BEHAVIORAL AND NEUROPHYSIOLOGICAL RESPONSES TO 
MECHANOSENSORY SIGNALS ASSOCIATED WITH THE HULA BEHAVIOR IN 
MALE AXOLOTLS (AMBYSTOMA MEXICANUM) 

By 

Taylor M. Rupp 

A DISSERTATION 

Submitted to 
Michigan State University 
in partial fulfillment of the requirements   
for the degree of 

Zoology – Doctor of Philosophy 
Ecology, Evolutionary Biology, and Behavior – Dual Major 

2024 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ABSTRACT 

Studying communication signals from the perspective of both senders and 

receivers allows us to create a greater understanding of how signals are exchanged 

between animals. For my doctoral dissertation, I leveraged the courtship ritual of the 

axolotl, Ambystoma mexicanum, a fully aquatic salamander, to investigate mechanisms 

of signal generation and detection. I focused specifically on a courtship behavior called 

the “hula”, which involves a rhythmic swaying motion of the pelvic region combined with 

an undulating movement of the tail. Although the hula has been historically been 

considered a male behavior, female axolotls play a critical role in the hula behavior as 

the receiver, and may potentially participate in the male hula by nudging the cloaca or 

changing proximity to the hula-ing male. Additionally, both males and females perform 

the hula behavior during courtship; here, I focused on male to female communication 

mediated by the hula. The hydrodynamic stimuli generated by the hula behavior may 

serve as communication signals; female axolotls can presumably detect these stimuli 

via their lateral line system, which is capable of sensing both mechanical and electrical 

signals in aquatic environments. For my dissertation research, I characterized the 

motion patterns that males displayed while hula-ing and measured how females 

responded to a range of hydrodynamic stimuli, both through a behavioral and 

neurophysiological lens.  

Critically, I also aimed to determine if a principle known as the sender-receiver 

matching hypothesis occurs in the mechanosensory lateral line system; this hypothesis 

posits that the physical properties of signals are shaped by the sensory responses of 

the receiver, and vice versa. This principle has been most notably demonstrated in 

 
auditory communication; for example, the frequency and rise time of calling songs in 

tree crickets and songbirds (respectively) have both been shown to match the sensitivity 

of their receiver’s auditory system. However, it is currently unclear if sender-receiver 

matches occur within vibratory communication. Importantly, my research serves as a 

building block to ascertain whether sender-receiver matching occurs in the 

mechanosensory lateral line system. 

In Chapter 1, I characterized courtship interactions between male and female 

axolotls. I accomplished this by placing pairs of male and female axolotls into an 

aquarium together and then observing their behaviors over a 24 hr period. Chapter 2 

describes the typical motion patterns that male axolotls exhibit during the hula behavior, 

as well as female behavioral responses to specific combinations of hula parameters. 

Chapter 3 describes neurophysiological responses of the anterodorsal lateral line nerve 

to the hydrodynamic stimuli that are generated during the hula behavior.  

I was able to demonstrate a moderate degree of sender-receiver matching in the 

female lateral line system; I found that male axolotls were most likely to display a sweep 

angle of 30° while hula-ing, and that the anterodorsal lateral line nerve of females had a 

significant excitatory response to 30°. However, given that I only tested sweep angles 

and speeds within the range of what male axolotls actually display during courtship, it is 

unknown if females exhibit significant excitatory responses to more extreme sweep 

angles and speeds. Thus, my findings serve to progress research on the lateral line 

system by providing insight into the level of matching between hydrodynamic signals 

generated by male axolotls as well as the behavioral and neurophysiological responses 

exhibited by female axolotls.  

Copyright by  
TAYLOR M. RUPP 
2024

 
 
 
 
 
 
 
 
 
 
 
 
 
 
For Rick, 
whose contributions to my research are immeasurable. 

v 

 
 
 
 
 
 
 
 
 
 
 
 
 
ACKNOWLEDGEMENTS 

I would like to thank my advisor, Dr. Heather Eisthen, for supporting my wild 

ideas and guiding me through the toughest and most rewarding journey of my life. 

Thank you to my committee members: Dr. Jeanette McGuire, for her incredible support 

with all things related to statistical analysis and interpretation, and Drs. Kay Holekamp 

and Weiming Li, for helping to make my dissertation the best it could possibly be.  

Thank you to Sergio Acuna, William Burke, Aalayna Green, Ayley Shortridge, 

and Faith White for their support in data collection and analysis. Thank you to Drs. 

Sheryl Coombs and Chris Braun for helping me make a smooth transition into lateral 

line research. Thank you to Drs. Nathan Tykocki and Jason Gallant for offering their 

time and expertise to get my experiments off the ground. Thank you to my mentors, Drs. 

Art Martin and Daniel Wagenaar, who believed I could be a scientist and always held 

me to the highest standard. 

To my friends and cohort, thank you for cheering me on and for keeping my 

spirits uplifted. To my family, thank you for always being in my corner. And lastly, thank 

you to my fiancé, Richard Pease, for creating the brilliant Robotail that my entire 

dissertation hinged upon, and for being my anchor through it all. 

My research was supported by the US National Science Foundation (IOS - 

1354089), the BEACON Center for the Study of Evolution in Action, and MSU’s EEB 

Program and Department of Integrative Biology. Special thanks to Laura Muzinic of the 

Ambystoma Genetic Stock Genetic at the University of Kentucky (supported by NIH 

grant P40-OD019794) for providing the animals necessary for my experiments.   

vi 

TABLE OF CONTENTS 

CHAPTER 1: Characterization of Axolotl Courtship Behaviors………………………..….. 1 
REFERENCES……………………………………………………………….……….. 32 

CHAPTER 2: Behavioral Responses of Female Axolotls to the Male Hula Behavior…. 38
REFERENCES………………………………………………………………..…...….. 83 
APPENDIX…………………………………………………………………………….. 89 

CHAPTER 3: Lateral Line Responses to the Male Hula Behavior………………….…… 95 
REFERENCES…………………………………………………………………...….. 128 

vii 

 
 
 
 
 
CHAPTER 1: Characterization of Axolotl Courtship Behaviors 

ABSTRACT 

Animals exhibit a wide diversity of behaviors during their courtship rituals, which 

creates a wealth of opportunities to understand important topics in behavioral biology 

such as communication strategies, sexual selection, and the evolution of behavior. The 

axolotl, Ambystoma mexicanum, a fully aquatic salamander, has a dynamic repertoire of 

courtship behaviors as well as the ability to breed year-round, which makes them 

especially suitable for courtship studies. The goal of our behavioral survey was to 

describe the durations of time that male and female axolotls spend performing various 

courtship behaviors to establish a baseline for comparison in future studies; to 

accomplish this, we quantified courtship behaviors of both sexes, as well as female 

locomotion patterns specifically, throughout their entire courtship ritual.  

The axolotl “hula” behavior (an undulating motion of the tail and hips) was 

previously thought to be a male courtship behavior, but we found that males and 

females generally performed the hula to the same degree. Additionally, we observed 

that males performed the tactile “bump” behavior (making physical contact between the 

male’s cloacal gland and female’s snout) during spawning attempts more frequently 

than females did. Furthermore, female axolotls modulated their locomotion patterns 

based on their proximity to a hula-ing male. These results revealed the significance of 

both male and female hula-ing, the necessity of tactile cues in successful courtship and 

spawning, and a potential locomotion strategy for localizing a hula-ing male. Overall, our 

results allowed us to gain a greater understanding of the courtship dynamics between 

1 

male and female axolotls and also provided important context for the upcoming 

chapters of this dissertation. 

INTRODUCTION 

Animal courtship rituals provide researchers with a unique substrate for studying 

communication strategies and understanding complex behavioral interactions. Courting 

animals exchange countless cues and signals with conspecifics, which includes not only 

competitors, but potential mates as well. Males and females display a wide range of 

courtship behaviors that engage multiple senses (e.g., visual, tactical, auditory, 

olfactory) both alone, and in combination.  Therefore, there are many opportunities for 

researching specific communication events that include both signalers and receivers. 

Salamanders specifically have exceptionally diverse courtship behaviors which affords 

behavioral researchers with opportunities to evaluate inputs of multiple senses alone 

and in combination (Arnold, 1977; Arnold, 1987; Halliday, 1990; Houck & Arnold, 2003; 

Salthe, 1967; Verrell, 1999), especially within the family Ambystomatidae (Anderson, 

1961; Arnold, 1976; Shoop, 1960). 

The axolotl (Ambystoma mexicanum), a fully aquatic ambystomatid salamander, 

is an excellent model system to evaluate complex courtship behaviors. Axolotls can 

breed year-round (Armstrong & Malacinski, 1989) and do not adhere to a breeding 

seasonality, allowing for courtship evaluation studies to be conducted year round.  Both 

males and females exhibit a diverse repertoire of courtship behaviors (Arnold, 1972), 

but a full characterization of courtship behaviors and their frequencies has not yet been 

documented. The purpose of this chapter is to characterize both male and female 

courtship behaviors and to establish a contextual foundation for the other chapters in 

2 

this dissertation. To accomplish this aim, we staged 30 encounters between male and 

female axolotls and quantified their courtship behaviors as well as female locomotion 

patterns throughout the courtship ritual. 

In the section that follows, we describe the axolotl’s courtship ritual, which we 

divided into 2 stages for our survey: the “preliminary stage” and the “mating stage”. This 

labeling scheme is a modified version of Salthe's (1967) general description of 

salamander courtship, which includes 5 distinct stages of courtship that are labeled 

stages A - E. Salthe also labels “stage A” as “a preliminary stage”; here, we also used 

the phrase “preliminary stage” but consolidated Salthe’s remaining stages into what we 

call the “mating stage”. Thus, we will refer to male and female interactions across both 

stages overall as the “courtship ritual” or simply “courtship”. 

Axolotl courtship ritual 

The function of the axolotl’s courtship ritual is to coordinate the process of 

transferring sperm from male to female. Female axolotls fertilize their eggs internally but 

rely on external sperm transfer, which is facilitated by the use of a spermatophore; each 

one, produced by the male, consists of a gelatinous base with a short stalk capped by a 

round pellet of sperm (sperm cap) that the female will pick up with her cloaca. Females 

are active receivers at the sperm collection phase, and successful copulation 

necessitates female involvement during courtship and while receiving sperm (i.e., 

picking up the sperm cap). Axolotl courtship tends to progress in a cyclical fashion, both 

within and between stages. Thus, a courting pair may alternate between preliminary 

stage behaviors several times before transitioning to the mating stage, and the pair may 

then return to the preliminary stage to begin the ritual over again (Salthe, 1967). 

3 

The preliminary stage begins with a male approaching a female and is 

punctuated by the male performing a “push” behavior; the male presses his snout 

against the female, typically against her lateral torso but sometimes against her head. 

Pushing can either involve a slight tapping motion or it may escalate to shoving; males 

often move females across the arena by forcefully walking or swimming into them. The 

preliminary stage also includes a behavior known as the “hula” (Verrell, 1982), which is 

exhibited by both sexes. Hula-ing involves a rhythmic swaying of the hips and posterior 

legs combined with an undulating motion of the tail; we have observed both sexes either 

holding their tails horizontally or upwards at an angle. Axolotls may hula at a distance 

from each other (which we have defined here as occurring at least one male tail-length 

away) or at a closer range (which we defined as within one male tail-length). 

After the preliminary stage ends, the pair typically (but not always) proceeds to 

the mating stage. The mating stage constitutes an escalation of the preliminary stage in 

that it involves all of the behaviors from the preliminary stage, but also includes 

additional behaviors that are necessary for successful spawning such as the “follow” 

behavior (Shoop, 1960). The follow behavior facilitates spawning by coordinating the 

process of external sperm transfer between male and female. The female initiates a 

follow bout by walking closely behind the male as he slowly walks forward. The female’s 

snout remains close to the male’s cloacal gland as he walks and elevates his tail slightly 

to accommodate the female’s head. During the mating stage, male axolotls will typically 

deposit at least one spermatophore onto the substrate; then, the male will move forward 

about one body length and stop, positioning the female’s cloaca directly over the 

4 

spermatophore. Finally, the female will remove the sperm cap with her cloaca, leaving 

the gelatinous base behind.  

Several additional courtship behaviors may occur while the female is following 

behind the male. The mating pair’s forward progress is often interrupted by a behavior 

called “bumping” (Shoop, 1960), which can be initiated either by the male or female. A 

male axolotl initiates a bump when he steps backwards and touches either his cloacal 

gland or the proximal half of his tail to the female’s snout; female bumping is similar, 

except that the female initiates physical contact. Males also perform frequent bouts of 

hula-ing during following periods and can be observed raising and lowering their tails 

onto the female’s snout, which comprises a “tail-tapping” behavior (Arnold, 1976). 

Additionally, we have defined a “nudge tip” behavior such that either the male or female 

initiates contact between the female’s snout and distal half of the male’s tail. Once the 

mating stage has ended, the pair will separate from one another and may begin part or 

all of the courtship ritual over from the beginning.  

MATERIALS AND METHODS 

Subjects 

Adult axolotls (Ambystoma mexicanum) were obtained from the Ambystoma 

Genetic Stock Center at the University of Kentucky. Animals were housed in ~114 L 

aquaria in 40-100% Holtfreter’s solution (Armstrong et al., 1989) supplemented with 

Replenish™ solution at temperatures between 18 and 22°C; animals were separated by 

sex and each aquarium contained 1-3 animals. We programmed the lights in our facility 

to match the natural sunrise and sunset of Mexico City, Mexico (the native habitat of 

5 

axolotls) with monthly updates to the axolotls’ photoperiod. Housing and experiments 

were conducted with the approval of the Institutional Animal Care and Use Committee 

of Michigan State University (approval number: 10/15-154-00). 

Behavioral trials 

We recorded behavioral interactions from 30 pairs of male and female axolotls (N 

= 10 males, 19 females); each male was paired with 3 different females. We randomly 

chose 10 males from our colony for the first 10 trials and then repeated the order of 

males twice to complete the set of 30 trials. Female axolotls were selected at random 

for the entire study, with two caveats: 1) no female was used twice in a row and 2) 

females that laid eggs were not used again later in the survey. No male was paired with 

the same female twice, but some females were paired with multiple males; 10 females 

were used once, 7 females were used twice, and 2 females were used three times. 

For each trial, we placed a male and a female together into a ~114 L aquarium 

(approximately 90 cm long x 45 cm wide x 30 cm wide) filled with Holtfreter’s solution at 

the same concentration as their home aquaria. Because axolotls are nocturnal, we 

placed the pair together shortly before the lights in our animal facility turned off for the 

evening. We allowed the courting pair to interact overnight and then returned them to 

their home aquaria the following morning. We used two Sony Nightshot camcorders 

(model: CMOS) to record top-down and side views of each trial for roughly 12 hrs.   

Because the infrared (IR) light on the camcorders did not sufficiently illuminate 

the recording area, we placed a lightbox equipped with a series of waterproof infrared 

LED lights underneath the aquarium. We also placed two IR lamps behind the aquarium 

6 

and attached a layer of light-colored paper to the outer surfaces of the aquarium to 

diffuse the IR light. We do not believe that the behavior of our study animals was 

impacted by our use of infrared lighting. The absorbance spectra of isolated rod cells 

from a closely related species, the tiger salamander (Ambystoma tigrinum), indicate that 

their absorbance drops to approximately 0 between 640 and 700 nm (Cornwall et al., 

1984). Given that infrared radiation has wavelengths between 700 nm and 1 mm, it is 

unlikely that our subjects were able to detect the infrared lighting that we used in our 

study.    

Definition of courtship behaviors 

We conducted a preliminary behavioral survey of 12 complete trials (out of 30 

total) to 1) understand which courtship behaviors were relevant to our study and 2) 

determine an appropriate amount of video footage to score for each trial. The subset of 

12 trials consisted of data from 4 male and 10 female axolotls, and we examined 9 hrs 

of each trial. Two of these males were in trials that led to a spawning event, which were 

the only two trials in the entire survey that resulted in spawning; the other 2 males were 

selected at random and we examined each male’s trio of trials. We used Shoop’s (1960) 

description of mole salamander (Ambystoma talpoideum) courtship to guide our 

behavioral survey because it is similar to axolotl courtship.  

We first quantified a small number of courtship behaviors described by Shoop 

(1960), e.g., push, follow, male bump, and female bump. Shoop’s “follow” is the same 

as our follow behavior. However, Shoop (1960) refers to what we call “push” and 

“female bump” behaviors simply as “bumping”; in our survey, we distinguished them to 

avoid duplicates in our ethogram. While surveying the subset of trials, we noted 

7 

additional behaviors that we included in our final coding (female walk, swim, and pause, 

female locomote/pause near male, orient, male and female close/distant hula, nudge 

tip, deposit or pick up spermatophore, nip, bite, and attack) and described which 

behaviors can and cannot co-occur (Fig. 1.1). Additionally, we found that courting pairs 

are most active during the first 2.5 hrs after the lights turned off; thus, we used this time 

window for all subsequent analyses. After completing our initial survey, we then 

quantified the behaviors that are defined in our ethogram (Table 1.1) for all 30 trials.  

Categorization of courtship behaviors 

We organized our axolotl behaviors of interest into several categories: female 

locomotion, preliminary stage, mating stage, and aggressive behaviors. The co-

occurrence patterns of these categories (and the behaviors within them) are 

summarized in Figure 1 below. The “female locomotion” category refers to patterns of 

walking, swimming, and pausing, which is the absence of locomotion (Coombs et al., 

2014) (Table 1.1); we recorded these behaviors throughout the trial except during the 

mating stage. The females in our study were motionless for highly variable amounts of 

time (0.25 sec up to 2.5 hr), and although the term “pause” is typically reserved for 

bouts up to 50 sec (Kramer & McLaughlin, 2001) we will use “pause” to signify that the 

female was motionless. “Locomote/pause near male” (Park et al., 2008) is meant to 

differentiate female locomotion behaviors that occur near and farther away from the 

male when the pair is not engaged in the preliminary or mating stage (Fig. 1.1); all of the 

behaviors within these stages are already defined such that they designate the 

proximity of the pair to one another. Although “orient” is not technically a locomotion 

behavior, we included it in the female locomotion category because it occurs during 

8 

periods of “locomote/pause near male”; a female axolotl orients toward a male when 

she is “near” the male and swivels her head to face him (Park et al., 2008). 

The “preliminary stage” category includes push behaviors as well as male and 

female hula behaviors; all hula bouts were measured when the courting pair was not 

engaged in following, except for “female close hula” which may occur at any point 

during the trial. We defined the end of a preliminary stage such that the pair fails to 

exhibit pushing and/or hula-ing for at least 5 min or they start to engage in a follow bout 

(Fig. 1.1). The “mating stage” category consists of all preliminary stage behaviors, plus 

“follow” and behaviors associated with following; for example, we recorded the number 

of times a male deposited a spermatophore onto the substrate as well when a female 

picked one up. We did not independently measure bouts of male hula-ing or female 

locomotion during follow bouts because they both consistently co-occur with following. 

We defined the mating stage as starting when a pair begins to follow, and ending when 

the pair stops following, pushing, and/or hula-ing for 5 min or more (Fig. 1.1).  

The last behavioral category, aggression, includes three rare but notable 

behaviors that we have defined for this survey. Females may exhibit aggressive 

behaviors at any point during the courtship ritual. “Nip” occurs when the female bites the 

male’s tail and then immediately releases it. “Bite” is similar to “nip” except that the 

female keeps her jaws engaged on the male’s tail for a few seconds. “Attack” is an 

escalated version of “bite”; a female bites the male’s tail for an extended period of time 

while thrashing her head from side to side.  

We also recorded several non-behavioral events that we did not include in the 

ethogram (Table 1.1). We observed how many spermatophores were floating in the 

9 

aquarium during the trial, as they can detach from the glass surface easily and we could 

not always see when a male axolotl was actively depositing a spermatophore. We also 

counted the number of spermatophores that were present in the aquarium once the trial 

had ended the following morning; this value encompasses the total number of 

spermatophore depositions (floating and still attached to the aquarium floor). 

Additionally, we noted if the female laid eggs, which typically occurred within 24 hrs of 

the trial ending.  

Behavioral coding 

We used Behavioral Observation Research Interactive Software (BORIS; version 

7.9.RC1) to quantify axolotl courtship behaviors (Friard & Gamba, 2016). BORIS aids in 

behavioral coding by allowing users to assign individual behaviors to particular 

computer keys; the user then watches a video and presses the corresponding keys 

when they observe a behavior of interest occurring. Once the user has finished coding a 

trial, BORIS will output the number of occurrences and the duration (where applicable) 

of each behavior into a data sheet. 

Importantly, BORIS allows users to set an “exclusion matrix”, which designates 

which behaviors can co-occur with one another and which behaviors are mutually 

excluded. For example, male bumps can be coded while a follow period is occurring; 

the user would press the key for “follow” to begin the behavior, then press the key for 

“male bump” any time one occurred, and finally press the key for “follow” again once the 

follow period had ended. In contrast, a female axolotl cannot walk and swim at the same 

time, so these behaviors are mutually excluded. If a swim behavior is followed by a walk 

behavior BORIS will automatically toggle off swimming once the walking bout begins.  

10 

Statistical analyses 

We calculated summary statistics (mean, standard deviation, minimum, 

maximum) on all courtship behaviors (Table 1.2) using the “psych” package (version 

2.4.3; Revelle, 2009) in R (R Core Team, 2020), with zeros removed for total durations 

and average bout durations (i.e., total duration divided by count) but retained for count 

variables. Summary statistics are written as mean(±SD) in the text that follows. We also 

assessed data on female locomotion during male hula-ing, which we obtained by 

performing conditional behavior analyses in BORIS; we then calculated transitions 

between walking, swimming, and pausing (Table 1.3), as well as the proportions of time 

that females spent performing these behaviors (Table 1.4). Because females were 

repeated multiply in the dataset an unequal number of times, we also evaluated 

behaviors using general linear mixed-effects models in R (using packages “lme4” 

(version 1.1.35.3; Bates et al., 2015) and ”lmerTest” (version 3.1.3; Kuznetsova et al., 

2017)) with female ID as the random effect and restricted maximum likelihood (REML) 

set to “true”. We evaluated sex-based differences in hula durations (total and average 

bout duration), and occurrence of “bump” behaviors; in addition, we assessed 

differences in female locomotion transitions based on the presence or absence of a 

hula-ing male as well as the female’s proximity to a hula-ing male (Table 1.5). 

RESULTS 

Female locomotion  

Of the 150 min of each trial, female axolotls spent an average (±SD) of 50(±20) 

min locomoting (i.e., walking and swimming). Females spent the most amount of time 

11 

pausing (90(±27) min), followed by walking (43(±21) min), and spent the least amount of 

time swimming (7(±10) min) (Fig. 1.2). One female in our study did not locomote 

(“paused”) for the entirety of the trial, but performed non-locomotive behaviors and 

therefore was not excluded from analyses.  . Females paused on average for 3(±13.6) 

min per bout. Female axolotls that spent any amount of time near the male spent a total 

of 24(±10) min near him when he was not in the preliminary or mating stage. Females 

spent on average 0.5(±1.4) min near the male per bout. Females oriented towards the 

male an average of 6(±7) times during a trial. 

