RED BLOOD CELL ABNORMALITIES IN WHITE-BLOODED 

ICEFISHES AND RELATED ANTARCTIC SPECIES 

 
 
 
 
 
 
 
 

by 

ZOE NUNEZ 

 
 
 
 
 
 
 
 

A THESIS 

 
 
 
 
 
 
 

Presented to the Department of Honors Biology 
and the Robert D. Clark Honors College  

in partial fulfillment of the requirements for the degree of  
Bachelor of Science 

 
November 2023 

 



 

2 
 

An Abstract of the Thesis of 

Zoe Nunez for the degree of Bachelor of Science 
in the Department of Honors Biology to be taken June 2024 

 
Title:  Red Blood Cell Abnormalities in White-Blooded Icefishes and Related Antarctic Species 
 
 

Approved:  John H Postlethwait Ph.D.  
    Primary Thesis Advisor 

 

The process of cell differentiation is crucial for organism viability, especially the 

formation of red blood cells (RBCs) that contain hemoglobin and provide essential oxygen 

transportation to the tissues. Antarctic white-blooded icefishes are, however, evolutionary 

oddities that lack mature RBCs and hemoglobin. An outstanding question is if the loss of 

hemoglobin genes in the icefish ancestor could have led to the arrest of RBC maturation in 

icefishes, or inversely if the arrest of RBC maturation could have triggered the loss of 

hemoglobin genes. Here, we aimed to test these two conflicting hypotheses. The ‘genes-first 

hypothesis’ predicts that the icefish ancestors would start to lose hemoglobin genes while having 

normal blood cells, but the ‘cells-first’ hypothesis predicts that icefish ancestors would have 

damaged RBCs but normal hemoglobin genes. To test these predictions, we analyzed the blood 

cell composition and RBC morphology in 13 red-blooded Antarctic fish species and in seven 

icefishes by microscopy, semi-automated digital image analyses, and phylogenetic comparative 

analyses. My investigation revealed that some species closely related to icefishes display 

abnormal RBCs that are larger and rounder than in other red-blooded species and have off-

centered nuclei. My results suggest that RBC maturation was likely already altered before the 

loss of hemoglobin genes in icefishes. Close analysis of cellular abnormalities arising in these 

Antarctic fish species may inform us on hereditary human diseases such as anemias arising from 

improper RBC formation. 



 

3 
 

 
Acknowledgements 

 

I would like to thank my mentor Thomas Desvignes for providing me with guidance and 

support along every step up my research journey. I would also like to thank Professor John H. 

Postlethwait for allowing me to join his lab and always providing feedback throughout my entire 

research process. Additionally, I would like to thank Dr. Elizabeth Raisanen for taking the time 

to be my honors college representative and always answering any questions I had about the 

thesis process. Over the past couple years I have gained an endless amount of guidance and 

experience I am incredibly thankful for. Although the research process has been challenging it 

has also been highly rewarding working in the Postlethwait lab and seeing my research come 

together.  

 

 

 

  



 

4 
 

Table of Contents 

Introduction 6 
Notothenioid Evolution: 6 
Relevancy and Goals: 12 

Methods 14 
Blood Smear Staining: 14 
Microscopy and Cellular Analyses 15 
Phylogenetic Comparative Analysis 18 

Results 20 
Evolution of the Antarctic Notothenioid Blood 20 

Evolution of notothenioid hematocrit 20 
Notothenioid blood cellular makeup 26 

Cellular abnormalities in the Antarctic notothenioid Blood 28 
Evolution of red-blood cell area 28 
Evolution of red-blood cell roundness 35 
Integration of red-blood cell area and roundness evolution 38 

Abnormally Located Nuclei in Antarctic Notothenioids 42 
Discussion 50 

Can discontinuing RBC production be energetically advantageous? 51 
If maladaptive, why did RBC production become discontinued? 53 

Conclusion 58 
Bibliography 61 
 

 

 

 

 

 

  



 

5 
 

List of Figures  

 
Figure 1. Phylogenetic tree of notothenioids and their traits 6 

Figure 2. Blood smear fixation 14 

Figure 3. Cellular classification 16 

Figure 4. Digital analysis of cells 17 

Figure 5. Comparison of nuclei position 18 

Figure 6. Evolution of Antarctic notothenioid hematocrit 22 

Figure 7. Proportions of main blood cell lineages in Antarctic notothenioid blood 27 

Figure 8. Cellular Abnormalities in the Antarctic notothenioid blood 31 

Figure 9. Area and Roundness Model and Phylogenetic Signal Testing 34 

Figure 10. Phylomorphospace of Red Blood Cell Area and Roundness 40 

Figure 11. Abnormally Located Nuclei in Antarctic Notothenioids 46 
  



 

6 
 

Introduction 

Notothenioid Evolution: 

Notothenioids, the Antarctic fishes that inhabit the freezing waters surrounding 

Antarctica have evolved many unique adaptations and also diversified into a vast variety of 

species (Matschiner et al., 2015). Notothenioids are classified into eight families and ~140 

different species (Eastman & Eakin, 2021). Five of the eight notothenioid families, referred to as 

cryonotothenioids, diversified from a single common ancestor and inhabit the Antarctic waters. 

These families, as shown below in Figure 1, include the Nototheniidae (notothens), 

Artedidraconidae (barbeled plunderfishes), Harpagiferidae (spiny plunderfishes), 

Bathydraconidae (dragonfishes), and Channichthyidae (icefishes). Cryonotothenioids are a prime 

example of adaptive radiation, which is the rapid origin of a variety of distinct species due to 

ecomorphological divergence among close relatives (Matschiner et al., 2015). These five 

Antarctic families are where I center my research, specifically focusing on the icefishes. 

 
Figure 1. Phylogenetic tree of notothenioids and their traits - Simplified phylogenetic tree 

showing the relationship of the eight notothenioid fish families and differences in hemoglobin 

expression and properties. The white and black branches of the tree identify the non-Antarctic and 

Antarctic notothenioids, respectively (Reproduced from Verde et al., 2006). 

 



 

7 
 

The Antarctic waters typically remain close to freezing throughout the year (about 

negative 1.9℃ for sea water), meaning there can be sea ice at the surface and nearly freezing 

temperatures in the water column (Near et al., 2006). Without necessary adaptations, these 

thermal conditions would result in the death and subsequent extinction of many fishes as their 

blood and body would freeze upon contact with the sea ice. However, notothenioids have 

evolved antifreeze glycoproteins (AFGPs) that lower their internal freezing point. AFGPs occur 

in all five of the Antarctic families (Cheng & DeVries, 2005).  

Other evolutionary differences occurring in the Antarctic notothenioids include the loss 

of heat-shock response, reduction in hemoglobin multiplicity (presence of multiple different 

hemoglobin isoforms), reduced hematocrit (volume percentage of cells in the blood), reduced 

hemoglobin affinity for oxygen, and the loss of hemoglobin in the Channichthyidae family 

(Matschiner et al., 2015; Verde et al., 2006). The heat-shock response is the cell's ability to 

produce higher levels of heat-shock proteins inside the cell in response to stress, such as changes 

in temperature, that could damage proteins. These heat-shock response proteins either refold 

misfolded proteins or target the damaged ones for degradation (Le Breton & Mayer, 2016). 

Hemoglobin multiplicity is the presence of multiple different hemoglobin isoforms, each 

possessing different physicochemical properties that can help adapt to changing environmental 

or physiological conditions (Desvignes et al., 2023). Hematocrit is the ratio of cell volume to the 

total blood volume. Thus, there has been a multitude of changes in notothenioids related to their 

blood composition and oxygen transportation as they evolved and adapted to the icy Antarctic 

environment. 

Icefishes, or Channichthyidae, diversified from the other Antarctic notothenioids about 

5.5 MYA (Bista et al., 2022). Icefishes are a group of Antarctic fish species that survive despite 



 

8 
 

lacking mature red blood cells and hemoglobin. They are an evolutionary and biological oddity 

because hemoglobin and red blood cells are an essential part of oxygen transport in all other 

vertebrates. However, despite this seemingly detrimental physiology, icefish have evolved 

adaptations that allow them to maintain normal function and survive in the frigid Antarctic 

waters. 

To accommodate for the lack of hemoglobin, which reduces the oxygen-carrying capacity 

of the blood by a factor of 10 (Matschiner et al., 2015), the Channichthyidae family evolved even 

more adapted features than their relatives. Some of their compensatory features include increased 

blood volume, a larger heart stroke volume and cardiac output, and a larger diameter of the 

arteries and capillaries (Eastman & Eakin, 2021). Along with the high levels of oxygen 

saturation in the frigid Antarctic Ocean, these unique adaptations allow icefishes to remain fully 

functional in their Antarctic environment despite what could be seen as a debilitating loss of 

function. 

Existing Literature: 

The icefish family is such a unique evolutionary anomaly living without hemoglobin and 

mature red blood cells that their physiology and adaptations have fascinated physiologists and 

evolutionary biologists for decades. However, the origin of this white-blooded phenotype 

remains unknown. Thus, I have chosen to focus my research on the gap of knowledge revolving 

around how the loss of hemoglobin and mature red blood cells occurred and which may have 

come first. Two hypotheses can be made 1) the loss of hemoglobin genes in the icefish ancestor 

led to the arrest of red blood cell (RBC) maturation because producing cells that have no 

function is energetically costly for the organism, or, inversely, 2) that detrimental alteration to 

RBC formation triggered the loss of hemoglobin genes in the icefish ancestor. This hypothesis is 



 

9 
 

plausible because hemoglobin that is not contained in RBCs and is instead dissolved directly in 

the blood plasma do not assemble into heterotetramers (four subunits) but instead in 

heterodimers (two subunits) that are toxic for the organism (Vallelian et al., 2022). Previous 

research on icefishes has mainly focused on the genomic loss of hemoglobin genes in icefishes 

and their phylogeny among notothenioids; however, few have studied red blood cell 

morphology, so there is still a large gap in knowledge concerning which loss of function came 

first. 

