 was used for (Moore & Fierro; Nyer, 2008) . Eq. 2 Finally, particle velocity , v particle was calculated using Equation 3. Using MatLab ® (Mathworks, Natick, MA) , Particle velocity was plotted versus radius from the speaker in meters to determine if any abrupt drop - off would occur that would justify using distance intervals for analysis (Figure 3 . 5 ). Eq. 3 Figure 3 . 5 : The particle velo city is asymptotic about zero as radius from the speaker increases. The only notable change in particle velocity happens between 0 and 5 m radius from the speaker. There was not any logical reason to run a zoned analysis, because traps were spaced 5 m 26 apa rt, and a separate zone for the significant change would only contain one trap. Therefore, a single factor ANOVA and Kruskal Wallis Rank Sum Test was used to determine if distance from the speakers had a significant effect on CPUE on a subset of data using sound treatments. The coding language R was used to perform all statistical analysis (Appen dix B ). Data Recording and Manipulation Data were intended to be recorded following a biweekly schedule. During the first week, the sound and food bait treatment w as used Monday through Tuesday, and sound was turned off for Wednesday and Thursday. In the second week, food bait was removed for Monday and Tuesday, and the sound treatment was used until Wednesday, where the no treatment condition was used u ntil Friday. Weekend data was not used for analysis because they were not actively trapped until Monday morning. Due to numerous schedule changes, the actual trapping schedule was modified as challenges arose. E quipment failure and unforeseen events compr omised much of the collected data. Originally, an additional infested pond was used to provide a different type of physiography; however, the water level dropped so low during the dry season that the perimeter of the pond was reduced and the linear trap de nsity became too variable for a reasonable comparison. Therefore, only one pond was used for the analysis. The intended trapping schedule was modified when flooding events washed traps into unsafe depths of the pond. July 8 th and 9 th saw 30 traps missing f rom analysis . Treatments changed schedule when technological failures occurred. When the MP3 player failed to loop the 10 - minute sound file on the first week of testing, the treatment type was switched to food instead of food and sound . The trapping schedule can be fo und in Appendix B . 27 RESULTS A s ingle factor ANOVA showed the effect of treatment on daily catch rate was statistically significant (p = 3.57e - 4 ) . Similarly, the Kruskal Wallis Rank Sum Test confirmed statistical significance ( p=6.892e - 6). Table 3.1 shows a data summary of the treatment type with associated CPUE. The pairwise test results can be viewed in Appendix B. The largest difference occurred between no treatment and food and sound . Table 3 . 1 : Summary of T reatment and CPUE . Treatm e nt Total RSC CPUE n Food 711 0.6745731 1054 Sound 56 0.6436782 87 Food & Sound 547 0.8200900 667 No Treatment 136 0.4874552 279 Distance from the speaker was not statistically significant as a factor of daily catch rate for Sound , Sound & Food, or a combination of both treatments as seen in table 3 . 2 . The Kruskal Wallis Rank Sum Test results agree in all ANOVA tests. Table 3 . 2 : Statistical Results of Speaker Proximity as a Factor of Daily Catch . Treatment Test p - value 2 df p<0.05? Sound ANOVA 0.5960 - 43 No Sound Kruskal Wallis 0.3579 45.769 43 No Sound &Food ANOVA 0.4760 - 96 No Sound &Food Kruskal Wallis 0.3505 100.73 96 No Both ANOVA 0.0971 - 96 No Both Kruskal Wallis 0.0710 117.06 96 No Trap type was found to be a statistically significant factor of daily catch rate during conventional use (Food Treatment) . The ANOVA and Kruskal Wallis Rank Sum Tests yielded p= 4.22e - 11 and p= 8.94e - 15 , respectively. The CPUE for each trap was calculated as shown in Table 3 . 3. 28 baited. A pairwise comparison was completed using a Wilcoxon Rank Sum Test. Every trap comparison was significantly different from each other(p<0.05), excluding the pyramid an d - value was 0.153. Table 3 . 3 : CPUE During Conventional Use . Trap Type CPUE Apartment 1. 69 1.5 6 ART 1. 10 1. 49 0.67 1.12 Juvenile 0. 53 1.95 Pyramid 0.84 1. 23 DISCUSSION The treatment had a remarkable impact o Minnow Trap without sound still captures crayfish with a CPUE of 0.487. This implies that without bait incentive (food, sound or otherwise), RSC are still captured effectively . Food bait showed a 38% i ncrease i n CPUE, while sound alone added a 32% increase. The food and sound combination increased CPUE by 68% from the control. Therefore, the increase in CPUE from sound and food seem to be additive. Both ART and Apartment trap types had high CPUE relati ve to th e baited traps. The creator of the ART trap found similarly larger CPUE relative to baited traps. However, she was trapping signal crayfish, which also tend to be aggressive. Despite the aggression, multiple adults, namely females and smaller crayf ish woul d be found sharing the same refuge tube (Green et al., 2018) . This was not true for the RSC adults, who were not observed sharing tubes. Juvenile crayfish were found beginning in early August, and late into Sept ember, s h aring tubes . This wave of juveniles may have skewed the mean CPUE . However, the refuge traps show the value of refuge as an attractant. Similarly, the no treatment similar attractant, since no food or sound bait was used. 29 CONCLUSION Two conclusions were drawn from this analysis. First, t he effectiveness of refuge as a means of attraction is important to RSC capture , shown by the refuge trap types and further reinforced by the control treatment for T raps. Se c ond, both sound and food bait traps. The se conclusions will inform the de sign process to determine a n enhanced capture solution. Further studies should measure part i cle velocity as an analog for sound intensity , to find out what the minimum energy requirements will be for a speaker system in the future. Additionally, more variety of ponds should be used to determine the effects of refuge and sound in the presence of a bundant refuge , vegetation , and effects on no n - target species . 30 CHAPTER 4 : ENGINEERING DESIGN OF SOFTVALVE TRAP DATA DRIVEN DESIGN Based on the data from both lab oratory and field tests, an engineering solution of RSC capture was designed. It is import an T rap is not the industry standard for crayfish farmers in the southern United States. Crayfish farmers need more storage volume and an incr eased speed of harvest . They use a galvanized steel pyramid geometry trap with hole s where the corners would be, and a smooth, PVC tube coming out of the top to keep crayfish from escaping (Figure 4.1 ) . Traps can be emptied in one fluid motion, turning the m upside - down and giving them a shake. These traps work well for uniform depth wate r levels, such as flooded rice fields. However, the water depths of infested Michigan sites vary greatly , and their ability to stack for transportation and seasonal storage is poor. Figure 4.1 : Rice farmer showing pyramid style trap (Boyd, n.d.) . c a tch rate without any bait , perhaps because they supply shelte r . Additionally, t he PVC traps show how 31 effective refuge traps can be, but their design leaves a major limitation : trapping capacity. Therefore, the design process used principals from both the m etal mesh style baited trap types, and the PVC refuge trap types . The first step was to generate many potential solutions which can be found in Appendix C . After considering Soft V prototyped . A c oncept model was made, to be tested in an aquarium. The Soft V alve model was made from clear PVC hose, soft PVC bristles, and a waterproof epoxy as seen in F igure 4 . 