Female locomotion during male hula-ing  

The presence of a hula-ing male impacted female locomotive activity during the 

courtship ritual; however, the proximity of the male had little bearing on female 

locomotion. For reference, the males that hula’d for any amount of time spent 5(±4) min 

close hula-ing, 4(±4) min distant hula-ing, and 8(±8) min hula-ing in total when the pair 

was not engaged in a bout of following. Female transitioned between walking, 

swimming, and pausing significantly more often (7.90(±8.62) times/min) when a male 

was hula-ing versus when there was no hula-ing occurring (2.76(±1.27) times/min; t = -

3.474, p = 0.00128; Fig. 1.3). Females transitioned between walking, swimming, and 

pausing an average of 9.99(±18.61) times/min when a male was hula-ing close by and 

transitioned 10.09(±18.32) times/min when the male was hula-ing at a distance (t = 

0.025, p = 0.9803). Thus, the presence of a hula-ing male caused females to increase 

their transitions between locomotive states, but females exhibited overall the same 

activity levels regardless of whether the male was hula-ing close to them or at a 

distance. 

12 

Although the proximity of a male (close vs far) did not affect how often a female 

transitioned between locomotive states, the proximity of a hula-ing male did impact the 

proportions of time that female axolotls spent locomoting and pausing. When a male 

axolotl was hula-ing close to a female, she spent an average of 44% of the time 

walking, 0.01% of the time swimming, and 55% of the time pausing. In contrast, when a 

male was hula-ing at a distance, females spent an average of 59% of the time walking, 

2% of the time swimming, and 39% of the time pausing. Thus, females allocated time 

differently to locomotive and pausing behaviors when a male was hula-ing at a distance, 

compared to when a male was hula-ing at close proximity to the female.  

Preliminary and mating stages 

The 27 pairs of axolotls that courted spent an average of 31(±29) min in the 

preliminary stage (pushing and hula-ing), and the 2 pairs that mated spent an average 

of 75(±44) min in the mating stage (preliminary stage behaviors plus following) per 150-

min trial. Among males that exhibited pushing behaviors, males pushed females for an 

average of 8(±7) min total; each bout of pushing lasted on average for 0.1(±0) min. The 

pairs in our study invested a substantial amount of time in the preliminary stage and an 

even greater period of time in the mating stage, suggesting that the axolotl courtship 

ritual involves a high level of effort and coordination between the sexes.  

In general, males and females did not differ substantially in the overall amount of 

time spent hula-ing, but the duration of each bout of hula-ing was longer for males than 

females. Among individuals that exhibited the hula (10 males, 14 females), males hula’d 

in total (close + distant) for 8(±8) min and females hula’d for 5(±13) min (t = 1.43, p = 

0.16). However, male axolotls hula’d for longer bouts of time for each hula event 

13 

(0.1(±0.1) min) than females did (0.1(±0) min; t = 2.38, p = 0.02). At distant ranges, 

males hula’d for 4(±4) min and female hula’d for 3(±9) min (t = 0.55, p = 0.59). Male 

axolotls hula’d distantly for longer bouts of time (0.2(±0.1) min) than females did 

(0.1(±0.1) min; t = 3.80, p = 0.0008).  

When in close proximity, males hula’d more than females. Males hula’d for an 

average of 5(±4) min, whereas females hula’d for an average of 2(±5) min (t = 2.45, p = 

0.02; Fig. 1.4). On average, each bout of close hula-ing that males performed lasted 

0.1(±0.1) min; for females the average was similar at 0.1(±0) min (t = 0.87, p = 0.39; 

Fig. 1.5).  

The 27 courting pairs in our survey spent approximately 15% of their time in the 

mating stage, and frequently exhibited behaviors related to spawning during those 

periods, including following, bumping, and spermatophore depositions. The males and 

females that participated in the mating stage engaged in following for an average of 

23(±14) min per trial; each bout of following lasted an average of 0.8(±0.3) min. We 

found that males bumped females (90(±145) significantly more often than females 

bumped males (51(±71), t = 2.45, p = 0.02; Fig. 1.6). Courting pairs that exhibited 

“nudge tip” did so an average of 7(±13) times. Male axolotls deposited spermatophores 

in 40% of trials (i.e., 12 out of 30 trials) determined by both observation on the video 

and through inspection of the aquaria after each trial.  Males deposited a maximum of 3 

spermatophores per trial where deposition was visible in the video footage, but we 

counted up to 5 in the aquarium when each trial was completed; certain 

spermatophores were not visible in our footage because they are clear, mostly 

colorless, and small. Due to the recording angle of the video camera we were unable to 

14 

observe females picking up a spermatophore; however, 2 females laid eggs after their 

trials had ended. 

Aggression 

Female aggression towards the male during the courtship ritual was minimal and 

restricted to nipping behavior. Nipping occurred in 27% of our trials (8 out of 30), 

however the number of nips was limited; of the trials where nipping happened, we 

observed an average of 1.25 nips. We never observed females biting or attacking the 

male.  

DISCUSSION  

Female locomotion behaviors 

Over the course of their entire courtship ritual, female axolotls spent the greatest 

amount of time pausing, followed by walking and then swimming (Fig. 1.2). When the 

male axolotl was performing the hula behavior, females transitioned between walking, 

swimming, and pausing significantly more often (~8 times/min) than when the male was 

not hula-ing (~3 times/min; Fig. 1.3). In contrast, the proximity of a hula-ing male (i.e., 

close vs distant) did not have a significant impact on how many times the female 

transitioned between locomotive states, but did impact the allocation of time in each 

locomotive behavior.  

Females modified the proportions of time that they spent walking, swimming, and 

resting in response to the proximity of a hula’ing male. When a male was hula-ing in 

close proximity, females spent ~40% of the time walking and ~60% of the time pausing, 

whereas the female spent ~60% of the time walking and ~40% of the time pausing is 

15 

the male was hula-ing at a distance. Therefore, females spent a larger proportion of 

time locomoting when the male was hula-ing at a distance, and paused rested more 

when the male was hula-ing at close proximity.  

Thus, it appears that females change their rates of transitioning between 

locomotive states depending on the presence or absence of a hula-ing male, but they 

modify the proportions of time spent on walking, swimming, and resting relative to the 

proximity of the hula-ing male. The male’s hula behavior likely generates sensory cues 

that can be propagated over short and long distances, including electrical cues (via 

muscle contractions (Soleymani et al., 2017)) and mechanosensory cues (i.e. water 

perturbations (Satou et al., 1991)), visual stimuli (Verrell, 1982), and the dispersal of 

odorants (Maex et al., 2016). The combination of these cues would allow for the female 

to detect the hula-ing male’s presence over multiple distances and adjust her 

locomotion levels accordingly. 

A possible explanation for the modulation of female locomotion patterns during 

male hula-ing is that females may increase their level of searching (i.e. walking more 

than pausing) if they detect vibrations in the water. From a distance, a male's hula 

movements may indicate the presence of another animal in the water, potentially a 

conspecific or a prey item, which may increase the female’s searching efforts. However, 

once the hula-ing male gets close to the female she may walk less and pause more 

often to gather sensory information from the male’s tail motions more effectively. 

Analogously, Kramer & McLaughlin (2001) posit that stillness reduces blur in the visual 

field, allows for detection of faint sounds, and decreases vibrations that may interfere 

with object localization across many invertebrate and vertebrate species; thus, pauses 

16 

in locomotion can help animals detect objects in their sensory fields more easily by 

reducing self-generated noise (Montgomery et al., 2014).  

The mate-search strategy of courting animals are known to be influenced by the 

presence of stimuli that are produced by conspecifics. Park et al. (2004) demonstrated 

that male axolotls, regardless of their prior experience level with females, increased 

their general activity levels when exposed to whole-body odorants from gravid (egg-

laden) or recently spawned females; similarly, the searching effort of male guppies, 

Poecilia reticulata, increased when they were exposed to female odorants (Guevara-

Fiore et al., 2010). Male axolotls have also been shown to decrease their general 

activity levels when presented with gravid female odorants; however, the experience 

levels of these males are unknown (Eisthen, unpublished). Park et al. (2004) also 

demonstrated that female axolotls did not change their activity levels when exposed to 

male whole-body odorants. We speculate that the females in our survey received other 

sensory signals from the hula-ing male in addition to any male odorants present in the 

aquarium, and that the presence of these multimodal signals may have increased the 

female’s activity levels, whereas isolated male odorants alone may not have been 

sufficient to increase the female’s search efforts in the experiments of Park et al. (2004). 

In contrast, the female may be prompted to pause more often once she is in 

closer proximity to the hula-ing male. An animal’s ability to sense their surroundings can 

be modulated by their own locomotion behaviors. For example, when the motor 

(efferent) nerve fibers of the mechanosensory lateral line system in the African clawed 

frog, Xenopus laevis, are electrically stimulated the corresponding sensing (afferent) 

fibers are inhibited from firing (Russell, 1968); under normal circumstances, the activity 

17 

of these afferent fibers would be suppressed by the animal’s own body motions 

(Lunsford & Liao, 2021; Plazas & Elgoyhen, 2021; Wullimann & Grothe, 2014). Given 

that X. laevis and axolotls share a similar lateral line system, female axolotls may 

experience a decrease in the sensitivity of their mechanosensory organs (i.e., 

neuromasts) during periods of walking or swimming. Thus, it may be advantageous for 

the female to pause more often once the hula-ing male is close to her to more 

effectively gather sensory information from the male without the interference of her own 

motions.  

Preliminary and mating stages 

Our pairs of male and female axolotls spent about 31 min out of each 150-min 

trial (approximately 21% of their time) in the preliminary stage, which was dominated by 

push behaviors as well as male and female hula behaviors. Tactile feedback via 

pushing appears to play an important role in the mating success of axolotls. The total 

duration of pushing had statistically significant, positive correlations with both 

spermatophore depositions (i.e., those that we observed during behavioral coding) and 

the total number of spermatophores we counted in the aquarium after each trial; in other 

words, males that pushed more often also tended to produce more spermatophores. In 

Ambystoma laterale, pushing behaviors may aid in the male’s recognition of the species 

and/or sex of other animals in the area. When placed in an arena with both sexes, A. 

laterale males will nudge males and females initially but then shift to nudging females 

only; it is thought that males may be able to detect a female-specific skin secretion or 

the presence of eggs (Storez, 1969). Thus, it appears pushing may be associated with 

18 

 
more frequent spawning attempts that are also aimed towards the correct species and 

sex. 

Although males hula’d more than females at close ranges (Fig. 1.4) and hula’d 

for longer bouts of time at distant ranges (Fig. 1.5), males and females hula’d to 

approximately the same degree in many other instances. Despite our findings, hula-ing 

and similar behaviors in axolotls and other salamanders have typically been regarded 

as a male behavior (Halliday, 1990; Maex et al., 2016; Verrell, 1983). It appears that 

researchers tend to assign more importance to male hula-ing in their descriptions of the 

behavior; in males, tail undulations are generally discussed in greater detail and given 

explicit labels like “hula” or “tail fan”. Although researchers have described hula-ing in 

female axolotls (Park et al., 2004) and hula-like behaviors in other female 

ambystomatids like Ambystoma macrodactylurn columbianum (Verrell & Pelton, 1996) 

as well as female plethdontids like Desmognathus wrighti (Verrell, 1999), female tail 

undulations are often given simpler descriptions and typically are not included in 

courtship ethograms. Observers often have preconceived notions about the behavioral 

roles of male and female animals (Ahnesjö et al., 2020), and the gender biases of 

observers may explain the discrepancy in the treatment of male and female hula-ing in 

the literature. 

 Pierotti et al. (1997) suggest that the reason male behaviors are emphasized in 

ethology journals is because both male and female researchers have largely been 

trained to focus on male-centered narratives. Given that many of the researchers who 

laid the groundwork for salamander courtship behaviors did so before gender biases in 

ethology were called into question starting in the mid-1990s, it is unsurprising that a 

19 

focus on male hula-ing has taken precedence over female hula-ing. In contrast, the 

observer’s gender appears to make little to no difference in the outcomes of behavioral 

coding of non-courtship behaviors, aggression and foraging, in salamanders specifically 

(Marsh & Hanlon, 2004). Although the function of hula-ing in female salamanders is 

unclear, it appears to play a significant role in the courtship dynamics of axolotls given 

that males and females hula to approximately the same degree. 

The axolotls in our study spent approximately 75 min in the mating stage 

(approximately 50% of the trial), and roughly 36% of that period was dedicated to bouts 

of following. All pairs that transitioned to the mating stage at least once exhibited the 

follow behavior multiple times over the course of a trial. Multiple spawning attempts are 

also a common feature in the courtship rituals of other ambystomatid species (Petranka, 

1982; Shoop, 1960; Verrell & Pelton, 1996). The bump behavior (which occurs during 

following) appears to be an important factor in the outcome of each spawning attempt. 

The function of courtship bumping has been described in the Alpine newt, Ichthyosaura 

alpestris (Denoël & Doellen, 2010), which exhibits a similar behavior during the 

spermatophore transfer process. In the Alpine newt’s courtship ritual, a female “tactile 

stimulus” (which we call a “female bump” in axolotls) serves to 1) communicate her level 

of responsiveness to the male and 2) lower the male’s chance of breaking contact with 

the female. Further, male Alpine newts perform a “push-back” behavior (which we call 

“male bump”), which pushes the female back toward the spermatophore in the event 

that she failed to locate it with her cloaca.  

We posit that male bumping in axolotls may also serve to correct the female’s 

lateral movements as she positions herself over the spermatophore; in our survey, we 

20 

observed males bumping their cloacal glands into the sides of the female’s snout as 

well as the center. Although male axolotls bumped females more than vice versa (Fig. 

1.6), both types of bumping seem to be important to overall spawning success; one of 

the two trials that resulted in spawning had the greatest number of combined bumps 

(772) out of any trial. The other trial with a spawning event had 175 combined bumps, 

which is moderate compared to other trials; however, this trial had the highest overall 

duration of male hula-ing. This observation suggests that the interaction between 

physical contact and male hula-ing may be an important component of successful 

mating.  

Aggression 

Occurrences of serious aggressive behaviors (biting, attacking) in our survey 

were not observed.  Nipping, while occurring in 27% of trials, was not a consistent 

feature of the axolotl courtship ritual; of the trials that featured nipping, females nipped 

at males on average ~1 time per trial. We were unable to discern if females were 

directly aggressive toward males or if they were instead exhibiting a predatory response 

to the male’s tail or limbs. 

CONCLUSION 

The courtship ritual of the axolotl is complex and dynamic, as the signaler and 

receiver respond not only to a range of touch and motion-based cues, but also modify 

their behaviors depending on the proximity to one another. Axolotl courtship involves a 

substantial time investment where the courtship behaviors are repeated and allows for 

changes and dynamic response based on specific current conditions. In general, 

21 

courting pairs spent more time in the mating stage than in the preliminary stage. We 

found that female axolotls transitioned between locomotion states more frequently when 

males performed hula-ing behaviors. This suggests that females are active receivers of 

male motion-based behaviors and modify their behavior in response. Additionally, 

females alter the time spent in each locomotive state depending on whether males are 

in close or far proximity, suggesting that they can respond to rapidly to changing 

signals. This survey allowed us to establish a baseline of courtship behaviors, define 

common and contrived male behaviors, and understand the context for how females 

naturally respond, serving as a foundation for future work on female behavioral and 

neurophysiological responses.   

22 

 
 
 
 
 
 
 
 
 
 
TABLES 

Behavior 

Description 

n
o

i
t

o
m
o
c
o

l

l

e
a
m
e
F

e
g
a
t
s

y
r
a
n
m

i

i
l

e
r
P

e
g
a

t
s
g
n

i
t

a
M

Walk 

Swim 

Pause 
Locomote/pause 
near male 

Orient 

Push 

Hula 
Male close hula 

Male distant hula 

Female close hula 

Female distant hula 

Follow 

Male bump 

Female bump 

Nudge tip 

  Deposit  
  Pick up 

i

n
o
s
s
e
r
g
g
A

Nip 

Bite 

Attack 

Female walks forward or backward at least 3 steps, with all 
four legs moving.  
Female swims forward using her tail, with legs pressed 
against the body. 
Female is not walking or swimming.  
Female is walking, swimming, or pausing and her head is 
within one male tail length of the male. 
Female has her head within one male tail length of the male 
and she turns her head to face the male. 
Male pushes against the female’s torso or head with his 
snout.  
Animal sways hips and posterior legs while undulating tail. 
Male performs hula near the female, within one tail length. 
Male performs hula at least one tail length away from the 
female. 
Female performs hula near the male, within one male tail 
length. 
Female performs hula at least one male tail length away from 
the male. 
Female walks closely behind the male as he steps forward. 
Starts when female snout is within one female head-length of 
the male's cloacal gland and ends when the male cloacal 
gland is at least one tail length away from the female's snout. 
Male walks backwards and touches his cloacal gland or 
proximal half of the tail to the female's snout. 

Female walks forward and touches the male's cloacal gland 
or proximal half of the tail with her snout. 

Contact is made between the female’s snout and distal half 
of the male’s tail. May be initiated by male or female. 

Male deposits a spermatophore onto the substrate. 
Female picks up a spermatophore with her cloacal gland.  

Female bites the male's tail and immediately releases it. 

Female bites the male's tail and briefly keeps jaws engaged. 

Female bites the male's tail and thrashes her head side to 
side. 

Table 1.1: Ethogram of male and female courtship behaviors. We classified 
courtship behaviors into four categories: female locomotion, preliminary stage, mating 
stage, and aggression. All of the preliminary stage behaviors may occur during the 
mating stage, whereas “follow” (and all behaviors associated with following) may only 
occur during the mating stage.  

23 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Behavior 

Count 

Total Duration 

NA 
Walk 
NA 
Swim 
NA 
Locomote (walk + swim) 
Pause 
NA 
Locomote/pause near male  NA 
Orient 

6(7) 0-32 

Push 
Male close hula 
Male distant hula 
Male all hula 
Female close hula 
Female distant hula 
Female all hula 
Total preliminary stage 

NA 
NA 
NA 
NA 
NA 
NA 
NA 
NA 

l

e
a
m
e
F

n
o

i
t

o
m
o
c
o

l

e
g
a
t
s

y
r
a
n
m

i

i
l

e
r
P

e
g
a

t
s
g
n

i
t

a
M

  Follow 

  Male Bump 

Female Bump 

  Nudge Tip 
  Deposit Spermatophore 
  Pick Up Spermatophore 
  Observe Spermatophore 
Total Spermatophores 

  Eggs 

Total mating stage 

i

n
o
s
s
e
r
g
g
A

Nip 

Bite 

Attack 

NA 
90(145) 0-583 
51(71) 0-297 
7(13) 0-59 
1(1) 0-3 
0(0) 0-0 
0(0) 0-2 
1(2) 0-5 
0(0) 0-1 
NA 

0(1) 0-2 

0(0) 0-0 

0(0) 0-0 

Average 
Duration 

0.2(0.1) 0.1-0.3 
0.2(0.1) 0.1-0.3 
0.2(0.1) 0.1-0.3 
3(13.6) 0.2-75 
0.5(1.4) 0-7.6 
NA 

0.1(0) 0-0.1 
0.1(0.1) 0-0.2 
0.2(0.1) 0.1-0.5 
0.1(0.1) 0-0.3 
0.1(0) 0-0.2 
0.1(0.1) 0-0.2 
0.1(0) 0-0.2 
NA 

0.8(0.3) 0.2-1.3 

NA 
NA 
NA 
NA 
NA 
NA 
NA 
NA 
NA 

NA 

NA 

NA 

43(21) 11-76 
7(10) 0-44 
50(20) 11-83 
90(27) 49-150 
24(10) 6-61 
NA 

8(7) 0-27 
5(4) 0-14 
4(4) 0-15 
8(8) 0-29 
2(5) 0-19 
3(9) 0-39 
5(13) 0-58 
31(29) 0-99 

23(14) 0-54 
NA 
NA 
NA 
NA 
NA 
NA 
NA 
NA 
75(44) 0-146 

NA 

NA 

NA 

Table 1.2: Summary statistics of male and female courtship behaviors. Data are 
displayed as mean(SD) minimum-maximum; all durations are listed in minutes. Zeroes 
have been removed from duration data but left in count data. Count and total duration 
data were rounded to the nearest whole number, and average duration was rounded to 
the nearest tenth, meaning that some durations show a minimum value of zero. 

24 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Analysis period 

Transitions / min 

Entire trial 
During close male hula-ing 
During distant male hula-ing  10.09(18.32) 0-75 
7.90(8.62) 0-37.43 
During all male hula bouts 
2.76(1.27) 0.01-5.15 
No male hula-ing 

3.06(1.22) 0.01-5.31 
9.99(18.61) 0-93 

Table 1.3: Summary statistics of female transitions between locomotive states. 
Data represent the number of times per minute that females transitioned among 
walking, swimming, and pausing states and are displayed as mean(SD) minimum-
maximum. All values were rounded to the nearest tenth.  

Close Hula 

Distant Hula 

All Hula 

Walk 

Swim 

0.44(0.25) 0.10-1.00 

0.59(0.26) 0-0.91 

0.47(0.25) 0-1 

0.01(0.05) 0-0.21 

0.02(0.08) 0-0.33 

0.03(0.08) 0-0.33 

Pause 

0.55(0.25) 0-0.90 

0.39 (0.23) 0.09-0.82 

0.50(0.22) 0-0.82 

Table 1.4: Summary statistics of proportions of time females spend locomoting 
during male hula-ing. Data represent the proportions of time that females spent 
walking, swimming, and pausing while the male was hula-ing at close proximity or at a 
distance and are displayed as mean(SD) minimum-maximum. All values were rounded 
to the nearest tenth. 

Dependent variable 

Close hula total duration 
Close hula average duration 
Distant hula total duration 
Distant hula average duration 
All hula total duration 
All hula average duration 
Bump count 
Female locomotor transitions 
Female locomotor transitions 

Independent variable 
Sex 
Sex 
Sex 
Sex 
Sex 
Sex 
Sex 
Proximity to hula-ing male 
Presence/absence of male hula 

t statistic 
2.45 
0.87 
0.55 
3.80 
1.43 
2.38 
2.45 
0.025 
-3.474 

p - value 

0.02 * 
0.39 
0.59 
0.0008 *** 
0.16 
0.02 * 
0.02 * 
0.98 
0.00128*** 

Table 1.5: Results of general linear mixed-effects models. Analysis of sex 
differences in hula durations, bump counts, as well as female locomotion patterns 
during bouts of male hula-ing revealed significant differences in close hula-ing, distant 
hula-ing, bumping, and female transitions between locomotive states. Female ID 
number was used as the random effect in all analyses. All t statistics and p-values were 
rounded to 2 decimals, except when the p-value was less than 0.01. 

25 

 
 
  
  
 
 
 
FIGURES 

Female locomotion 

Locomote/pause near ♂ 

Push 

Hula 

Follow 

Stages 

Total trial duration 

0 Hr 

Preliminary 
stage 

Mating 
stage 

2.5 Hr 

Figure 1.1: Example time budget of courtship stages. Illustration displaying which 
behavioral categories may co-occur and which are mutually excluded: includes female 
behaviors (female locomotion, locomote/pause near male), male behaviors (push, hula), 
and follow, in which both sexes are engaged in the behavior. All behaviors displayed in 
this illustration are listed in the ethogram (Table 1).  

26 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 1.2: Female locomotion throughout the courtship ritual. Overall, females 
spent the greatest amount of time pausing, followed by walking and then swimming. 
Each box and whisker plot represents five statistical values: the lower end of the bottom 
vertical line is the minimum, the bottom edge of the box is the lower quartile, the line in 
the center of the box is the median, the upper edge of the box is the upper quartile, and 
the upper end of the top vertical line is the maximum. Dots represent outlier values and 
zeroes have been removed.  

27 

 
 
 
*** 

Figure 1.3: Female transitions between locomotive states with the presence or 
absence of a hula-ing male. Female axolotls transitioned between walking, swimming, 
and resting more often when the male was hula-ing versus when he was not hula-ing (p 
= 0.00128). Each box and whisker plot represents the minimum, the lower quartile, the 
median, the upper quartile, and the maximum value; dots represent outlier values and 
zeroes have been removed. 