Along with discovering that the loss of hemoglobin genes occurred in the last common 

ancestor of icefishes (Near et al., 2006), a few studies have focused on characterizing what this 

gene loss entails and how it happened. Under normal circumstances, hemoglobin is composed of 

four subunits: two α- and two β-globin protein subunits (Near et al., 2006). However, research 

has proven that the absence of hemoglobin in icefish is the result of the complete loss of 

hemoglobin genes, except for a truncated and inactive remnant of one α-globin gene (Near et al., 

2006; Bista et al., 2022; Desvignes et al., 2023).  

Several studies and hypotheses have been advanced concerning why this deletion may 

have occurred. One hypothesis is that the loss of hemoglobin genes may have occurred as an 

incidental consequence to alterations to hemoglobin gene expression due to the stability of 

temperatures and constant high oxygenation of the Antarctic regions (Desvignes et al., 2023; 

Verde et al., 2006). Indeed, red-blooded Antarctic notothenioids express only one major 

hemoglobin isoform representing from 85% to 100% of circulating hemoglobin while temperate 

notothenioids and other perciformes express multiple hemoglobin isoforms in comparable 

proportions. One of the main advantages of expressing multiple hemoglobin is that each isoform 

may have different physicochemical properties thus helping adapt and respond to changing 



 

10 
 

environmental or physiological conditions (Desvignes et al., 2023). For example, in the human 

fetus instead of expressing α- and β-globin genes, the hemoglobin is made of α- and γ-globin 

protein subunits (Craig et al., 2021). The gamma subunits give fetal hemoglobin an increased 

affinity for oxygen, which is necessary in the fetal environment to obtain oxygen from the 

mother’s alpha/beta hemoglobin, however, this gamma subunit is not necessary after the first 

year of a baby’s life (Craig et al., 2021). Thus expressing different hemoglobin isoforms with 

differing physicochemical properties are highly important in environments that are constantly 

fluctuating; however, the frigid waters of the Antarctic are consistently cold, hovering at freezing 

temperatures. Thus, relaxed selective pressure in the environmental conditions of icefishes and 

other Antarctic notothenioids may have reduced the need for expressing multiple hemoglobin 

isoforms (Desvignes et al., 2023). Furthermore, one of the main functions of hemoglobin is 

oxygen transport; however, there are abundantly high oxygen levels in the Antarctic waters. 

Since the solubility of oxygen in water is inversely correlated to water temperature, the Antarctic 

Ocean reaches almost 100% oxygen saturation year-round. (Verde et al., 2006). This means that 

Antarctic icefishes may not be as limited by oxygen uptake and transportation as fish in normal 

ocean temperatures despite their lack of hemoglobin. Thus, a lack of necessity for hemoglobin 

may have been what permitted the loss of hemoglobin genes in icefishes (Verde et al., 2006). 

In contrast, another study argues that the loss of hemoglobin was not selectively neutral 

and instead was maladaptive (Near et al., 2006). Multiple adaptations that have evolved in 

icefish, such as enlarged hearts and increased cardiac output, provide evidence that the loss of 

hemoglobin was unfit and needed to be compensated for. That study argued that if the loss of 

hemoglobin was selectively neutral, icefishes would then not have needed to compensate for its 



 

11 
 

loss. Other research additionally supports the conclusion that the loss of hemoglobin was 

maladaptive (Beck et al., 2022). 

Studies have also been conducted focusing more broadly on the genetics underlying the 

red blood cell maturation in icefish. Investigation of genome evolution in connection with red 

blood cell loss suggested that “the genomic importance of Southern Ocean climate change is 

biased toward erythrocyte-associated conserved noncoding elements (CNEs) rather than to 

coding regions” (Daane et al., 2020). CNEs are gene regulatory elements essential for gene 

expression while coding regions are the portions of the gene that encode for the protein. 

Furthermore, the drift in CNEs is intensified near genes that are preferentially expressed late in 

erythropoiesis. Also, the icefish hematopoietic marrow still possesses proerythroblasts (immature 

RBC), which further indicates that the early differentiation of red blood cells remains intact even 

though they do not continue to maturation (Daane et al., 2020). Furthermore, when comparing 

the blood of Parachaenichthys charcoti (Dragonfish family) to Notothenia coriiceps (Notothens 

family) differences were found in cellular shape, with the P. charcoti blood possessing spherical 

erythrocytes while the N. coriiceps blood possessed oval erythrocytes (Daane et al., 2020). 

Additionally, although icefish lack the ability to produce red blood cells, the “erythroid” 

genes (genes used to make RBCs) appear to remain intact, potentially because they perform 

other essential functions elsewhere in the organism (pleiotropy) (Daane et al., 2020). Thus, the 

authors concluded that mutations have accumulated in gene regulatory regions near genes that 

control terminal RBC maturation so that icefishes continue to produce red cell progenitors; 

however, they do not produce mature erythrocytes (Daane et al., 2020). This research also looked 

into the same question addressed in other articles (Beck et al., 2022; Near et al., 2006): whether 

or not RBC loss is maladaptive or adaptive. They suggest that although icefish save energy 



 

12 
 

through a lack of RBC production, this may be counteracted by the costs of physiological 

adaptations needed to overcome anemia and the reduced capability to adjust to environmental 

changes (Daane et al., 2020). 

Thus, although existing literature provides a basis for explaining the fascinating 

physiological adaptations concerning the loss of hemoglobin and red blood cells in icefish, it 

fails to provide a complete picture of how these changes occurred. There is still much to be 

discovered on the reasoning for this loss of function and to answer the question: Did the loss of 

hemoglobin or the loss of red blood cells come first? I, therefore, decided to focus my scientific 

research on the gaps in knowledge that exist in these evolutionary processes to provide a more 

complete picture of the icefish developmental adaptations. Specifically, I hypothesize that 

alterations to red blood cell maturation triggered the loss of hemoglobin genes in icefish. 

Relevancy and Goals: 

Outside of the intriguing evolutionary steps underlying the loss of hemoglobin genes and 

the arrest of red blood cell maturation in icefishes, the Channichthyidae family may also serve as 

a research model for human diseases. Icefish are what we would classify as evolutionary mutant 

models (EMMs), which are species possessing naturally evolved phenotypes that mimic human 

diseases (Beck et al., 2022). “Erythrocytes and hemoglobin appear to be dispensable in red-

blooded notothenioid lineages, which suggests an inherent resiliency within cryonotothenioids to 

accommodate extreme anemia” (Daane et al., 2020). Thus, studying the cellular morphology of 

icefish could lead to important therapeutic developments for anemias in humans, which affect 

about 27% of the world’s population at some point in their lives (Kassebaum, 2016).  

Looking more closely into the genes associated with the lack of red blood cell production 

in icefish as well as their association with anemia can also be highly useful in characterizing 



 

13 
 

human diseases. A study investigating the genetic sequences of two dragonfish species, 

Parachaenichthys charcoti and Gerlachea australis, found that both have missense mutations in 

the erythrocytic beta spectrin (sptb) gene at sites corresponding to pathogenic mutations in the 

same human SPTB gene, which plays a crucial role in red blood cell shape and stability (Daane 

et al., 2020). Thus, the authors demonstrate convergence between the genes causing improper 

cell formation in Antarctic fish and in humans. Therefore, because of the genetic and 

physiological similarities, the cellular morphology of icefish and related Antarctic fish species 

may help further a mechanistic understanding of red blood cell alterations in humans and the 

development of new treatments for patients suffering from diseases such as anemia. 



 

14 
 

Methods 

Blood Smear Staining: 

The blood of 13 red-blooded Antarctic fish species and seven icefish species were 

sampled in Antarctica by Thomas Desvignes. A drop of blood was deposited on a microscope 

slide and smeared using a second slide, as shown in figure 2. Blood smears were then fixed for 3 

min in 100% methanol, air-dried, and kept at room temperature until analysis. The slides were 

stained using a Modified Wright-Giemsa stain following the manufacturer’s protocol (Sigma-

Adrich, St. Louis, USA). Briefly, slides were first placed into a Coplin Jar containing about 50 

mL of Wright-Giemsa solution for 30 seconds. Slides were then transferred for 5 minutes to a 

second Coplin Jar containing about 50 mL of distilled water. Finally, slides were transferred to a 

third Coplin Jar containing about 50 mL of distilled water for 2 minutes. Slides were then left to 

air dry for at least 24 hours before observation under the microscope. 

 
Figure 2. Blood smear fixation - Example of the blood smear fixation process. Image obtained 

from agric.wa.gov.au 

 



 

15 
 

Microscopy and Cellular Analyses 

High-magnification images of the blood cells for each of the 21 fish species were 

obtained using a Leica microscope equipped with a Leica digital camera operated by the Leica 

VARs application suite. Images were taken at 40x magnification as well as at 100x 

magnification with an oil immersion lens. Images were obtained from across the slide in order to 

capture the cellular diversity of each species, avoiding dense areas where cells overlap with one 

another thus preventing accurate measurements. 

After the images were collected the blood cell composition of each fish species was 

analyzed using ImageJ Cell Counter (NIH, Bethesda, USA). The cells of the 40x images of the 

cells were categorized and counted as either from the red blood cell lineage, the white blood cell 

lineage, or the thrombocyte lineage as shown in Figure 3. Thrombocytes in fish are the 

equivalent of blood platelets in humans (they are involved in responding to blood vessel damage 

and stopping bleeding) (Stosik et al., 2019). The varying types of cells were marked based on 

visual indicators evident from our staining protocol. The indicators centered on cellular shape 

and coloring, and many of the classifications were made based on the descriptions of fish blood 

cell types (Claver & Quaglia, 2009). Fish erythrocytes (red blood cells) appear oval to rounded 

in shape and possess a centrally located nucleus. Leukocytes (white blood cells) are typically 

larger than erythrocytes, with much of their area being composed of a large nucleus and appear 

to stain more blue than purple. Thrombocytes are much smaller than both erythrocytes and 

lymphocytes. They are oval in shape and often possess tails or spindles (Claver & Quaglia, 

2009). A total of at least 1000 cells were classified for each species. An example of this process 

is shown below in Figure 3 The proportion of each cell type was then utilized to create a graph 

showcasing the cellular makeup of the blood of each of the Antarctic fish species analyzed. 



 

16 
 

 
Figure 3. Cellular classification - Example of cellular classification in the Notothenia coriiceps 

species. RBC.) Red Blood Cells (Erythrocytes), WBC.) White Blood Cell (Leukocyte), TBC.) 