2 . The conc ept is that a crayfish could enter the non - bristled side, pass the bristles as t h ey easily bend out of the way, but cannot re - enter the bristled ends, as the bristles support each other to resist the movement. The Soft V alve was attached to an aquarium divi der to test the efficacy. Food was placed on bristled side of the aquarium and c r ayfish were placed on the other side of the divider. While no formal experiment was conducted, the Soft V alve was able to allow crayfish to enter the side with food without ret urning. Figure 4 . 2 : The top view shows two RSC that moved through the SoftVal v e to the bristle side of the aquarium (left). The p rofile view (middle) show the directionality of the Softvalve . A 3D rendering was (right) . INITIAL DESIGN The initial design was a box - shaped trap made from galvanized ste e l frame and surrounded by a galvanized steel wire mesh. The major diameter sides contained 11 SoftV alves 32 for points of entry into the trap. This trap is already an improvement on the previously used artificial refuge traps, because there is now storage fo r the captured crayfish, so as soon as a tube is emptied, it can be filled again . It also utilize d the added benefit of the food bait. Using this trap in conjunction with the speakers, will also lend the added benefit of the auditory stimuli. The basic sha p e and Soft V alve placement can be seen in F igure 4 . 3 . The design process is iterative, however, and after pre liminary testing, more modifications will be tested. Figure 4 . 3 : A n AutoCAD rendering of the initial trap design. SoftV alves are placed low enou g h on the frame for crayfish to have access. The galvanized steel mesh is shown as a clear surface to better view the SoftV alves. VISION FOR FINAL DESIGN The full design incorporated the SoftV alve, and can include other modular com ponents, such as speakers , heat sources, space for food bait, a smooth - tube chimney for fish to escape, and drone trap retrieval systems. Figure 4. 4 shows a rendering of what these components look like. Manual labor is time consuming and expensive, especia lly when the invasive spe c ies spreads 33 rapidly , as they have in Michigan. Drone retrieval will allow for rapid trap emptying and Figure 4. 4 : A floating mat is attached to top of the trap. M o dular systems such as battery bank, heating elements, speakers and anchors for drone retr i eval are attached. 34 CHAPTER 5 : CONCLUSIONS Objective s i) and ii) were completed by research on frequency testing and locomotion responses. High range white noise a n d 12 kHz pure tone were most able to affect a locomotion response in RSC. Additionally, sound was found to enhance CPUE with or without baited traps. ithout food or sound, which helped to inform t h e engineering design of a solution. This thesis has addressed a possible solution to controlling invasive populations of RSC. While the solutions to the RSC infestation will be varied, depending on the charact eristics of the water body, this research show s that improvements can be made on population control using acoustic stimuli, food bait , and different trapping gear s. Such improvements led to a preliminary engineering design of next - generation large trap, which will alleviate the expensive practice of i n tensive trapping. Thus, objective iv) was met. O bjective iii) was not able to be completed because of the unforeseen circumstances surrounding the pandemic which began to sprea d rapidly in the United States in March 2020 . Most university research was halt e d during the spring, which was when cold water heat trials were to resume. Research on heat stimuli has therefore been moved to future work in Chapter 6 . 35 CHAPTER 6 : CURRENT AND FUTURE WORK While this thesis helped lay the groundwork for a solution to t h e RSC infestations in Michigan, it cannot be discussed without the context of events in the 2020 year. The pandemic has hampered much of the research which was planned for the spring and summer trapping season. Much of that research has been moved to Chap t er 8 and will be brought into another thesis or dissertation. More work is necessary to inform the engineering solutions to the RSC infestation; the design process is iterative and will continue to be updated with new information. CURRENT WORK Investigati o ns are already underway to use manganese enhanced magnetic resonance imaging (ME - MRI) to determine how crayfish are affected by each sensory stimuli, including heat, s ounds, and olfactory. Crayfish are injected with the contrast agent, then stimulated, be f ore being anesthetized on ice prior to the imaging (Figure 6.1). The results should elucidate the cause of their behavior and help to further inform the engineering de sign process. Figure 6.1 : A canula is installed using a gel glue (left) and the crayfi s h is placed in a 50 mL tube before anesthetizing . During the acoustic frequency response trials (Chapter 2), the HVAC system failed for one week in early February, causing a temporary halt to trials. Small 300 - Watt aquarium heaters were placed each habita t to slow the heat loss. During this time, it was o bserved that most of the crayfish were in close proximity to the heaters. The anecdotal evidence suggests that heat may be 36 an effective attractant in colder temperature waters. The possibility of extending the crayfish capture season would also increase th e total seasonal capture of RSC; this could also aid in achieving the goal of the research. Heat stimuli trials began in the late fall of 2019 with promising preliminary observations. Aquarium heaters wer propane generator (Figure 6.2 ) . After the first 24 - hour trial, ten RSC were captured in 3 heated traps. However, more time will be needed to collect sufficient data to draw any conclusions. If successful, it may be possible to extend the trapping season i n Michigan by use of a thermal stimulus , and therefore increase yearly capture. Figure 6.2 Traps (right). FUTURE WORK Additi onal frequency testing should be completed, upwards of 15 kHz to help find the optimum frequency for locomotion response. In addition, the sound intensity and pressure level s should be better tested to find the minimum amounts necessary to enhance trapping . Finding this quantity will help to reduce energy usage for crayfish capture using a sound system, and therefore lower costs of associated trapping methods. 37 APPENDICES 38 APPENDIX A : CHAPTER 2 SUPPLEMENTAL MATERIALS ADDITIONAL TA BLES AND FIGURES Table A.1: Number of Trials . Sound Treatment Number of Trials 0 35 1000 8 10000 2 11000 7 12000 2 13000 2 14000 2 15000 2 2000 4 3000 6 4000 4 500 2 5000 4 6000 4 7000 4 8000 2 9000 6 PN1k15k 2 PN6k10k 2 WN10k15k 2 WN1 k 15k 4 WN1k5k 2 WN6k10k 2 39 ACOUSTIC FREQUENCY RESPONSE ANALYSIS IN R The following is an R Markdown code output for the frequency response trials. Lab Trial Analysis Douglas Clements & Wei Liao 7/13/2020 Setup library (MASS) library (ggplot2) library (gri d) library ( gridExtra) library (ggpubr) library (readxl) library (conover.test) Add windows fonts to get Times New Roman windowsFonts ( A = windowsFont ( "Times New Roman" )) Add in a custom color for Kelley Green (MSU green) color< - rgb ( 24 , 69 , 59 , maxColorValue = 25 5 ) Set the working directory and add in the two data files. One is used for frequency analysis, and another used for a grouped frequency analysis. setwd ( "G:/School/Thesis Work/Sound Pond Trials/Data Files" ) LabData1< - as.data.frame ( read_xlsx ( "FrequencyR ange Analysis.xlsx" , sheet = "Sheet1" )) head (LabData1) ## row.names Frequency Frequency_range Hours Zone3 Zone2 Zone1 Sound ## 1 1 0 Control 24 27.77778 38.88889 33.33333 FALSE ## 2 2 0 Control 24 38.88889 27.77778 33.33333 FALSE ## 3 3 0 Control 24 27.77778 38.88889 33.33333 FALSE ## 4 4 0 Control 24 33.33333 33.33333 33.33333 FALSE ## 5 5 0 Control 24 2 7.77778 33. 3333 3 38.88889 FALSE ## 6 6 0 Control 24 41.17647 29.41176 29.41176 FALSE Data Visualization Check the distribution of population distribution data for normal distribution. 