28 

 
 
* 

Figure 1.4: Total duration of male versus female hula behavior. Male axolotls 
performed “close” hula-ing longer than did females (p = 0.02). Each box and whisker 
plot represents the minimum, the lower quartile, the median, the upper quartile, and the 
maximum value; dots represent outlier values and zeroes have been removed.  

29 

 
 
 
 
*** 

* 

Figure 1.5: Average bout duration of male versus female hula behavior. Male 
axolotls exhibited significantly longer bouts of distant hula-ing than females did (p = 
0.0008); the same was true when we pooled all bouts of hula-ing (p = 0.02). Each box 
and whisker plot represents the minimum, the lower quartile, the median, the upper 
quartile, and the maximum value; dots represent outlier values and zeroes have been 
removed. 

30 

 
 
 
* 

Figure 1.6: Male versus female bumping behaviors. Male axolotls performed the 
bump behavior more frequently than did females (p = 0.02). Each box and whisker plot 
represents the minimum, the lower quartile, the median, the upper quartile, and the 
maximum value for each data set; dots represent outliers.  

31 

 
 
 
 
 
 
 
 
 
 
 
 
REFERENCES 

Ahnesjö, I., Brealey, J. C., Günter, K. P., Martinossi-Allibert, I., Morinay, J., Siljestam, 

M., Stångberg, J., Vasconcelos, P. (2020). Considering gender-biased assumptions 
in evolutionary biology. Evolutionary Biology, 47(1), 1–5. 

Altmann, J. (1974). Observational study of behavior: sampling methods. Behaviour, 

49(3–4), 227–266. 

Anderson, J. D. (1961). The courtship behavior of Ambystoma macrodactylum croceum. 

Copeia, 1961(2), 132–139. 

Armstrong, J. B., Duhon, S. T., & Malacinski, G. M. (1989). Raising the axolotl in 

capitivity. In J. B. Armstrong & G. M. Malacinski (Eds.), Developmental Biology of 
the Axolotl (pp. 220–227). New York: Oxford University Press. 

Armstrong, J. B., & Malacinski, G. M. (1989). Developmental biology of the axolotl (7th 

ed.). Oxford University Press. 

Arnold, S. J. (1987). The comparative ethology of courtship in salamandrid 
salamanders. 1. Salamandra and Chioglossa. Ethology, 74, 133–145. 

Arnold, S. J. (1976). Sexual behavior, sexual interference and sexual defense in the 

salamanders Ambystoma maculatum, Ambystoma tigrinum and Plethodon jordani. 
Zeitschrift Für Tierpsychologie, 42(3), 247–300. 

Arnold, S. J. (1972). The evolution of courtship behavior in salamanders. University of 

Michigan. 

Baker, C. A., Clemens, J., & Murthy, M. (2019). Acoustic Pattern Recognition and 

Courtship Songs: Insights from Insects. Annual Review of Neuroscience, 42, 129–
147.  

Baker, C. V. H., Modrell, M. S., & Gillis, J. A. (2013). The evolution and development of 
vertebrate lateral line electroreceptors. Journal of Experimental Biology, 216, 2515–
2522. 

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects 

models using lme4. Journal of Statistical Software, 67(1), 1–48. 

Bazar, M. A., Quinn, M. J., Mozzachio, K., Bleiler, J. A., Archer, C. R., Phillips, C. T., & 
Johnson, M. S. (2009). Toxicological responses of red-backed salamanders 
(Plethodon cinereus) to soil exposures of copper. Archives of Environmental 
Contamination and Toxicology, 57(1), 116–122. 

32 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Bolker, B. M. (2015). Linear and generalized linear mixed models. In G. A. Fox, S. 
Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary 
Theory and Application (1st ed., pp. 309–333). Oxford University Press. 

Butler, J. M., & Maruska, K. P. (2015). The mechanosensory lateral line is used to 

assess opponents and mediate aggressive behaviors during territorial interactions 
in an African cichlid fish. Journal of Experimental Biology. 

Butler, Julie M., & Maruska, K. P. (2016). Mechanosensory signaling as a potential 

mode of communication during social interactions in fishes. Journal of Experimental 
Biology, 219(18), 2781–2789. 

Coombs, S., Bleckmann, H., Fay, R. R., & Popper, A. N. (Eds.). (2014). The lateral line 

system. New York: Springer. 

Coombs, S., Görner, P., & Münz, H. (1989). A brief overview of the mechanosensory 

lateral line system and the contributions to this volume. In S. Coombs, P. Görner, & 
H. Münz (Eds.), The Mechanosensory Lateral Line : Neurobiology and Evolution 
(pp. 3–5). New York: Springer-Verlag. 

Coombs, S., & Montgomery, J. (2014). The role of flow and the lateral line in the 

multisensory guidance of orienting behaviors. In H. Bleckmann, J. Mogdans, & S. 
Coombs (Eds.), Flow Sensing in Air and Water: Behavioral, Neural and 
Engineering Principles of Operation (pp. 65–101). 

Cornwall, M. C., MacNichol Jr, E. F., & Fein, A. (1984). Absorptance and spectral 
sensitivity measurements of rod photoreceptors of the tiger salamander, 
Ambystoma tigrinum. Vision Research, 24(11), 1651–1659. 

De Luca, P. A., & Morris, G. K. (1998). Courtship communication in meadow katydids : 

female preference for large male vibrations. Behaviour, 135(6), 777–794. 

Denoël, M., & Doellen, J. (2010). Displaying in the dark: light-dependent alternative 
mating tactics in the Alpine newt. Behavioral Ecology and Sociobiology, 64(1), 
1171–1177. 

Eisthen, H. (1989). Courtship and mating behavior in the axolotl. Axolotl Newsletter, 18, 

18–19. 

Friard, O., & Gamba, M. (2016). BORIS: a free, versatile open-source event-logging 
software for video/audio coding and live observations. Methods in Ecology and 
Evolution, 7(11), 1325–1330. 

Fritzsch, B., & Wahnschaffe, U. (1983). The electroreceptive ampullary organs of 

urodeles. Cell Tissue Research, 229, 483–503. 

33 

 
 
 
 
 
 
 
 
 
 
 
 
Green, A. J. (1989). The sexual behaviour of the great crested newt, Triturus cristatus 

(Amphibia: Salamandridae). Ethology, 83, 129–153. 

Gresens, J. (2004). An introduction to the Mexican Axolotl (Ambystoma mexicanum). 

Lab Animal, 33(9), 41–47.  

Guevara-Fiore, P., Stapley, J., Krause, J., Ramnarine, I. W., & Watt, P. J. (2010). Male 

mate-searching strategies and female cues: how do male guppies find receptive 
females? Animal Behaviour, 79(6), 1191–1197. 

Halliday, T. R. (1990). The evolution of courtship behavior in newts and salamanders. 

Advances in the Study of Behavior, 19, 137–169. 

Houck, L. D., & Arnold, S. J. (2003). Courtship and mating behavior. In Reproductive 

Biology and Phylogeny of Urodela (1st ed., pp. 383–424). 

Hrouzková, E., Dvořáková, V., Jedlička, P., & Šumbera, R. (2013). Seismic 

communication in demon African mole rat Tachyoryctes daemon from Tanzania. 
Journal of Ethology.  

Joel, A. C., Habedank, A., & Mey, J. (2017). Vibratory movements in contests between 

females of the feather-legged spider (Uloborus plumipes). Zoology.  

Kernighan, B. W., & Ritchie, D. M. (2002). The C programming language. 

Kim, J. K., Lee, J. H., Ra, N. Y., Lee, H. J., Eom, J., & Park, D. (2009). Reproductive 

function of the body and tail undulations of Hynobius leechii (Amphibia: 
Hynobiidae): A quantitative approach. Animal Cells and Systems, 13, 71–78. 

Kramer, D. L., & McLaughlin, R. L. (2001). The behavioral ecology of intermittent 

locomotion. American Zoologist, 41(2), 137–153. 

Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: 

tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. 

Lauder, G. V, & Shaffer, H. B. (1985). Functional Morphology of the Feeding 
Mechanism in Aquatic Ambystomatid Salamanders, 326, 297–326. 

Lunsford, E. T., & Liao, J. C. (2021). Activity of posterior lateral line afferent neurons 

during swimming in zebrafish. Journal of Visualized Experiments, 2021(168), 1–17. 

Maex, M., Bocxlaer, I. Van, Mortier, A., Proost, P., & Bossuyt, F. (2016). Courtship 

pheromone use in a model urodele, the Mexican axolotl (Ambystoma mexicanum). 
Scientific Reports, 6(20184), 1–9. 

34 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Marsh, D. M., & Hanlon, T. J. (2004). Observer gender and observation bias in animal 

behaviour research: experimental tests with red-backed salamanders. Animal 
Behaviour, 68(6), 1425–1433. 

Montgomery, J., Bleckmann, H., & Coombs, S. (2014). Sensory ecology and 

neuroethology of the lateral line. In S. Coombs, H. Bleckmann, R. R. Fay, & A. N. 
Popper (Eds.), The Lateral Line System (pp. 121–150). New York: Springer. 

Münz, H., Claas, B., & Fritzsch, B. (1984). Electroreceptive and mechanoreceptive units 
in the lateral line of the axolotl Ambystoma mexicanum. Journal of Comparative 
Physiology A, 33–44. 

O’Connell-Rodwell, C. E. (2024). Keeping an “ Ear ” to the Ground : Seismic 

Communication in Elephants, (45), 287–294. 

Osorio, D., & Vorobyev, M. (2008). A review of the evolution of animal colour vision and 

visual communication signals. Vision Research, 48, 2042–2051.  

Park, D., McGuire, J. M., Majchrzak, A. L., Ziobro, J. M., & Eisthen, H. L. (2004). 

Discrimination of conspecific sex and reproductive condition using chemical cues in 
axolotls (Ambystoma mexicanum). Journal of Comparative Physiology A, 190, 415–
427. 

Park, Daesik, Lee, J.-H., Ra, N.-Y., & Eom, J. (2008). Male salamanders Hynobius 
leechii respond to water vibrations via the mechanosensory lateral line system. 
Journal of Herpetology, 42(4), 615–625. 

Petranka, J. (1982). Courtship behavior of the small-mouthed salamander (Ambystoma 

texanum) in Central Kentucky. Herptologica, 38(2), 333–336. 

Pierotti, R., Annett, C., & Hand, J. (1997). Male and female perceptions of pair-bond 
dynamics: monogamy in western gulls, Larus occidentalis. In Feminism and 
Evolutionary Biology: Boundaries, Intersections and Frontiers (pp. 261–275). 
Boston, MA: Springer US. 

Plazas, P. V., & Elgoyhen, A. B. (2021). The cholinergic lateral line efferent synapse: 

structural, functional and molecular similarities with those of the cochlea. Frontiers 
in Cellular Neuroscience, 15, 1–15. 

Proctor, H. C. (1991). Courtship in the water mite Neumania papillator: males capitalize 

on female adaptations for predation. Animal Behaviour, 42(4), 589–598. 

Revelle, W. (2023). psych: procedures for psychological, psychometric, and personality 
research. Northwestern University, Evanston, Illinois. R package version 2.3.6. 

35 

 
 
 
 
 
 
 
 
 
 
 
 
 
Russell, I. J. (1968). Influence of efferent fibres on a receptor. Nature, 219(5150), 177–

178. 

Salthe, S. N. (1967). Courtship patterns and the phylogeny of the urodeles. Copeia, 1, 

100–117. 

Satou, M., Takeuchi, H. A., Takei, K., Hasegawa, T., Matsushima, T., & Okumoto, N. 

(1994). Characterization of vibrational and visual signals which elicit spawning 
behavior in the male himé salmon (landlocked red salmon, Oncorhynchus nerka). 
Journal of Comparative Physiology A, 174, 527–537. 

Satou, Masahiko, Shiraishi, A., Matsushima, T., & Okumoto, N. (1991). Vibrational 
communication during spawning behavior in the himé salmon (landlocked red 
salmon, Oncorhynchus nerka). Journal of Comparative Physiology A. 

Shoop, C. R. (1960). The breeding habits of the mole salamander, Ambystoma 

talpoideum (Holbrook), in southeastern Louisiana. Master’s Thesis, Tulane Studies 
in Zoology and Botany, 8(3), 65–82. 

Soleymani, M., Villaro-Dixon, F., Pun, T., & Chanel, G. (2017). Toolbox for Emotional 

feAture extraction from Physiological signals (TEAP). Frontiers in ICT. 

Storez, R. A. (1969). Observations on the courtship of Ambystoma laterale. Journal of 

Herpetology, 3(1), 87–95. 

Topp, G. (1983). Primary lateral line response to water surface waves in the topminnow 

Aplocheilus lineatus (Pisces, Cyprinodontidae). Pflügers Archiv European Journal 
of Physiology.  

Verrell, P. A. (1983). The influence of the ambient sex ratio and intermale competition 
on the sexual behavior of the red-spotted newt, Notophthalmus viridescens 
(Amphibia: Urodela: Salamandridae). Behavioral Ecology and Sociobiology, 13, 
307–313. 

Verrell, P., & Pelton, J. (1996). The sexual strategy of the central long-toed salamander, 
Ambystoma macrodactylum columbianum, in south-eastern Washington. Journal of 
Zoology, 240(1), 37–50. 

Verrell, P. (1982). The sexual behaviour of the red-spotted newt, Notophthalmus 

viridescens (Amphibia: Urodela: Salamandridae). Animal Behaviour, 30(4), 1224–
1236. 

Verrell, P. (1999). Bracketing the extremes: courtship behaviour of the smallest- and 

largest-bodied species in the salamander genus Desmognathus (Plethodontidae: 
Desmognathinae). Journal of Zoology, 247(1), 105–111. 

36 

 
 
 
 
 
 
 
 
 
 
 
 
Voss, S. R., Woodcock, M. R., & Zambrano, L. (2015). A tale of two axolotls. 

BioScience, 65(12), 1134–1140. 

Woolley, S. M. N., & Moore, J. M. (2011). Coevolution in communication senders and 

receivers: vocal behavior and auditory processing in multiple songbird species. 
Annals of the New York Academy of Sciences, 1225(1), 155–165.  

Wullimann, M. F., & Grothe, B. (2014). The central nervous organization of the lateral 

line system. In The Lateral Line System (pp. 195–251). 

37 

 
 
 
 
 
CHAPTER 2: Behavioral Responses of Female Axolotls to the Male Hula Behavior 

ABSTRACT 

Hydrodynamic communication signals are important for aquatic vertebrates to 

coordinate complex schooling behaviors, mediate aggressive encounters, and 

synchronize mating. Some fully aquatic salamanders, including the axolotl, Ambystoma 

mexicanum, create hydrodynamic stimuli by performing a rhythmic wriggling or fanning 

motion of the tail. Unlike amphibians that metamorphose into terrestrial adults, axolotls 

maintain functionality of the lateral line systems throughout their lives. The lateral line 

system detects electrical and mechanical stimuli in aquatic environments, and tail 

movements can be detected readily by the lateral line.  During courtship, both male and 

female axolotls exhibit a behavior known as the “hula”, which involves a rhythmic 

undulation of the tail combined with a swaying motion of the pelvic region. We 

investigated the potential communicative role of the hydrodynamic stimuli generated 

during the male hula behavior and response of the female to these signals.   

We first characterized the range of speeds and angles that males can perform 

with their tails (sweep angle, speed, and elevation angle), and investigated females’ 

behavioral responses to particular motion parameters at both close proximity and at a 

distance. We then designed a robotic device (the “Robotail”) to mimic the hula, which 

we programmed to produce speeds and angles that reflect 27 potential combinations of 

these parameters that reflect both normal behaviors and abnormal male behaviors that 

males were unlikely to perform or were never observed naturally.  

38 

 
We found that when males were in close proximity to females, males performed 

predominantly moderate combinations that represented a wider sweep angle (~20° to 

93°), slightly slower speed (~0.25 Hz to 1.4 Hz), and higher elevation angle (~15° to 

75°) compared to when they were hula-ing at a distance from a female. When we 

examined the behavioral effects of single hula parameters, we found that female 

axolotls responded to the Robotail such that wider sweep angles (90°) and faster 

speeds (1.5 Hz) caused females to exhibit shorter bouts of walking, and wide to 

moderate sweep angles (30° to 90°) caused females to exhibit shorter bouts of 

swimming. Concordantly, females transitioned between locomotive states (walking, 

swimming, pausing) more often when we stimulated them with sweep angle/speed 

combinations of 90°/1.0 Hz, 30°/1.5 Hz, or 90°/1.5 Hz. Additionally, faster speeds (1.5 

Hz) caused females to spend more time near the tail overall. Importantly, we found that 

while males generally performed moderate behaviors, females responded to more 

extreme behaviors from the Robotail. These results suggest that females would respond 

strongly to real males that exhibit very fast or wide hula patterns; perhaps males are 

physically limited in their ability to perform these extreme behaviors.  

Our experiments serve to describe how males generate hydrodynamic stimuli 

during courtship and how females behaviorally respond to a range of hula motion 

parameters. This research represents an important first step in understanding whether 

the lateral line system is physiologically tuned to hydrodynamic signals generated 

during courtship. 

39 

 
 
INTRODUCTION 

The roles of auditory (Baker et al., 2020; Woolley & Moore, 2011) and visual 

signals (Osorio & Vorobyev, 2008) in animal communication are well established; 

however, a comparatively small amount of research exists on the communicative 

function of vibrational signals. Although vibrational communication has been 

documented in insects like katydids (De Luca & Morris, 1998) and spiders (Joel et al., 

2017) as well as terrestrial vertebrates such as elephants (O’Connell-Rodwell, 2024) 

and mole rats (Hrouzková et al., 2013), an even smaller body of literature exists on 

vibrational communication in aquatic systems. Hydrodynamic stimuli (i.e., water 

disturbances) are known to play an important communicative role in the lives of aquatic 

animals. In particular, some animals produce body movements to direct hydrodynamic 

stimuli towards conspecifics during social interactions, such as aggression (Butler & 

Maruska, 2015) and mating (Butler & Maruska, 2016). For example, the body 

movements of some teleost fishes generate vibrational stimuli that facilitate schooling 

behaviors (Coombs & Montgomery, 2014) as well as coordinate mating efforts between 

males and females (Satou et al., 1994). Aquatic invertebrates such as male water mites, 

Neumania papillator, are also known to wave their front appendages near females 

during courtship to attract them (Proctor, 1991).  

Many male salamanders perform a courtship behavior that involves a repetitive, 

undulating motion of the tail; variations of this include the “tail fan” in which just the tip of 

the tail is moved, as seen in the great crested newt, Triturus cristatus (Green, 1989) as 

well as “pelvic wagging” exhibited by the mole salamander, Ambystoma talpoideum, in 

which the male sways the hips and tail simultaneously (Shoop, 1960). These types of 

40 

salamander behaviors may serve to stimulate the lateral line system of conspecifics; the 

lateral line is a sensory system that can detect both electrical (Baker et al., 2013) and 

mechanical stimuli (Coombs et al., 1989) in aquatic environments. Park et al. (2008) 

demonstrated that male Korean salamanders (Hynobius leechii) detect vibrational 

stimuli via their lateral line systems by testing salamanders’ behavioral responses to a 

model that mimics the tail motions of rival males in courtship settings. Males oriented 

toward and approached the model less often when vibrations were blocked with an 

acrylic barrier or when the lateral line was chemically blocked with cobalt chloride. Here, 

we aim to further elucidate the role of hydrodynamic stimuli in salamander courtship by 

studying the behavioral responses of the female axolotl, Ambystoma mexicanum, to a 

variety of stimuli produced by males.  

We chose to use the fully aquatic axolotl in our experiments for several reasons: 

axolotls breed year-round in laboratory settings (Voss et al., 2015) and males reliably 

perform a behavior known as the “hula” throughout their courtship ritual, which involves 

a swaying motion of the hips combined with an undulating motion of the tail (Eisthen, 

1989). Although hula-ing in axolotls is similar to the pelvic wagging behavior in A. 

talpoideum (Shoop, 1960), this type of behavior in axolotls has customarily been called 

the “hula” (Park et al., 2004). Additionally, given their neotenic life cycle, axolotls 

maintain their lateral line systems throughout their entire lives, whereas salamanders 

that undergo metamorphosis tend to lose their lateral line functionality upon the 

transition to terrestrial life (Fritzsch & Wahnschaffe, 1983). Here, we designed two 

experiments to investigate the potential mechanosensory role of the hula behavior 

during axolotl courtship. Specifically, we aimed to characterize the typical range of 

41 

motions and speeds that males can perform with their tails during the hula and assess 

female behavioral responses to various combinations of hula motion parameters.  

We first defined three natural motion parameters of the male’s tail during the hula 

behavior, which we called the “sweep angle” (the degree of side-to-side motion), the 

“speed” of undulation (measured in Hz), and the “elevation angle”; males periodically 

raise and lower their tails during courtship, so we described this axis of motion by 

measuring the angle between the substrate and the male’s tail. Video recordings of 

male and female courtship interactions helped to establish the range of natural 

movements and behaviors (Chapter 1). Once we established common motion patterns 

we assessed female behavioral responses to a range of hula motion parameters using 

the “Robotail”, a robotic device we created. We designed the Robotail to mimic the 

undulating tail motion of the male hula behavior; importantly, this device allowed us to 

test the effects of specific parameters on female behavior in a controlled and repeatable 

fashion. 

We tested three values from each of the three motion parameters for our 

experiment, which represented the minimum, median, and maximum values that males 

produced with their tails during courtship; we then evaluated all 27 potential 

combinations of these parameters. Some of these motion combinations were 

naturalistic (i.e., male axolotls perform them) whereas others were contrived (i.e., males 

were unlikely to perform them). Our research will allow us to understand how the 

individual parameters of the hula influence female behavior. Importantly, these 

experiments serve as a framework to support further studies to understand if the lateral 

42 

line system is physiologically tuned to detect the hydrodynamic stimuli generated by 

conspecifics in a courtship context. 

MATERIALS AND METHODS 

Subjects 

Adult axolotls (Ambystoma mexicanum) were obtained from the Ambystoma 

Genetic Stock Center at the University of Kentucky. Animals were housed in ~114 L 

aquaria in 40-100% Holtfreter’s (HF) solution (Armstrong et al., 1989) supplemented 

with Replenish™ solution (Seachem Laboratories, Madison, GA) at temperatures 

between 18 and 22°C; animals were separated by sex into aquaria containing 1-3 

animals each. We programmed the lights in our facility to match the natural sunrise and 

sunset of Mexico City, Mexico (the native habitat of axolotls) with monthly updates to 

the axolotls’ photoperiod. Housing and experiments were conducted with the approval 

of the Institutional Animal Care and Use Committee of Michigan State University 

(approval numbers: 10/15/-154-00, PROTO201800106). 

Hula motion parameters 

To assess the range of tail motions that male axolotls exhibited while hula-ing, 

we recorded behavioral interactions between 30 pairings of male and female axolotls (N 

= 10 males, 19 females); each male was paired with 3 different females. For each trial, 

we placed a pair together into a ~114 L aquarium (90 cm long x 45 cm wide x 30 cm 

wide) filled with HF solution at the same concentration as their home aquaria shortly 

before the lights in our animal facility turned off for the evening; we then allowed the pair 

43 

  
to interact overnight because axolotls are nocturnally active. We used 2 Sony Nightshot 

camcorders (model: CMOS) to record top-down and side views of each trial. 