Thrombocyte 

The 100x magnification images for each of the Antarctic fish species were analyzed 

using semi-automated digital analyses in FIJI (a computer software for image processing). The 

cellular area and roundness of the mature red blood cells were measured for a total of over 200 

cells in each of the red-blooded fish species. The steps conducted for my FIJI Macro are as 

follows: 1) Duplicate the original image; 2) Enhance contrast and set saturation to 0.3; 3) 

Subtract the background and set rolling to 50 light; 4) Split channels; 5) Select window and open 

up the image in the red channel; 6) Enhance contrast, set saturation to 0.3, and normalize; 7) 

Subtract background and set rolling to 50 light; 8) Enhance contrast, set saturation to 0.3, 

normalize; 9) Run with a median and radius of five; 10) Set the auto threshold to default and run; 

11) Set option as black background; 12) Convert to mask; 13) Open, close, and fill holes; 14) Set 

scale using distance = 9.78, known = 1, and unit = um; 15) Set measurements for area perimeter 



 

17 
 

bounding fit shape skewness using a variable of 3; 16) Analyze particles, set parameters of size 

50 - 100, circularity 0.5 - 1.00. Show outlines, exclude clear and include add; 17) Now, from the 

black and white image of the cells, remove any cells that do not appear circular or do not appear 

representative of the original image. Record measurements for the ones that do accurately 

measure the wanted variables. An overview of this process is shown in figure 4. 

 

Figure 4. Digital analysis of cells - Cells of Notothenia coriiceps showcase the semi-automated 

digital analysis process. First, the microscope image was opened in FIJI. The image was then 

turned into a black-and-white contrast image and the outline of the cells were obtained with the 

cellular quantifications of area and roundness of the cell. Roundness is a measure of the length-to-

width aspect ratio, an example of which is shown on cell A. 

Data were also obtained for the area and roundness in ~10 immature red blood cells for 

the seven icefish species following the same steps listed above. Due to the rarity of immature 

erythrocytes in the icefish blood, we were not able to measure more erythropoietic cells in 

icefishes. Individual measurements were then averaged to characterize each species because the 

phylogenetic comparative methods used allow only a single value per trait and per species. 

The 40x images were also utilized to quantify the number of off-centered vs centered cell 

nuclei in each of the 13 red-blooded fish species. No immature RBCs were quantified since they 

always have an overly large and off-centered nucleus. Thus the icefish species were excluded 

from this portion of my analysis since they only have immature RBCs. Off-centered cells in the 

red-blooded species have a different staining of the cytoplasm and a non-condensed nucleus so 



 

18 
 

they were easy to spot and remove. ImageJ Cell Counter was used to count the number of red 

blood cells with a centered nucleus and with an off-centered nucleus for a total of at least 500 

cells in each species. The proportion of centered nuclei was used to classify the species as having 

either mainly centered nuclei or mainly off-centered nuclei at a threshold of 50%. Cells were 

each individually classified as being centered only if their nucleus was almost perfectly centered; 

everything else was classified as off-centered if their nucleus was greater than ~25% off from the 

true center of the cell. An example of this process is shown below in Figure 5. 

 
Figure 5. Comparison of nuclei position - Image of cells in the species comparing centered 

nuclei (labeled with #1) and off-centered nuclei (labeled with #2). The image on the left is the 

species Trematomus bernacchii, possessing a majority of centered cell nuclei. The image on the 

right is the species Parachaenichthys charc2oti, possessing a majority of off-centered cell nuclei. 

The cell in the bottom left corner with a different color staining of its cytoplasm and an 

uncondensed nucleus is an immature RBC that was not counted. 

Phylogenetic Comparative Analysis 

After collecting and averaging data for the cellular composition and cellular morphology 

of the blood of each of the 20 fish species, phylogenetic comparative analyses of my quantitative 

measurements were performed using R studio. I utilized three R packages in combination with a 

published tree of species (Parker et al., 2022) to analyze the evolution of hematocrit, RBC area, 



 

19 
 

RBC roundness (RBC roundness values, which varies from 0 being completely flat and 1 being a 

perfect circle, are based on the length-to-width aspect ratio), and positioning of the cell nuclei. 

The package phytools was utilized for most of my phylogenetic analyses (Revell L, 2012). In this 

package, I used the functions “contMap” (to map continuous trait evolution on the tree), 

“phenogram” (to plot a phylogenetic traitgram), “rotateNodes” (to rotate nodes on a phylogenetic 

tree), “phylosig” (to compute phylogenetic signals using the methods of Blomberg’s K or Pagel’s 

λ), and “fastBM” (to produce quantitative trait simulation under Brownian motion on 

phylogenies). Blomberg’s K and Pagel’s λ are two different ways to estimate the strength of 

phylogenetic signal in a dataset relative to what is expected with Brownian motion. The package 

geiger was used to create disparity-through-time plots and perform model fitting (Pennell et al., 

2014). In this package, I utilized the functions “dtt” (to calculate and plot disparity-through-time 

for a phylogenetic tree/data) and “fitContinuous” (to perform model fitting for continuous data). 

Lastly, I used the package Ape in the construction of phylogenetic trees (Paradis & Schliep, 

2019). In this package, I utilized the function “drop.tip” (to remove tips in a phylogenetic tree). 

 



 

20 
 

Results 

Evolution of the Antarctic Notothenioid Blood 

As presented in the introduction, the notothen blood and the icefish blood are extremely 

different. To better understand these differences, I first investigated the blood cellular content. I 

started by analyzing the hematocrit (i.e., the proportion of cells in the blood) of various species 

of Antarctic notothenioid fishes based on existing literature and data collected by my lab. I then 

quantified the differences in the cellular makeup of the blood of red-blooded and white-blooded 

Antarctic fishes based on microscopic imaging of blood smear samples collected by my lab. 

 Evolution of notothenioid hematocrit 

Simply drawing blood from notothens and icefishes reveals outstanding differences. The 

blood of the two groups strikingly differs in their coloration: notothens have dark red blood, 

similar to ours (Figure 6A), and icefishes have clear white blood (Figure 6B). By centrifuging 

the blood of notothens and icefishes, dramatic differences in cellular content are further revealed 

(Figure 6A-B). Notothens have a high quantity of blood cells that appear dark red while icefishes 

have a small quantity of blood cells that appear white. 



 

21 
 

 



 

22 
 

Figure 6. Evolution of Antarctic notothenioid hematocrit - A) Image of a red-blooded 

Antarctic notothen, Notothenia rossii, alongside an image of its blood before (left) and after 

centrifugation (right). The upper, clear portion of the centrifuged blood is the blood plasma and 

the dark red portion at the bottom are the blood cells. B) Image of a white-blooded Antarctic fish, 

Cryodraco antarcticus, alongside an image of its blood before (left) and after centrifugation 

(right). The large upper, clear portion is the plasma and the small, white portion at the bottom are 

the blood cells. C) Ancestral state reconstruction of Antarctic fish hematocrit using the contmap 

function and a color gradient with yellow representing low hematocrit values (~1%), blue 

representing intermediate values (~10%), and green representing high hematocrit values (39%). 

Notothen species names are in red, plunderfishes are in green, dragonfishes are in purple, and 

icefishes are in cyan. D) Phenogram based on the hematocrit value of various Antarctic fish 

species. E) Disparity-through-time plot for the evolution of notothenioid hematocrit. F) Model 

testing of the Early Burst model, Ornstein-Uhlenbeck model, and the Brownian motion model for 

the hematocrit values. The solid line represents the observed disparity, and the dotted line is the 

evolution predicted by the BM model of evolution. The shaded gray area in the background 

represents the 95% confidence interval of the predicted evolution of disparity under the BM model 

of evolution. G) Model testing for a phylogenetic signal with Blomberg’s K and Pagel’s λ. The p-

value for Blomberg’s L is based on 10000 randomizations and the p-value for Pagel’s λ is based 

on LR-test. Significant values (p<0.05) are written in bold font. 

 

I first used the contmap function in the R phytools package to analyze the evolution of 

hematocrit in 36 species of Antarctic notothenioid fishes (including 17 notothens, 2 

plunderfishes, 9 dragonfishes, and 8 icefishes) and reconstructed the ancestral state of the 

character across the species phylogeny. Hematocrit values were obtained from existing literature 

(Kooyman 1963; Holeton, 1970; Kunzmann et al., 1992; Beers et al., 2010; etc) and 

measurements collected by my lab. The contmap analysis returned a tree that revealed the 

evolution of hematocrit using a color gradient varying from ~1% hematocrit in yellow to ~40% 

hematocrit in green with a transition by blue set at ~10% (Figure 6C). To visualize the evolution 

of hematocrit outside of the constraints of the branches of a phylogenetic tree, I created a 

phenogram of the hematocrit values (Figure 6D). A phenogram plots time on the x-axis and the 



 

23 
 

values of the studied character on the y-axis with phylogenetic relationships displayed based on a 

time-calibrated tree and the ancestral state of the character predicted under Brownian motion. 

The blue shading around the edges of the phylogenetic tree represents the 95% confidence 

interval of the predicted reconstruction. To help identify the different Antarctic notothenioid 

families, I further colored the names of notothen species in red, plunderfishes in green, 

dragonfishes in purple, and icefishes in cyan. Both visual representations of the evolution of 

hematocrit in notothenioids revealed that red-blooded Antarctic fish families possess much 

higher hematocrit levels than icefishes. Furthermore, my analysis showed that the majority of 

red-blooded notothenioids have hematocrit values ranging from 12% to 26% and that a few 

species of the genera Trematomus and Notothenia have much higher hematocrit between 31% 

and 39%. These analyses further highlighted a clear separation of the white-blooded icefishes 

whose hematocrit are around 1% in all species. 