40 # Check distribution using a histogram, Shapi ro test, an d Q - Q plot hist (LabData1 $ Zone1) # Looks Normal shapiro.test (LabData1 $ Zone1) # Confirmed Normal ## ## Shapiro - Wilk normality test ## ## data: LabData1$Zone1 ## W = 0.97639, p - value = 0.04777 qqnorm (LabData1 $ Zone1, pch= 1 , frame= FALSE ) qqline (LabData1 $ Zone 1, co l= color, lwd= 2 ) 41 hist (LabData1 $ Zone2) # Looks Normal shapiro.test (LabData1 $ Zone2) # Confirmed Normal 42 ## ## Shapiro - Wilk normality test ## ## data: LabData1$Zone2 ## W = 0.98725, p - value = 0.3853 qqnorm (LabData1 $ Zone2, pch= 1 , frame= FALSE ) qqline (LabDa ta1 $ Zone2, col= color, lwd= 2 ) hist (LabData1 $ Zone3) # Looks Normal 43 shapiro.test (LabData1 $ Zone3) # Confirmed Normal ## ## Shapiro - Wilk normality test ## ## data: LabData1$Zone3 ## W = 0.9881, p - value = 0.4451 qqnorm ( LabData1 $ Zone3, pch= 1 , frame= FALSE ) qqlin e (La bData1 $ Zone3, col= color, lwd= 2 ) 44 Analysis Kruskal - Wallis Rank Sum Test Zone 1 kruskal.test (Zone1 ~ as.factor (Frequency), data = LabData1) ## ## Kruskal - Wallis rank sum test ## ## data: Zone1 by as.factor(Frequency) ## Kruskal - Wallis chi - squared = 40 .767 , df = 22, p - value = 0.008784 Zone 2 kruskal.test (Zone2 ~ as.factor (Frequency), data = LabData1) ## ## Kruskal - Wallis rank sum test ## ## data: Zone2 by as.factor(Frequency) ## Kruskal - Wallis chi - squared = 38.206, df = 22, p - value = 0.01737 Zone 3 45 krus kal.test (Zone3 ~ as.factor (Frequency), data = LabData1) ## ## Kruskal - Wallis rank sum test ## ## data: Zone3 by as.factor(Frequency) ## Kruskal - Wallis chi - squared = 28.633, d f = 22, p - value = 0.1557 Zone 1 and 2 have p values < 0.05. Zone 3 does no t. C onover Tests #ZONE1 CoZ1< - conover.test (LabData1 $ Zone1, as.factor (LabData1 $ Frequency), method= "bh" , kw= FAL SE ) ## ## Comparison of x by group ## (Benjamini - Hochberg) ## Col Mean - | ## Row Mean | 0 1000 10000 11000 12000 13 000 ## --------- + ------------------------------------------------------------------ ## 1000 | - 1.620464 ## | 0.1764 ## | ## 10000 | 1.539085 2.218642 ## | 0.1853 0.1023 ## | ## 11000 | - 2.3131 49 - 0.623523 - 2.590093 ## | 0.0912 0.3955 0.0508 ## | ## 12000 | - 2.713330 - 1.691995 - 3.091630 - 1.2658 47 # # | 0.0508 0.1779 0.0338 0.2425 ## | ## 13000 | 0.357182 1.131731 - 0.859278 1.518385 2.232352 ## | 0.4433 0.2558 0.3473 0.1884 0.1018 ## | ## 14000 | 1.965545 2. 6108 27 0.310048 2.976792 3.401679 1.169327 ## | 0.1356 0.0498 0.4519 0.0341 0.0256* 0.2504 ## | ## 15000 | - 0.739429 0.123257 - 1.656547 0.524016 1.435083 - 0.797268 ## | 0.3743 0. 4836 0.1778 0.4181 0.2084 0.3654 ## | ## 2000 | - 2.613019 - 1.215138 - 2 .884567 - 0.672343 0.685340 - 1.892358 ## | 0.0514 0.2482 0.0417 0.3834 0.3795 0.1474 ## | ## 3000 | - 1.9 9185 7 - 0.453811 - 2.448359 0.139512 1.338099 - 1.395962 46 ## | 0.1333 0.4290 0 .0690 0.4808 0.2309 0.2168 ## | ## 4000 | - 0.917841 0.245920 - 1.851442 0.755124 1.718465 - 0.859232 ## | 0.3 218 0.4658 0.1525 0.3739 0.1711 0.3449 ## | ## 500 | - 0.520107 0. 324952 - 1.497093 0.722890 1.594537 - 0.637814 ## | 0.4155 0.4472 0.1918 0.3801 0.1723 0.3909 ## | ## 50 00 | - 0.288444 0.788393 - 1.467856 1.285124 2.102051 - 0.475646 ## | 0.4595 0. 3673 0.2004 0.2390 0.1131 0.4276 ## | ## 6000 | 2.111659 2.857022 - 0.005114 3.306192 3.564793 0.987094 ## | 0.1132 0.0423 0.4980 0.0217* 0.0752* 0.2991 ## | ## 7000 | - 1.6 89903 - 0.419512 - 2.321974 0.104990 1.247933 - 1.329765 ## | 0.1760 0.4340 0.0921 0.4852 0.2390 0.2320 ## | ## 8000 | - 0.410446 0.425800 - 1.417366 0.822327 1.674264 - 0.558088 ## | 0.4 360 0.4355 0.2130 0.3580 0.1740 0.4086 ## | ## 9000 | - 1.350308 0.071078 - 2.101177 0.649037 1.685281 - 1.048780 # # | 0.2282 0.4912 0.1107 0.3924 0.1751 0.2827 ## | ## PN1k1 5k | - 0.020539 0.784368 - 1.133893 1.175880 1.957737 - 0.274614 ## | 0.4957 0.3668 0.2569 0.2518 0.1353 0.4614 ## | ## PN6k10k | - 0.252046 0.571468 - 1.302205 0.965958 1.789425 - 0.442927 ## | 0.4651 0.4045 0.2387 0.3021 0.1624 0.4297 ## | ## WN10k15k | - 2.615854 - 1.602353 - 3.020762 - 1.177458 0.070868 - 2.16 1484 ## | 0.0531 0.1739 0.0322 0.2532 0.4892 0.1031 ## | ## WN1k15k | - 1.614375 - 0.354415 - 2.275944 0.168590 1.293963 - 1.283734 ## | 0.1762 0.4424 0.0970 0.4808 0.2376 0.23 74 ## | ## WN1k5k | - 1.056228 - 0.168079 - 1.886869 0.236754 1.204761 - 1.027590 ## | 0.2815 0.4768 0.1464 0.4656 0.2441 0.2877 ## | ## WN6k10k | - 2.664592 - 1.647174 - 3.056196 - 1.221653 0 .035 434 - 2.196918 ## | 0.0528 0.1787 0.0314 0.2477 0.4917 0.1021 ## Col Mean - | ## Row Mean | 14000 15000 2000 3000 4000 500 ## --------- + -------------------------------------------------- ---- ------------ ## 15000 | - 1.966596 ## | 0.1381 ## | 47 ## 2000 | - 3.24 2581 - 0.971751 ## | 0.0236* 0.3038 ## | ## 3000 | - 2.828090 - 0.419512 0.773095 ## | 0.0408 0.4362 0. 3699 ## | ## 4000 | - 2.209455 0.061373 1.265314 0.612987 ## | 0.1018 0.4911 0.2405 0.3959 ## | ## 500 | - 1.807142 0.159453 1.155872 0.614802 0.122747 ## | 0.1618 0.4784 0. 2519 0.3974 0.4818 ## | ## 5000 | - 1.825869 0.444959 1.735109 1.12762 1 0.469795 0.260838 ## | 0.1582 0.4309 0.1760 0.2555 0.4281 0.4633 ## | ## 6000 | - 0.363128 1.907701 3.5 26595 3.090095 2.261280 1.723580 ## | 0.4427 0.1453 0.0427* 0.0309 0.0976 0.1746 ## | ## 7000 | - 2.679988 - 0.409158 0.689032 - 0.018298 - 0.576281 - 0.593279 ## | 0.0530 0.4344 0.3 847 0.4947 0.4045 0.4008 ## | ## 8000 | - 1.727415 0.239180 1.247933 0.712447 0.214808 0.079726 ## | 0.1760 0.4666 0.2411 0.3828 0.4711 0.4916 ## | ## 9000 | - 2.480908 - 0.07 2329 1.212250 0.490990 - 0.173832 - 0.267619 ## | 0.0656 0.4927 0.2473 0.4226 0.4827 0.4624 ## | ## PN1k15k | - 1.443942 0.522653 1.575260 1.059629 0.542135 0.363200 ## | 0.2072 0.41 65 0.1768 0.2822 0.4117 0.4470 ## | ## PN6k10k | - 1.612254 0.354341 1.380909 0.853490 0.347784 0.194887 ## | 0.1748 0.4403 0.2205 0.3452 0.4411 0.4756 ## | ## WN10k15k | - 3 .330 811 - 1.364215 - 0.603508 - 1.251303 - 1.636633 - 1.523669 ## | 0.0230* 0.2249 0.3982 0.2419 0.1753 0.1886 ## | ## WN1k15k | - 2.633957 - 0.363128 0.745408 0.043458 - 0.519906 - 0.547249 ## | 0 .052 6 0.4449 0.3738 0.4944 0.4133 0.4116 ## | ## WN1k5k | - 2 .196918 - 0.230322 0.705798 0.137426 - 0.327326 - 0.389775 ## | 0.0995 0.4663 0.3789 0.4796 0.4483 0.4391 ## | ## WN6 k10k | - 3.366245 - 1.399649 - 0.644424 - 1.294701 - 1.677549 - 1.559103 ## | 0. 0240* 0.2177 0.3923 0.2396 0.1753 0.1803 ## Col Mean - | ## Row Mean | 5000 6000 7000 8000 9000 PN1k15k 48 ## ------ --- + ------------------------------------------------------------------ ## 6000 | 1.791 485 ## | 0.1644 ## | ## 7000 | - 1.046077 - 2.837562 ## | 0.2817 0.0421 ## | ## 8000 | - 0.168777 - 1.631519 0 .685340 ## | 0.4828 0.1748 0.3818 ## | ## 9000 | - 0.6 88466 - 2.650940 0.457453 - 0.365264 ## | 0.3826 0.0524 0.4295 0.4483 ## | ## PN1k15k | 0.158548 - 1.304192 1.012667 0.28 3473 0.712447 ## | 0.4768 0.2402 0.2921 0.4596 0.3804 ## | ## PN6k10k | - 0.035801 - 1.498542 0.818316 0.115161 0.506307 - 0.168312 ## | 0.4956 0.1934 0.3575 0.4829 0.4175 0.47 88 # # | ## WN10k15k | - 2.020220 - 3.482961 - 1.166101 - 1.603395 - 1.598486 - 1.88 6869 ## | 0.1277 0.0246* 0.2497 0.1757 0.1731 0.1438 ## | ## WN1k15k | - 0.989701 - 2.781186 0.056375 - 0.639310 - 0.395 696 - 0.966636 ## | 0.3002 0.0442 0.4912 0.3925 0.4385 0.3040 ## | ## WN1k5k | - 0.710912 - 2.173654 0.143205 - 0.469502 - 0.209756 - 0.752976 ## | 0.3787 0.1026 0.4813 0.4260 0.471 2 0.3725 ## | ## WN6k10k | - 2.061135 - 3.523877 - 1.207017 - 1.638830 - 1 .641883 - 1.922303 ## | 0.1188 0.0287* 0.2452 0.1769 0.1782 0.1435 ## Col Mean - | ## Row Mean | PN6k10k WN10k15k WN1k15k WN1k 5k # # --------- + -------------------------------------------- ## WN10k15k | - 1.718557 ## | 0.1737 ## | ## WN1k15k | - 0.772286 1.212132 ## | 0.3679 0.2452 ## | ## WN1k5k | - 0.584663 1.133893 0.09717 5 ## | 0.4027 0.2589 0.4864 ## | ## WN6k10k | - 1.753991 - 0.035434 - 1.253047 - 1.169327 ## | 0.1720 0.4937 0.2434 0.2525 ## 49 ## alpha = 0.05 ## Reject Ho if p <= alpha/2 #ZONE2 CoZ2< - conover.test (L abDa ta1 $ Zone2, as.factor (LabData1 $ Frequency), method= "bh" , kw= FAL SE ) ## ## Comparison of x by group ## (Benjamini - Hochberg) ## Col Mean - | ## Row Mea n | 0 1000 10000 11000 12000 13 000 ## --------- + ------------------------------------------------------------------ ## 1000 | - 1.155327 ## | 0.2319 ## | ## 10000 | - 1.824661 - 1.105317 ## | 0.1330 0.2424 ## | ## 11000 | - 0.7920 00 0.241200 1.245551 ## | 0.3222 0.4615 0.2206 ## | ## 12000 | 1.440252 1.897186 2.373687 1.714957 ## | 0.1921 0.1247 0. 