We analyzed hula motion parameters for 6 different males; 2 spawned with 

females and the other 4 were selected at random. For each male, we only analyzed 

trials that contained at least one bout of “close” hula-ing that was at least 2 min long 

and/or a bout of “distant” hula-ing that was at least 30 sec long; thus, we included 3 

trials for 1 male, 2 trials for 3 males, and a single trial for the remaining 2 males, 

resulting in 11 trials total. We defined close hula-ing as occurring within one male tail-

length of the female and distant hula-ing as occurring at least one male tail-length away 

from the female. For each trial, we analyzed 2 hrs of footage starting when the lights in 

our animal facility turned off for the evening.   

We first used BORIS (Behavioral Observation Research Interactive Software; 

version 7.9.RC1; Friard & Gamba, 2016) to catalog the frequency and duration of all 

hula bouts, which we trimmed to 2 min for close bouts and 30 sec for distant bouts. We 

used a modified version Altmann's (1974) “instantaneous sampling” method to analyze 

the males’ tail movements within each bout. For close hula bouts, we divided each bout 

into 4 30-sec windows and took measurements for the first 10 sec of each window; for 

distant bouts, we analyzed 2 15-sec windows by measuring the first 5 sec of each one. 

We used Kinovea software version 0.8.27 (www.kinovea.org) to measure the angle of 

the male’s tail as it moved from side to side (sweep angle), the frequency (in Hz) of tail 

movement, and the elevation angle of the tail. 

We measured tail sweep angles and frequencies in tandem from the overhead 

perspective by watching video footage frame-by-frame and measuring the lateral angle 

44 

whenever the proximal third of the tail changed direction. We recorded the maximum 

and minimum sweep angle for each measurement period (i.e. 10 sec for close bouts, 5 

sec for distant bouts), discarding any measurements less than 5° because the 

resolution of our videos footage made it difficult to accurately discern smaller lateral 

motions. The frequency of tail motion was calculated by dividing the total number of 

angle measurements by the duration of the measurement period. Elevation angles were 

measured from the side view by measuring the ventral edge of the male’s tail relative to 

the aquarium floor. We obtained a baseline measurement for the elevation angle 30 sec 

before any bout of hula-ing began and then recorded the maximum and minimum 

elevation angles during each measurement period.  

Construction of a hula-mimicking robot 

To assess the behavioral effects of specific motion parameters on females in a 

controlled and repeatable manner, we constructed a robotic tail, the “Robotail”, that 

mimics male hula behaviors. The Robotail consisted of a silicone “tail”, 3D printed 

components to mount the silicone tail on the floor of an aquarium, and a suite of 

electronics to program and drive the tail’s motions. We ensured that all components 

within the aquarium were made of non-metallic materials to 1) prevent metal ions from 

leaching into the HF solution, which can be harmful to salamanders (Bazar et al., 2009) 

and 2) prevent the generation of electrical cues and noise in the test arena, as the 

lateral line system of axolotls can detect electrical stimuli (Münz, Claas, & Fritzsch, 

1984).  

To create the silicone tail, we measured the tail dimensions of 5 adult male 

axolotls by anaesthetizing each male in a solution of 0.1% pH-corrected MS-222; we 

45 

 
measured the smallest and largest male in our colony (based on weight) and randomly 

chose an additional 3 males. The total length of the male’s tail (posterior edge of cloacal 

gland to tail tip) was divided into 5 equal segments; at each landmark, we measured the 

width and height of the dorsal and ventral fins, the width at the middle of the tail, and the 

total height (Fig. 2.1). We used the median value of each measurement to generate a 

“representative” tail,  which we modeled, along with a 2-part mold, in SolidWorks® 

software (version: SolidWorks 2016; Dassault Systèmes, Vélizy-Villacoublay, France). 

We made the silicone tail from Smooth-On Ecoflex™ 00-30 (Smooth-On, Inc., 

Macungie, Pennsylvania) using a 3D print of our custom mold from Shapeways 

(Livonia, Michigan); during the casting process, we embedded a custom 3D printed clip 

in the proximal end of the silicone tail to attach the tail to the mounting components (Fig. 

2.2A). 

The mounting components consisted of a 3D printed triangular base plate and a 

pivoting cylinder with a slot to hold the tail in place (Fig. 2.2); the triangular base was 

mounted to the aquarium floor using waterproof silicone. We also sourced a silicone 

salamander toy resembling an ambystomatid salamander and removed the tail; we 

mounted the remainder of the toy in front of the base plate to resemble the shape of a 

male axolotl (Fig. 2.3). Finally, we attached EVA foam tiles to the aquarium floor to 

make the floor flush with the mounting plate and rubber salamander. A length of braided 

nylon fishing line was used to connect the pivoting cylinder to the gimbal, which was 

outside of the aquarium, without the use of metal (Fig. 2.2). When the Robotail was 

submerged in water, the pivoting motion of the cylinder transferred an undulating motion 

to the silicone tail, simulating the hula behavior. 

46 

 
We used a series of electronic components to drive and program the oscillating 

motion of the silicone tail, including a power distribution channel, Victor SP speed 

controller, 104:1 NeveRest Sport gearbox, NeveRest gearmotor, and Thrifty Throttle 

3/Cypress PSoC4™ microcontroller from AndyMark Inc. (Kokomo, IN) as well as a 

PSoC4™ MiniProg3 programmer (Infineon Technologies AG, Munich, Germany) and a 

potentiometer (DigiKey, Thief River Falls, MN). Additionally, we created a metal gimbal 

to control the lateral position of the tail. The electronics and gimbal were mounted to an 

acrylic sheet attached to a wooden frame elevated approximately 60 cm above the 

aquarium floor (Fig. 2.4). We wrote custom software in C (Kernighan & Ritchie, 2002) 

using PSoC™ Creator (Infineon Technologies AG, Munich, Germany) to control the 

speed and sweep angle of tail oscillation. The elevation angle of the silicone tail was 

changed manually. 

Female behavioral responses to the Robotail 

We determined the parameters for our behavioral tests by simplifying each of the 

3 continuous variables of hula motions (sweep angle, hula speed, and elevation angle) 

into 3 discrete categories, which were based on the minimum, median, and maximum 

values of each parameter (Table 2.1); thus, we condensed our data set into 27 different 

combinations of parameters. We then created a series of triple-axis matrices to assess 

the occurrence rate of each motion combination; we discarded any hula bouts in which 

we were unable to measure all 3 parameters and then counted which combinations 

occurred “often” (occurred more often than the median value), “sometimes” (occurred 

less often than the median value), or “never” (not occurring). Importantly, the matrices 

allowed us to determine which combinations were natural (i.e., those that male axolotls 

47 

 
perform) and which were contrived (i.e., those that males were unlikely or unable to 

exhibit).  

 We programmed the Robotail to perform each of the 27 motion combinations to 

assess female behavioral responses to both natural and contrived hula motion patterns; 

thus, we tested sweep angles of 10°, 30°, and 90°, speeds of 0.5 Hz, 1.0 Hz, and 1.5 

Hz, and elevation angles of 0°, 20°, and 55°. In total, we staged 162 encounters with the 

Robotail by testing each motion combination with 6 different females. All trials were 

conducted after the lights in our animal facility turned off for the evening; between trials, 

we wore headlamps with red lights to minimize disrupting our subjects. All trials were 

recorded from an overhead angle using a Lorex Night Vision Security Camera system 

(model: HDIP82W). We placed a female in the test aquarium (filled with HF at the same 

concentration as their home aquarium) with the robot and began each trial with a 5 min 

acclimation period, after which we turned the robot on and allowed the female to interact 

with it for 15 min. We quantified behavioral responses to the robotic axolotl tail in BORIS 

by measuring locomotor behaviors (walk, swim, pause), hula behaviors, time spent near 

the robot (i.e., within a circular area around the tail, with the tail’s length being equal to 

its radius) orientation behaviors, physical contacts with the robot (nudge base, nudge 

tip), as well as aggressive behaviors (nip, bite, attack). Full definitions for all behaviors 

are provided in Table 2.2.  

In preliminary trials, female axolotls attacked the robot unless male odorants 

were present in the aquarium. In the absence of male odorants, it is possible that the 

Robotail resembled prey more than a male conspecific; thus we included male odorants 

throughout the experiment. We collected whole-body odorants from male axolotls by 

48 

placing 3 males into plastic bowls each filled with 1 L of HF roughly 24 hr before the 

start of a trial. The following day, we combined the HF solution from all 3 buckets into a 

single solution to minimize effects of individual variation. Odorants were delivered into 

the aquarium through a silicone tube mounted near the base of the robot; a peristaltic 

pump delivered odorant solution at a rate of approximately 33 mL/min while the robot 

was on. We replaced the HF solution in the aquarium between trials to remove male 

odorants as well as any odorants that the female test subject may have released.  

Statistical analyses 

We analyzed female behavioral responses to the Robotail (locomotor behaviors, 

physical contacts with the Robotail, time spent near the robot, and aggressive 

behaviors) using general linear mixed-effects models in R (R Core Team, 2020); we 

utilized the packages ”lme4” (version 1.1.35.3; Bates et al., 2015) and ”lmerTest” 

(version 3.1.3; Kuznetsova et al., 2017).  The three motion hula parameters (sweep 

angle, speed, and elevation, angle) were treated as fixed effects, with female ID as a 

random effect because females were repeated in the dataset an unequal number of 

times. Analyses were run on models alone and in combination. We set the restricted 

maximum likelihood (REML) to false because we compared models with different fixed 

effects (Bolker, 2015). Zeros were removed for total durations and average bout 

durations (i.e., total duration divided by count) but retained for count variables. 

Additionally, we compared models of increasing complexity by calculating corrected 

Akaike information criterion (AICc) values using the R package “glmulti” (version 1.0.8). 

Models with a delta AICc of 3 or less were considered equally weighted models, and 

among weighted models we considered the simplest model to be the best fitting model. 

49 

RESULTS 

Hula motion parameters 

Male axolotls performed the hula behavior using moderate sweep and elevation 

angles and speeds and rarely moved their tails at the upper and lower extremes of their 

motion range. Among close hula combinations, males sometimes used moderate (30°) 

to wide (90°) sweep angles, moderate speeds (1.0 Hz), and moderate elevation angles 

(20°). Additionally, we found that males performed two close hula combinations the 

most often: 30° sweep/1.0 or 1.5 Hz/20° elevation (Table 2.3A). In contrast, distant hula 

bouts that fell into the “sometimes” category generally featured narrow to moderate 

(10°-30°) sweep angles, moderate speeds (1.0 Hz), and low to moderate (0°-20°) 

elevation angles. Males performed 4 combinations of distant hula motions most often; 

10° sweep/1.5 Hz/0° elevation, 30° sweep/1.0 Hz/0° elevation, 30° sweep/1.5 

Hz/0°elevation, and 30° sweep/1.0Hz/20° elevation (Table 2.3B). Thus, the motion 

patterns of close hula bouts were typically wider and higher than distant hula bouts, but 

males tended to hula with moderate speeds regardless of their proximity to a female.  

When we pooled all hula bouts (close plus distant) and examined the 

combinations that occurred sometimes, we found that males tended to perform all 

sweep angles, but only moderate speeds (1.0 Hz) and low to moderate (0°-20°) 

elevation angles. We additionally found 3 motion combinations that males performed 

most often overall; 30° sweep/1.0 Hz/0° or 20° elevation and 30° sweep/1.5 Hz/20° 

elevation (Table 2.3C). Six of the hula motion combinations we tested in our experiment 

were never performed by males, including 1.5 Hz/55° elevation at any sweep angle 

(10°, 30°, or 90°). Additionally, males did not exhibit 0.5 Hz/55° elevation at the 

50 

 
extremes of their sweep angle range (10° or 90°), nor did they move their tails with a 

combination of 90° sweep/0.5 Hz/0° elevation. Overall, males did not lift their tails high if 

they were moving at the extremes of their speed range (0.5 or 1.5 Hz) or lower their tails 

if they were moving slowly and widely.  

Female behavioral responses to the Robotail 

Locomotor behaviors 

We found that female axolotls initiated and terminated bouts of locomotion more often 

when the Robotail was oscillating quickly or at a wide angle. Females walked for 

significantly shorter bouts of time as the sweep angle increased, with a sweep angle of 

90° being significantly different from 10° (t = -2.91, p = 0.004; Fig. 2.5). Females also 

walked for shorter bouts of time when the tail was moving at its fastest speed of 1.5 Hz 

compared to the slowest speed of 0.5Hz (t = -2.27, p = 0.03; Fig. 2.6). Females 

exhibited shorter bouts of swimming as the sweep angle increased (10° vs 30°; t = -

2.21, p = 0.03; 10° vs 90°; t = -2.21, p = 0.03; Fig. 2.7) and they also exhibited shorter 

bouts of locomotion (walking + swimming) as the sweep angle increased (10° vs 90°; t = 

-2.15, p = 0.03; Table 2.4; Fig. 2.8). The Robotail parameters we tested did not affect 

the total durations of time that female axolotls spent performing walking, swimming, and 

pausing behaviors. 

The sweep angle of the Robotail best explained the duration of bouts of walking 

and swimming; however, the duration of pausing bouts was not explained by a single 

best model. Additionally, the overall amounts of time that females spent walking, 

swimming, and pausing were not explained by a single best model. Every variable of 

total locomotion duration had multiple, equally weighted models (i.e., within 3 AICc units 

51 

of the top ranked model); in all cases, the equally weighted models included a single 

parameter (sweep angle, speed, or elevation angle), so parsimony could not be used to 

select a best model (Table S2.2).  

Time spent near the Robotail 

We found that females spent significantly more time overall near the Robotail 

when it was moving at its fastest speed of 1.5 Hz (t = 2.63, p = 0.009; Fig. 2.9). In 

contrast, females spent less time per visit to the tail when it was moving at its widest 

angle of 90° (t = -2.20, p = 0.03; Fig. 2.10). Additionally, we found a marginal negative 

effect of elevation angle on the number of times that a female oriented toward the tail 

(0° vs 20°; t = -1.92, p = 0.06; 0° vs 55°; t = -1.92, p = 0.06; Table 2.5).  

The statistical model that best described the total amount of time female axolotls 

spent near the tail included the speed of the Robotail as the only predictor, whereas the 

duration of each visit to the tail was not explained by a single best model. Accordingly, 

the statistical model that best described orient count included elevation angle only 

(Table S2.3). Thus, speed played an important role in determining how long females 

stayed near the Robotail overall, the sweep angle impacted how long each visit to the 

tail lasted, and elevation angle best explained how often females directed themselves 

towards it.  

Physical contacts with the Robotail 

Female axolotls touched the tip of the Robotail with their snouts more often as 

the speed increased but made fewer physical contacts when the tail was moving at a 

moderate sweep angle or was positioned at a steep elevation angle. Females nudged 

52 

the tip of the tail significantly more often when the tail was moving at its fastest speed  (t 

= 2.11, p = 0.04) but less often at its highest elevation (t = -2.85, p = 0.005; Table 2.6). 

Additionally, we found that females nudged the base of the tail marginally less often 

when the sweep angle was increased from 10° to 30° (t = -1.89, p = 0.06).  

The number of times that females nudged the base of the robot was best 

described by the model with only sweep angle as a predictor, whereas the model that 

best explained the number of tip nudges included only the elevation angle. We found 

that four models for the total number of nudges (count of nudges anywhere on the 

Robotail) were equally weighted, and rules of parsimony did not distinguish among 

models (Table S2.4).  

Female aggression 

Females behaved less aggressively towards the Robotail as the sweep angle 

and elevation angles were increased. Specifically, females nipped at the tail significantly 

less often as the sweep angle of the Robotail increased (10° vs 30°; t = -3.00, p = 0.003; 

10° vs 90°; t = -2.55, p = 0.01) and marginally less often when the tail was elevated to 

20° (t = -1.73, p = 0.09; Table 2.7). Additionally, female axolotls attacked the tail slightly 

less often as the sweep angle increased (10° vs 30°; t = -1.77, p = 0.08; 10° vs 90°; t = -

1.77, p = 0.08) and as the elevation angle increased (0° vs 20°; t  = -1.77, p =  0.08; 20° 

vs 55°; t = -1.77, p = 0.08; Table 2.7).  

The statistical model that best described nip count included only sweep angle, 

however bite count and attack count did not have a single best model; bite and attack 

53 

 
count both had multiple models with equal weighting and rules of parsimony did not 

distinguish among these models (Table S2.5).  

Locomotion transitions per minute (combination analysis) 

Female axolotls transitioned between locomotive states per minute more often 

when the Robotail performed motion combinations featuring moderate to wide sweep 

angles and moderate to fast speeds. Females transitioned significantly more frequently 

when presented with a combination of 90°/1.0 Hz (t = 2.19, p = 0.03), 30°/1.5 Hz (t = 

2.37, p = 0.02), or 90°/1.5 Hz (t = 4.01, p = 9.51e-05; Table 2.8).  

DISCUSSION 

Hula motion parameters 

Although male axolotls are capable of performing the hula behavior at the 

extremes of their sweep angle, speed, and elevation angle ranges, we found that they 

generally did not hula in this fashion. Additionally, males did not combine fast hula 

speeds with high tail elevation angles, regardless of the sweep angle. Thus, we posit 

that fast speeds paired with high elevation angles may be too energetically costly for 

males to sustain during courtship, no matter the degree of side-to-side motion. 

Additionally, the males in our study only combined slow speeds with high elevation 

angles if the sweep angle was moderate (30°), but not if the sweep angle was very 

narrow or very wide, suggesting that males may have been more likely to perform a 

motion combination if at least one of the parameters had a moderate value. Lastly, we 

never observed males performing a motion pattern of 90° sweep combined with 0.5 Hz 

54 

 
and 0° elevation, suggesting that males may not be able to achieve their widest sweep 

angle at the slowest speed and lowest elevation angle.  

Similarly to axolotls, male Korean salamanders (Hynobius leechii) generate 

vibrational signals with their bodies and tails to communicate with both males and 

females during mating. Kim et al. (2009) demonstrated that H. leechii males exhibit two 

types of vibration behaviors during mating: body undulations, which involve a side-to-

side motion of the tail and the posterior region, and tail undulations, which feature a 

waving motion of only the distal third of the tail. H. leechii males performed body 

undulations at an average speed of 0.64 Hz to communicate with both females and 

other males; in contrast, tail undulations (performed on average at 1.7 Hz) are typically 

only displayed during aggressive interactions between males. Although male axolotls 

are known to court other males and deposit spermatophores when housed in all-male 

aquaria (Gresens, 2004), the function of these behaviors is currently unknown. It is 

possible that male axolotls perform slower hula motions in the presence of a female 

they are attempting to court and reserve faster tail motions for communicating with other 

males.  

Close hula bouts featured somewhat wider sweep angles and higher elevation 

angles than distant hula bouts, but males tended to hula with a moderate speed of 1.0 

Hz regardless of their proximity to a female (Table 2.3). Moderate vibration speeds play 

an important communicative role in other aquatic species as well.  In landlocked red 

salmon, (Oncorhynchus nerka) moderate whole-body vibration speeds generated by a 

female-like model induced significantly greater numbers of spawning behaviors 

(approach and spawning acts) in males compared with extreme speeds (Satou et al., 

55 

1994). However, lower frequencies resulted in male salmon performing greater numbers 

of courtship behaviors, suggesting that an increase in the female’s vibration speed may 

indicate a transition from courtship to spawning. Although we did not test how male hula 

parameters directly related to the preliminary and mating phases (see Chapter 1) of 

axolotl pairs, we did find that males were slightly more likely to hula at slow speeds of 

0.5 Hz if the female was in close proximity. Thus, males may modulate their hula 

speeds depending on the which phase of courtship they are in, and this idea warrants 

further investigation. Overall, we found that male axolotls generally hula’d with moderate 

angles and speeds and tended to modulate their sweep and elevation angles, but not 

the speed of tail motions, depending on their proximity to a female. The fact that males 

mainly performed moderate behaviors either suggests that males are unable to perform 

the extreme behaviors, or that it is costly; alternatively, this may suggest that females do 

not respond (or respond negatively) to extreme behaviors and therefore there is no 

reason for males at the extremes of their motion ranges. To evaluate these hypotheses, 

we used the Robotail to test female response to single hula parameters as well as 

combinations of sweep angles and speeds; we excluded elevation angle from our 

combination analyses because females did not exhibit meaningful responses to 

changes in this parameter. 

Female behavioral responses to the Robotail 

Locomotor behaviors 

Female axolotls modulated their locomotor behaviors in response to changes in 

the sweep angle and speed of the Robotail. Specifically, females altered the rate at 

which they started and stopped locomoting, but they did not change the overall amounts 

56 

of time that they spent walking, swimming, and pausing. Females exhibited shorter 

bouts of walking and swimming as the sweep angle widened to 90º (Fig. 2.5, Fig. 2.7) 

and exhibited shorter bouts of walking as the speed increased to 1.5 Hz (Fig. 2.6). 

Accordingly, when we examined female responses to combinations of sweep angles 

and speed, we found that females transitioned between locomotive states more 

frequently if the sweep angle was moderate to wider and the speed was moderate to 

fast (Table 2.8).  

The average duration of time spent walking and time spent swimming in each 

bout were both explained best by the model that included only the sweep angle 

parameter. However, the total amount of time spent in all locomotion categories 

(walking, swimming, and combined) all had multiple models that were equally weighted 

(i.e., within 3 AICc units of the top ranked model) and were not nested models (Table 

S2.2). Thus, the sweep angle and the speed of the Robotail influenced the duration on 

individual bouts of walking and swimming, but the overall amounts of time that females 

spent locomoting were not influenced by any single parameter. 

The male axolotls in our study were slightly more likely to hula at 1.5 Hz when 

they were at a distance from a female, but more likely to exhibit a 90° sweep angle 

when they were in close proximity. We found that female axolotls in a courtship setting 

paused significantly more often and for longer periods of time when they were closer to 

a hula-ing male (see Chapter 1). Interestingly, in our experiments with the Robotail, 

females responded to both fast speeds and wide sweep angles by starting and stopping 

locomotion bouts more frequently. We recognize that, in a courtship setting, males and 

females can move independently of one another, whereas in our current experiment the 

57 

Robotail was stationary and only the female was free to move around the aquarium. 

Thus, a faster Robotail speed may not have been indicative of a distant male, just like a 

wider sweep angle may not have suggested the presence of a nearby male. We 

speculate that these behaviors may have represented an increased search effort to 

localize the source of hydrodynamic stimuli, given that females spent more time near 

the Robotail when it was moving at 1.5 Hz. 

Time spent near the Robotail 

Female axolotls spent less time per visit to the tail when it was moving at a wider 

angle (Fig. 2.10) but spent more time near the Robotail overall when it was moving at a 

faster pace (Fig. 2.9). Additionally, females tended to avoid orienting towards the tail 

when it was raised at higher elevation angles. Wider sweep angles occupy more 

physical space in the designated “near” zone around the Robotail than narrower angles, 

which may have made it difficult for females to enter or stay in the zone for extended 

periods of time. Accordingly, the statistical model that best explained the total duration 

of time spent near included only speed. However, the average duration of time spent 

near had multiple models that were equally weighted (i.e., within 3 AICc units of the top 

ranked model; Table S2.3), suggesting that while speed encouraged more time spent in 

close proximity, the time spent for each visit near the robotail was not influenced by any 

particular set of variables that we tested (Table S2.3).  

The male axolotls in our experiment were slightly more likely to hula at a speed 

of 1.5 Hz when they were at a distance from a female, suggesting that a faster 

undulation speed may serve to draw females towards a courting male. Although 

research on female preferences for the speed of hydrodynamic cues is limited, some 

58 

 
female insects are known to prefer faster rates of substrate-borne vibration signals 

during courtship. For example, female meadow katydids (Conocephalus nigropleurum) 

prefer male tremulation signals with shorter inter-pulse intervals (i.e. a faster rate of 

tremulation) because shorter intervals correspond to larger males; large males typically 

produce larger spermatophores, which help to increase female fecundity (De Luca & 

Morris, 1998). Thus, faster hula speed could potentially be indicative of a higher quality 

potential mate.  