To further explore how the hematocrit changed over time across the different families, I 

utilized the dtt function in the R geiger package to create a disparity-through-time plot of the 

character. Disparity-through-time plots estimate the relative trait disparity both within and 

between subclades and compares it to the null hypothesis that the trait follows an unconstrained 

model of random, incremental evolution (Brownian Motion, BM) along a time-calibrated 

phylogenetic tree. The solid black line represents the observed disparity, and the dotted line is 

the evolution predicted by the BM model of evolution. High disparity values mean that the 

disparity observed is very different from the disparity seen at the present time (relative time of 

1.0 at the present time and 0 representing the stem of the phylogeny at 10.73 MYA) and low 

disparity means it is similar or the same. The morphological disparity index (MDI) statistics 

associated with the disparity-through-time analysis sums the deviations of the observed 



 

24 
 

disparity-through-time curve (solid line) from the median of the null model simulations (dotted 

line) to test whether the observed disparity is significantly different from what would be 

predicted under a BM model of evolution. In the plot, the shaded gray area represents the 95% 

confidence interval of the predicted evolution of disparity under the BM model of evolution. A 

significant positive MDI value signifies that the disparity tends to be distributed among 

subclades (i.e., the differences in the trait values are large between families) while a significant 

negative MDI value means that the disparity tends to be distributed within the subclades (i.e., the 

differences in the trait values are large between species in the same families). Looking at the 

disparity-through-time plot for hematocrit (Figure 6E), we first observed that there was a change 

in disparity occurring at a relative time of 0.3, which corresponds to about 6.7 million years ago 

(MYA), which is around when icefishes diverged from red-blooded notothenioids. While the 

solid line deviates from the dotted line indicative of BM model, this change is not significant 

since it is still within the gray shaded area representative of the confidence interval. However, 

the solid line did deviate outside of our shaded gray area around a relative time of 0.7, which is 

about 4 MYA, indicating there may be evolution of the trait outside of the BM model happening 

at this time; however, this deviation was very slight. Overall, the observed disparity remained 

within the confidence interval of the BM model of evolution, meaning that the hematocrit in 

notothenioids largely evolved in a random, unconstrained manner across the phylogeny. Finally, 

our hematocrit MDI value of -0.11 with a non-significant p-value of 0.13 suggests that the 

hematocrit in notothenioids varied as much between subclades as within clades. 

As mentioned above, Brownian Motion is a model for character evolution that involves 

random, incremental, and unconstrained evolution; however, there are other models such as the 

Early Burst (EB) and the Ornstein-Uhlenbeck (OU) models that make different evolutionary 



 

25 
 

hypotheses. The EB model assumes that organisms evolved randomly after a sudden burst or 

change in character. Thus, the EB model is suitable to depict the rapid evolution of traits often 

seen in the context of adaptive radiations. The OU model, in contrast, predicts that organisms 

evolved randomly and incrementally like in the BM model; however, it is constrained and 

assumes that the trait, while able to deviate, will evolve toward an optimal value. To compare the 

fit of each model to the data, I used the corrected Akaike Information Criterion (AICc), which 

estimates the prediction of error used for a statistical model in a given set of data. The lower the 

AICc value, the better the model is at fitting the given data. In our case, for hematocrit, since the 

BM model has the lowest AICc score it is the best at predicting the hematocrit evolution in 

notothenioids (Figure 6F). This suggests that the notothenioid hematocrit evolved randomly and 

incrementally in an unconstrained manner. 

Finally, to determine whether there is a significant phylogenetic signal in my hematocrit 

data (i.e., whether closely related species are more similar to one another than expected by 

chance, due to their shared evolutionary history), I calculated the Blomberg’s K and Pagel’s λ 

(lambda), which are utilized to approximate the strength of phylogenetic signal in a dataset 

relative to Brownian motion predictions. Based on my model testing on the hematocrit data 

(Figure 6G), I found that both Blomberg’s K and Pagel’s λ were significant, suggesting that there 

is a phylogenetic signal in the evolution of hematocrit in our data. This indicates that the 

evolution of hematocrit in our species is correlated with their evolutionary relationships and that 

closely related species tend to be more similar to one another than to distantly related species. 

 



 

26 
 

 Notothenioid blood cellular makeup 

Beyond cell quantity in the blood (i.e., hematocrit), in order to understand how blood 

cells differ between species, I scored over 1000 randomly-chosen blood cells per species (with 

one to three fish scored per species) and counted up the distinct number of red blood cells, white 

blood cells, and thrombocytes in the blood of 21 different Antarctic fish species (Figure 7). The 

red-blooded families of notothens, plunderfishes, and dragonfishes have the majority of their 

blood composed of red blood cells (85-96%) and a small portion of white blood cells (~6%) and 

thrombocytes (~3%). In striking contrast, the icefish blood is primarily composed of white blood 

cells (70-85%) with only a small portion of red blood cells (~6%) and thrombocytes (~14%). 



 

27 
 

 
Figure 7. Proportions of main blood cell lineages in Antarctic notothenioid blood - The 

proportion of main blood cell lineages is illustrated through the different colors inside the bars. 

The red coloring represents the proportion of red blood cells in the blood. The light blue coloring 

represents the proportion of white blood cells in the blood. The purple represents the proportion of 

thrombocytes in the blood. The species names are color-coded based on their families. Notothen 

species names are in red, plunderfish names are in green, dragonfish names are in purple, and 

icefish names are in cyan. The images to the right of the bar graph are examples of microscopic 

images of Antarctic fish species (a notothen, a dragonfish, and an icefish) and how the blood cells 

were quantified. The 1’s are the red blood cells, 2’s are the white blood cells, and 3’s are the 

thrombocytes. 

 



 

28 
 

Not only does cellular makeup of the blood between red-blooded families and white-

blooded families look dramatically different, but so do the red blood cells themselves. In the red-

blooded families, the majority of red blood cells appear to be fully mature, but in the white-

blooded icefish, the few red blood cells seen are immature and appear larger and rounder than in 

red-blooded notothens (compare “1” in the blood smear of the icefish Pagetopsis macropterus to 

the notothen Lepidonotothen squamifrons in Fig.2). Intriguingly, a few species of red-blooded 

fish seem to have red blood cells that are quite round compared to other notothens (See the 

dragonfish Gymnodraco acuticeps compared to the notothen Lepidonotothen squamifrons in 

Fig.2). These qualitative observations of red blood cell size and shape variations between white 

blooded icefishes and red-blooded notothens, and between red-blooded species prompted my 

following endeavor in research: quantifying the differences in cell area and roundness in 

icefishes and red-blooded notothenioids. 

Cellular abnormalities in the Antarctic notothenioid Blood 

As mentioned above, the qualitative cellular differences observed by microscope imaging 

of the blood of the various Antarctic notothenioids families prompted me to examine these 

abnormal cellular dimensions quantitatively. Not only were the immature red blood cells in 

icefishes abnormally large and round, but seemingly so were the red blood cells in a few closely 

related species 

Evolution of red-blood cell area 

I first used the contmap function in R to analyze the evolution of the cellular area and 

roundness in 20 Antarctic notothenioid species (including seven notothens, two plunderfishes, 

four dragonfishes, and seven icefishes) and reconstructed the ancestral state of the character 

across the species phylogeny. The contmap analysis returned a tree that revealed the evolution of 



 

29 
 

red blood cell area using a color gradient. The gradient varied from high area values of 148 µm2 

in blue to low area values of 74 µm2 in red (Figure 8A). To further visualize the evolution of red 

blood cell area without the constraints of a phylogenetic tree, I created a phenogram of the red 

blood cells’ dimensional values (Figure 8B). To help identify the different Antarctic notothenioid 

families I used the family color-coding previously described in Figure 6. Both visual 

representations of the evolution of red blood cell area in notothenioids revealed that icefishes and 

the closely related dragonfish family displayed abnormally large red blood cells (RBCs). In 

icefishes the average value of cellular area was 122 µm2, it was 105 µm2 in dragonfishes, 92 µm2 

in plunderfishes, and 91 µm2 in notothens. The dragonfish species with the largest areas were 

Akarotaxis nudiceps, Gymnodraco acuticeps, and Vomeridens infuscipinnis, which had an 

average cellular area of ~111 µm2. However, these were not general rules as Parachaenichthys 

charcoti (88 µm2) from the dragonfish family and Chaenodraco wilsoni (87 µm2) from the 

icefish family both had smaller areas, more in the notothen area range. 



 

30 
 

 



 

31 
 

Figure 8. Cellular Abnormalities in the Antarctic notothenioid blood - A) Ancestral state 

reconstruction of Antarctic fish blood using the contmap function for cell area values. The color 

gradient shows the high values of 148 µm2 in blue and the low values of 74 µm2 in red. Notothen 

species names are written in red, plunderfishes are in green, dragonfishes are in purple, and 

icefishes are in cyan. B) Ancestral state reconstruction of Antarctic fish blood using the contmap 

function for cell roundness values. The color gradient shows the high values of 0.96 in blue and 

the low values of 0.60 in red. Notothen species names are highlighted using the same color 

scheme previously mentioned. C) Phenogram based on the red blood cell area values of various 

Antarctic fish species. D) Phenogram based on the red blood cell roundness values of various 

Antarctic fish species. E) Disparity-through-time plots for the evolution of notothenioid red blood 

cell area plotted with and without icefishes. The solid line represents the observed disparity, and 

the dotted line is the evolution predicted by the BM model of evolution. The shaded gray area in 

the background represents the 95% confidence interval of the predicted evolution of disparity 

under the BM model of evolution. F) Disparity-through-time plots for the evolution of 

notothenioid red blood cell roundness plotted with and without icefishes. 

 

To investigate further how the red blood cell area changed over time across the different 

families, I utilized the dtt function in R to create disparity-through-time plots across the 

character, with one plot for area values with icefishes and another for the values without the 

icefishes. I made two separate plots to test whether the changes occurring in the cellular area 

were caused by icefishes or were occurring even without the icefishes.  

Looking at the disparity-through-time plots of area both the graphs plotted with icefishes 

and without the icefishes have a spike in disparity around 0.75 relative time, which is about 2.5 

million years ago (Figure 8C). Looking at the phenogram, this time, which is fairly close to the 

present time (Figure 8B), seems to correspond to when sister species of red-blooded notothenioid 

species diversified from each other with some dramatically different cell areas, as shown by the 

many branches splitting off from common ancestors and particularly marked between the sister 

species pairs Trematomus bernacchii and T. hansoni, and Notothenia coriiceps and N. rossii 

(Figure 8B). The fact that this spike in disparity appears in both the plot with icefishes and the 



 

32 
 

one without icefishes tells us that this change in cellular area was occurring even without the 

icefishes' influence. Thus red-blooded notothenioid species have different RBC areas just as 

icefishes do. 