0835 0.1458 ## | ## 13000 | - 1.884458 - 1.160308 - 0.043474 - 1.299773 - 2.417161 ## | 0.1262 0.2317 0.4944 0.2095 0.0863 ## | ## 14000 | - 2.027971 - 1.292286 - 0.147812 - 1.429905 - 2.521499 - 0.1 0433 7 ## | 0.1069 0.2105 0.4752 0.1920 0.0899 0.4834 ## | ## 15000 | - 0.222102 0.368439 1.165106 0.207591 - 1.208580 1.208580 ## | 0.4637 0.4318 0.2351 0.4680 0.2256 0.2 274 ## | ## 2000 | 1.992147 2.456356 2.745918 2.200716 0.005019 2.796118 ## | 0.1098 0.0921 0.0773 0.0836 0.4980 0.0732 ## | ## 3000 | 1.149751 1.779015 2.246927 1.502556 - 0.66 0234 2.300172 ## | 0.2323 0.1403 0.0799 0.1800 0.3610 0.0815 ## | ## 4000 | 0.072955 0.802220 1.576267 0.584611 - 1.164630 1.626467 ## | 0.4864 0.3216 0.1667 0.3711 0.23 18 0.1618 ## | ## 500 | 0.603095 1 .127313 1.765049 0.955852 - 0.608637 1.808523 ## | 0.3727 0.2374 0.1424 0.2789 0.3742 0.1356 ## | 50 ## 5000 | 0.262403 0.965504 1.691726 0.744 141 - 1.049171 1.741926 ## | 0.4563 0 .2768 0.1454 0.3336 0.2522 0.1435 ## | ## 6000 | - 2.216897 - 1.171383 0.180718 - 1.343616 - 2.560179 0.230918 ## | 0.0822 0.2344 0.4714 0.206 2 0.1028 0.4640 ## | ## 7000 | - 1. 796816 - 0.809319 0.436736 - 0.989876 - 2.304161 0.486936 ## | 0.1370 0.3224 0.4074 0.2705 0.0829 0.3989 ## | ## 8000 | - 2.016011 - 1.281288 - 0.1391 17 - 1.419061 - 2.512804 - 0.095643 ## | 0. 1078 0.2110 0.4729 0.1921 0.0874 0.4850 ## | ## 9000 | 0.952972 1.618018 2.140437 1.346273 - 0.766724 2.193682 ## | 0.2783 0.1607 0.0907 0.2071 0.3294 0.0832 ## | ## PN1k 15k | 2.133897 2.535081 2.877987 2.343929 0.504299 2.921461 ## | 0.0902 0.1030 0.0708 0.0844 0.3951 0.1123 ## | ## PN6k10k | - 0.030752 0.54441 0 1.304224 0.381101 - 1.069463 1.347698 ## | 0.4976 0.3812 0.2115 0.4303 0.2511 0.2084 ## | ## WN10k15k | 1.452211 1.908184 2.382382 1.725801 0.008694 2.425856 ## | 0.1937 0.1279 0.0876 0.1445 0.5025 0.0914 ## | ## WN1k15k | - 0.182388 0.582142 1.420648 0.369593 - 1.320249 1.470848 ## | 0.4727 0.3703 0.1934 0.4334 0.2092 0.1890 ## | ## WN1k5k | 1.236942 1 .710217 2.225875 1.530603 - 0.147812 2.269349 ## | 0.2203 0.1436 0.0822 0.1743 0.4773 0.0856 ## | ## WN6k10k | 1.081470 1.567241 2.112842 1.389626 - 0.260844 2.156317 ## | 0.2499 0 .1659 0.0895 0.1970 0.4550 0.0891 # # Col Mean - | ## Row Mean | 14000 15000 2000 3000 4000 500 ## --------- + ------------------------------------------------------------------ ## 15000 | 1.312918 ## | 0.2101 ## | ## 2000 | 2.9 16597 1.400569 ## | 0.0949 0.1968 ## | ## 3000 | 2.427959 0.819968 - 0.841872 ## | 0.0948 0.3197 0.3121 ## | ## 4000 | 1.746946 0.2 30918 - 1.432523 - 0.727378 ## | 0.143 9 0.4619 0.1929 0.3390 51 ## | ## 500 | 1.912861 0.599943 - 0.707814 - 0.085191 0.461836 ## | 0.1288 0.3721 0.3437 0.4853 0.4062 ## | ## 5000 | 1.862405 0.346377 - 1.291115 - 0.5724 73 0.141407 - 0.346377 ## | 0.1303 0.4335 0.2092 0.3707 0.4759 0.4355 ## | ## 6000 | 0.351397 - 1.164630 - 3.141715 - 2.599703 - 1.709191 - 1.857385 ## | 0.4374 0.2335 0.1452 0.1067 0.1421 0.1297 ## | ## 7000 | 0.607415 - 0.908612 - 2.828158 - 2.256219 - 1.395634 - 1.601367 ## | 0.3728 0.2894 0.0735 0.0820 0.1967 0.1642 ## | ## 8000 | 0.008694 - 1.304224 - 2.90655 7 - 2.417310 - 1.736906 - 1.904167 ## | 0.5005 0.2097 0.0838 0.0897 0.1430 0.1269 ## | ## 9000 | 2.321469 0.713479 - 0.976572 - 0.150598 0.592678 - 0 .021 297 ## | 0.0866 0.3432 0.2741 0.4781 0.3714 0.4994 ## | ## PN1k15k | 3.025799 1.712880 0.577295 1.277873 1.746946 1.112937 ## | 0.1375 0.1446 0.3705 0.2105 0.1459 0 .241 2 ## | ## PN6k10k | 1.452036 0.139117 - 1.239930 - 0.649585 - 0.070279 - 0.460825 ## | 0.1918 0.4749 0.2210 0.3618 0.4855 0.4026 ## | ## WN10k15k | 2.530194 1.217275 0.005019 0.670883 1. 1746 70 0.617332 ## | 0.0982 0.2259 0.5000 0.3582 0.2350 0.3723 ## | ## WN1k15k | 1.591327 0.075299 - 1.623117 - 0.936162 - 0.190593 - 0.617455 ## | 0.1655 0.4874 0.1610 0.2816 0. 4733 0.3743 ## | ## WN1k5k | 2 .373687 1.060768 - 0.175698 0.479202 0.993952 0.460825 ## | 0.0864 0.2510 0.4715 0.4004 0.2706 0.4046 ## | ## WN6k10k | 2.260654 0.947736 - 0.306217 0.3 4076 6 0.863433 0.347793 ## | 0 .0852 0.2787 0.4411 0.4319 0.3048 0.4370 ## Col Mean - | ## Row Mean | 5000 6000 7000 8000 9000 PN1k15k ## --------- + ----------------------------------------- ---- --------------------- ## 6000 | - 1.850 599 ## | 0.1277 ## | ## 7000 | - 1.537042 0.313556 ## | 0.1740 0.4399 ## | ## 8000 | - 1.852365 - 0.341357 - 0.597375 52 ## | 0.1291 0.43 37 0.3713 ## | ## 9000 | 0.4 37773 2.465004 2.121519 2.310820 ## | 0.4089 0.0944 0.0893 0.0839 ## | ## PN1k15k | 1.631487 3.142495 2.886477 3.017104 1.384362 ## | 0.1622 0.28 98 0.0777 0.1058 0.1970 ## | ## PN6k10k | - 0.185738 1.325269 1.069251 1.443341 - 0.543096 - 1.573763 ## | 0.4733 0.2092 0.2495 0.1929 0.3798 0.1657 ## | ## WN10k15k | 1.059211 2 .570 219 2.314201 2.521499 0.777372 - 0.49 5605 ## | 0.2500 0.1072 0.0856 0.0949 0.3267 0.3970 ## | ## WN1k15k | - 0.332001 1.518598 1.205041 1.581287 - 0.801463 - 1.902565 ## | 0.4338 0 .176 4 0.2252 0.1669 0.3201 0.1253 ## | ## WN1k5k | 0.878493 2.389501 2.133483 2.364992 0.585691 - 0.652112 ## | 0.3002 0.0892 0.0886 0.0827 0.3726 0.3627 ## | ## WN6k10k | 0. 7479 74 2.258982 2.002964 2.251960 0 .447255 - 0.765144 ## | 0.3338 0.0834 0.1091 0.0808 0.4067 0.3282 ## Col Mean - | ## Row Mean | PN6k10k WN10k15k WN1k15k WN1k5k ## --------- + ----------------------------- ---- ----------- ## WN10k15k | 1.078158 ## | 0.2494 ## | ## WN1k15k | - 0.085339 - 1.330289 ## | 0.4873 0.2092 ## | ## WN1k5k | 0.921651 - 0.156506 1.149570 ## | 0.2858 0.4776 0.2 307 ## | ## WN6k10k | 0.808618 - 0.269539 1.019052 - 0.113032 ## | 0.3208 0.4553 0.2623 0.4818 ## ## alpha = 0.05 ## Reject Ho if p <= alpha/2 #ZONE3 CoZ3< - conover.test (LabData1 $ Zone3, as.factor (LabData1 $ Frequency), me thod = "bh" , kw= FAL SE ) ## ## Comparison of x by group ## (Benjamini - Hochberg) 53 ## Col Mean - | ## Row Mean | 0 1000 10000 1100 0 12000 13 000 ## --------- + ------------------------------------------------------------------ ## 1000 | 2.003124 ## | 0.2908 ## | ## 10000 | 0.167726 - 0.838694 ## | 0.4919 0.4154 ## | ## 11000 | 2.3865 55 0.392503 1.080323 ## | 0.4043 0.4783 0.3808 ## | ## 12000 | 0.549360 - 0.487732 0.277459 - 0.734271 ## | 0.4562 0.4665 0.4850 0.4454 ## | ## 13000 | 1.44 7323 0.338058 0.930304 0.079970 0.652845 ## | 0.3909 0.4875 0.3836 0.4875 0.4529 ## | ## 14000 | - 0.056764 - 1.045142 - 0.163211 - 1.283883 - 0.440670 - 1.093515 ## | 0.4930 0.3858 0.48 74 0.3610 0.4775 0.3770 ## | ## 15000 | 0.672830 - 0.374186 0.367225 - 0.622312 0.089766 - 0.563079 ## | 0.4480 0.4747 0.4756 0.4607 0.4895 0.4545 ## | ## 2000 | 0.393385 - 0 .942 825 0 .098941 - 1.245246 - 0.221440 - 0.975281 ## | 0.4804 0.3799 0.4939 0.3555 0.4878 0.3854 ## | ## 3000 | 0.608414 - 0.955734 0.179903 - 1.292886 - 0.159913 - 0.959482 ## | 0.4592 0 .376 0 0 .4931 0.3711 0.4845 0.3773 ## | ## 4000 | 0.772192 - 0.616334 0.329805 - 0.926262 0.009423 - 0.744417 ## | 0.4369 0.4609 0.4816 0.3794 0.5042 0.4463 ## | ## 500 | - 0. 4047 24 - 1. 365136 - 0.416188 - 1.599402 - 0.693648 - 1.346493 ## | 0.4826 0.3768 0.4794 0.3774 0.4489 0.3706 ## | ## 5000 | - 0.147767 - 1.409240 - 0.230863 - 1.700938 - 0.551246 - 1.305086 ## | 0. 4856 0. 