In contrast, a moderate or high elevation angle had a repelling effect; female 

axolotls oriented toward the tail marginally less often when the tail was raised to 20° or 

55°. Correspondingly, the statistical model that best explained orient count included 

elevation angle as the only predictor (Table S2.3). Although the male axolotls in our 

study did not commonly raise their tails to the highest elevation angle, we still observed 

males holding their tails aloft during courtship, particularly during process of 

spermatophore transfer. However, females may not have equated a high elevation 

angle with a courting male, given that many of the factors that contribute to mating, such 

as courtship-specific pheromones (Maex et al., 2016) and tactile feedback in the form of 

the “male bump” behavior (see Chapter 1), were absent from our experiment. Thus, 

moderate or high elevation angles may have repelled the females in our study because 

the Robotail’s undulations may have been too dissimilar to an actual male while 

performing these elevation angles.  

Physical contacts with the Robotail 

Female axolotls were more likely to nudge the robot’s tail tip if it was moving at a 

fast speed, but less likely to make physical contact in general if it was positioned at a 

59 

 
high elevation angle or moving at a moderate sweep angle. Importantly, we did not 

differentiate physical contacts (nudge tip or nudge base) that were initiated by the 

female from occurrences in which the robot hit the female in the snout; thus, the 

relationship between speed and tip nudges could be a simple consequence of a faster 

rate of motion leading to more opportunities for the Robotail to collide with the female’s 

snout. In contrast, females nudged the tip of the tail significantly less often when the tail 

was raised to its highest elevation angle of 55°; appropriately, the statistical model that 

best explained the number of tip nudges only included elevation angle (Table S2.4). At 

its highest elevation angle, the tip of the silicone tail can only be accessed by females 

while they are swimming, meaning that tip nudges at this elevation may have been 

coincidental. Thus, it seems likely that the physical contacts that we observed between 

the female’s snout and the tail tip happened accidentally.   

Females nudged the base of the tail marginally less often when the tail was 

moving at a moderate sweep angle of 30°. Accordingly, the statistical model that best 

described the number of base nudges included only sweep angle as a predictor (Table 

S2.4). The “nudge base” behavior is important for coordinating spermatophore transfer 

between male and female ambystomatid salamanders (Shoop, 1960), yet none of the 

hula parameters that we tested significantly influenced the number of times that females 

nudged the base of the Robotail with their snouts. This is likely a result of the fact that 

our experimental design omitted the tactile interactions that are necessary for 

successful courtship and mating, such as “pushing” and “male bump” behaviors (see 

Chapter 1). Additionally, although we delivered male whole-body odorants near the 

base of the robotail, which was intended to mimic odorant secretions from the cloacal 

60 

region (Maex et al., 2016), females did not perform the nudge base behavior to the 

same degree that they might in an actual courtship encounter. Given that we designed 

the Robotail to only deliver hydrodynamic stimuli and male whole-body odorants (rather 

than courtship-specific pheromones), females axolotls in our study were not provided 

with all the appropriate sensory stimuli to perform their entire suite of courtship 

behaviors.  

Female aggression 

Female axolotls nipped at the Robotail significantly less often as the sweep angle 

widened and attacked the tail marginally less often as the sweep and the elevation 

angles increased. The statistical model that best explained nip count included only 

sweep angle, whereas bite and attack count had multiple equally weighted models 

(Table S2.5). A moderate or wide sweep angle may have discouraged female axolotls 

from nipping or attacking the tail simply because the tail should have been easier for the 

female to secure in her mouth while it was moving at a narrower angle. Similarly to the 

“nudge tip” behavior, when the tail was raised to a moderate or high elevation angle it 

could only be accessed by a female while she was swimming; perhaps females are not 

physically able to coordinate aggressive behaviors during bouts of swimming. Thus, a 

narrower sweep angle and a lower elevation angle may have led females to behave 

more aggressively toward the tail simply because these parameters would have made 

the tail more accessible to the female’s mouth.  

We found that the adding a solution of male whole-body odorants to the 

aquarium during our experiments drastically reduced the severity of aggression that 

female axolotls exhibited, compared to our pilot trials in which no male odorants were 

61 

present. An experiment conducted by Proctor (1991) demonstrated a similar 

phenomenon in the courtship ritual of the water mite, Neumania papillator. Male water 

mites vibrate their first and second pairs of legs during courtship (called “trembling”) 

when they approach a female. Females often respond to trembling with prey-seeking 

behaviors unless the male has deposited a spermatophore, which may be embedded 

with pheromones; at this point, the female begins to respond to the male as a potential 

mate instead. Thus, the combination of vibrational stimuli and spermatophore odorants 

enables female water mites to recognize the presence of male conspecifics.  

Given that the addition of male axolotl odorants did not completely eliminate 

nipping, biting, and attacking behaviors in our experiment, it is possible that female 

axolotls may have perceived the Robotail as a potential prey item while it was moving 

with particular motion combinations. For example, the lateral line system of the surface-

feeding topminnow, Aplocheilus lineatus, is particularly sensitive to vibration frequencies 

between 70 and 120 Hz, which correspond to the hydrodynamic stimuli generated by 

flying insects that fall onto the water’s surface (Topp, 1983). Although male axolotls hula 

at much slower speeds (≤ 1.5 Hz), axolotls are known to strike at earthworm pieces 

dropped in front of the mouth (Lauder & Shaffer, 1985). Additionally, we have observed 

females infrequently nipping at males’ tails during actual courtship encounters, 

particularly when the tail glanced against the females’ snout. Thus, the amount of 

nipping, biting, and attacking we observed in our experiment may reflect a typical 

response to a moving object near the snout. 

62 

 
 
CONCLUSION 

Our study demonstrated that courting male axolotls modulated their hula motions 

depending on female proximity, such that they exhibited wider sweep angles, slower 

undulations, and higher elevations when they were close to a female. Notably, wider 

sweep angles (performed by the Robotail) caused females to exhibit shorter bouts of 

walking and swimming as well as longer bouts of spending time near the Robotail, and 

reduced aggressive behaviors from the female. Similarly, faster speeds led females to 

shorten their bouts of walking and spend more time overall near the Robotail. When we 

examined female responses to combinations of sweep angle and speeds, we found that 

females transitioned between locomotive states more frequently when the Robotail was 

operating at moderate to wide sweep angles combined with moderate to fast speeds. 

Overall, our experiments demonstrated that the sweep angle and speed of male hula 

motions are important aspects of axolotl courtship dynamics, whereas elevation angle 

plays a more minor role. Importantly, we found that while male axolotls are more likely 

to exhibit moderate hula patterns, female axolotls exhibit stronger behavioral responses 

to more extreme hula parameters.  

63 

 
 
TABLES 

Raw data value 

< 30° 

≥ 30° – 60° ≤ 

≥ 60° 

Bin 

10° 

30° 

90° 

< 0.6 Hz 

0.5 Hz 

≥ 0.6 Hz – 1.1 Hz ≤ 

1.0 Hz 

≥ 1.1 Hz 

1.5 Hz 

< 20° 

≥ 20° – 55° ≤ 

≥ 55° 

0° 

20° 

55° 

)
°
(

p
e
e
w
S

y
c
n
e
u
q
e
r
F

)
z
H

(

)
°
(

n
o

i
t

a
v
e
E

l

Table 2.1: Binning criteria for tail motion matrices. For each of the 3 tail motion 
parameters (sweep angle, frequency, and elevation angle), we organized raw data 
values into 3 bins. Bin values represent the minimum, median, and maximum angles for 
each parameter that male axolotls demonstrated during hula behaviors.  

64 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Behavior 

Description 

Walk 

Swim 

Pause 
Hula 

Time Spent Near  

Orient 
Nudge Base 
Nudge Tip 
Nip 
Bite 
Attack 

Female walks forward or backward at least 3 steps, with all four legs 
actively moving. 
Female swims forward using her tail only, with legs pressed against the 
body. 
Female is not walking or swimming. 
Female sways hips and posterior legs while undulating tail. 
Female is within the designated zone around the tail. Female must have at 
least her head (up to the gills) within the zone. 
Female is within the designated zone and turns her head to face the tail. 
Female makes contact with the proximal half of the tail with her snout. 
Female makes contact with the distal half of the tail with her snout. 
Female bites the tail and immediately releases it. 
Female bites the tail and briefly keeps jaws engaged.  
Female bites the tail and thrashes her head side to side. 

Table 2.2: Ethogram of female behaviors. We measured female behavioral 
responses to the robot by quantifying locomotor behaviors, instances of hula-ing, time 
spent near the robot, physical contact with the robot, and aggressive behaviors.  

65 

 
 
 
A) Close Bouts 

Sweep Angle 

Hz 
 

10° 

30° 

90° 

  Occurrences 

0° 

0.5  1.0  1.5  0.5  1.0  1.5  0.5  1.0  1.5 

Never 

20° 

0.5  1.0  1.5  0.5  1.0  1.5  0.5  1.0  1.5 

Sometimes 

55° 

0.5  1.0  1.5  0.5  1.0  1.5  0.5  1.0  1.5 

Often 

n
o

i
t

a
v
e
E

l

l

e
g
n
A

B) Distant Bouts 

Sweep Angle 

Hz 
 

0° 

20° 

55° 

10° 

1.0 

1.0 

1.0 

0.5 

0.5 

0.5 

n
o

i
t

a
v
e
E

l

l

e
g
n
A

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

C) All Bouts 

Sweep Angle 

10° 

1.0 

1.0 

1.0 

0.5 

0.5 

0.5 

Hz 
 

0° 

20° 

55° 

n
o

i
t

a
v
e
E

l

l

e
g
n
A

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

30° 

1.0 

1.0 

1.0 

30° 

1.0 

1.0 

1.0 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

90° 

1.0 

1.0 

1.0 

90° 

1.0 

1.0 

1.0 

1.5 

1.5 

1.5 

1.5 

1.5 

1.5 

Table 2.3: Matrices displaying the occurrence rate of tail motion combinations. 
Each hula bout (A; Close, B; Distant, C; All) was binned (Table 1) and then categorized 
into one of the 27 motion combinations. Combination frequencies above the median 
value were deemed as “often”, and values below the median were labeled as 
“sometimes”, and values of 0 were labeled as “never”.  

66 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable 

Model 

Fixed Effects 

t statistic 

p - value 

Walk Total Duration 

Walk Average Duration 

Swim Total Duration 

Swim Average Duration 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

1.64 
0.81 
0.72 
0.18 
0.47 
-1.02 

-0.20 
-2.91 
-0.23 
-2.27 
1.34 
0.97 

-0.84 
0.03 
0.62 
-0.48 
0.42 
0.80 

-2.21 
-2.21 
0.62 
-0.78 
0.31 
0.63 

0.10 
0.42 
0.48 
0.86 
0.64 
0.31 

0.84 
0.004 ** 
0.82 
0.03 * 
0.18 
0.34 

0.40 
0.98 
0.54 
0.63 
0.68 
0.43 

0.03 * 
0.03 * 
0.54 
0.44 
0.76 
0.53 

Table 2.4: General linear mixed-effects models for female locomotor data 
(continued on next page).  

67 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable 

Model 

Fixed Effects 

t statistic  p - value 

Locomote Total Duration 

Locomote Average Duration  

Pause Total Duration 

Pause Average Duration 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

0.58 
30° sweep 
0.64 
90° sweep  
1.22 
1.0 Hz 
1.5 Hz 
-0.34 
20° elevation   0.61 
-0.04 
55° elevation  

-1.58 
30° sweep 
-2.15 
90° sweep  
0.78 
1.0 Hz 
1.5 Hz 
-0.97 
20° elevation   0.62 
55° elevation   0.88 

-0.57 
30° sweep 
-0.61 
90° sweep  
-1.14 
1.0 Hz 
0.38 
1.5 Hz 
20° elevation  
-0.52 
55° elevation   0.09 

-0.70 
30° sweep 
-1.39 
90° sweep  
-1.47 
1.0 Hz 
-1.79 
1.5 Hz 
20° elevation   0.76 
55° elevation   0.57 

0.56 
0.53 
0.22 
0.74 
0.54 
0.97 

0.12 
0.03 * 
0.44 
0.34 
0.54 
0.38 

0.57 
0.54 
0.26 
0.71 
0.60 
0.93 

0.49 
0.17 
0.15 
0.08 
0.45 
0.57 

Table 2.4 (continued): Changes in female locomotor behaviors in response to the 3 
levels of each hula motion parameter. Average durations of walk, swim, and locomote 
bouts differed significantly with sweep angle, speed, or both. Female ID number was 
used as the random effect in all analyses; all t statistics and p-values were rounded to 2 
decimals, except when the p-value was less than 0.01. 

68 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable 

Model 

Fixed Effects 

t statistic 

p - value 

Time Spent Near  
Total Duration 

Time Spent Near  
Average Duration 

Orient Count 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

30° sweep 

90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  
30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  
30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

-0.51 

-1.38 
1.48 
2.63 
-1.08 
-1.30 
-1.04 
-2.20 
0.43 
1.90 
-0.29 
-0.09 
-1.14 
-1.14 
0 
0.38 
-1.92 
-1.92 

0.61 

0.17 
0.14 
0.009 **  
0.28 
0.20 
0.30 
0.03 * 
0.67 
0.06 
0.77 
0.93 
0.26 
0.26 
1 
0.71 
0.06 
0.06 

Table 2.5: General linear mixed-effects models for time spent near the Robotail. 
Changes in duration and count of times that the female spent near the tail, as well as 
orient count, in response to the 3 levels of each hula motion parameter. Duration of time 
spent near (total and average) differed significantly with speed or sweep angle. Female 
ID number was used as the random effect in all analyses; all t statistics and p-values 
were rounded to 2 decimals, except when the p-value was less than 0.01. 

69 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable  Model 

Fixed Effects 

t statistic 

p - value 

Nudge Base Count 

Nudge Tip Count 

Total Nudge Count 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  
30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

-1.89 
-1.49 
-0.48 
0.45 
-0.59 
-0.54 
0.50 
0.25 
0.58 
2.11 
-1.43 
-2.85 

-1.56 
-1.26 
-0.35 
0.77 
-0.86 
-1.09 

0.06  
0.14 
0.63 
0.65 
0.56 
0.59 
0.61 
0.80 
0.57 
0.04 * 
0.15 
0.005 **  

0.12 
0.21 
0.72 
0.44 
0.39 
0.28 

Table 2.6: General linear mixed-effects models for physical contacts with the 
Robotail. Changes in the number of times females physically interacted with the tail, 
relative to the 3 levels of each hula motion parameter. Nudge tip count, but not nudge 
base count or the total number of nudges performed, differed significantly with speed 
and elevation angle. Female ID number was used as the random effect in all analyses; 
all t statistics and p-values were rounded to 2 decimals, except when the p-value was 
less than 0.01. 

70 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable  Model 

Fixed Effects 

t statistic 

p - value 

Nip Count 

Bite Count 

Attack Count 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

Sweep 

Speed 

Elevation 

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  
30° sweep 
90° sweep  
1.0 Hz 
1.5 Hz 
20° elevation  
55° elevation  

-3.00 
-2.55 
0.25 
0.15 
-1.73 
0.58 

-1.24 
-1.24 
1.24 
0 
-1.24 
-1.24 
-1.77 
-1.77 
-0.87 
0 
-1.77 
-1.77 

0.003 **  
0.01 *   
0.80 
0.88 
0.09 
0.56 

0.22 
0.22 
0.22 
1 
0.22 
0.22 
0.08 
0.08 
0.38 
1 
0.08 
0.08 

Table 2.7: General linear mixed-effects models for female aggression data. 
Changes in female aggressive behaviors relative to the 3 levels of each hula motion 
parameter. The number of times females nipped at the tail, but not the number of bites 
or attacks differed significantly with sweep angle. Female ID number was used as the 
random effect in all analyses; all t statistics and p-values were rounded to 2 decimals, 
except when the p-value was less than 0.01. 

71 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent 
Variable 

Random 
Effect 

Model 

Fixed Effects 

t statistic 

p - value 

Locomotion 
transitions 
per minute 

Female 
ID 

Combo 

Sweep (°)  Speed (Hz) 
30 
90 
10 
30 
90 
10 
30 

0.5 
0.5 
1.0 
1.0 
1.0 
1.5 
1.5 

90 

1.5 

0.52 
1.69 
0.69 
1.79 
2.19 
-0.28 
2.37 

4.01 

0.61 
0.09 
0.49 
0.08 
0.03 * 
0.78 
0.02 * 
9.51e-05 
*** 

Table 2.8: General linear mixed-effects models for locomotion transitions. 
Changes in the number of times that females transitioned between locomotive states 
per minute relative to each combination of sweep angle and speed that the Robotail 
performed. Females locomoted significantly more often if the Robotail was programmed 
with a combination of 90°/1.0 Hz, 30°/1.5 Hz, or 90°/1.5 Hz compared to a combination 
of 10°/1.5 Hz. Female ID number was used as the random effect in this analysis; all t 
statistics and p-values were rounded to 2 decimals, except when the p-value was less 
than 0.01. 

72 

  
  
  
  
  
 
 
 
FIGURES 

Figure 2.1: Measurement parameters for male axolotl tails. Green dashed lines 
indicate the 5 positions along the tail at which measurements were taken to create the 
Robotail. At each position, we measured the width and height of the dorsal fin (A), the 
width and height of the ventral fin (B), the width at the middle of the tail (C) and the total 
height (D).  

73 

 
 
 
4 

3 

2 

1 

A 

B 

1 

2 

3 

4 

Figure 2.2: Diagram of the silicone tail and mounting components. Overhead (A) 
and side (B) views of the tail and mounting components; parts consisted of (1) the 
silicone tail (2) the clip embedded in the tail (3) the pivoting cylinder and (4) the 
triangular base. The embedded clip slid into a slot on the pivoting cylinder and was held 
in place by a nylon bolt and nut and 2 rubber washers. Green lines indicate the position 
of the fishing lines that were used to turn the pivoting cylinder from side to side. 

74 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1 

2 

1 

2 

3 

3 

5 

4 

3 

4 

A 

B 

Figure 2.3: Robotic components contained within the aquarium. The Robotail with 
dry components (A) and with the components submerged in water (B). Components 
attached to the aquarium floor included tubing carrying male odorants (1), the silicone 
salamander toy (2), triangular base plate (3), and EVA foam tiles (4); the silicone tail (5) 
was attached to the base plate via a pivoting cylinder and is shown in panel B only.  

75 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
A 

B 

E 

D 

F 

G 

C 

Figure 2.4: Electronic components of the Robotail. Components consisted of (A) 
power source (B) power distribution channel (C) Victor SP speed controller (D) Thrifty 
Throttle 3/Cypress PSOC4™ microcontroller (E) potentiometer (F) 104:1 NeveRest 
Sport gearbox and NeveRest gearmotor and (G) custom gimbal. The gimbal swiveled to 
the left and right to drive the position of the silicone tail; red arrows indicate the 
attachment points on the gimbal for the fishing lines. 

76 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 2.5: Average walk duration of females in response to sweep angle.  Female 
axolotls exhibited shorter bouts of walking when stimulated by the widest sweep angle 
of 90° (mixed model analysis, t = -2.91, p = 0.004). Each dot represents the average 
duration of a walking bout for a given trial, and bars represent the standard error of each 
group. 

77 

 
 
 
Figure 2.6: Average walk duration of females in response to speed.  Female 
axolotls exhibited shorter bouts of walking when stimulated by the fastest speed of 1.5 
Hz (mixed model analysis, t = -2.27, p = 0.03). Each dot represents the average 
duration of a walking bout for a given trial, and bars represent the standard error of each 
group. 

78 

 
 
Figure 2.7: Average swim duration of females in response to sweep 
angle.  Female axolotls exhibited shorter bouts of walking as the sweep angle of the 
Robotail increased (mixed model analysis, 10° vs 30°, t = -2.21, p = 0.03, 10° vs 90°, t = 
-2.21, p = 0.03). Each dot represents the average duration of a swimming bout for a 
given trial, and bars represent the standard error of each group. 

79 

 
 
Figure 2.8: Average locomotion duration of females in response to sweep 
angle.  Female axolotls exhibited shorter bouts of locomotion (walking plus swimming) 
when stimulated by the widest sweep angle of 90° (mixed model analysis, t = -2.15, p = 
0.03). Each dot represents the average duration of a locomotor bout for a given trial, 
and bars represent the standard error of each group. 

80 

 
 
 
Figure 2.9: Total duration of time spent near for females in response to 
speed.  Female axolotls exhibited more time near the Robotail overall when stimulated 
by the fastest speed of 1.5 Hz (mixed model analysis, t = 2.63, p = 0.009). Each dot 
represents the total duration of TSN for a given trial, and bars represent the standard 
error of each group. 

81 

 
 
Figure 2.10: Average duration of time spent near for females in response to 
sweep angle.  Female axolotls exhibited shorter bouts of spending time near the 
Robotail when stimulated by the widest sweep angle of 90° (mixed model analysis, t = -
2.20, p = 0.03). Each dot represents the average duration of a TSN bout for a given trial, 
and bars represent the standard error of each group. 

82 

 
 
 
 
REFERENCES 

Ahnesjö, I., Brealey, J. C., Günter, K. P., Martinossi-Allibert, I., Morinay, J., Siljestam, 

M., Stångberg, J., Vasconcelos, P. (2020). Considering gender-biased assumptions 
in evolutionary biology. Evolutionary Biology, 47(1), 1–5. 

Altmann, J. (1974). Observational study of behavior: sampling methods. Behaviour, 

49(3–4), 227–266. 

Anderson, J. D. (1961). The courtship behavior of Ambystoma macrodactylum croceum. 

Copeia, 1961(2), 132–139. 

Armstrong, J. B., Duhon, S. T., & Malacinski, G. M. (1989). Raising the axolotl in 

capitivity. In J. B. Armstrong & G. M. Malacinski (Eds.), Developmental Biology of 
the Axolotl (pp. 220–227). New York: Oxford University Press. 

Armstrong, J. B., & Malacinski, G. M. (1989). Developmental biology of the axolotl (7th 

ed.). Oxford University Press. 

Arnold, S. J. (1987). The comparative ethology of courtship in salamandrid 
salamanders. 1. Salamandra and Chioglossa. Ethology, 74, 133–145. 

Arnold, S. J. (1976). Sexual behavior, sexual interference and sexual defense in the 

salamanders Ambystoma maculatum, Ambystoma tigrinum and Plethodon jordani. 
Zeitschrift Für Tierpsychologie, 42(3), 247–300. 

Arnold, S. J. (1972). The evolution of courtship behavior in salamanders. University of 

Michigan. 

Baker, C. A., Clemens, J., & Murthy, M. (2019). Acoustic Pattern Recognition and 

Courtship Songs: Insights from Insects. Annual Review of Neuroscience, 42, 129–
147.  

Baker, C. V. H., Modrell, M. S., & Gillis, J. A. (2013). The evolution and development of 
vertebrate lateral line electroreceptors. Journal of Experimental Biology, 216, 2515–
2522. 

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects 

models using lme4. Journal of Statistical Software, 67(1), 1–48. 

Bazar, M. A., Quinn, M. J., Mozzachio, K., Bleiler, J. A., Archer, C. R., Phillips, C. T., & 
Johnson, M. S. (2009). Toxicological responses of red-backed salamanders 
(Plethodon cinereus) to soil exposures of copper. Archives of Environmental 
Contamination and Toxicology, 57(1), 116–122. 

83 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
Bolker, B. M. (2015). Linear and generalized linear mixed models. In G. A. Fox, S. 
Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary 
Theory and Application (1st ed., pp. 309–333). Oxford University Press. 