Both plots for RBC areas with icefishes or without icefishes had positive MDI values of 

0.15 and 0.34 respectively with non-significant p-values of 0.76 and 0.95 respectively, 

suggesting that the RBC area in notothenioids is more pronounced within clades than between 

subclades (Figure 8C). The cellular area values appeared to have a large change in disparity as 

individual species begin to diversify within their families rather than in distant common 

ancestors. This suggestion is also reflected in the phenogram for cell area values (Figure 8B). 

Looking at the phenogram, you can see that when sorted by their cellular area values many of the 

different species from varying families are intermixed with one another, further showcasing that 

the amount of variability within the subclades is comparable to the amount of variability between 

clades (Figure 8B).  

The following undertaking for analyzing my cell size quantitative data was comparing the 

fit of each evolutionary model using the AICc. I found that the best fit for the area data both with 

and without the icefishes was the OU model (Figure 9A). This indicates that the area of RBCs in 

notothenioids evolved randomly and incrementally around an optimal value. In looking at the 

phenogram, it appears the optimal value is around 95 µm2 as this is the predicted ancestral value 

the majority of area measurements in the studied Antarctic notothenioid species vary around 

(Figure 8B). 



 

33 
 

 



 

34 
 

Figure 9. Area and Roundness Model and Phylogenetic Signal Testing - A) Model testing of 

the Early Burst mode, Ornstein-Uhlenbeck model, and the Brownian motion model for the red 

blood cell area values both with and without the icefishes. B) Model testing of the Early Burst 

mode, Ornstein-Uhlenbeck model, and the Brownian motion model for the red blood cell 

roundness values both with and without the icefishes. C) Model testing for a phylogenetic signal 

of area values with Blomberg’s K and Pagel’s λ. The p-value for Blomberg’s K is based on 10000 

randomizations and the p-value for Pagel’s λ is based on LR-test. Significant values (<0.05) are 

written in bold font. B) C) Model testing for a phylogenetic signal of roundness values with 

Blomberg’s K and Pagel’s λ. The p-value for Blomberg’s L is based on 10000 randomizations and 

the p-value for Pagel’s λ is based on LR-test. Significant values (<0.05) are written in bold font. 

 

To assess whether there is a significant phylogenetic signal in my data I calculated 

Blomberg's K and Pagel’s λ for cell area (Figure 9B). I found that neither the Blomberg’s K nor 

Pagel’s λ were significant for the cellular area data, indicating that there is not a phylogenetic 

signal in the evolution of this character. This means that the cellular area in our species is not 

correlated with their evolutionary relationships and closely related species do not tend to be more 

similar to one another than to distantly related species. This idea was also shown in our 

phenogram and disparity plots as I previously mentioned (Figure 8B & 8C), where there was a 

large amount of disparity and diversification closer to the present time and the species cellular 

area values diversified both between and within clades. However, even though the cellular area 

values between clades were not statistically significant there is still a trend for many species in 

the icefish and dragonfish families having larger cellular areas. Despite the two outliers 

(Parachaenichthys charcoti and Chaenodraco wilsoni), species in the icefish and dragonfish 

families generally grouped together because of their larger cellular areas (Figure 8B). 



 

35 
 

Evolution of red-blood cell roundness 

The RBCs of stereotypical fish are nucleated flat ovals (See image of Lepidonotothen 

squamifrons in Figure 7), but initial observations suggested that some notothenioids had more 

circular or even spherical RBCs. I initially started my analysis of the evolution of the cellular 

roundness values in the same 20 Antarctic notothenioid species I utilized for my analysis of 

cellular area. I reconstructed the ancestral state of the character across the species phylogeny 

using the contmap function in R. The contmap analysis returned a tree that displayed the 

evolution of red blood cell roundness using a color gradient. The gradient varied from high 

roundness values of 0.96 in blue to low roundness values (more oval) of 0.60 in red (Figure 8D).  

For visualization of the evolution of RBC roundness in the absence of the restrictions of a 

phylogenetic tree, I created a phenogram using my collected RBC roundness values (Figure 8E). 

I colored the names of the species families in the same manner with notothen species in red, 

plunderfishes in green, dragonfishes in purple, and icefishes in cyan. The visual representation of 

the evolution of red blood cell roundness in notothenioids revealed that icefishes and their 

closely related species of dragonfishes, specifically Parachaenichthys charcoti and Gymnodraco 

acuticeps, displayed round RBCs (Figure 8E). The average cellular values for roundness were 

0.92 for icefishes, 0.80 for dragonfishes, 0.70 for plunderfishes, and 0.70 for notothens. This plot 

highlights the abnormal cellular roundness found in icefishes and their related species of 

dragonfishes compared to the oval RBCs in other species. One of the plunderfish species, 

Dolloidraco longedorsalis, appears to represent an intermediate value between the overall highly 

round RBC of icefishes and more oval RBC of notothens. However, it is important to note that 

there were some outliers to this statement. From the dragonfish family, Vomeridens infuscipinnis 

and Akarotaxis nudiceps displayed RBC that were more oval (Figure 8E). The species 



 

36 
 

Harpagifer antarcticus from the plunderfish family also displayed more oval RBC (Figure 8E). 

Thus, icefishes have very round RBC, dragonfish are split with some species having more round 

cells (Parachaenichthys charcoti and Gymnodraco acuticeps) and some having more oval cells 

(Vomeridens infuscipinnis and Akarotaxis nudiceps), and plunderfish and notothen RBCs 

displaying typical oval proportions (Figure 8E). 

The next step I took in inspecting how the RBC dimensions of roundness changed over 

time across the different families was to create a disparity-through-time plot across the character 

by utilizing the dtt function. Both disparity-through-time plots evolved fairly close to the dotted 

line predicative of the BM model of evolution, suggesting that cell roundness evolved in a 

random, unconstrained manner across the phylogeny. However, the plot for roundness without 

the icefishes appears to be evolving with a higher rate of disparity. It is interesting that the 

disparity is higher without the icefishes (MDI = 0.07) then with the icefishes (MDI = -0.04) 

because this indicates that it is not the icefish data causing the larger differences in roundness, 

but rather the red-blooded notothenioid lineages themselves (Figure 8F). This is also reflected in 

the phenogram for roundness, where there are variations in the roundness values within the 

families and between species such as in the dragonfish Parachaenichthys charcoti  & 

Gymnodraco acuticeps and Vomeridens infuscipinnis & Akarotaxis nudiceps. Both the MDI 

values of the disparity-through-time plots for the cell roundness with the icefishes and without 

had non-significant p-values of 0.36 and 0.70 respectively (Figure 8F). This finding reflects the 

variations within the families and between species I listed above, which is shown in the 

phenogram where two dragonfish species have very round RBC (Parachaenichthys charcoti  & 

Gymnodraco acuticeps averaging at 0.85) and two have more oval RBC (Vomeridens 

infuscipinnis & Akarotaxis nudiceps averaging at 0.66) (Figure 8E). When focusing on the two 



 

37 
 

dragonfish species with highly round RBC, these two dragonfish appear to be grouped together 

as having abnormally round values that are more similar to the icefishes and separate from the 

other lineages of notothens and plunderfishes (Figure 8E). 

Comparing the fit of each evolutionary model using the AICc was the subsequent step I 

took in analyzing my cell shape quantitative data. I found that the best fit for the roundness data 

with the icefishes was the BM model and the best fit for the roundness data without the icefishes 

was the OU model (Figure 9C). For the roundness data with icefishes, the BM model indicates 

that the notothenioid roundness values evolved randomly and incrementally in an unconstrained 

manner (Figure 9C). For the roundness data without the icefishes, the OU model indicates that 

the notothenioid roundness values evolved randomly and incrementally around an optimal value, 

presumably 0.75 for roundness since this is the reconstructed ancestral roundness value predicted 

at the base of the species studied and where the values appear to evolve around (Figure 8E & 

9C). The distinction between the best fitting model for the two different data sets indicates that 

while the roundness values of the Antarctic species excluding the icefishes likely evolved around 

a set optimal value, the icefishes evolved outside of this constraint. This idea is also illustrated in 

the phenogram for cell roundness where there is a clear branching off of the icefishes occurring 

around 7.1 MYA that led them to have more round values than the rest of the Antarctic 

notothenioids (Figure 8E). This intense deviation of the icefish roundness values leads to them 

having a clear separation in their group in both the contmap color gradient scale and the 

roundness phenogram that is reflected in the distinct best fit model (Figure 8D, 8E, & 9C). The 

intensity of the deviation further supports the idea that two of the dragonfish species 

(Parachaenichthys charcoti and Gymbodraco acuticeps) evolved to have roundness values 

evidently more extreme than the majority of the other Antarctic fish species (Figure 8D & 8E). 



 

38 
 

I calculated Blomberg's K and Pagel’s λ for the cell roundness to assess whether there is a 

significant phylogenetic signal in my data (Figure 9D). I found that both Blomberg’s K and 

Pagel’s λ were significant for the cellular roundness that included icefishes but were non-

significant for the cellular roundness data that did not include icefishes. This difference is 

important because it shows that when we include the roundness values for icefishes in our data 

set, closely related species are more similar to each other than to distantly related species, 

suggesting an outsized role of icefish species in this test. This is indicated by the clear separation 

of the icefish RBC high roundness values by their coloring in the contmap and their grouping in 

the phenogram plots (Figure 8D & 8E). However, when looking at the roundness values for the 

data excluding the icefishes, closely related species do not tend to be more similar to one another 

than to distantly related species. This is evident in the intermixing of species from different 

families in the phenogram for roundness, where the roundness values have variability within 

their subclades (Figure 8E). The differences for values of cell roundness become significant 

when the icefishes are taken into account as this creates more of a clear separation between the 

cellular roundness of the different families. 