3734 0.4881 0.3547 0.4580 0.3860 ## | ## 6000 | - 0.008614 - 1.289304 - 0.146056 - 1.583760 - 0.466438 - 1.220279 ## | 0.5025 0.3627 0.4842 0.3606 0.4668 0.3481 ## | ## 7 000 | 2.9 67727 1.275978 1.667872 0.922542 1.347490 0.593649 ## | 0.2450 0.3559 0.3681 0.3782 0.3760 0.4613 ## | ## 8000 | 2.423858 1.236106 1.640273 0.965457 1.362814 0.709969 54 ## | 0.4 409 0.3475 0.3674 0.3875 0.3720 0.4494 ## | ## 9000 | 0.602257 - 0.960771 0.176571 - 1.297775 - 0.163245 - 0.962814 ## | 0.4596 0.3799 0.4924 0.3849 0.4895 0.3856 ## | ## PN1k1 5k | - 1.414933 - 2.294151 - 1.150639 - 2.515423 - 1.428099 - 2.080944 ## | 0.3835 0.3825 0.3637 0.4341 0.3890 0.3405 ## | ## PN6k10k | - 0.000641 - 0.993530 - 0.122408 - 1.232993 - 0.399867 - 1.052713 ## | 0.4997 0.3931 0.4881 0.3450 0.4798 0.3933 ## | ## WN10k15k | 1.537120 0.420637 0.995589 0.161394 0.718129 0.065284 ## | 0.3762 0.4825 0.3957 0.4860 0.4514 0.4915 ## | ## WN1k15k | 1.723075 0.203223 0.909320 - 0.125547 0.588938 - 0.164902 ## | 0.3608 0.4871 0.3823 0.4909 0.4579 0.4910 ## | ## WN1k5k | - 0.292479 - 1.261912 - 0.334583 - 1.497622 - 0.612042 - 1.2 64887 ## | 0.4850 0.3501 0.4867 0.3710 0.4602 0.3530 ## | ## WN6k10k | 1.705488 0.575473 1.117997 0.314064 0.840538 0.187693 ## | 0.3624 0.4535 0.3790 0.4819 0.4178 0.4 919 ## Col Mean - | ## Row Mean | 14000 15000 2000 3000 4000 500 ## --------- + ------------------------------------------------------------------ ## 15000 | 0.530436 ## | 0.4578 ## | ## 200 0 | 0.28 7401 - 0.325093 ## | 0.4850 0.4790 ## | ## 3000 | 0.379795 - 0.269854 0.094817 ## | 0.4795 0.4838 0.4936 ## | ## 4000 | 0.518265 - 0.094230 0.282749 0.214919 ## | 0.4560 0.4917 0.4848 0.4863 ## | ## 500 | - 0.252977 - 0.783414 - 0.579515 - 0.689628 - 0.810378 ## | 0.4894 0.4338 0.4542 0.4480 0.4216 ## | ## 5000 | - 0.042403 - 0.654899 - 0.403927 - 0 .53729 7 - 0.686676 0.249709 ## | 0.4949 0.4549 0.4803 0.4598 0.4465 0.4886 ## | ## 6000 | 0.042403 - 0.570092 - 0.300060 - 0.423516 - 0.582809 0.334516 ## | 0.4969 0.4536 0.4862 0 .4837 0.4553 0.4842 ## | 55 ## 7000 | 1.856332 1.243837 1.921540 2.010124 1.638791 2.148446 ## | 0.3520 0.3518 0.3187 0.3005 0.3585 0.3632 ## | ## 8000 | 1.803485 1.273048 1. 795083 1.829013 1.564219 2.056462 ## | 0.3378 0.3528 0.3320 0.3446 0.3656 0.3379 ## | ## 9000 | 0.376463 - 0.273186 0.090603 - 0.004711 - 0.219133 0.686296 ## | 0.4760 0.4846 0. 4912 0.5021 0.4866 0.4435 ## | ## PN1k15k | - 0.987428 - 1.517865 - 1.427586 - 1.589143 - 1.658449 - 0.734450 ## | 0.3856 0.3649 0.3819 0.3751 0.3644 0.4487 ## | ## PN6k10k | 0.040802 - 0.4 89633 - 0.240286 - 0.329822 - 0.471150 0.293780 ## | 0.4935 0.4683 0.4860 0.4840 0.4670 0.4868 ## | ## WN10k15k | 1.158800 0.628363 1.050665 1.039439 0.819801 1.411777 ## | 0.3631 0.4 604 0.3864 0.3814 0.4229 0.3785 ## | ## WN1k15k | 1.097780 0.485285 0.992507 0.992420 0.709758 1.389893 ## | 0.3827 0.4624 0.3900 0.3864 0.4462 0.3731 ## | ## WN1k5k | - 0.17 1371 - 0.701808 - 0.485285 - 0.589682 - 0.716148 0.081605 ## | 0.4925 0.4475 0.4651 0.4605 0.4492 0.4888 ## | ## WN6k10k | 1.281208 0.750772 1.192010 1.189358 0.961147 1.534186 ## | 0.35 76 0.4460 0.3562 0.3535 0.3830 0.3698 ## Col Mean - | ## Row Mean | 5000 6000 7000 8000 9000 PN1k15k ## --------- + ------------------------------------------------------------------ ## 6000 | 0.10 3867 ## | 0.4939 ## | ## 7000 | 2.325467 2.221600 ## | 0.4043 0.3324 ## | ## 8000 | 2.124888 2.040081 0.226152 ## | 0.3292 0.3119 0.4880 ## | ## 9000 | 0. 5330 83 0.419302 - 2.014338 - 1.832345 ## | 0.4592 0.4804 0.3133 0.3558 ## | ## PN1k15k | - 1.097780 - 1.182587 - 2.996517 - 2.790913 - 1.585811 ## | 0.3786 0.3533 0.4501 0.2724 0.3681 ## | ## PN6k10k | 0.089518 0.004711 - 1.809217 - 1.762682 - 0.326490 1.028231 ## | 0.4876 0.5001 0.3461 0.3435 0.4807 0.3841 56 ## | ## WN10k15k | 1.380470 1.295663 - 0.518265 - 0.644684 1.042771 2.1 4622 8 ## | 0.3729 0.3748 0.4587 0.4544 0.3833 0.3371 ## | ## WN1k15k | 1.396434 1.292567 - 0.929033 - 0.984704 0.996634 2.237964 ## | 0.3753 0.3659 0.3810 0.3836 0.3990 0.3 514 ## | ## WN1k5k | - 0.155479 - 0.240286 - 2.054216 - 1.974856 - 0.586350 0.816056 ## | 0.4843 0.4907 0.3196 0.2959 0.4563 0.4217 ## | ## WN6k10k | 1.521816 1.437009 - 0.376920 - 0.522276 1 .19 2690 2.268637 ## | 0.3702 0.3904 0.4784 0.4594 0.3600 0.3621 ## Col Mean - | ## Row Mean | PN6k10k WN10k15k WN1k15k WN1k5k ## --------- + -------------------------------------------- ## WN10k15k | 1.117997 ## | 0.3748 ## | ## WN1k15k | 1.050665 - 0.240286 ## | 0.3905 0.4883 ## | ## WN1k5k | - 0.212174 - 1.330172 - 1.295663 ## | 0.4853 0.3754 0.3805 ## | ## WN6k10k | 1.240406 0 .122 408 0.381631 1.452580 ## | 0.3493 0.4902 0.4812 0.3951 ## ## alpha = 0.05 ## Reject Ho if p <= alpha/2 Plot box and whisker for each fit box_ 1 < - ggboxplot (LabData1, x = "Frequency" , y = "Zone1" ) + xlab ( "Frequency (Hz)" ) + ylab ( "Crayfish population percentage in Zone 1 (%)" ) + ylim ( 0 , 60 ) + theme_bw () + theme ( text= element_text ( family= "A" , size= 12 )) #geom_hline(yintercept = 33.33,color ='red') box_ 1 57 box_ 2 < - ggboxplot (LabData1, x = "Frequency" , y = "Zone2" ) + xl ab ( " Frequency (Hz)" ) + ylab ( "Crayfish population percentage in Zone 2 (%)" ) + ylim ( 0 , 60 ) + theme_bw () + theme ( text= element_text ( family= "A" , size= 12 )) #geom_hline(yintercept = 33.33,color ='red') box_ 2 58 box_ 3 < - ggboxplot (LabData1, x = "Frequency" , y = "Zone3" ) + xlab ( "Frequency (Hz)" ) + ylab ( "Crayfish population percentage in Zone 3 (%)" ) + ylim ( 0 , 60 ) + theme_bw () + theme ( text= element_text ( family= "A" , size= 12 )) #geom_hline(yintercept = 33.33,color ='red') box_ 3 59 ga< - grid.arrange (box_ 1 , box _ 2 , box_ 3 , ncol= 3 ) 60 Create bar charts First, calculate sd using the following function. data_summary < - function (data, varname, groupnames){ require (plyr) summary_func < - function (x, col){ c ( mean = mean (x[[col]], na.rm= TRUE ), sd = sd (x [[co l]], na.rm= TRUE )) } data_sum < - ddply (data, groupnames, .fun= summary_func, varname) data_sum < - rename (data_sum, c ( "mean" = varname)) return (data_sum) } Zones by frequency treatment Zone 1 by Frequency MovementToZone1 < - data_summary (LabData1, varn ame= "Zone1" , groupnames= c ( "Frequenc y" )) MovementToZone1 $ Frequency< - factor (MovementToZone1 $ Frequency, levels= c ( "0" , "500" , "10 00" , "2000" , "3000" , "4000" , "5000" , "6000" , "7000" , "8000" , "9000" , "10000" , "11000" , "12000" , "130 00" , "14000" , "15000" , "PN1k15k" , "PN6k10k" , " WN1k 15k" , "WN1k5k" , "WN6k10k" , "WN10k15k " )) box_ 11 < - ggplot (MovementToZone1, aes ( x= Frequency, y= Zone1, alpha= Frequency != 0 )) + ge om_bar ( stat= "identity" , position= position_dodge ( 0.9 ), fill= color) + geom_errorbar ( aes ( ymin= Zone1 - sd, ymax= Zone1 + sd), width= 0.2 , po sition= position_dodge ( 0.9 )) + xlab ( "Frequency (Hz)" ) + ylab ( "Population in Zone 1 (%)" ) + labs ( fill= "" ) + ylim ( 0 , 60 ) + labs ( title = "" , subtitle= NULL ) + #geom_hline(yintercept = 33.33,color ='red')+ scale_alpha_manual ( values = c ( 0.75 , 1 )) + guid es ( a lpha= F) + theme_bw () + theme ( text= element_text ( family= "A" , size= 18 )) + theme ( title= element_text ( size= 18 ), axis.text.x = element_text ( size= 18 , face= "italic" , angle= 9 0 , vjust= 0.5 , hjust= 0 ), axis.text.y= element_text ( size= 18 ), axis.title.y = element_text ( s ize = 18 ), axis.title.x= element_text ( size= 18 )) box_ 11 61 Zone 2 by Frequency MovementToZone2 < - data_summary (LabData1, varname= "Zone2" , groupnames= c ( "Frequency" )) MovementToZone2 $ Frequency< - factor (MovementToZone2 $ Frequency, l evel s= c ( "0" , "500" , "10 00" , "2000" , "3000" , "4000" , "5000" , "6000" , "7000" , "8000" , "9000" , "10000" , "11000" , "12000" , "130 00" , "14000" , "15000" , "PN1k15k" , "PN6k10k" , "WN1k15k" , "WN1k5k" , "WN6k10k" , "WN10k15k " )) head (MovementToZone2) ## Frequency Zone2 sd ## 1 0 35.26773 8.092236 ## 2 1000 38.88889 8.195106 ## 3 10000 45.40441 2.339692 ## 4 11000 38.36354 7.584991 ## 5 12000 23.33333 14.142136 ## 6 13000 45.55556 1.571348 box_ 21 < - ggplot (MovementToZone2, aes ( x= Frequency, y= Zone2, al pha= Frequency != 0 )) + ge om_bar ( stat= "identity" , position= position_dodge ( 0.9 ), fill= color) + geom_errorbar ( aes ( ymin= Zone2 - sd, ymax= Zone2 + sd), width= 0.2 , position= position_dodge ( 0.9 )) + xlab ( "Frequency (Hz)" ) + ylab ( "Population in Zone 2 (%)" ) + labs ( fill = "" ) + ylim ( 0 , 60 ) + labs ( title = "" , subtitle= NULL ) + 62 scale_alpha_manual ( values = c ( 0.75 , 1 )) + guides ( alpha= F) + #geom_hline(yintercept = 33.33,color ='red')+ theme_bw () + theme ( text= element_text ( family= "A" , size= 18 )) + theme ( title= element_text ( siz e= 18 ), axis.text.x = element_text ( size= 18 , face= "italic" , angle= 9 0 , vjust= 0.5 , hjust= 0 ), axis.text.y= element_text ( size= 18 ), axis.title.y = element_text ( size = 18 ), axis.title.x= element_text ( size= 18 )) box_ 21 Zone 3 by Frequency MovementToZone3 < - data_summ ary ( LabData1, varname= "Zone3" , groupnames= c ( "Frequency" )) MovementToZone3 $ Frequency< - factor (MovementToZone3 $ Frequency, levels= c ( "0" , "500" , "10 00" , "2000" , "3000" , "4000" , "5000" , "6000" , "7000" , "8000" , "9000" , "10000" , "11000" , "12000 " , "1 30 00" , "14000" , "15000" , "PN1k15k" , "PN6k10k" , "WN1k15k" , "WN1k5k" , "WN6k10k" , "WN10k15k " )) head (MovementToZone3) ## Frequency Zone3 sd ## 1 0 31.93596 8.842344 ## 2 1000 23.14134 11.238741 ## 3 10000 30.33088 1.299829 ## 4 1 1000 22.44473 9.181065 63 ## 5 12000 27.77778 7.856742 ## 6 13000 23.88889 5.499719 box_ 31 < - ggplot (MovementToZone3, aes ( x= Frequency, y= Zone3, alpha= Frequency != 0 )) + ge om_bar ( stat= "identity" , position= position_dodge ( 0.9 ), fill= color) + geom_err orba r ( aes ( ymin= Zone3 - sd, ymax= Zone3 + sd), width= 0.2 , position= position_dodge ( 0.9 )) + xlab ( "Frequency (Hz)" ) + ylab ( "Population in Zone 3 (%)" ) + labs ( fill= "" ) + ylim ( 0 , 60 ) + labs ( title = "" , subtitle= NULL ) + scale_alpha_manual ( values = c ( 0.75 , 1 )) + gu ides ( alpha= F) + #geom_hline(yintercept = 33.33,color ='red')+ theme_bw () + theme ( text= element_text ( family= "A" , size= 18 )) + theme ( title= element_text ( size= 18 ), axis.text.x = element_text ( size= 18 , face= "italic" , angle= 9 0 , vjust= 0.5 , hjust= 0 ), axis.text.y= el emen t_text ( size= 18 ), axis.title.y = element_text ( size = 18 ), axis.title.x= element_text ( size= 18 )) box_ 31 Side - by - side visualization ka< - grid.arrange (box_ 11 , box_ 21 , box_ 31 , nrow= 3 , ncol= 1 ) 64 65 LABORATORY CRAYFISH STANDARD OPERATING PROCEDURES Standard Operat ing Procedure for Red Swamp Crayfish Research Created by Douglas Clements 10/01/18 All individuals working with Red Swamp Crayfish ( Procambarus clarkii ) must read and agree to follow the standard operating procedure (SOP) outlined below. R equirements All per sonnel working with P. clarkii must be trained in proper handling procedure. If personnel are uncomfortable holding crayfish, they may use a skimmer/net provided for catching crayfish safely. Rubber gloves are provided for protection a gainst claw pin chin g if required. Personnel must check the water temperature daily; safe levels are between 6°C and 22°C. The habitats must be gravel syphoned weekly; syphon and pump are provided. Effluent must be disposed of in the designated sink. Sponge filters will b e cl eaned weekly; rinsed in the designated sink. Crayfish must be fed every two days according to instructions on food label; sinking algae pellets will be provided. Dead crayfish must be disposed of in a biohazard bag and placed in designat ed freezer prio r to EHS incineration disposal. The same applies to euthanization. Lids on habitats must remain sealed to prevent unintended escape of P. clarkii . Weights must be placed on the lid to prevent movement. Close attention must be paid to open habitats. If pers onne l must leave the area, lids must be closed. First aid kits are available in the adjacent laboratory. Habitats must be monitored every day (including weekends and holidays) by personnel Facility Information ADREC is on a septic sy stem which is pumped a nd f ed into the South Campus Anaerobic Digestor (SCAD). ADREC uses well water not treated with chlorine or fluoride. ADREC has a power redundancy and back - up generator available. Experimental Procedure 1. Open the container by lifting a ll lids simultaneously ; re cord the population distribution of the crayfish among the three zones. 2. Check and record water temperature. Make sure all air hoses are making bubbles. 3. Cover with lids and play sounds for 5 minutes, record population distribution. 4. Cover with lids and p lay sounds for 24 hours, record population distribution. 5. Feed P. clarkii according to instructions on container and allow 24 hours of silence before beginning the next experiment. 66 HABITAT CLEANING PROCEDURE Last updated 6/21/2019 Two cleaning regimes we re c ompleted weekly to control nutrient levels and pH in the crayfish habitats. The first was a gravel siphon and 50% water change, while the second was sponge filter cleaning. Two of the sponge filters in each habitat were new, whil - ch arge with microbes from established aquariums in Preuss Pets, a Lansing , MI pet store. This ensured the establishment of a healthy microbial colony to help manage filtering and nutrient levels. The order of cleaning was important t o prevent escape of Red Swa mp Crayfish (RSC). Since the tap water was untreated well - water, there was no need to dechlorinate water prior to refilling habitats. The following include the steps taken during each weekly cleaning. 1. Following an experimental tr ial on Monday or Tuesda y, a ll lids would be removed from Habitat #1. 2. Any crayfish remains would be removed (i.e. chela, legs, molt) and placed in a freezer bag prior to incineration. 3. Gravel siphoning would be completed until bottom area was complete or un til 50% of water was re move d (which ever happened first). 4. Sponge filters would be removed and rinsed under tap water and returned to their original position. 5. The number of RSC were counted and checked against the total number recorded for the habitat, duri ng the course of the ex peri mental trial. 6. Lids would be replaced, and steps 1 - 6 would be repeated for Habitat #2 and #3. 67 APPENDIX B : CHAPTER 3 SUPPLEMENTAL MATERIALS TREATMENT SCHEDULE Figure B.1 : Recorded schedule used for tracking the treatment being used. 68 ACOUSTIC POND TRIA L ANALYSIS IN R The following is an R Markdown code output for the statistical analysis of the field trials. Pond Trial Analysis Douglas Clements 8/4/2020 Set - up Load in libraries library (readxl) #reads excel documents library (ggplot2) #makes plots more ea sily customizable library (doBy) #adds the summaryby function library (dplyr) # Data manipulation library (tidyverse) library (ggpubr) library (writexl) # Writes excel files library (extrafont) #Adds in Times N ew Roman Set Theme and add custom color #Theme theme_set ( theme_pubr ()) # Creates Kelly Green (Spartan Green) SpartanGreen < - rgb ( 24 , 69 , 59 , maxColorValue = 255 ) # Import Times New Roman windowsFonts ( Times= windowsFont ( "Times New Roman" )) Read in Data # setwd("H:/School/Thesis Work/Sound Pond Trials /Data Fi les") DALL< - as.data.frame ( read_xlsx ( "MasterDataCrayfish.xlsx" , sheet = "Sheet1" )) head (DALL) #shows a sampleof data ## Trap_Number row.name Sample_Number Set_Date Pull_Date Day_Number ## 1 1 1 19ST005 2019 - 07 - 09 2019 - 07 - 1 0 1 ## 2 1 58 19ST006 2019 - 07 - 10 2019 - 07 - 11 2 ## 3 1 114 19ST007 2019 - 07 - 11 2019 - 07 - 12 3 ## 4 1 171 19TR260 2019 - 07 - 15 2019 - 07 - 16 5 ## 5 1 22 8 19TR813 2019 - 07 - 17 2019 - 07 - 18 6 ## 6 1 274 19TR810 2019 - 07 - 18 2019 - 07 - 19 7 ## Trap_Type Daily_Catch Sound Food Week_Number Volume Amp AnySp_InBerry ## 1 Gee MinnowTrap 0 1 1 1 0.5 1 0 ## 2 GeeMinnowTrap 1 0 1 1 0.5 1 0 ## 3 GeeMinnowTrap 1 0 0 1 0.5 1 0 ## 4 GeeMinnowTrap 0 0 1 2 0.5 1 0 69 ## 5 GeeMinnowTrap 0 0 1 2 0.5 1 NA ## 6 GeeMinnowTrap 1 0 1 2 0.5 1 NA ## CommentsTrap CommentsSample WaterTemp V olTreat VolLevel Treatment ## 1 NA NA NA Treatment1 Treatment1 Food+Sound ## 2 NA NA NA Treatment1 NA Food ## 3 NA NA Treatment1 NA Non e ## 4 NA NA 77 Tr eatment1 NA Food ## 5 82.4 Treatment1 NA Food ## 6 80.3 Treatment1 NA Food ## Distance ## 1 2 2 ## 2 22 ## 3 22 ## 4 22 ## 5 22 ## 6 22 Subset data by trap type geesonly< - DALL[ which (DALL $ Trap_Type == "GeeMinnowTrap" ),] #Gees Traps only, all treat ments PVConly< - DALL[ which (DALL $ Trap_Type == "Apartment" | DALL $ Trap_Type == "ART" ),] # P VC refuge traps ARTonly< - DALL[ which (DALL $ Tr ap_Type = = "ART" ),] # ART traps APARTonly< - DALL[ which (DALL $ Trap_Type == "Apartment" ),] # APART traps soundgees< - geesonly[ which (geesonly $ Treatment == "Sound" | geesonly $ Treatment == "Food+So und" ),] #Both