Butler, J. M., & Maruska, K. P. (2015). The mechanosensory lateral line is used to 

assess opponents and mediate aggressive behaviors during territorial interactions 
in an African cichlid fish. Journal of Experimental Biology. 

Butler, Julie M., & Maruska, K. P. (2016). Mechanosensory signaling as a potential 

mode of communication during social interactions in fishes. Journal of Experimental 
Biology, 219(18), 2781–2789. 

Coombs, S., Bleckmann, H., Fay, R. R., & Popper, A. N. (Eds.). (2014). The lateral line 

system. New York: Springer. 

Coombs, S., Görner, P., & Münz, H. (1989). A brief overview of the mechanosensory 

lateral line system and the contributions to this volume. In S. Coombs, P. Görner, & 
H. Münz (Eds.), The Mechanosensory Lateral Line : Neurobiology and Evolution 
(pp. 3–5). New York: Springer-Verlag. 

Coombs, S., & Montgomery, J. (2014). The role of flow and the lateral line in the 

multisensory guidance of orienting behaviors. In H. Bleckmann, J. Mogdans, & S. 
Coombs (Eds.), Flow Sensing in Air and Water: Behavioral, Neural and 
Engineering Principles of Operation (pp. 65–101). 

Cornwall, M. C., MacNichol Jr, E. F., & Fein, A. (1984). Absorptance and spectral 
sensitivity measurements of rod photoreceptors of the tiger salamander, 
Ambystoma tigrinum. Vision Research, 24(11), 1651–1659. 

De Luca, P. A., & Morris, G. K. (1998). Courtship communication in meadow katydids : 

female preference for large male vibrations. Behaviour, 135(6), 777–794. 

Denoël, M., & Doellen, J. (2010). Displaying in the dark: light-dependent alternative 
mating tactics in the Alpine newt. Behavioral Ecology and Sociobiology, 64(1), 
1171–1177. 

Eisthen, H. (1989). Courtship and mating behavior in the axolotl. Axolotl Newsletter, 18, 

18–19. 

Friard, O., & Gamba, M. (2016). BORIS: a free, versatile open-source event-logging 
software for video/audio coding and live observations. Methods in Ecology and 
Evolution, 7(11), 1325–1330. 

Fritzsch, B., & Wahnschaffe, U. (1983). The electroreceptive ampullary organs of 

urodeles. Cell Tissue Research, 229, 483–503. 

84 

 
 
 
 
 
 
 
 
 
 
 
 
Green, A. J. (1989). The sexual behaviour of the great crested newt, Triturus cristatus 

(Amphibia: Salamandridae). Ethology, 83, 129–153. 

Gresens, J. (2004). An introduction to the Mexican Axolotl (Ambystoma mexicanum). 

Lab Animal, 33(9), 41–47.  

Guevara-Fiore, P., Stapley, J., Krause, J., Ramnarine, I. W., & Watt, P. J. (2010). Male 

mate-searching strategies and female cues: how do male guppies find receptive 
females? Animal Behaviour, 79(6), 1191–1197. 

Halliday, T. R. (1990). The evolution of courtship behavior in newts and salamanders. 

Advances in the Study of Behavior, 19, 137–169. 

Houck, L. D., & Arnold, S. J. (2003). Courtship and mating behavior. In Reproductive 

Biology and Phylogeny of Urodela (1st ed., pp. 383–424). 

Hrouzková, E., Dvořáková, V., Jedlička, P., & Šumbera, R. (2013). Seismic 

communication in demon African mole rat Tachyoryctes daemon from Tanzania. 
Journal of Ethology.  

Joel, A. C., Habedank, A., & Mey, J. (2017). Vibratory movements in contests between 

females of the feather-legged spider (Uloborus plumipes). Zoology.  

Kernighan, B. W., & Ritchie, D. M. (2002). The C programming language. 

Kim, J. K., Lee, J. H., Ra, N. Y., Lee, H. J., Eom, J., & Park, D. (2009). Reproductive 

function of the body and tail undulations of Hynobius leechii (Amphibia: 
Hynobiidae): A quantitative approach. Animal Cells and Systems, 13, 71–78. 

Kramer, D. L., & McLaughlin, R. L. (2001). The behavioral ecology of intermittent 

locomotion. American Zoologist, 41(2), 137–153. 

Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: 

tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. 

Lauder, G. V, & Shaffer, H. B. (1985). Functional Morphology of the Feeding 
Mechanism in Aquatic Ambystomatid Salamanders, 326, 297–326. 

Lunsford, E. T., & Liao, J. C. (2021). Activity of posterior lateral line afferent neurons 

during swimming in zebrafish. Journal of Visualized Experiments, 2021(168), 1–17. 

Maex, M., Bocxlaer, I. Van, Mortier, A., Proost, P., & Bossuyt, F. (2016). Courtship 

pheromone use in a model urodele, the Mexican axolotl (Ambystoma mexicanum). 
Scientific Reports, 6(20184), 1–9. 

85 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Marsh, D. M., & Hanlon, T. J. (2004). Observer gender and observation bias in animal 
behaviour research: experimental tests with red-backed salamanders. Animal 
Behaviour, 68(6), 1425–1433. 

Montgomery, J., Bleckmann, H., & Coombs, S. (2014). Sensory ecology and 

neuroethology of the lateral line. In S. Coombs, H. Bleckmann, R. R. Fay, & A. N. 
Popper (Eds.), The Lateral Line System (pp. 121–150). New York: Springer. 

Münz, H., Claas, B., & Fritzsch, B. (1984). Electroreceptive and mechanoreceptive units 
in the lateral line of the axolotl Ambystoma mexicanum. Journal of Comparative 
Physiology A, 33–44. 

O’Connell-Rodwell, C. E. (2024). Keeping an “ Ear ” to the Ground: Seismic 

Communication in Elephants, (45), 287–294. 

Osorio, D., & Vorobyev, M. (2008). A review of the evolution of animal colour vision and 

visual communication signals. Vision Research, 48, 2042–2051.  

Park, D., McGuire, J. M., Majchrzak, A. L., Ziobro, J. M., & Eisthen, H. L. (2004). 

Discrimination of conspecific sex and reproductive condition using chemical cues in 
axolotls (Ambystoma mexicanum). Journal of Comparative Physiology A, 190, 415–
427. 

Park, Daesik, Lee, J.-H., Ra, N.-Y., & Eom, J. (2008). Male salamanders Hynobius 
leechii respond to water vibrations via the mechanosensory lateral line system. 
Journal of Herpetology, 42(4), 615–625. 

Petranka, J. (1982). Courtship behavior of the small-mouthed salamander (Ambystoma 

texanum) in Central Kentucky. Herptologica, 38(2), 333–336. 

Pierotti, R., Annett, C., & Hand, J. (1997). Male and female perceptions of pair-bond 
dynamics: monogamy in western gulls, Larus occidentalis. In Feminism and 
Evolutionary Biology: Boundaries, Intersections and Frontiers (pp. 261–275). 
Boston, MA: Springer US. 

Plazas, P. V., & Elgoyhen, A. B. (2021). The cholinergic lateral line efferent synapse: 

structural, functional and molecular similarities with those of the cochlea. Frontiers 
in Cellular Neuroscience, 15, 1–15. 

Proctor, H. C. (1991). Courtship in the water mite Neumania papillator: males capitalize 

on female adaptations for predation. Animal Behaviour, 42(4), 589–598. 

Revelle, W. (2023). psych: procedures for psychological, psychometric, and personality 
research. Northwestern University, Evanston, Illinois. R package version 2.3.6. 

86 

 
 
 
 
 
 
 
 
 
 
 
 
 
Russell, I. J. (1968). Influence of efferent fibres on a receptor. Nature, 219(5150), 177–

178. 

Salthe, S. N. (1967). Courtship patterns and the phylogeny of the urodeles. Copeia, 1, 

100–117. 

Satou, M., Takeuchi, H. A., Takei, K., Hasegawa, T., Matsushima, T., & Okumoto, N. 

(1994). Characterization of vibrational and visual signals which elicit spawning 
behavior in the male himé salmon (landlocked red salmon, Oncorhynchus nerka). 
Journal of Comparative Physiology A, 174, 527–537. 

Satou, Masahiko, Shiraishi, A., Matsushima, T., & Okumoto, N. (1991). Vibrational 
communication during spawning behavior in the himé salmon (landlocked red 
salmon, Oncorhynchus nerka). Journal of Comparative Physiology A. 

Shoop, C. R. (1960). The breeding habits of the mole salamander, Ambystoma 

talpoideum (Holbrook), in southeastern Louisiana. Master’s Thesis, Tulane Studies 
in Zoology and Botany, 8(3), 65–82. 

Soleymani, M., Villaro-Dixon, F., Pun, T., & Chanel, G. (2017). Toolbox for Emotional 

feAture extraction from Physiological signals (TEAP). Frontiers in ICT. 

Storez, R. A. (1969). Observations on the courtship of Ambystoma laterale. Journal of 

Herpetology, 3(1), 87–95. 

Topp, G. (1983). Primary lateral line response to water surface waves in the topminnow 

Aplocheilus lineatus (Pisces, Cyprinodontidae). Pflügers Archiv European Journal 
of Physiology. 

Verrell, P. A. (1983). The influence of the ambient sex ratio and intermale competition 
on the sexual behavior of the red-spotted newt, Notophthalmus viridescens 
(Amphibia: Urodela: Salamandridae). Behavioral Ecology and Sociobiology, 13, 
307–313. 

Verrell, P., & Pelton, J. (1996). The sexual strategy of the central long-toed salamander, 
Ambystoma macrodactylum columbianum, in south-eastern Washington. Journal of 
Zoology, 240(1), 37–50. 

Verrell, P. (1982). The sexual behaviour of the red-spotted newt, Notophthalmus 

viridescens (Amphibia: Urodela: Salamandridae). Animal Behaviour, 30(4), 1224–
1236. 

Verrell, P. (1999). Bracketing the extremes: courtship behaviour of the smallest- and 

largest-bodied species in the salamander genus Desmognathus (Plethodontidae: 
Desmognathinae). Journal of Zoology, 247(1), 105–111. 

87 

 
 
 
 
 
 
 
 
 
 
 
 
Voss, S. R., Woodcock, M. R., & Zambrano, L. (2015). A tale of two axolotls. 

BioScience, 65(12), 1134–1140. 

Woolley, S. M. N., & Moore, J. M. (2011). Coevolution in communication senders and 

receivers: vocal behavior and auditory processing in multiple songbird species. 
Annals of the New York Academy of Sciences, 1225(1), 155–165. 

Wullimann, M. F., & Grothe, B. (2014). The central nervous organization of the lateral 

line system. In The Lateral Line System (pp. 195–251). 

88 

 
  
 
 
A) Close Bouts 

APPENDIX 

Sweep Angle 

Hz 
 

0° 

10° 

30° 

90° 

0 males 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

1 male 

20° 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

2 males 

55° 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

0.5 

1.0 

1.5 

3 males 

n
o

i
t

a
v
e
E

l

l

e
g
n
A

10° 

1.0 

1.0 

1.0 

10° 

1.0 

1.0 

1.0 

0.5 

0.5 

0.5 

0.5 

0.5 

0.5 

Sweep Angle 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

30° 

1.0 

1.0 

1.0 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

Sweep Angle 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

30° 

1.0 

1.0 

1.0 

1.5 

1.5 

1.5 

0.5 

0.5 

0.5 

4 males 

5 males 

90° 

1.0 

1.0 

1.0 

90° 

1.0 

1.0 

1.0 

1.5 

1.5 

1.5 

1.5 

1.5 

1.5 

B) Distant Bouts 

Hz 
 

n
o

i
t

a
v
e
E

l

l

  0° 
e
g
n
A

20° 

55° 

C) All Bouts 

Hz 
 

n
o

i
t

a
v
e
E

l

l

  0° 
e
g
n
A

20° 

55° 

Table S2.1: Matrices displaying the number of males that performed each tail 
motion combination. We organized hula bouts by proximity (A; Close, B; Distant, C; 
All), and counted the total number of males that performed each of the 27 motion 
combinations. No single combination was performed by all 6 males. 

89 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Dependent Variable 

Model 

AICc 

Δ AICc 

Walk Total  
Duration 

Walk Average 
Duration 

Swim Total 
Duration 

Swim Average 
Duration 

837.4518  0 
839.0042  1.5524 
2.3002 

2053.646  0 
Sweep 
2053.983  0.337 
Elevation  
2055.664  2.018 
Sweep+Elevation 
2055.758  2.112 
Speed 
2057.400  3.754 
Speed+Sweep 
2057.749  4.103 
Speed+Elevation 
Speed+Sweep+Elevation  2059.512  5.866 
Speed+Sweep 
Sweep 
Speed+Sweep+Elevation  839.752 
Sweep+Elevation 
Speed 
Speed+Elevation 
Elevation 

841.1188  3.667 
843.0461  5.5943 
845.5645  8.1127 
847.1814  9.7296 
1981.949  0 
Speed 
1982.210  0.261 
Sweep 
1982.516  0.567 
Elevation  
1985.427  3.478 
Speed+Sweep 
1985.694  3.745 
Speed+Elevation 
Sweep+Elevation 
1985.867  3.918 
Speed+Sweep+Elevation  1989.221  7.272 
Sweep 
1238.13 
Speed+Sweep 
1241.036  2.906 
Sweep+Elevation 
1242.004  3.874 
Speed 
1242.577  4.447 
Elevation 
1244.118  5.988 
Speed+Sweep+Elevation  1245.051  6.921 
Speed+Elevation 
1246.567  8.437 

0 

Table S2.2: AICc rankings of mixed-effects models for female locomotion data 
(continued on next page). 

90 

 
 
Dependent Variable 

Model 

AICc 

Δ AICc 

Locomote Total 
Duration 

Locomote Average 
Duration 

Pause Total  
Duration 

Pause Average 
Duration 

Speed 

2152.368  0 

Elevation  
2154.483  2.115 
Sweep 
2154.522  2.154 
Speed+Sweep 
2156.163  3.795 
Speed+Elevation 
2156.181  3.813 
Sweep+Elevation 
2158.361  5.993 
Speed+Sweep+Elevation  2160.126  7.758 

Sweep 

1230.631  0 

1.379 

Speed+Sweep 
Speed 
Sweep+Elevation 
Elevation 
Speed+Sweep+Elevation  1235.631  5 
Speed+Elevation 

1232.01 
1232.531  1.9 
1234.137  3.506 
4.089 
1234.72 

1236.07 

5.439 

Speed 

2153.560  0 

Sweep 
2155.540  1.980 
Elevation  
2155.563  2.003 
Speed+Sweep 
2157.388  3.828 
Speed+Elevation 
2157.469  3.909 
Sweep+Elevation 
2159.468  5.908 
Speed+Sweep+Elevation  2161.442  7.882 
1499.972  0 
Speed 
Sweep 
1501.661  1.689 
Speed+Sweep 
1502.495  2.523 
Elevation 
2.978 
1502.95 
Speed+Elevation 
1503.736  3.764 
Sweep+Elevation 
1505.337  5.365 
Speed+Sweep+Elevation  1506.325  6.353 

Table S2.2 (cont’d): Seven statistical models were built for each locomotion variable 
and their corresponding AICc values calculated to assess which model(s) best 
explained our data set. Equally weighted models (i.e., within 3 AICc values of the top 
ranked model) are indicated by gray shading. The model that best described each 
behavioral variable, which is the simplest model with the lowest AICc value, is indicated 
by bold text.  

91 

 
 
Dependent Variable 

Model 

AICc 

Δ AICc 

Time Spent Near 
Total Duration  

Time Spent Near 
Average Duration 

Orient Count 

Speed 
1764.445 
Speed+Sweep 
1766.572 
Speed+Elevation 
1766.768 
Speed+Sweep+Elevation  1769.153 
Sweep 
1769.252 
Elevation  
1769.294 
Sweep+Elevation 
1771.826 

Speed+Sweep 
1210.355 
Sweep 
1210.456 
Speed 
1211.277 
Sweep+Elevation 
1214.821 
Speed+Sweep+Elevation  1214.856 
Elevation 
1215.148 
Speed+Elevation 
1215.587 

Elevation  
377.085 
Sweep+Elevation 
379.641 
Sweep 
380.211 
Speed+Elevation 
381.230 
Speed 
381.752 
Speed+Sweep+Elevation  383.899 
Speed+Sweep 
384.360 

0 
2.127 
2.323 
4.708 
4.807 
4.849 
7.381 

0 
0.101 
0.922 
4.466 
4.501 
4.793 
5.232 

0 
2.557 
3.127 
4.145 
4.668 
6.814 
7.276 

Table S2.3: AICc rankings of mixed-effects models for time spent near (TSN) and 
orient data. Seven statistical models were built for the duration and frequency of TSN 
as well as orient count, and their corresponding AICc values calculated to assess which 
model(s) best explained our data set. Equally weighted models (i.e., within 3 AICc 
values of the top ranked model) are indicated by gray shading. The model that best 
described each behavioral variable, which is the simplest model with the lowest AICc 
value, is indicated by bold text.  

92 

 
 
Dependent Variable 

Model 

AICc 

Δ AICc 

Nudge Base Count 

Nudge Tip Count 

Total Nudge Count 

0 

599.138 

1119.697  0 
Sweep 
Speed 
1122.730  3.033 
Speed+Sweep 
1123.090  3.393 
Elevation  
1123.168  3.471 
Sweep+Elevation 
1123.644  3.947 
Speed+Elevation 
1126.653  6.956 
Speed+Sweep+Elevation  1127.155  7.458 
Speed+Elevation 
Elevation  
599.704 
Speed 
602.893 
Sweep+Elevation 
603.750 
Speed+Sweep+Elevation  603.320 
Speed+Sweep 
606.974 
Sweep 
607.319 
1148.241  0 
Sweep 
Elevation  
1149.645  1.404 
Speed 
1149.646  1.405 
Speed+Sweep 
1151.174  2.933 
Sweep+Elevation 
1151.306  3.065 
Speed+Elevation 
1152.691  4.450 
Speed+Sweep+Elevation  1154.357  6.116 

0.566 
3.755 
4.612 
4.182 
7.836 
8.182 

Table S2.4: AICc rankings of mixed-effects models for physical contacts with the 
Robotail. Seven statistical models were built for nudge base, nudge tip, and total 
nudges and their corresponding AICc values calculated to assess which model(s) best 
explained our data set. Equally weighted models (i.e., within 3 AICc values of the top 
ranked model) are indicated by gray shading. The model that best described each 
behavioral variable, which is the simplest model with the lowest AICc value, is indicated 
by bold text.  

93 

 
 
Dependent Variable 

Model 

AICc 

Δ AICc 

Nip Count 

Bite Count 

Attack Count 

Sweep+Elevation 
472.933  0 
474.195  1.263 
Sweep 
Speed+Sweep+Elevation  477.322  4.390 
478.446  5.514 
Speed+Sweep 
478.664  5.732 
Elevation  
482.955  10.022 
Speed+Elevation 
484.277  11.345 
Speed 

Sweep 
Speed 
Elevation  
Sweep+Elevation 
Speed+Sweep 
Speed+Elevation 
Speed+Sweep+Elevation 

Elevation  
Sweep 
Sweep+Elevation 
Speed 
Speed+Elevation 
Speed+Sweep 
Speed+Sweep+Elevation 

-357.098  0 
-357.098  0 
-357.098  0 
-354.806  2.292 
-354.806  2.292 
-354.806  2.292 
-352.426  4.672 

-247.894  0 
-247.894  0 
-247.759  0 
-244.808  3.086 
-244.593  3.301 
-244.593  3.301 
-244.372  3.522 

Table S2.5: AICc rankings of mixed-effects models for aggression data. Seven 
statistical models were built for measures of aggression and their corresponding AICc 
values calculated to assess which model(s) best explained our data set. Equally 
weighted models (i.e., within 3 AICc values of the top ranked model) are indicated by 
gray shading. The model that best described each behavioral variable, which is the 
simplest model with the lowest AICc value, is indicated by bold text. 

94 

 
 
 
 
 
CHAPTER 3: Lateral Line Responses to the Male Hula Behavior 

ABSTRACT 

Many physical features of the communication signals exchanged between 

animals are governed by a hypothesis known as sender-receiver matching. This 

hypothesis posits that communication signals generated by senders are shaped by the 

tuning properties of the receiver’s corresponding sensory system.  While sender-

receiver matches have been extensively documented within auditory communication, it 

is currently unclear if this hypothesis also pertains to vibratory communication strategies 

within mechanosensory systems. We aimed to determine whether sender-receiver 

matching occurs within the mechanosensory lateral line system specifically, and we 

leveraged the courtship ritual of the axolotl, Ambystoma mexicanum, a fully aquatic 

salamander, to achieve this aim. 

Male axolotls reliably perform a courtship behavior known as the “hula”, which 

features a swaying motion of the hips paired with an undulation of the tail; this behavior 

generates disturbances in the surrounding water, which likely stimulate the lateral line 

systems of female receivers. We used a custom device called the “Robotail” that mimics 

the male hula behavior to deliver a range of hydrodynamic stimuli to our test subjects 

and recorded multi-unit activity from the peripheral aspect of the anterodorsal lateral line 

nerve (ADLLn). This nerve innervates all neuromasts (mechanically sensitive hair cells) 

on the anterior dorsal surface of the axolotl’s head; given that females position their 

heads underneath the hula-ing male’s tail at several key points during the courtship 

ritual, this cluster of neuromasts is likely to be highly stimulated by the hula behavior.  

95 

 
We found that the ADLLn of female axolotls exhibited excitatory responses to all 

sweep angles (i.e., angle of side-to-side motion) and speeds that we tested and 

exhibited the strongest responses to moderate angles and fast speeds. In contrast, the 

ADLLn of male axolotls exhibited inhibitory responses to narrow angles and slow 

speeds, but excitatory responses to all other parameters that we tested. In our previous 

research, we determined that male axolotls most often hula’d during courtship using 

moderate sweep angles; thus, the results of our current experiment act as preliminary 

evidence that sender-receiver matching may occur within the vibratory communication 

modality in axolotl courtship.  

INTRODUCTION 

Animals exchange countless communication signals with conspecifics to 

coordinate social processes like aggressive contests (Butler & Maruska, 2015), 

cooperative behaviors (Noë, 2006), and mating (Denoël & Doellen, 2010). The sender-

receiver matching hypothesis seeks to identify and explain physical constraints on 

communication signals and their detection (Capranica & Moffat, 1983); this hypothesis 

posits that senders generate communication signals that are tuned to the sensory 

systems of receivers and vice versa, such that receivers also play a role in shaping 

signal characteristics (Henry et al., 2016). Sender-receiver matching has been most 

notably demonstrated in the acoustic communication modality (Gall et al., 2012; Henry 

et al., 2016; Mhatre et al., 2011), but it also occurs in electric organ discharges and their 

detection by the lateral line system of Apteronotus (Allen & Marsat, 2019).  

However, considerably less research has been conducted on sender-receiver 

matching that potentially occurs in the mechanosensory aspect of the lateral line 

96 

system. Although Satou et al. (1994) demonstrated that the lateral line is responsible for 

mediating vibratory communication during the courtship ritual of the himé salmon 

(Oncorhynchus nerka), little evidence demonstrates that the lateral line system is 

physiologically tuned to hydrodynamic signals that are generated by conspecifics. Here, 

we aim to investigate the degree of sender-receiver matching that occurs in the 

mechanosensory lateral line system as it specifically relates to communication signals 

generated by conspecifics in courtship scenarios.  

Salamanders that court in aquatic environments are useful model organisms for 

understanding the sensory tuning properties of the mechanosensory lateral line system. 