Integration of red-blood cell area and roundness evolution 

After looking at the red-blood cell area and roundness values separately, my next 

endeavor was to explore how they evolved together. In order to merge the two characters into 

one plot I used the phylomorphospace function in phytools package to create a plot for the area 

and roundness using the same data I used in the previous two sections from 20 Antarctic 

notothenioid species. A phylomorphospace plot is highly useful for representing data from two 

or more continuous parameters in a visual way. In short, it is similar to a Principal Component 

Analysis on which species are linked by lines representing their phylogenetic relationships. It 



 

39 
 

gives a graphical representation of the changing morphology of two or more characters as they 

evolve in a group of related species. Here, one character, the cell area, is represented on the x-

axis and the other, the cell roundness, on the y-axis. The black dots represent the internal nodes 

based on the phylogenetic tree given. In other words, it is the point where one or more branches 

of the tree meet, indicating the hypothetical common ancestor of certain species. The colored 

dots represent the tips, which indicates the individual species. Each colored point on the 

phylomorphospace plot represents a species character average and the distance between the 

points represents how alike or different the species are from one another based on the character 

values. The more closely the points are to one another the more similar the data values of the 

species are. For easier distinction in the Antarctic fish families, I have plotted the notothens in 

red, the plunderfishes in green, the dragonfishes in purple, and the icefishes in cyan (Figure 10).  



 

40 
 

 
Figure 10. Phylomorphospace of Red Blood Cell Area and Roundness - The morphological 

differences between various Antarctic notothenioids are illustrated on the phylomorphospace plot. 

The black dots represent the internal nodes based on the phylogenetic tree given. These are the 

points where more than one branch of the tree meet, indicating the hypothetical common ancestor 

of certain species. The different colored dots represent the different species, with notothens in red, 

plunderfishes in green, dragonfishes in purple, and icefishes in cyan. Located next to some of the 

dots are a cellular image of a representative cell from the species. The x-axis represents the area of 

the cells, ranging from 74 µm2 - 148 µm2 and the y-axis represents the roundness of the cells 

ranging from 0.60 - 0.96. Located on the x-axis are visual representations of the differences in 

cellular area between the low end and high end of the axis. Located on the y-axis for roundness are 

visual representations of the differences in cellular roundness between the low end and high end of 

the axis.  

 



 

41 
 

RBC area is represented on the x-axis with values ranging from 74 µm2 in 

Lepidonotothen squamifrons (notothen family) to values of 148 µm2 in Chaenocephalus aceratus 

(icefish family). The cellular roundness is represented on the y-axis with values ranging from 

0.60 in Akarotaxis nudiceps (dragonfish family) to 0.96 in Chaenodraco wilsoni (icefish family) 

(Figure 10). Overall, the notothens are grouped near the bottom left, indicating that they have 

smaller RBC areas (averaging 91 µm2) and a more oval shape (averaging 0.7) (Figure 10). 

Lepidonotothen squamifrons and Trematomus bernacchii are notable outliers to the general 

notothen grouping (Figure 10). Lepidonotothen squamifrons is among the species having the 

most oblong RBCs (roundness of 0.65) and the species having the smallest RBCs (73.9 µm2). 

Trematomus bernacchii also has small RBCs compared to the rest of the notothens (74.6 µm2) 

but in contrast to Lepidonotothen squamifrons, Trematomus bernacchii produces RBCs that are 

among very round (0.83) (Figure 10). The family most closely related to the notothens is the 

plunderfishes, which appear similar to notothens on these characters because they are placed near 

the bottom left of the plot and grouped with the notothens (Figure 10). The plunderfishes have 

cell areas averaging at 92 µm2 and roundness averaging at 0.7, meaning their RBC are small and 

oval (Figure 10). In terms of roundness, the dragonfishes appear extremely variable with 

Akarotaxis nudiceps having RBCs that are the most oval cells observed among all studied 

cryonotothenioids, Vomeridens infuscipinnis having cells as oval as the majority of notothens 

and plunderfishes, and the other two species (Parachaenichthys charcoti & Gymnodraco 

acuticeps) having RBCs more round than notothens, especially Parachaenichthys charcoti 

(Figure 10). The area values of species Akarotaxis nudiceps, Vomeridens infuscipinnis, and 

Gymnodraco acuticeps are relatively grouped together around an average of 110.6 µm2, which is 

in the upper end of what was observed among red-blooded species. However, there is still some 



 

42 
 

variability as Parachaenichthys charcoti has much smaller cells than the other dragonfishes with 

an area of 87.6 µm2 (Figure 10). Overall, these measurements place the dragonfishes in between 

the notothens/plunderfishes and the icefishes on the plot, although their area and roundness 

values still more closely resemble the red-blooded Antarctic notothenioids than the icefishes 

(Figure 10). The icefishes are located near the upper portion of the graph throughout the upper 

middle and upper right area, indicating that the majority of them have large area values 

(averaging at 122 µm2) and large roundness values (averaging at 0.92) (Figure 10). However, the 

species Chaenodraco wilsoni is an outlier to this statement because although it has the largest 

roundness value of 0.96 it has a lower area value of 87.2 µm2, which is more similar to the red-

blooded Antarctic notothenioids than the other icefishes (Figure 10). Overall, the 

phylomorphospace indicates that there is a clear distinction between the morphology of large and 

round immature red blood cells of icefishes and the smaller and oval red blood cells of the other 

Antarctic species, with cellular shape and size variability beginning to emerge in the dragonfish 

lineage that is phylogenetically closely related to the icefishes. 

Abnormally Located Nuclei in Antarctic Notothenioids 

While utilizing cellular microscopy to quantify differences in cellular area and cellular 

roundness I noticed abnormalities in the position of the nuclei in some of the Antarctic 

notothenioids. Not only do a few of the dragonfish species closely related to icefishes display 

abnormally large (Akarotaxis nudiceps & Gymnodraco acuticeps) and/or abnormally round red 

blood cells (Parachaenichthys charcoti & Gymnodraco acuticeps), but the nuclei of their red 

blood cells also appear off-centered (Figure 8B, 8E, & 10). To better understand these 

differences, I utilized cellular microscopy of blood smears from various red-blooded Antarctic 

notothenioid fishes collected by my lab to count the number of cells in each species that had 



 

43 
 

centered nuclei and that had off-centered nuclei among a minimum of 500 cells per species. 

After collecting data on the nuclei placement, I found that many species closely related to 

icefishes had a large portion of off-centered nuclei placements. 

I first used the contmap function to analyze the evolution of nuclei placement in the 

RBCs of 13 Antarctic notothenioid species (including 7 notothens, 2 plunderfishes, and 4 

dragonfishes). Icefishes were excluded from this analysis due to the rarity and lack of mature red 

blood cells in their blood. Immature red blood cells, whether from red-blooded or white-blooded 

species, typically have abnormally large and off-centered nuclei so I excluded them from my 

analysis and only scored mature cells with compact nuclei. The contmap analysis provided a tree 

conveying the evolution of the proportion of centered nuclei in the species' blood using a color 

gradient. The gradient varied from high proportions of centered nuclei (~98.7%) in blue to low 

proportions of centered nuclei (~14.4%) in red (Figure 11A). To help identify the different 

Antarctic notothenioid families, I placed the species more closely related to icefishes 

(plunderfishes and dragonfishes) in bold font (Figure 11A). I also included microscopic images 

of three species (the notothen Trematomus eulepidotus, the plunderfish Dolloidraco 

longedorsalis, and the dragonfish Gymnodraco acuticeps) in order to help with the visualization 

of this character (Figure 11A). To showcase the evolution of the changing proportions of 

centered nuclei without the limitations of a phylogenetic tree, I created a phenogram using the 

proportions of centered and off-centered nuclei in various Antarctic notothenioid species (Figure 

11B). To assist with the identification of the distinct Antarctic notothenioid families I followed 

the same color scheme utilized in the previous sections, with notothens written in red, 

plunderfishes in green, and dragonfishes in purple (Figure 11B). Both visual depictions of the 

evolution of the proportion of centered nuclei in notothenioids revealed that many species 



 

44 
 

closely related to icefishes display abnormally low proportions of centered nuclei and thus a 

higher proportion of off-centered nuclei (Figure 11B). Three of the four dragonfish species 

(Gymnodraco acuticeps, Parachaenichthys charcoti, & Vomeridens infuscipinnis) and one of the 

two plunderfish species (Dolloidraco longedorsalis) I analyzed displayed abnormally low 

proportions of centered nuclei with values ranging from 14.4% (Vomeridens infuscipinnis) to 

22.2% (Gymnodraco acuticeps) (Figure 11B). The two outliers to this statement were the 

dragonfish Akarotaxis nudiceps with a value of 76.0% centered nuclei and the plunderfish 

Harpagifer antarcticus with a value of 85.0% centered nuclei (Figure 11B). Both of the two 

aforementioned species had high proportions of centered nuclei similar to the notothens, which 

had an average of 82.0% percent of centered nuclei and ranged from 59.6% (Trematomus 

bernacchii) to 98.7% (Gobionotothen gibberifrons) (Figure 11B). These analyses display a clear 

separation between the high proportions of centered nuclei in the notothens and the low 

proportions of centered nuclei in one of the plunderfishes and the majority of studied 

dragonfishes (which are evolutionarily closely related to icefishes).  



 

45 
 

 



 

46 
 

Figure 11. Abnormally Located Nuclei in Antarctic Notothenioids - A) Ancestral state 

reconstruction of proportion of centered nuclei in Antarctic fish using the contmap function and a 

color gradient with red representing low proportion (~14.4%) and blue representing high 

proportions (~98.7%). Located to the left of the plot are microscopic images of cells from the 

species Trematomus eulepidotus (top image - possessing centered cells), Dolloidraco 

longedorsalis (middle image - possessing mainly off-centered cells), and Gymnodraco acuticeps 

(bottom image - possessing mainly off-centered nuclei). The species closer in evolutionary 

relationship to icefishes (dragonfish and plunderfish families) are written in bold font. B) 

Phenogram based on the proportion of centered nuclei values of various Antarctic fish species. 

Notothen species names are in red, plunderfishes are in green, dragonfishes are in purple, and 

icefishes are in cyan. The line through the middle represents the threshold of 50% used to classify 

a species as having centered cells or off-centered cells. C) Based on the 50% threshold in the 

proportion of centered nuclei data, species were classified as having either centered or off-centered 

cells. Each node marks the likelihood that the ancestral nucleus placement was either centered 

(black) or off-centered (pink). The final nodes next to the species names represent the present 

classification of the species as having centered or off-centered nuclei. D) Disparity-through-time 

plot for the evolution of notothenioid proportions of centered nuclei. The solid line represents the 

observed disparity, and the dotted line is the evolution predicted by the BM model of evolution. 