Sound and Sound+Food soundonlyg ees< - gee sonly[ which (geesonly $ Treatment == "Sound" ),] #Sound soundfoodgees< - geesonly[ which (geesonly $ Treatment == "Food+Sound" ),] #Food+Sound FoodGees< - geesonly[ which (geesonly $ Treatment == "Food" ),] #Food NoneGees< - geesonly[ which (geesonly $ Treatment == "None" ),] #Silence Subset data by Treatment FoodOnly< - DALL %>% filter (Treatment == "Food" ) FoodandSound< - DALL %>% filter (Treatment == "Food+Sound" ) SoundOnly< - DALL %>% filter (Treatment == "Sound" ) None< - DALL %>% filter (Treatment == "None" ) 70 Data summary Combined Data Summa ry # Bas ic summaryBy with Daily catch as a factor against trap type (Daily catch mean is CPUE) DataSummary< - summaryBy (Daily_Catch ~ Treatment + Trap_Type, data = DALL, FUN = c ( mean ,length,sum)) DataSummary ## Treatment Trap_Type Daily_Catch.mean Daily_Ca tch.leng th Daily_Catch.sum ## 1 Food Apartment 1.6891892 74 125 ## 2 Food ART 1.1014493 69 76 ## 3 Food GeeMinnowTrap 0.6745731 1 054 711 ## 4 Food Juvenile 0.5333333 60 32 ## 5 Food Pyramid 0.8378378 74 62 ## 6 Food+Sound Apartment 1.4255319 47 67 ## 7 Food+Sound ART 1.2888889 45 58 ## 8 Food+Sound GeeMinnowTrap 0.8200900 667 547 ## 9 Food+Sound Juvenile 0.2857143 42 12 ## 10 Food+Sound Pyramid 1.2830189 53 68 ## 11 None Apartment 1.5882353 17 2 7 ## 12 None ART 1.4444444 18 26 # # 13 None GeeMinnowTrap 0.4874552 279 136 ## 14 None Juvenile 2.2307692 13 29 ## 15 None Pyramid 0.9230769 13 12 ## 16 Sound Apartment 1.5000000 6 9 ## 17 Sound ART 1.6666667 6 10 ## 18 Sound GeeMinnowTrap 0.6436782 87 56 ## 19 Soun d J uvenile 0.0000000 4 0 ## 20 Sound Pyramid 1.7000000 10 17 Summari # Basic summaryBy with Daily catch as a factor against trap t ype (Dai ly catch mean is CPUE) GeesSummary< - summaryBy (Daily_Catch ~ Treatment, data = geesonly, FUN = c (mean,length,su m)) GeesSummary ## Treatment Daily_Catch. mean Daily_Catch.length Daily_Catch.sum ## 1 Food 0.6745731 1054 71 1 ## 2 Food+Sound 0.8200900 667 547 ## 3 None 0.4874552 279 136 ## 4 So und 0.6436782 87 56 ggplot ( data= GeesSummary, aes ( x= Treatment, y= Daily_Ca tch.mean)) + geom_col ( fill= SpartanGreen) + labs ( y= "CPUE" , title = "Gee's Minnow Traps Performance by Tr eatment" ) + theme ( plot.title = element_text ( hjust = 0.5 ), 71 text= element_text ( family= "Times" , face= "bold" , size= 14 )) + scale_x_discrete ( lim its = c ( "None" , "Food" , "Sound" , "Food+Sound" )) Daily_Catch.mean is CPUE. The highest CPUE on a treatment basis is Food & Sound . Note that Food and Sound are approximately equal. None represents an empty trap and is still nearly 2/3 of Food or Sound . Also, note tha t when Food - None and Sound - None are added to None , the results are approximately equal to Food & Sound . It seems the effects are addative, and if so, likely work on different mechanisms. Analysis on Treatment w Traps table (geesonly $ Daily_Catch) ## ## 0 1 2 3 4 5 6 7 8 11 ## 1252 474 219 82 37 10 4 5 3 1 Note the high number of Zeros in the daily catch data. Daily catch data should be checked for normality. ggpl ot ( data= geesonly, aes ( x= Daily_Catch)) + geom_histogram ( fill= "#18453b" ) + labs ( x= "Daily Catch" , y= "Frequency" , title = "Histogram of Gee's Minnow Trap Daily Catch" ) + 72 theme ( text= element_text ( family= "Times" , face= "bold" , size= 12 )) + scale_y_continuous ( expa nd= c ( 0 , 0 ), limits = c ( 0 , 1400 )) ggsave ( "GeesHistogram.png" , width = 8 , height = 8 ) ggqqplot (geesonly $ Daily_Catch, color = SpartanGreen) 73 Non - normal, since most dots are outside the CI. Finally, lets check with a Shapir o - Wilk Test; if shapiro.test (geesonly $ Daily_Catch) ## ## Shapiro - Wilk normality test ## ## data: geesonly$Daily_Catch ## W = 0.65835 , p - value < 2.2e - 16 Non - normal distribution. However, depending on the sa mple siz e, we can use a parametric test anyway because of the Central Limit Theorem. Statistical Tests kruskal.test (Daily_Catch ~ Treatment, data = geesonly) ## ## Kruskal - Wallis rank sum test ## ## data: Daily_Catch by Treatment ## Kruskal - Wallis chi - squared = 26.674, df = 3, p - value = 6.892e - 06 summary ( aov (Daily_Catch ~ Treatment, data= geesonly)) 74 ## Df Sum Sq Mean Sq F value Pr(>F) ## Treatment 3 23.1 7.708 6.172 0.000357 *** ## Residuals 2083 2601.4 1.249 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 pairwise.wilcox.test (geesonly $ Daily_Catch, as.factor (geesonly $ Treatment), p.adjust= "BH" ) ## ## Pairwise comparisons using Wilcoxon rank sum test ## ## data: geesonly$Dail y_Catch and as.factor(geesonly$Treatment) ## ## Food Food+Sound None ## Food+Sound 0.0042 - - ## None 0.0042 3.5e - 06 - ## Sou nd 0.9425 0.2098 0.1307 ## ## P value adjustment method: BH Analysis on Dist ance to Speaker kruskal.test (Daily_Catch ~ Distance, data = soundfoodgees) ## ## Kruskal - Wallis rank sum test ## ## data: Daily_Catch by Distance ## Kruskal - Wallis chi - squared = 100.73, df = 96, p - value = 0.3505 kruskal.test (Daily_Catch ~ Distance, dat a = soun donlygees) ## ## Kruskal - Wallis rank sum test ## ## data: Daily_Catch by Distance ## Kruskal - Wallis chi - squared = 45.769, df = 43, p - value = 0.3579 kruskal.test (Daily_Catch ~ Distance, data = soundgees) ## ## Kruskal - Wallis rank sum test ## ## data: Daily_Catch by Distance ## Kruskal - Wallis chi - squared = 117.06, df = 96, p - value = 0.071 summary ( aov (Daily_Catch ~ Distance, data= soundfoodgees)) 75 ## Df Sum Sq Mean Sq F value Pr(>F) ## Distance 1 3.0 3.039 2.124 0.145 ## R esiduals 665 951.4 1.431 summary ( aov (Daily_Catch ~ Distance, data= soundonlygees)) ## Df Sum Sq Mean Sq F value Pr(>F) ## Di stance 1 0.43 0.4260 0.444 0.507 ## Residuals 85 81.53 0.9592 summary ( aov (Daily_Catch ~ Distance, data= sou ndgees)) ## Df Sum Sq Mean Sq F value Pr(>F) ## Distance 1 3.2 3.184 2.312 0.129 ## Residuals 752 1035.6 1.3 77 treatment with sound . Analys is on Trap Type for Food Treatment kruskal.test (Daily_Catch ~ as.factor (Trap_Type), data = FoodOnly) ## ## Kruskal - Wallis rank sum test ## ## data: Daily_Catch by as.factor(Trap_Type) ## Kruskal - Wallis chi - squared = 71.915, df = 4, p - value = 8. 943e - 15 summary ( aov (Daily_Catch ~ as.factor (Trap_Type), data= FoodOnly)) ## Df Sum Sq Mean Sq F value Pr(>F) ## as.factor(Trap_Type) 4 83.1 20.784 13.89 4.22e - 11 *** ## Residuals 1326 1984.5 1.497 ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 76 APPENDIX C : CHAPTER 4 SUPPLEMENTAL MATERIALS GENERATED DESIGN CONCEPTS The following are images of generated design concepts, transferred from engineering paper. Figure C .1 : PVC p ipes are tipped upwards using electromagnets to empty crayfish into a cage. 77 Figure C.2 : Crayfish enter the refuge style traps before a pump intermittently empties traps into a cage through a flapper valve. 78 Figure C.3 : The third design uses a combinati on of semi - cylindrical trap and refuge trap to direct crayfish toward a central floating cage. 79 Figure C.4 : This design uses refuge tubes with an angle at one end to act as a one - way valve for crayfish entry into a cage. An entrance similar to Minnow Trap us used on each end. 80 Figure C.5 : The Dragon design is made of a corrugated plastic tube that coils up for storage and is uncoiled during deployment. Refuge tubes with flapper valves allow crayfish to enter into the corrugated tube. 81 Figure C.6 : Later termed the SoftValve, the concept is to use flexible bristles or fib ers to allow a crayfish in one - way but not the opposing direction. 82 BLUEPRINTS FOR CURRENT DESIGN Figure C.7 : Blue prints of the current design were created using A utodesk I nventor. 