Many male salamanders perform a courtship behavior that involves an oscillating 

motion of the pelvic region and/or tail, such as the “tail fan” behavior, which generates 

mechanical stimuli (Green, 1989). We leveraged the courtship ritual of the fully aquatic 

axolotl, Ambystoma mexicanum, to examine how both senders and receivers shape the 

physical properties of mechanosensory signals. Male axolotls reliably perform a 

behavior known as the “hula”, a swaying motion of the hips paired with tail undulations, 

throughout their courtship ritual (Arnold, 1976; Park et al., 2004). The hula behavior 

generates hydrodynamic stimuli in the form of water disturbances, which likely act as a 

communication signal during mating encounters by stimulating the mechanosensory 

lateral line system of female axolotls.  

Importantly, axolotls maintain the functionality of their lateral line systems 

throughout their life cycles, unlike amphibians that undergo metamorphosis and typically 

lose their lateral line functions upon transitioning to a terrestrial environment 

(Wahnschaffe et al., 1987). Hydrodynamic stimuli likely play an important role for female 

97 

axolotls during courtship, as they may indicate the male’s location as well as his quality 

as a potential mate. Additionally, these stimuli may aid the female in coordinating sperm 

transfer with the male. Thus, axolotl courtship provides a unique opportunity to study 

both sender and receiver dynamics of mechanical signals.  

Hydrodynamic signals generated by conspecifics can be highly complex; we 

previously captured multiple aspects of the hula by measuring the sweep angle (i.e., 

degree of side-to-side motion), oscillation speed, and the elevation angle of tail 

movements. Using these data, we built a programmable device called the “Robotail”, 

which mimics the motions of the male hula (see Chapter 2 for methods). Importantly, 

the Robotail allowed us to evaluate how factors influencing components of vibrational 

communication (including sweep angle, speed, elevation) signals are received by 

mechanosensory lateral line systems. We tested 3 sweep angles (10°, 30°, 90°) paired 

with 3 speeds (0.5 Hz, 1.0 Hz, 1.5 Hz) for a total of 9 motion combinations; because 

hula elevation angle did not impact female behavior significantly (see Chapter 2), we 

excluded this parameter from our experiment. Using multi-unit recordings, we measured 

the firing rates of the anterodorsal lateral line nerve (ADLLn) in response to the 9 

different hula motion combinations.  

We chose to study the ADLLn specifically because it innervates dozens of 

neuromasts (i.e., mechanically sensitive hair cells) on the dorsal surface of the axolotls’ 

head (Northcutt, 1992); given that female axolotls position their heads directly 

underneath the male’s tail during spawning (Salthe, 1967), these neuromasts are likely 

to be highly stimulated during courtship. We created a semi-natural setting for our 

recordings by 1) positioning the recording preparation behind the raised Robotail, 

98 

mimicking the orientation of the female relative to the male during courtship, and 2) 

recording inside an aquarium filled with several inches of water, well above the top of 

the recording preparation; typical recording dishes hold less than a centimeter of fluid. 

Although we were primarily focused on studying female responses to hydrodynamic 

stimuli, we also tested males because their responses may differ from females, 

providing additional insight into sender-receiver matching and potentially into male-male 

competition in axolotls (Park et al., 2008).  

We found that the ADLLn of female axolotls exhibited an excitatory response to 

all sweep angles and all speeds; however, it responded most strongly to moderate 

sweep angles (30°) and moderate speeds (1.0 Hz). In contrast, the ADLLn of males 

exhibited inhibitory responses to narrow sweep angles (10°) and slow speeds (0.5 Hz), 

but excitatory responses to all other hula parameters, suggesting that the tuning of the 

lateral line systems in the two sexes may differ. Furthermore, we found that males most 

often hula with moderate combinations of motions (i.e., 30° paired with 1.0 Hz) during 

courtship (see Chapter 2), indicating that a high level of alignment exists between the 

hydrodynamic cues generated by males and the tuning properties of the female lateral 

line. Our experiments thus suggest that sender-receiver matching occurs within the 

mechanosensory lateral line system of axolotls. 

MATERIALS AND METHODS 

Subjects 

Adult axolotls (Ambystoma mexicanum) were obtained from the Ambystoma 

Genetic Stock Center at the University of Kentucky. Animals were housed in ~114 L 

aquaria in 40-100% Holtfreter’s (HF) solution (Armstrong et al., 1989) supplemented 

99 

with Replenish™ solution at temperatures between 18 and 22°C; animals were 

separated by sex and each aquarium contained 1-3 animals. We programmed the lights 

in our facility to match the natural sunrise and sunset of Mexico City, Mexico (the native 

habitat of axolotls) with monthly updates to the axolotls’ photoperiod. Housing protocols 

and experiments were conducted with the approval of the Institutional Animal Care and 

Use Committee of Michigan State University (approval numbers: PROTO201800106, 

PROTO202100239). 

Electrophysiology 

We used the “Robotail” (Chapter 2) to assess responses of the anterodorsal 

lateral line nerve to a range of mechanosensory stimuli similar to those produced by 

male axolotls during the hula behavior. We tested 3 sweep angles (10°, 30°, 90°) paired 

with 3 speeds (0.5 Hz, 1.0 Hz, 1.5 Hz) for a total of 9 motion combinations and rotated 

the order of stimulus presentation across animals. We used a constant elevation angle 

of ~45° because in earlier work we found that the elevation angle of the Robotail did not 

affect female behavior (Chapter 2); additionally, at lower angles the silicone tail 

interfered with the preparation and electrodes. The testing protocol, which we ran in 

triplicate for each combination, included a preparation period of 15 sec followed by a 

stimulus period of 30 sec, with an intertrial interval, or recovery period, of 2 min. Our 

preparations generally remained viable for a period of 3-4 hr; we analyzed data from 

recordings with robust action potentials (APs; generally above 1 mV) and stopped 

recording when nerve activity dropped below 0.5 mV. Using these criteria, we obtained 

viable recordings from 14 females and 5 males in total.  

100 

Experiments were conducted in a glass aquarium (71 cm x 22 cm x 18 cm) that 

held approximately 20 L of HF solution (~15 cm depth) at the same concentration as the 

animals’ home aquaria; the aquarium was housed inside a Faraday cage to shield our 

recordings from electrical noise. The mounting plate for the Robotail’s silicone tail was 

attached to the aquarium floor with waterproof silicone and the electronics were housed 

outside the Faraday cage; the tail was controlled via fishing lines fed through openings 

in the cage wall (Fig. 3.1). Importantly, we fabricated all submerged components of the 

robot from non-metallic materials to ensure that the Robotail did not generate electrical 

stimuli that might stimulate the electrosensory receptors innervated by the ADLLn. 

We performed rapid decapitation and exposed the peripheral aspect of the right 

ADLLn on all subjects by removing a ~1cm2 piece of skin posterior to the eye. The 

preparation was secured with pins to a Petri dish lined with Sylgard™ compound (Dow 

Inc., Midland, MI), which we affixed to the aquarium floor with dental wax. The 

preparation was placed underneath the silicone tail such that the anterior edge of the 

axolotl’s snout was ~4 cm from the mounting plate and ~8 cm from the tail tip (Fig. 3.2) 

to mimic courtship postures and to maximize stimulation of the neuromasts innervated 

by the ADLLn. We fabricated a pair of waterproof electrodes with long shafts that 

allowed us to record electrical activity deep under water; an Ag/AgCl wire suction 

electrode with a borosilicate glass pipette tip was used for recording and an Ag/AgCl 

pellet electrode acted as the reference electrode. We performed multiunit recordings 

(Fig. 3.3) by lowering the recording electrode onto the surface of the nerve as 

posteriorly as possible and drawing gentle suction, moving the recording site anteriorly 

101 

 
when the signal deteriorated. The reference electrode was placed in the HF solution 

approximately 1 cm away from the tip of the suction electrode.   

Signals were amplified through a differential amplifier (DP-304, Warner 

Instruments LLC, Holliston, MA), high-pass filtered (60 Hz), and digitized using a 

Digidata® 1550A digitizer (Molecular Devices LLC, San Jose, CA). We used pCLAMP™ 

10 software (Molecular Devices LLC, San Jose, CA) with Clampex and Clampfit 

programs (version 10.7.0.3) for data recording and analysis, respectively. Automated 

spike data extraction was performed using MATLAB code (MathWorks Inc, Natick, MA, 

version R2023b), which was modified from a classroom handout created by Wagenaar 

& Wright (2008) for the Neural Systems and Behavior Course at the Marine Biological 

Laboratory in Woods Hole, MA.  

Statistical analyses 

We calculated summary statistics (mean, standard deviation, minimum, 

maximum) for female (Table 3.1) and male (Table 3.2) firing rates (FR; i.e., the number 

of action potentials per second) using the “psych” package (version 2.4.3; Revelle, 

2009) in R (R Core Team, 2020). We also calculated the “percent change” (PC) for 

each trial, which is the percent change in mean firing rate between the preparation and 

stimulation periods (Table 3.3). To calculate the PC, we used the following equation: 

((Stimulation FR – Preparation FR)/Preparation FR)*100. For example, a trial with a 

Preparation FR of 5.93 APs/sec and a Stimulation FR of 10.2 APs/sec would yield a PC 

value of ~72%. Furthermore, a positive PC value represents an excitatory response 

from the nerve, whereas a negative value represents an inhibitory response. We chose 

to standardize our data in this fashion to reduce variation due to individual differences in 

102 

 
baseline nerve activity, and because males typically had higher firing rates than 

females. PC values lent context to our FR summary statistics and acted as the 

dependent variable in our statistical models. Summary statistics are written as 

mean(±SD) throughout the rest of this chapter. 

Because females and males were used an unequal number of times, we also 

evaluated percent change in the nerve FR with general linear mixed-effects models 

(Table 3.4) using the R packages ”lme4” (version 1.1.35.3; Bates et al., 2015) and 

”lmerTest” (version 3.1.3; Kuznetsova et al., 2017). Sweep angle and speed were 

treated as fixed effects. In models testing for differences between the sexes, animal ID 

was included as a random effect (Table 3.5). Due to limited power and unbalanced 

sample sizes between males and females (n = 5,14 respectively) we assessed 

responses to sweep angles and speeds within each sex and used either female ID or 

male ID (as appropriate) as a random effect. We set the restricted maximum likelihood 

(REML) to false because we compared models with different fixed effects (Bolker, 

2015).  

RESULTS 

Responses of the female ADLLn to individual parameters 

Responses to sweep angles 

The ADLLn of female axolotls exhibited excitatory responses (i.e., a positive 

percent change in FR) to all 3 sweep angles that we tested, but displayed the largest 

PC when stimulated with a moderate sweep angle of 30°. Our summary statistics 

indicate that a stimulus of 10° caused the firing rate (FR) of the ADLLn to increase from 

103 

4.33(±4.57) APs/sec during the preparation period (i.e., before the Robotail was turned 

on) to 5.35(±5.88) APs/sec during the stimulation period (i.e., while the Robotail was in 

operation). A 30° stimulus caused the FR to increase from 6.57(±9.39) to 12.23(±19.52) 

APs/sec, and a 90° stimulus resulted in an increase in FR from 6.46(±5.64) to 

9.98(±7.67) APs/sec (Table 3.1). Additionally, the PC (i.e., change in FR from 

preparation to stimulus period) for female ADLLns exposed to a sweep angle of 10° was 

75.42(±272.48)%; a stimulus of 30° resulted in a PC of 203.75(±461.37)%, whereas a 

stimulus of 90° resulted in a PC of 111.05 (±223.11)% (Table 3.3; Fig. 3.4). The results 

of our general linear mixed-effects models corroborated these findings; the FR exhibited 

a significantly greater increase (i.e., we observed a higher PC) when the Robotail was 

moving at a sweep angle of 30° compared to 10°, but showed no difference when we 

presented our preparation with extreme (90°) angles (10° vs 30°, t = 2.001, p = 0.049; 

10° vs 90°, t = 0.686, p = 0.494; Table 3.4; Fig. 3.4).  

Responses to speed 

The female ADLLn exhibited excitatory responses to all 3 speeds that we tested 

and displayed the greatest increase in FR when stimulated by a moderate speed of 1.0 

Hz. A stimulus of 0.5 Hz caused the FR to increase from 5.82(±6.59) APs/sec during the 

preparation period to 8.01(±10.61) APs/sec during the stimulation period; 1.0 Hz 

resulted in an increase in FR from 6.01(±6.59) to 9.80(±12.02) APs/sec, whereas 1.5 Hz 

resulted in an increase of 5.59(±7.26) to 9.91(±14.56) APs/sec (Table 3.1). Additionally, 

the PC value for female ADLLn stimulated by a speed of 0.5 Hz was 94.05(±317.70)%; 

a stimulus of 1.0 Hz resulted in a PC of 196.24(±423.06)% , whereas a stimulus of 1.5 

Hz resulted in a mean PC of 90.80(±196.39)% (Table 3.3; Fig. 3.5). Although female 

104 

nerve responses to the 3 speeds did not differ significantly, we observed a slightly 

elevated PC when presented with a stimulus speed of 1.0 Hz compared to 0.5 Hz (t = 

1.653, p = 0.100; Table 3.4; Fig. 3.5).  

Responses of the male ADLLn to single parameters 

Responses to sweep angle 

ADLLn activity in males was inhibited when stimulated with a sweep angle of 10°, 

but we observed excitatory responses when we tested sweep angles of 30° and 90°. 

Specifically, a stimulus of 10° resulted in a decrease in FR from 12.35(±14.50) APs/sec 

during the preparation period to a FR of 10.45(±12.97) APs/sec during the stimulation 

period. In contrast, a 30° stimulus caused the FR to increase from 8.50 (±11.77) to 

9.54(±13.46) APs/sec, and a 90° stimulus resulted in an increase from 33.90(±18.68) to 

40.57(±20.57) APs/sec (Table 3.2). Furthermore, the PC value for male ADLLn exposed 

to a sweep angle of 10° was -7.07(±90.14)%, whereas 30° resulted in a PC of 

33.00(±91.78)% and 90° resulted in a PC of 5.71(±63.05)% (Table 3.3; Fig. 3.6). We 

found no significant differences in male ADLLn PC responses to the 3 sweep angles 

that we tested, but the nerve did display a slightly elevated response to a sweep angle 

of 30° (t = 1.642, p = 0.106; Table 3.4; Fig. 3.6).  

Responses to speed 

In males, the ADLLn exhibited an inhibitory response to a slow speed of 0.5 Hz, 

but excitatory responses to moderate and fast speeds. A speed of 0.5 Hz caused a 

decrease in nerve activity from a FR of 14.16(±18.48) APs/sec during the preparation 

period to a FR of 13.92(±16.68) APs/sec during the stimulation period; 1.0 Hz resulted 

105 

 
in an increase in FR from 15.89(±17.47) to 16.38(±20.19) APs/sec, whereas 1.5 Hz 

resulted in an increase in FR from 14.12(±15.56) to 17.12(±20.20) APs/sec (Table 3.2). 

Accordingly, male ADLLn stimulated by 0.5 Hz exhibited a PC value of 26.41(±95.16)%, 

a stimulus 1.0 Hz resulted in a PC of 6.48(±97.46)%, whereas a stimulus of 1.5 Hz 

resulted in a PC of 4.69(±69.35)% (Table 3.3; Fig. 3.7). We did not find any significant 

differences in male PC responses to the 3 speeds that we tested (Table 3.4; Fig. 3.7). 

Female vs male responses to single parameters 

Overall, PC values were lower in the ADLLn of male axolotls compared with that 

of females (t = -2.095, p = 0.056). Additionally, when presented with a moderate sweep 

angle or faster speeds, the ADLLn of males exhibited a marginally lower PC than 

females (30°, t = -1.81, p = 0.108; 1.5 Hz, t = -1.914, p = 0.071; Table 3.5).  

Female responses to combinations of sweep angles and speeds 

The ADLLn of female axolotls exhibited a significantly higher PC value when 

stimulated by a combination of 30°/1.0 Hz, which represented a moderate sweep angle 

and speed, compared to a combination of 10°/0.5 Hz (t = 2.962, p = 0.004). Additionally, 

the female ADLLn displayed a marginally higher PC to a combination of 10°/1.0 Hz (t = 

1.467, p = 0.144; Table 3.6). 

Male responses to combinations of sweep angles and speeds 

We found no significant differences in response of the ADLLn of males to the 9 

different combinations of sweep angles and speeds that we tested (Table 3.7). 

However, our ability to detect differences in nerve responses among our male subjects 

is likely limited by our sample size (N = 5).  Additionally, the ADLLn response to the 

106 

Robotail stimulus was inhibitory when presented with narrow sweep angle parameters 

or slow speeds, but was excitatory when presented with other sweep angles and 

speeds; therefore, combination effects may be obscured in the overall PC. 

Comparison of the male hula to female behavioral and physiological responses 

We found a high level of alignment between male hula behaviors and responses 

of the female ADLLn to the Robotail, During courtship, male axolotls most often hula’d 

with a combination of 30°/1.0 Hz, which represents both a moderate sweep angle and 

speed, and 30°/1.5 Hz, which represents a moderate sweep angle combined with a 

relatively high speed. We also found that female axolotls exhibited heightened ADLLn 

responses to a hula combination of moderate speed and sweep angle. When we 

stimulated the female ADLLn using the Robotail, we found that the nerve exhibited a 

significantly higher PC to a combination of 30°/1.0 Hz (compared to 10°/0.5 Hz). We 

also observed a marginally higher female ADLLn response to 10°/1.0 Hz, which males 

only performed “sometimes” (Fig. 3.8).  

In contrast, we observed a  moderate level of alignment between male hula 

patterns and female behavioral responses to the Robotail. Female axolotls transitioned 

between locomotive states (i.e., walking, swimming, and pausing) significantly more 

times per minute when the Robotail was oscillating with a pattern of 30°/1.5 Hz or 

90°/1.5 Hz compared to 10°/0.5 Hz (see Chapter 2). Additionally, females transitioned 

between locomotive states marginally more often when the Robotail operated with a 

pattern of 30°/1.0 Hz (t = 1.785, p = 0.076) or 90°/0.5 Hz (t = 1.693, p = 0.093; Fig. 3.8) 

compared to 10°/0.5 Hz. Thus, we found a moderate level of alignment between male 

and female behavior for the combination of 30°/1.0 Hz; male axolotls performed this 

107 

pattern often during courtship, but females only exhibited a slightly higher rate of 

locomotion transitions, which may indicate a reduced searching effort by the female to 

locate the source of mechanical stimuli. Although we observed a greater level of 

alignment between the male hula and female behavioral response for the combination 

of 30°/1.5 Hz, female axolotls also transitioned between locomotive states more rapidly 

when the Robotail was moving at wider angles, which male axolotls only performed 

“sometimes” during courtship.  

DISCUSSION 

Female ADLLn responses to single parameters 

The ADLLn of female axolotls exhibited excitatory responses to all sweep angles 

and speeds that we tested but displayed the largest increase in FR when exposed to 

moderate sweep angles (30°) and moderate speeds (1.0 Hz; Table 3.1).  Behaviorally, 

male axolotls generally hula using intermediate sweep angles during courtship (Chapter 

2); thus, we found a high level of matching between the sweep angle of the male’s hula 

and the physiological response of the female AD LL nerve. Interestingly, in our 

behavioral experiments (Chapter 2), we also observed that females responded to wide 

sweep angles by initiating bouts of locomotion more frequently, compared to narrower 

angles; we speculate that this result may represent an increased search effort by the 

female to localize the male. One potential explanation is that a wider sweep angle may 

deflect the neuromast hair cells at an intermittent rate, whereas a narrower sweep angle 

likely results in more consistent stimulation, which may partially explain our behavioral 

results.  

108 

Based on the work of Mogdans & Bleckmann (1999) in goldfish (Carassius 

auratus) lateral line nerves, we expected the firing rate of the ADLLn to increase as the 

stimulus amplitude increased; instead, we observed a more subdued response when 

the sweep angle was increased to 90°. However, Mogdans & Bleckmann (1999) used a 

small vibrating sphere that moved in a single axis to generate their stimuli, whereas we 

used a flexible robot with greater degrees of freedom in its motion patterns, and so 

there were perhaps other hydrodynamic factors at play in our experiment that may 

explain the discrepancy in our results. Additionally, the mechanical stimuli that arise 

from the Robotail are naturalistic, so we may have observed an aspect of sender-

receiver matching occurring, given that the female ADLLn generally responds more 

strongly to motions that males display often (i.e., 30° sweep or 1.0 Hz) compared to 

wider sweep angles.  

Similarly, the female ADLLn exhibited a larger percent change in FR when we 

stimulated the nerve with a moderate speed of 1.0 Hz, and a smaller PC when tested 

the fastest speed of 1.5 Hz (Table 3.3). Thus, we also observed a high level of 

alignment between male hula behaviors and female neurophysiological responses 

within the speed parameter, given that male axolotls generally hula’d with moderate 

speeds of 1.0 Hz (see Chapter 2). In contrast, our previous behavioral research 

demonstrated that female axolotls spent more time near the Robotail when it was 

moving at 1.5 Hz (Chapter 2). It is unclear why female axolotls behaviorally preferred 

faster hula speeds but the female ADLLn responded with a faster FR when presented 

with a moderate speed stimulus, although we speculate that faster hula speeds may be 

indicative of a higher quality mate or may serve to draw the female closer to the male 

109 

during courtship. Meadow katydid (Conocephalus nigropleurum) females also prefer 

male substrate-borne vibrational signals with shorter inter-pulse intervals (i.e., faster 

frequencies), which are indicative of a larger, and likely higher-quality, potential mate. 

Additionally, our results from Chapter 2 suggest that male axolotls are somewhat more 

likely to hula at 1.5 Hz when they are at a distance from a female, thus, it is possible 

that faster hula speeds may serve to draw the female towards the male during 

courtship. 

Male ADLLn response to single parameters 

The ADLLn of male axolotls exhibited inhibitory responses to narrow sweep 

angles (10°) and slow speeds (0.5 Hz), but excitatory responses to all other sweep 

angles and speeds. Although male ADLLn responses did not differ significantly with 

sweep angle, we found that the nerve displayed a marginally elevated FR when 

stimulated by a sweep angle of 30° (Table 3.4). Male axolotls, when hula-ing at a 

distance from a female, are more likely to perform the hula with a narrow sweep angle 

of 10° (see Chapter 2). Romer (1993) proposed that male insects in the order 

Orthoptera (i.e., grasshoppers, locusts, crickets) experience an inhibitory effect in the 

auditory pathway when other calling males are nearby. Thus, an inhibitory response of 

the male ADLLn to 10° may represent the presence of another male that is attempting 

to advertise to a nearby female. In contrast, male axolotls hula with a slightly wider 

sweep angle when they are closer to females; thus, a heightened FR within the male 

ADLLn may indicate that the male should engage in competition with the rival male 

rather than avoid him. 

110 

 
Female vs male responses to single parameters 

Overall, the ADLLn of male axolotls exhibited significantly smaller increases in 

FR (i.e., between preparation and stimulus periods) than that of female axolotls. In the 

blue-spotted fantail stingray (Taeniura lymma), females possess a greater number of 

electrosensory axons within the anterior lateral line nerve than males do (Kempster et 

al., 2013). Given that the lateral line system in axolotls is composed of both 

mechanosensory and electrosensory modalities (Northcutt, 1992), female axolotls may 

possess higher numbers of mechanically sensitive axons within the ADLLn, which could 

contribute to the higher PCs that we observed in females. Alternatively, females 

exhibited lower FRs overall than males did, so this result may be explained by the fact 

that there is a greater potential for an increase in female FR compared to males.  