The shaded gray area in the background represents the 95% confidence interval of the predicted 

evolution of disparity under the BM model of evolution. E) Model testing of the Early Burst 

model, Ornstein-Uhlenbeck model, and the Brownian motion model for the proportion of centered 

nuclei values F) Model testing for a phylogenetic signal with Blomberg’s K and Pagel’s λ. The p-

value for Blomberg’s L is based on 10000 randomizations and the p-value for Pagel’s λ is based 

on LR-test. Significant values (<0.05) are written in bold font. 

 

On the phenogram for the proportion of centered nuclei, I placed a dashed gray line to 

indicate the threshold of 50% centered nuclei used to determine if the species had mainly 

centered or off-centered nuclei (Figure 11B). This threshold was used to characterize the nucleus 

placement in the various families as a discrete character. After classifying the nucleus placement 

as a discrete character, I made an ancestral state reconstruction of the nucleus placement to 

analyze how this trait evolved over time using the function phenogram from the phytools 

package (Figure 11C). I forced the first node on the tree to be black, assuming that the first 



 

47 
 

ancestor to the Antarctic notothenioids had centered cells because this is the normal case in all 

fishes (Claver & Quaglia, 2009) (Figure 11C). The rest of the nodes display a prediction of the 

ancestral state of the species as they branched off from each other. The black portion of the 

nodes represents the likelihood the ancestor had mainly centered cells and the pink portion of the 

nodes represents the likelihood the ancestor had mainly off-centered cells (Figure 11C). This tree 

shows a clear distinction between the evolution of nucleus placement in notothens (where all 

have centered nuclei) and the evolution of nucleus placement in the plunderfish and dragonfish 

families (where most have off-centered nuclei) (Figure 11C). There appears to be a switch from 

centered nuclei to off-centered nuclei in the plunderfishes, with Harpagifer antarcticus classified 

as having centered nuclei and Dolloidraco longedorsalis classified as having off-centered nuclei 

(Figure 11C). All except one of the dragonfishes (Akarotaxis nudiceps) have off-centered nuclei 

as well (Figure 11C). This highlights the abnormalities that arise in nucleus placement in species 

evolutionarily most closely related to the icefishes (specifically, the plunderfish and dragonfish 

families). 

To further explore how the proportion of centered nuclei changed over time across the 

different families, I utilized the dtt function to generate a disparity-through-time plot of the 

character. Looking at the disparity-through-time plot of the proportion of centered nuclei (Figure 

11D), it is noticeable that the disparity is slightly higher than the dotted line (which predicts 

disparity based on the Brownian motion model). However, the observed disparity does remain 

within the confidence interval of the BM model of evolution, meaning that the proportion of 

centered nuclei in our notothenioid dataset largely evolved in a random, unconstrained manner 

across the phylogeny. Furthermore, our proportion of centered nuclei MDI value of 0.19 with a 

non-significant p-value of 0.78 suggests that proportion of centered nuclei in notothenioids 



 

48 
 

varied as much between subclades as within clades. This is apparent in the large variation of 

centered nuclei proportions observed in the plunderfish and dragonfish families where two 

species (Harpagifer antarcticus & Akarotaxis nudiceps) are separated from their families and 

mixed in with the proportion of centered nuclei averages characteristic of the notothens (Figure 

11B). 

Next, I compared the fit of each of the three evolutionary models to the data using the 

AICc. In the case of the proportion of centered nuclei, since the Brownian Motion model has the 

lowest AICc value it is the best at predicting the proportion of centered nuclei evolution in 

notothenioids (Figure 11E). Thus, this further indicates that the proportion of centered nuclei 

evolved randomly, incrementally, and in an unconstrained manner among the studied species. 

Finally, to assess whether there is a significant phylogenetic signal in my data, I 

calculated Blomberg's K and Pagel’s λ to estimate the strength of the phylogenetic signal in my 

data set relative to what is expected with Brownian motion. Based on my model testing on the 

proportion of centered nuclei data (Figure 11F), I found that neither Blomberg’s K nor Pagel’s λ 

were significant, suggesting that there is not a phylogenetic signal in the evolution of the 

proportion of centered cells in my data. This indicates that the evolution of the proportion of 

centered nuclei in the observed species is not strongly correlated with their evolutionary 

relationships and that closely related species may be more distinct to one another than to 

distantly related species. This is apparent when looking at the phenogram of centered nuclei 

(Figure 11B), where although the majority of the dragonfishes are grouped near the bottom of the 

plot, there is an outlier intermixed with the notothens. A similar trend is observed in the 

plunderfishes where although one species is placed at the bottom of the phenogram (Dolloidraco 

longedorsalis) due to its low proportion of centered nuclei, the other plunderfish (Harpagifer 



 

49 
 

antarcticus) is intermixed with the notothens due to its high proportion of centered nuclei (Figure 

11B). Thus, there is a small group of species evolutionarily closely related to icefishes (in the 

plunderfish and dragonfish family) that are distinct from other species in their family and outside 

of the normal center nuclei placement of mature red blood cells (Figure 11B & 11F). 

 

 

 



 

50 
 

Discussion 

While it is known that icefishes have abnormal blood composition due to their lack of 

mature red blood cells (RBCs) and hemoglobin (Beck et al., 2022), little to no research has been 

done on the specific morphology of their remaining immature RBCs outside a few analyses 

observing only a few species at a time (e.g., Spillman & Hureau, 1967). Research into the blood 

morphology of their Antarctic red-blooded relatives is also mostly lacking. The scientific 

literature on icefish blood has mainly focused on the physiological adaptations (such as a larger 

heart and increased cardiac output) that have developed in the Antarctic notothenioids and 

specifically in the icefishes to compensate for their lack of RBC production (Eastman, 1993; di 

Prisco et al., 2007; Verde et al., 2006; Matschiner et al., 2015), but there has not yet been much 

investigation into the characterization of erythropoiesis arrest in icefishes and the steps that led to 

it, outside of one study on the evolution of erythropoietic genes in notothenioids (Daane et al., 

2020). However, there is still much to be discovered about the changes that occurred in the 

icefish ancestor to cause the loss of hemoglobin genes and RBC maturation, specifically 

concerning which of the two might have happened first. If the loss of hemoglobin genes occurred 

first in the icefish ancestor, then the RBC would have become mostly useless without their 

oxygen-carrying unit and thus could have triggered the arrest of RBC maturation. Inversely, if 

RBC maturation was detrimentally altered first then the blood cells would potentially be unable 

to transport oxygen and thus the hemoglobin genes could have become unnecessary. By looking 

at the close Antarctic relatives to icefishes I hoped to find evidence for my research question of 

whether the loss of RBC production could have triggered the loss of hemoglobin, or inversely, if 

the loss of hemoglobin genes triggered the arrest of RBC production in icefishes. 



 

51 
 

Can discontinuing RBC production be energetically advantageous? 

The Antarctic waters remain close to freezing temperatures year-round (at about negative 

1.8℃ for sea water), which has a well-known adverse effect on blood flow (Egginton, 1955; 

Hureau, 1977; Wells et al., 1989; Near et al., 2006; Sidell & O’Brien, 2006). Frigid temperatures 

present a problem for cardiovascular systems as delivery of oxygen to tissues may be impaired 

by the high viscosity of bodily fluids (Wells et al., 1989). Thus, it could be beneficial to reduce 

the overall blood proportion (hematocrit) in Antarctic species so that it is less viscous and 

facilitates blood flow. Polar fishes typically possess lower hematocrits than warmer-bodied 

fishes and even temperate zone species decrease their hematocrit during winter months (Powers, 

1974; Scholander & Van Dam, 1957; Sidell & O’Brien 2006). Hence, it is reasonable to wonder 

if diminishing the level of hematocrit and RBCs in icefishes would be a beneficial adaptation for 

the Antarctic environment. Along this line of thinking, a decrease in cell proportions and the 

eventual loss of RBC maturation would lower the amount of energy needed to push the blood 

throughout the body. As shown in my data, icefish have significantly lower hematocrit values 

(values ranging from 1.23% to 2.58%) than the red-blooded Antarctic notothenioid (values 

ranging from 12.4% to 39.0%) (Figure 6). If the hypothesis that a loss of RBC production is 

energetically favorable is true, then we would likely see a continuous reduction in hematocrit in 

lineages phylogenetically similar to the icefish family. In looking at my data, species belonging 

to the dragonfish family (Racovitizia glacialis, Akarotaxis nudiceps, & Vomeridens 

infuscipinnis) do appear to have the lowest hematocrit outside of the icefish family (values 

ranging from 12.4% in Vomeridens infuscipinnis to 17.1% in Gerlachea australis) (Figure 6). 

However, there are also notothens with lowered hematocrit values (ranging from 14.0% 

Aethotaxis mitopteryx to 19.5% in Trematomus pennellii) and dragonfish with high hematocrit 



 

52 
 

values (ranging from 20.0% in Bathydraco macrolepis to 29.5% in Parachaenichthys charcoti) 

(Figure 6). Furthermore, the range of hematocrit values does appear to mostly range from 12% to 

26%, with some species from various families presenting at higher levels, so dragonfish do not 

have abnormally low values compared to the other species (Figure 6).  

Due to the wide variation of hematocrit values occurring in species from the same family, 

the red-blooded Antarctic notothenioids do not appear to be experiencing a reduction in 

hematocrit as species get evolutionarily more closely related to icefishes (Figure 6). The 

significant Blomberg’s K and Pagel’s λ value found for my hematocrit data, indicates that 

closely related species have more similar hematocrit values than to distantly related species and 

is reflective of the clear distinction between the icefishes and the rest of the lineages but does not 

indicate that families evolutionary more similar to icefishes begin to experience lowered 

hematocrit (Figure 6). Additionally, since Antarctic notothenioids already present a generally 

lowered hematocrit, reduced hemoglobin concentrations, and lowered hemoglobin affinity for 

oxygen compared to other warmer environment fishes (Desvignes et al., 2023; Eastman, 1993; di 

Prisco et al., 2007; Verde et al., 2006), the need to reduce hematocrit further by eliminating RBC 

maturation may not have been necessary. Moreover, even though the lack of RBCs in icefishes 

causes their blood to be reduced to half the viscosity of the red-blooded notothenioid blood 

(Twelves, 1972), the notothenioid plasma already has a lower viscosity than both human and 

icefish plasma (Wells, 1990) which may already compensate for this difference. Also, due to 

their frigid environment, the Antarctic waters have almost 100% oxygen saturation (Verde et al., 

2006), which may be another reason why having more viscous blood may not present a large 

problem if the Antarctic fish already have an abundance of oxygen available and thus do not 

need to move blood around as quickly. (Egginton, 1996). Therefore, there appears to be a lack of 



 

53 
 

evidence for a complete reduction in RBC being advantageous (Wells, 1990; Egginton, 1996; 

Sidell & O’Brien., 2006). 