83 REFERENCES 84 REFERENCES Aquiloni, L., Becciolini, A., Berti, R., Porciani, S., Trunfio, C., & Gherardi, F. (2009). Managing invasive crayfish: use of X - ray sterilisa tion of m ales. Freshwater Biology , 54 (7), 1510 1519. https://doi.org/1 0.1111/j.1365 - 2427.2009.02169.x Aquiloni, L., Brusconi, S., Cecchinelli, E., Tricarico, E., Mazza, G., Paglianti, A., & Gherardi, F. (2010). Biological control of invasive populations o f crayfis h: the European eel (Anguilla anguilla) as a predator of Proc ambarus clarkii. Biological Invasions , 12 (11), 3817 3824. https://doi.org/10.1007/s10530 - 010 - 9774 - z Ausmus, L, Kontorousis, A, Li, B, Tang, X. (2018). Crayfish Will . East Lansing. Barb aresi, S., Tricarico, E., & Gherardi, F. (2004). Factors inducing the intense burrowing activity of the red - swamp crayfish, Procambarus clarkii, an invasive species. Naturwissenschaften , 91 (7), 342 345. https://doi.org/10.1007/s00114 - 004 - 0533 - 9 Beaver, M. , Clements , D., Frost, H., & Ma, X. (2019). Invasive Crustacean Eradication . East Lansing. Boets, P., Lock, K., Cammaerts, R., Plu, D., & Goethals, P. L. M. (2009). Occurrence of the invasive crayfish Procambarus clarkii (Girard, 1852) in Belgiu m (Crustac ea: Cambar idae). Belgian Journal of Zoology , 139 (2), 173 176. Booy, O., Cornwell, L., Parrott, D., Sutton - Croft, M., & Williams, F. (2017). Impact of Biological Invasions on Infrastructure. In M. Vilà & P. E. Hulme (Eds.), Impact of Biological I nvasions o n Ecosyste m Services (pp. 235 247). https://doi.org/10.1007/978 - 3 - 319 - 45121 - 3_15 Boyd, V. (n.d.). Mudbug Season. Rice Farming . Retrieved from https://www.ricefarming.com/uncategorized/mudbug - season/ Breithaupt, T., & Tautz, J. (1990). The Sensi tivity of Crayfish M echanoreceptors to Hydrodynamic and Acoustic Stimuli. In K. Wiese, W. - D. Krenz, J. Tautz, H. Reichert, & B. Mulloney (Eds.), Frontiers in Crustacean Neurobiology (pp. 114 120). https://doi.org/10.1007/978 - 3 - 0348 - 5689 - 8_12 Buscaino, G., Filiciott o, F., Buf Maccarrone, V. (2012). The underwater acoustic activities of the red swamp crayfish Procambarus clarkii . The Journal of the Acoustical Society of America , 132 (3), 1792 1798. http s://doi.or g/10.1121/ 1.4742744 Correia, A. M. (2003). Food choice by the introduced crayfish Procambarus clarkii. Annales Zoologici Fennici , 40 (6), 517 528. Retrieved from 85 http://www.jstor.org.proxy1.cl.msu.edu/stable/23736508 De Palma - Dow, A. A., Curti, J. N., & E mi Fergus, different passive trap types to effectively catch and control the invasive red swamp crayfish (Procambarus clarkii) in streams of the santa monica mountains. Management of Biological Invasions , 11 (1), 44 62. https: //doi.org/10.3391/mbi.2020.11.1.04 Denissenko, P., Lukaschuk, S., & Breithaupt, T. (2007). The flow generated by an active olfactory system of the red swamp crayfish. Journal of Experimental Biology , 210 (23), 4083 LP 4091. https://do i.org/10.1 242/jeb.00 8664 Edmonds, N. J. (7AD). A review of crustacean sensitivity to high amplitude underwater noise: Data needs for effective risk assessment in relation to UK commercial species. Marine Pollution Bulletin , 108 (1 2), 5 11. https://doi.org /10.1016/j .marpolbul .2016.05.006 Fanjul - Moles, M. L., & Fuentes - Pardo, B. (1988). Spectral sensitivity in the course of the ontogeny of the crayfish Procambarus clarckii. Comparative Biochemistry and Physiology Part A: Physiology , 91 (1), 61 66. https://doi.org/http s://doi.or g/10.1016/0300 - 9629(88)91592 - 7 Freeman, M. A., Turnbull, J. F., Yeomans, W. E., & Bean, C. W. (2010). Prospects for management strategies of invasive crayfish populations with an emphasis on biological control. Aquatic Conservation: Marine and F reshwater Ecosystems , 20 (2), 211 223. https://doi.org/10.1002/aqc.1065 Gherardi, F., & Barbaresi, S. (2007). Feeding Preferences of the Invasive Crayfish, Procambarus Clarkii. Bull. Fr. Pêche Piscic. , (387), 7 20. Retrieved from https://doi.org/10.1051/km ae:2007014 Gherardi, F., & Cioni, A. (2004). Agonism and Interference Competition in Freshwater Decapods. Behaviour , 141 (10), 1297 1324. Retrieved from http://www.jstor.org.proxy1.cl.msu.edu/stable/4536201 Goodall, C., Chapman, C., & Neil, D. (1990). The Acoustic Re sponse Threshold of the Norway Lobster, Nephrops Norvegicus (L.) in a Free Sound Field. In K. Wiese, W. - D. Krenz, J. Tautz, H. Reichert, & B. Mulloney (Eds.), Frontiers in Crustacean Neurobiology (pp. 106 113). https://doi.org/10.1007/978 - 3 - 034 8 - 5689 - 8_1 1 Goudkamp, J. E., Seebacher, F., Ahern, M., & Franklin, C. E. (2004). Physiological thermoregulation in a crustacean? Heart rate hysteresis in the freshwater crayfish Cherax destructor. Comparative Biochemistry and Physiology Part A: Molecular & Integrat iv e Physiology , 138 (3), 399 403. https://doi.org/https://doi.org/10.1016/j.cbpb.2004.06.002 Green, N., Bentley, M., Stebbing, P., Andreou, D., & Britton, R. (2018). Trapping for invasive crayfish: comparisons of efficacy and selectivity of baite d traps ve rs us novel artificial refuge traps. Knowl. Manag. Aquat. Ecosyst. , (419). Retrieved from 86 https://doi.org/10.1051/kmae/2018007 Greenspan, M., & Tschiegg, C. E. (10AD). Speed of sound in water by a direct method. Journal of Research of the Nationa l Bureau o f Standards (1934) , 59 (4). https://doi.org/10.6028/jres.059.028 Hobbs III, H. (1993). Trophic Relationships of North American Freshwater Crayfishes and Shrimps. Contributions in Biology and Geology / Milwaukee Public Museum , 85 , 1 110. Holdich, D. M., Gy de mo, R., & Rogers, W. D. (2017). A review of possible methods for controlling nuisance populations of alien crayfish. In Crayfish in Europe as Alien Species (pp. 245 270). Routledge. Horner, A. J., Schmidt, M., Edwards, D. H., & Derby, C. D. (2 008). Role o f the olfactory pathway in agonistic behavior of crayfish, Procambarus clarkii. Invertebrate Neuroscience , 8 (1), 11 18. https://doi.org/10.1007/s10158 - 007 - 0063 - 1 Ingle, R. (1997). Crayfishes, lobsters, and crabs of Europe: an illustrated guide to common a nd traded species (Academic). Springer, Dordrecht. Kilian, J. V., Klauda, R. J., Widman, S., Kashiwagi, M., Bourquin, R., Weglein, S., & Schuster, J. (2012). An assessment of a bait industry and angler behavior as a vector of invasive species. Biological Invasions , 14 (7), 1469 1481. https://doi.org/10.1007 /s10530 - 012 - 0173 - 5 Moonga, K., & Musuka, C. G. (2014). The effect of accidentally introduced red swamp crayfish ( Procambarus clarkii ) in Kafue fishery . 2 (1), 8 15. Moore, R. J., & Fierro; N and Internet Resources. Experimental Agriculture , 44 (4), 576 577. https://doi.org/http://dx.doi.org/10.1017/S0014479708006911 Nagy, R., Fusaro, A., Conard, W., & Morn ingstar, C. (2019). Nonin digenous Aquatic Species Database. Retrieved April 11, 2019, from USGS - NAS website: https://nas.er.usgs.gov/queries/FactSheet.aspx?SpeciesID=217 - R. (2018). Crayfish arts: an evaluation into t he efficacy of artificial refuge traps for monitoring lotic white - clawed crayfish Austropotamobius pallipes (Lereboullet, 1858) (Decapoda, Astacidae) populations. Crustaceana , 91 (3), 297 309. https://doi.org/10.1163/15685403 - 00003772 Offutt, G. C. (1970). Acoustic stimulus percep tion by the American lobsterHomarus americanus (Decapoda). Experientia , 26 (11), 1276 1278. https://doi.org/10.1007/BF01898016 Paillisson, J. - M., Soudieux, A., & Damien, J. - P. (2011). Capture efficiency and siz e selectivity of sampling gears targeting red - swamp crayfish in several freshwater habitats. Knowl. Managt. 87 Aquatic Ecosyst. , (401). Retrieved from https://doi.org/10.1051/kmae/2011015 Parkyn, S. M., Distefano, R. J., & Imhoff, E. M. (2011). Comparison of constructed microhabitat and Baited traps in Table Rock Reservoir, Missouri, USA. Freshwater Crayfish , 18 (1), 69 74. https://doi.org/10.5869/fc.2011.v18.69 Plummer, M. R., Tautz, J., & Wine, J. J. (1986). Frequency coding of waterborne vibrations by abdo minal mechanosensory interneurons in the crayf ish,Procambarus clarkii. Journal of Comparative Physiology A , 158 (6), 751 764. https://doi.org/10.1007/BF01324819 Popper, A. N., Salmon, M., & Horch, K. W. (2001). Acoustic detection and communication by decap od crustaceans. Journal of Comparative Physiol ogy A , 187 (2), 83 89. https://doi.org/10.1007/s003590100184 Stancliffe - Vaughan, A. (2015). Sampling UK Pacifastacus leniusculus (Dana, 1852): the effect of trapping on population structure . https://doi.org/10. 13140/RG.2.1.279 3.6721