Although we did not detect any significant differences between female and male 

nerve responses when we examined specific parameters, we found that the female 

ADLLn exhibited a marginally higher increase in FR than the male ADLLn when 

stimulated by a moderate angle of 30° or a fast speed of 1.5 Hz. A tail oscillation speed 

of 1.5 Hz may indicate the presence of a particularly robust male, whereas a sweep 

angle of 30° may represent a male that is preferable for females. The male ADLLn may 

experience an inhibitory effect  when stimulated by these parameters. Within 

bushcrickets, male Hemisaga denticulata produce a discontinuous song, whereas 

males of the sympatric species Mygalopsis marki sing continuously. The auditory 

pathway of H. denticulate exhibits an inhibitory response when stimulated by the song 

of M. marki; additionally, researchers have observed that discontinuously singing 

bushcrickets reduce their singing activity in the presence of heterospecific crickets that 

111 

are singing continuously (Romer, 1993). Although we did not examine behavioral 

interactions between different salamander species, it is possible that the hula behavior 

in male axolotls may inhibit activity in the ADLLn of rival males and possibly hula-ing as 

well; this effect may allow males to conserve time and energy during courtship. 

Female responses to combinations of sweep angles and speeds 

The results of our experiments revealed a high level of alignment between male 

hula motion patterns and female physiological responses to those motions. The firing 

rate within the female ADLLn was significantly elevated when stimulated with a hula 

combination of 1.0 Hz and a 30° sweep angle (Table 3.6). This finding is corroborated 

by our single-parameter general linear mixed-effects statistical models; the female 

ADLLn exhibited a significantly greater increase in FR when stimulated by 30°, and a 

marginally higher response when stimulated by 1.0 Hz (Table 3.4). 30°/1.0 Hz is also 

one of the hula combinations that males most often display during actual courtship (Fig. 

3.8). The female ADLLn also exhibited a marginal increase in FR when stimulated by 

10°/1.0 Hz, although this particular hula combination was only “sometimes” displayed by 

the males in our behavioral experiments. Our experimental results suggest that 

hydrodynamic stimuli play an important role in axolotl courtship, and that females are 

attuned to detect mechanical communication signals that arise from the male hula 

behavior.  

Male responses to combinations of sweep angles and speeds 

We found no marginal or significant differences the responses of the male ADLLn 

to the 9 different combinations of sweep angles and speeds that we tested. The male 

112 

 
 
ADLLn may be more sensitive to changes in single hula parameters, given that the 

nerve showed a slightly elevated response to a 30° sweep angle (Table 3.4). These 

results  could be partially due to our small sample size (5 males). Alternatively, given 

that the male ADLLn exhibited both excitatory and inhibitory responses, it is possible 

that any differences we observed with single parameters may have been “canceled out” 

when we examined responses to combinations of parameters. 

Comparison of the male hula to female behavioral and physiological responses 

We observed a high level of alignment between male hula behaviors and female 

ADLLn responses to simulated hula motions generated by the Robotail. In contrast, we 

found moderate alignment between the male’s hula and female behavioral responses to 

the Robotail. During courtship, male axolotls performed a combination of 30°/1.0 Hz 

“often”; the female ADLLn exhibited a significantly elevated PC to this combination, and 

females behaviorally responded by exhibiting a marginally elevated number of 

locomotion transitions per minute. We speculate that more frequent locomotion 

transitions (e.g., switching between walking, swimming, and pausing) may represent 

more frequent search attempts by the female to localize the stimulus source, but we did 

not test this hypothesis directly. Males also performed a combination of 30°/1.5 Hz 

“often”; we found that with this combination of stimuli females exhibited a significantly 

higher number of locomotion transitions per minute, but we found no significant 

physiological responses to this particular combination of sweep angle and speed.  

Female axolotls also exhibited more frequent locomotion transitions in response 

to combinations that males only performed “sometimes” during courtship. We speculate 

that female axolotls exhibit broad behavioral responses but more narrowly tuned ADLLn 

113 

responses to our hula combinations because the male hula is likely associated with 

other sensory stimuli, such as sex pheromones, which may cause females to respond 

with more complex behaviors; in contrast, the mechanosensory lateral line only 

functions to detect hydrodynamic stimuli and may be more sensitive to hula 

combinations that males perform often. Interestingly, females always transitioned 

between locomotive states more frequently (either significantly or marginally) when the 

Robotail was moving at the widest sweep angle of 90°.  A wider sweep angle is likely to 

displace more water during the hula behavior than a narrower sweep angle would, 

which may deflect the hair cells of the lateral line neuromasts more drastically; this 

phenomenon may contribute to the stronger behavioral responses exhibited by females. 

Besides the marginally elevated female nerve response to a motion pairing of 10°/1.0 

Hz, females did not show heightened behavioral or physiological responses to any of 

the other combinations that featured a narrow sweep angle of 10°.  

Male axolotls may avoid moving their tails at the extremes of their sweep angle 

and speed ranges because these motions could be too energetically costly to sustain 

over the course of an entire courtship ritual. However, more vigorous tail motions from 

the Robotail were effective at eliciting an increased behavioral response from female 

axolotls through more frequent locomotion transitions. In contrast, the female ADLLn 

exhibited heightened, excitatory responses to more moderate hula combinations (i.e., 

30°/1.0 Hz or 10°/1.0 Hz). More extreme hula motions performed in short bursts may 

therefore serve to draw females toward the courting male, whereas more moderate 

combinations of motions may confirm the presence of a conspecific male. Our results 

thus suggest that sender-receiver matching occurs within the mechanosensory aspect 

114 

of the lateral line system. To our knowledge, our suite of experiments is the first to 

demonstrate this neuroethological phenomenon within the mechanosensory lateral line.  

CONCLUSION 

Our electrophysiology experiment demonstrated that the ADLLn of female and 

male axolotls exhibited different physiological responses to various sweep angles and 

speeds associated with the hula behavior. We observed excitatory responses within the 

female ADLLn to all sweep angles and speeds that we tested, and we found that the 

female ADLLn responded most strongly to moderate sweep angles and moderate 

speeds. In contrast, the male ADLLn exhibited inhibitory responses to narrow sweep 

angles, but excitatory responses to all other parameters that we tested. Additionally, we 

found a high level of alignment between actual male hula behaviors and female ADLLn 

responses, and a moderate level of alignment between the male hula and female 

behavioral responses. Importantly, these experiments indicate that sender-receiver 

matching occurs within the mechanosensory lateral line system, such that the female 

ADLLn is highly tuned to the hydrodynamic stimuli generated by males during courtship. 

115 

 
 
TABLES 

Stimulus 

Preparation FR 

Stimulation FR 

Recovery FR 

10° 
30° 
90° 

0.5 Hz 
1.0 Hz 
1.5 Hz 

4.33(4.57) 0.13-18.67 
6.57(9.39) 0.13-29.53 
6.46(5.64) 0.13-20.93 

5.82(6.59) 0.13-23.33 
6.01(6.59) 0.13-25.33 
5.59(7.26) 0.13-29.53 

5.53(5.88) 0-23.25 
2.94(4.08) 0-20.32 
12.23(19.52) 0-63.63  4.64(7.86) 0-23.33 
3.97(5.12) 0-21.11 
9.98(7.67) 0-27.82 

8.01(10.61) 0-44.85 
9.80(12.02) 0-55.04 
9.91(14.56) 0-63.63 

4.61(6.76) 0.03-21.48 
3.92(5.60) 0.01-23.25 
2.96(4.92) 0-23.33 

Table 3.1: Summary statistics for female ADLLn firing rates. The ADLLn of female 
axolotls were stimulated using 3 sweep angles and 3 speeds; we then quantified the 
firing rates (FRs) during the preparation, stimulus, and recovery periods of each trial. FR 
is defined here as the number of action potentials per second, and values are displayed 
as “mean(SD) min-max”. 

Stimulus 

Preparation FR 

Stimulation FR 

Recovery FR 

10° 
30° 
90° 

0.5 Hz 
1.0 Hz 

1.5 Hz 

12.35(14.50) 0-35.87 
8.50(11.77) 0.13-34.73 
33.90(18.68) 0.13-51.33 

10.45(12.97) 0-33.14  9.02(11.32) 0-24.23 
9.54(13.46) 0-40.25 
40.57(20.57) 0-57.40  31.64(15.71) 0-40.42 

5.78(8.87) 0-27.17 

14.16(18.48) 0.07-51.33 
15.89(17.47) 0-49.67 

13.29(16.68 )0-44.78  10.83(14.53) 0.03-39.74 
16.38(20.19) 0-57.40  11.55(14.93) 0-39.92 

14.12(15.56) 0.13-45.67 

17.25(20.20) 0-54.41  13.16(15.38) 0-40.42 

Table 3.2: Summary statistics for male ADLLn firing rates. The ADLLn of male 
axolotls were stimulated using 3 sweep angles and 3 speeds; we then quantified the 
FRs during the preparation, stimulus, and recovery periods of each trial. FR is defined 
here as the number of action potentials per second, and values are displayed as 
“mean(SD) min-max”. 

116 

 
 
 
 
 
 
 
 
 
 
Stimulus 

♀ PC (%) 

10° 
30° 
90° 
0.5 Hz 
1.0 Hz 
1.5 Hz 

75.24(272.48) -100.00-1787.22 
203.75(461.37) -100.00-1585.80 
111.05(223.11) -100.00-1454.31 
94.05(317.70) -100.00-1832.33 
196.24(423.06) -100.00-1858.80 
90.80(196.39) -100.00-1038.14 

♂ PC (%) 

-7.07(90.14) -100.00-250.00 
33.00(91.78) -100.00-303.12 
5.71(63.05) -100.00-119.42 
26.41(95.16) -100.0-205.00 
6.48(97.46) -100.00-303.12 
4.69(69.35) -100.00-127.27 

Table 3.3: Summary statistics for female and male ADLLn PCs. The ADLLn of 
female and male axolotls were stimulated using 3 sweep angles and 3 speeds, and PC 
of the nerve to each stimulus was then quantified. PC is defined here as the percent 
change in firing rate between the preparation and stimulus phase of each trial; for 
example, a PC of 75% means that the FR of the ADLLn nerve increased by 75% from 
the preparation to the stimulus phase. Values are displayed as “mean(SD) min-max”. 

Dependent 
Variable 

Random 
Effect 

Model  Fixed Effects 

t statistic 

p - value 

Percent Change 

Female ID 

Percent Change 

Male ID 

Sweep 

Speed 

Sweep 

Speed 

30° sweep 
90° sweep 
1.0 Hz 

1.5 Hz 

30° sweep 

90° sweep 
1.0 Hz 
1.5 Hz 

2.001 
0.686 
1.653 

-0.034 

1.642 

0.413 
-0.725 
-0.79 

0.049 * 
0.494 
0.100 

0.973 

0.106 

0.681 
0.471 
0.433 

Table 3.4: General linear mixed-effects models for female and male PCs. We 
evaluated percent changes (PC) in female and male ADLLn firing rate in response to 
hula motion parameters (sweep angle and speed). PC reflects the percent change in 
firing rate from the preparation period to the stimulation period. Females exhibited a 
significantly higher PC to a sweep angle of 30° compared to 10°. 

117 

 
 
 
 
 
 
 
 
 
Dependent 
Variable 

Random 
Effect 

Model 

Data Set 

t statistic 

p - value 

Percent Change 

Animal ID 

Sex 

All 
10° 
30° 
90° 
0.5 Hz 
1.0 Hz 
1.5 Hz 

-2.095 
-1.207 
-1.81 
-1.55 
-0.663 
-1.387 
-1.914 

0.056* 
0.256 
0.108 
0.149 
0.517 
0.187 
0.071 

Table 3.5: General linear mixed-effects models for female vs. male PCs. We 
evaluated percent changes (PC) in ADLLn firing rate between males and females in 
response to combinations of stimuli across all 3 sweep angles and all 3 speeds. Overall, 
male axolotls exhibited lower PCs compared to females. Additionally, males displayed 
marginally lower PCs to a moderate sweep angle of 30° and fast speed of 1.5 Hz, 
compared to females. 

118 

 
 
Dependent 
Variable 

Random 
Effect 

Model 

Fixed Effects 

t statistic 

p - value 

Sweep(°)  Speed(Hz) 

Percent 
Change 

Female ID  Combo 

30 
90 
10 
30 
90 
10 
30 
90 

0.5 
0.5 
1.0 
1.0 
1.0 
1.5 
1.5 
1.5 

1.023 
1.171 
1.467 
2.962 
0.788 
0.111 
0.171 
0.904 

0.308 
0.243 
0.144 
0.004** 
0.432 
0.911 
0.243 
0.368 

Table 3.6: General linear mixed-effects models for female responses to 
combinations of sweep angles and speeds. We evaluated percent changes (PC) in 
female ADLLn firing rate in response to 9 combinations of sweep angles and speeds. 
Females exhibited a significantly higher PC to a combination of 30° and 1.0 Hz, and a 
marginally higher PC to a combination of 10° and 1.0 Hz.  

Dependent 
Variable 

Random 
Effect 

Model 

Fixed Effects 

t statistic 

p - value 

Sweep(°)  Speed(Hz) 

Percent 
Change 

Male ID 

Combo 

30 
90 
10 
30 
90 
10 
30 
90 

0.5 
0.5 
1.0 
1.0 
1.0 
1.5 
1.5 
1.5 

0.088 
-0.66 
-1.23 
-0.282 
0.061 
-1.309 
0.517 
-0.714 

0.930 
0.512 
0.223 
0.779 
0.952 
0.196 
0.607 
0.478 

Table 3.7: General linear mixed-effects models for male responses to 
combinations of sweep angles and speeds. We evaluated percent changes (PC) in 
male ADLLn firing rate when presented with 9 combinations of sweep angles and 
speeds. We found no significant differences in male nerve responses.  

119 

  
  
  
  
  
 
 
  
  
  
  
  
 
 
 
 
 
FIGURES 

A 

B 

Figure 3.1: Electrophysiology setup. Photograph showing the position of the Robotail 
electronic components (A) relative to the aquarium (B), which was inside a Faraday 
cage. Nylon fishing lines were fed through openings in the cage wall to connect the 
robot gimbal to the silicone tail within the aquarium.  

120 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.2: Diagram of preparation with Robotail. The preparation was placed on a 
Petri dish underneath the silicone tail to to mimic the position of the female relative to 
the male during mating and to maximize stimulation of neuromasts on the surface of the 
head. The recording electrode was lowered onto the right anterodorsal lateral line nerve 
and angled so that the tail did not interfere with the electrode while the robot was in 
operation.  

121 

 
 
 
 
 
 
 
 
 
 
 
 
Figure 3.3: Multi-unit recording from the ADLLn stimulated with a 90° sweep angle 
and a 1.5 Hz speed. Action potentials were recorded from multiple fibers in the ADLLn 
using a suction electrode.  

122 

 
 
 
Figure 3.4: Firing rate in the female ADLLn in response to stimulation with 
different sweep angles. A significantly higher excitatory response from the female 
ADLLn was observed when presented with a moderate sweep angle of 30° compared 
with a narrow angle of 10° (mixed model analysis, t = 2.001, p = 0.049). Each dot 
represents the percent change in FR (between preparation and stimulus periods) for a 
given trial and bars represent the standard error of each group. 

123 

 
 
 
 
 
Figure 3.5: Firing rate in the female ADLLn in response to stimulation with 
different speeds. Our mixed model analysis revealed no significant differences in 
female ADLLn responses when stimulated with different hula speeds. Each dot 
represents the percent change in FR (between preparation and stimulus periods) for a 
given trial and bars represent the standard error of each group. 

124 

 
 
 
Figure 3.6: Firing rate in the male ADLLn in response to stimulation with different 
sweep angles. Our mixed model analysis revealed no significant differences in male 
ADLLn responses to different sweep angles. Each dot represents the percent change in 
FR (between preparation and stimulus periods) for a given trial and bars represent the 
standard error of each group. 

125 

 
 
 
 
Figure 3.7: Firing rate in the male ADLLn in response to stimulation with different 
speeds. Our mixed model analysis revealed no significant differences in male ADLLn 
responses to different hula speeds. Each dot represents the percent change in FR 
(between preparation and stimulus periods) for a given trial and bars represent the 
standard error of each group. 

126 

 
 
 
 
0.5 
Hz 

1.0 
Hz 

d
e
e
p
S

Sweep Angle 

10° 
♂ Hula: Sometimes 

30° 
♂ Hula: Sometimes 

♀ 
Behavior 

♀ 
Physiology  

N/A 

N/A 

♀ 
Behavior 
t = 0.516 
p = 0.607 

♀ 
Physiology  
t = 1.023 
p = 0.308 

♂ Hula: Sometimes 

♀ 
Behavior 
t = 0.694 
p = 0.489 

♀ 
Physiology  
t = 1.467 
p = 0.144 . 

♂ Hula: Sometimes 

♀ 
Behavior 
t = -0.278 

♀ 
Physiology  
t = 0.111 

1.5 
Hz 

p = 0.782 

p = 0.911 

♂ Hula: Often 
♀ 
Physiology  
t = 2.962 

♀ 
Behavior 
t = 1.785 
p = 0.076.  p = 0.004 ** 
♂ Hula: Often 
♀ 
Physiology  
t = 1.171 

♀ 
Behavior 
t = 2.372 
p = 0.019 
* 

p = 0.243 

90° 
♂ Hula: Sometimes 
♀  
Behavior 
t = 1.693 
p = 0.093 . 

♀ 
Physiology  
t = 1.171 
p = 0.243 

♂ Hula: Sometimes 

♀ Behavior 

t = 2.187 
p = 0.030 * 

♀ 
Physiology  
t = 0.788 
p = 0.432 

♂ Hula: Sometimes 

♀ Behavior 

t = 4.014 
p = 9.51e-
05 *** 

♀ 
Physiology  
t = 0.904 

p = 0.368 

Male Hula 

Female Behavior 

Sometimes 
Often 

Not significant 
  Marginally significant 

Significant 

  Female Physiology 
  Not significant 
  Marginally significant 
  Significant 

Figure 3.8: Comparison of male hula behaviors with female behavioral and 
neurophysiological responses. Alignment between male hula behaviors, female 
behavioral responses to the hula behavior, and responses of the female ADLLn to hula 
parameters. We found a high level of alignment between male behaviors and female 
physiological responses, especially within the 30°/1.0 Hz combination. In contrast, 
females exhibited a broader behavioral response to the hula combinations we tested, 
such that females transitioned between locomotive states more often when exposed to 
wider sweep angles and faster speeds. 

127 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
REFERENCES 

Allen, K. M., & Marsat, G. (2019). Neural processing of communication signals: The 

extent of sender-receiver matching varies across species of apteronotus. ENeuro, 
6(2).  

Armstrong, J. B., Duhon, S. T., & Malacinski, G. M. (1989). Raising the axolotl in 

capitivity. In J. B. Armstrong & G. M. Malacinski (Eds.), Developmental Biology of 
the Axolotl (pp. 220–227). New York: Oxford University Press. 

Arnold, S. J. (1976). Sexual behavior, sexual interference and sexual defense in the 

salamanders Ambystoma maculatum, Ambystoma tigrinum and Plethodon jordani. 
Zeitschrift Für Tierpsychologie, 42(3), 247–300. 

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects 

models using lme4. Journal of Statistical Software, 67(1), 1–48. 

Bolker, B. M. (2015). Linear and generalized linear mixed models. In G. A. Fox, S. 
Negrete-Yankelevich, & V. J. Sosa (Eds.), Ecological Statistics: Contemporary 
Theory and Application (1st ed., pp. 309–333). Oxford University Press. 

Butler, J. M., & Maruska, K. P. (2015). The mechanosensory lateral line is used to 

assess opponents and mediate aggressive behaviors during territorial interactions 
in an African cichlid fish. Journal of Experimental Biology. 

Capranica, R. R., & Moffat, A. J. M. (1983). Neurobehavioral Correlates of Sound 

Communication in Anurans. Advances in Vertebrate Neuroethology, 701–730.  

Denoël, M., & Doellen, J. (2010). Displaying in the dark: light-dependent alternative 
mating tactics in the Alpine newt. Behavioral Ecology and Sociobiology, 64(1), 
1171–1177. 

Gall, M. D., Brierley, L. E., & Lucas, J. R. (2012). The sender-receiver matching 

hypothesis: Support from the peripheral coding of acoustic features in songbirds. 
Journal of Experimental Biology, 215(21), 3742–3751.  

Green, A. J. (1989). The sexual behaviour of the great crested newt, Triturus cristatus 

(Amphibia: Salamandridae). Ethology, 83, 129–153. 

Henry, K. S., Gall, M. D., Velez, A., & Lucas, J. R. (2016). Avian Auditory Processing at 
Four Different Scales: Variation Among Species, Seasons, Sexes, and Individuals. 
In Physiological Mechanisms in Animal Communication (pp. 17–55).  

128 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Kempster, R. M., Garza-Gisholt, E., Egeberg, C. A., Hart, N. S., O’Shea, O. R., & Collin, 

S. P. (2013). Sexual dimorphism of the electrosensory system: A quantitative 
analysis of nerve axons in the dorsal anterior lateral line nerve of the blue-
spotted fantail stingray (Taeniura lymma). Brain, Behavior and Evolution, 81(4), 
226–235.  

Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: 

tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. 

Mhatre, N., Bhattacharya, M., Robert, D., & Balakrishnan, R. (2011). Matching sender 
and receiver: Poikilothermy and frequency tuning in a tree cricket. Journal of 
Experimental Biology, 214(15), 2569–2578.  

Mogdans, J., & Bleckmann, H. (1999). Peripheral lateral line responses to amplitude-

modulated sinusoidal wave stimuli. Journal of Comparative Physiology - A Sensory, 
Neural, and Behavioral Physiology, 185(2), 173–180.  

Noë, R. (2006). Cooperation experiments: Coordination through communication versus 

acting apart together. Animal Behaviour, 71(1), 1–18.  

Northcutt, R. G. (1992). Distribution and innervation of lateral line organs in the axolotl. 

The Journal of Comparative Neurology, 325, 95–123. 

Park, D., McGuire, J. M., Majchrzak, A. L., Ziobro, J. M., & Eisthen, H. L. (2004). 

Discrimination of conspecific sex and reproductive condition using chemical cues in 
axolotls (Ambystoma mexicanum). Journal of Comparative Physiology A, 190, 415–
427. 

Park, Daesik, Lee, J.-H., Ra, N.-Y., & Eom, J. (2008). Male salamanders Hynobius 
leechii respond to water vibrations via the mechanosensory lateral line system. 
Journal of Herpetology, 42(4), 615–625. 

Revelle, W. (2023). psych: procedures for psychological, psychometric, and personality 
research. Northwestern University, Evanston, Illinois. R package version 2.3.6. 

Romer, H. (1993). Environmental and biological constraints for the evolution of long- 

range signalling and hearing in acoustic insects. Philosophical Transactions - Royal 
Society of London, B, 340(1292), 179–185.  

Salthe, S. N. (1967). Courtship patterns and the phylogeny of the urodeles. Copeia, 1, 

100–117. 

129 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Satou, M., Takeuchi, H. A., Nishii, J., Tanabe, M., Kitamura, S., Okumoto, N., & Iwata, 
M. (1994). Behavioral and electrophysiological evidences that the lateral line is 
involved in the inter-sexual vibrational communication of the himé salmon 
(landlocked red salmon, Oncorhynchus nerka). Journal of Comparative Physiology 
A.  

Wagenaar, D. A., & Wright, M. (2008). An electrophysiologist’s introduction to Matlab 

[class handout]. Neural Systems and Behavior Course, Marine Biological 
Laboratory. 

Wahnschaffe, U., Bartsch, U., & Fritzsch, B. (1987). Metamorphic changes within the 

lateral-line system of Anura. Anatomy and Embryology, (175), 431–442. 

130