Another question to be asked is: if a reduced hematocrit and lack of RBC maturation 

were indeed a beneficial adaptation then why are other adaptations needed to compensate for the 

loss? There are many energetically expensive physiological differences that developed in icefish, 

such as very large hearts compared to other red-blooded fishes of comparable body size (Sidell 

& O’Brien, 2006). Their increased heart size is necessary to pump their increased blood volume 

throughout their large blood vessels and extended capillary network, which are all physical 

modifications icefish have evolved (Desvignes et al., 2023). Altogether, the cardiac output of 

icefishes is four- to five times greater than that of red-blooded species of similar size (Sidell & 

O’Brien, 2006). These physiological differences seen in icefishes may be evidence that such 

energetically expensive adaptations were necessary to compensate for RBC loss and thus 

indicate that the detrimental effects of losing RBC outweigh any possible benefits a reduction in 

viscosity would bring. Furthermore, if the loss of hemoglobin genes and mature RBC in icefish 

was adaptive then it raises the question of why have no other lineages evolved a similar 

eradication? Therefore, the required adaptations needed to compensate for a lack of RBCs in 

icefishes and the lack of occurrence of this loss in other Antarctic notothenioid lineages means it 

is unlikely that the lack of RBC was an advantage over the other notothenioids, and the loss was 

probably maladaptive but occurred because of other reasons. 

If maladaptive, why did RBC production become discontinued? 

Due to the unlikelihood of icefishes stopping RBC production under the motivation that it 

was more energetically favorable, it is important to hypothesize other theories for why the loss of 

RBC production and hemoglobin genes occurred as well as the steps that led to this arrest. When 



 

54 
 

looking at my data, I noticed that some species closely related to icefishes have RBCs 

resembling cells characteristics of the human diseases’ spherocytosis and elliptocytosis. As seen 

in the literature work conducted by Daane (Daane et al., 2020) and Martins (Martins et al., 

2021), “normal” fish RBCs are typically relatively small (with a mean cell area of 76.64 µm2 

(Martins et al., 2021)) and oval (there is a lack of quantitative values in the literature but the 

notothens and plunderfish in my data averaged at 0.7 roundness (Figure 8 & 10)). The majority 

of the fish in my analyses had a slightly larger cellular area than the indicated value, with 

notothens averaging at 91 µm2 and plunderfishes averaging at 92 µm2 (Figure 8), but they still 

maintain the standard features of a small area and oval shape, especially exemplified in the 

notothens Trematomus bernacchii (with a cellular area of 73.9 µm2 and a roundness of 0.71) and 

Lepidonotothen squamifrons (with a cellular area of 74.6 µm2 and a roundness of 0.65) (Figure 8 

& 10). The plunderfishes also had relatively “normal” looking cells, similar to the notothens, as 

seen in the species Dolloidraco longedorsalis (cellular area of 91.8 µm2 and a roundness of 0.76) 

and Harpagifer antarcticus (cellular area of 91.9 µm2 and a roundness of 0.65) (Figure 8 & 10). 

However, in the dragonfish family, which is phylogenetically most closely related to icefish, 

variability in the cellular morphology is important. For instance, the cells in Gymnodraco 

acuticeps have a large area (108.4 µm2) and a slightly higher than average roundness (0.80), 

Akarotaxis nudiceps cells are the largest cells (116.0 µm2) and also the most oblong (roundness 

value of 0.60), and in contrast Parachaenichthys charcoti cells are relatively small and the most 

round (0.90). Therefore, some of the dragonfish species begin to have cells resembling the quite 

large and round immature RBCs of icefish. Furthermore, the abnormalities present in the cells of 

the species Akarotaxis nudiceps (which are very large and ovular) resemble the human disease 



 

55 
 

elliptocytosis and cells of the species Parachaenichthys charcoti (which are very round) 

resemble the human disease spherocytosis.  

The aforementioned human diseases of spherocytosis and elliptocytosis, which resemble 

the abnormal cellular morphology found in some dragonfish species, are often caused by genetic 

mutations that lead to improper formation of the cellular membrane. The genetic defects causing 

these two human diseases typically occur in the shared locations of α-spectrin and β-spectrin in 

the SPT1A and SPTB genes, which both play major roles in the shaping and stability of the RBC 

membrane (Perrotta et al., 2008; Zivot et al., 2017; He BJ et al., 2018). Additionally 

spherocytosis can also be caused by defects in Ankyrin 1 (in the gene ANK1 – which affects the 

attachment of the transmembrane proteins to the cell membrane skeleton) and/or the band 3 

protein (in the gene SLC4A1 – which affects the linkages to ankyrin and the cytoskeleton 

network) (Perrotta et al., 2008; Zivot et al., 2017; He BJ et al., 2018). All of the aforementioned 

mutations affect the cell membrane of the cytoskeleton which leads to malformed cells that have 

increased fragility and are more susceptible to bursting (Perrotta et al., 2008; Zivot et al., 2017; 

He BJ et al., 2018). The similarities between the abnormally formed RBC in certain dragonfish 

species and those formed in the aforementioned human diseases raise the question: could 

dragonfish be suffering from the same genetic mutations underlying those in elliptocytosis and 

spherocytosis? Furthermore, could additional genetic mutations have accumulated in the 

cytoskeleton genes of the icefish ancestors, making them even more fragile and leading to 

detrimentally high levels of hemolysis? 

Another abnormality I observed in the cells of Antarctic notothenioid closely related to 

icefishes is an improper placement of the cell nuclei, which could be indicative of a genetic 

mutation in the intracellular cytoskeleton. In addition to having a smaller area and an oval shape, 



 

56 
 

“normal” fish RBCs also stereotypically have a nucleus located in the middle of the RBC (Fig. 3, 

Martins et al., 2020). However, I noticed that some dragonfish and plunderfish species I analyzed 

had off-centered nuclei rather than the expected centered nuclei (Figures 10 & 11). This is a 

finding that has never been noticed before, although it is visible in one scientific paper (Daane et 

al., 2020). Furthermore, there is no literature on RBC nuclei placement in human RBC because 

they are enucleated, so we do not know if abnormally placed nuclei are detrimental to the 

function of RBC. However, there is evidence that in some cell types a poorly placed nucleus may 

be pathogenic, such as in the photoreceptor when the NESPRIN genes (specifically 

SYNE2/Nesprin2) and LINC complex proteins are mutated (Razafsky et al., 2012; Maddox et 

al., 2015). The SYNE2/Nesprin2 gene is a component of the linker of nucleoskeleton and 

cytoskeleton (LINC) complexes and performs a role in joining the nucleus to the cytoskeleton 

(Maddox et al., 2015). When mutations arise in NESPRIN genes and disrupt the LINC complex 

function then normal retinal function is altered (Razafsky et al., 2012). Thus, mislocalization of 

the nuclei could possibly disrupt neuronal functions and underlie both human retinal diseases and 

progressive neuronal disorders (Razafsky et al., 2012; Maddox et al., 2015). Consequently, it is 

possible that abnormally placed nuclei in RBCs of fish may decrease or alter their normal 

function as well. Moreover, the abnormally off-centered placement of the cell nuclei in certain 

species (belonging to the dragonfish and plunderfish families) suggests they may have mutations 

in the genes controlling nuclei placement. It is therefore possible that the genes/proteins 

controlling nuclei placement also influence the cellular shape and capability to handle 

deformations when traveling through capillaries, such as the disadvantages genetic mutations 

occurring in elliptocytosis and spherocytosis. Although we do not yet know whether the 

abnormal nuclei placement is pathogenic, this observation suggests that some alterations to RBC 



 

57 
 

maturation began in the common ancestor of plunderfish, dragonfish, and icefish. Additionally, 

the timing of the arising abnormalities also corresponds to when hemoglobin multiplicity in 

Antarctic notothenioids went from one major hemoglobin and on to three minor hemoglobins to 

a single hemoglobin isoform in adult plunderfishes and dragonfishes (Desvignes et al., 2023). 

Could these two timings be related to each other or are they purely coincidental? This new 

finding raises the question of whether or not genetic mutations in closely related species to 

icefishes began to accumulate until the cells started to present such extreme deformities that it 

was energetically more favorable to completely get rid of the RBC maturation process and 

hemoglobin genes then to try to fix the genes.  



 

58 
 

Conclusion 

To wrap up my findings, I found that there is a significant distinction between the 

hematocrit and blood composition found in red-blooded notothenioid and the icefish; I noticed 

that larger RBC area and RBC roundness begin to arise in species from the dragonfish family; 

and lastly, I observed abnormally placed nuclei  in species belonging to the plunderfish and 

dragonfish families (Figures 1-6). In my discussion, I also argue that although the low hematocrit 

observed in the icefish may have been beneficial in reducing blood viscosity, the adaptations 

needed to offset the loss of hemoglobin and RBCs were energetically costly and likely offset any 

advantages obtained. Thus, the loss of RBCs and hemoglobin in icefish was indeed likely 

maladaptive.  

The abnormal cellular area and roundness that arise in certain dragonfish species 

resemble cells in humans suffering from pathogenic diseases such as hereditary spherocytosis 

and elliptocytosis, which are caused by genetic mutations affecting the cellular membrane. The 

abnormally located nuclei in some plunderfish and dragonish species resemble the 

mislocalization of nuclei occurring in photoreceptors due to genetic mutations and having 

pathogenic effects. We do not yet know whether the alterations to the nuclei placement in fish 

red blood cells alter cell function but based on the consequences arising in photoreceptors, it has 

the potential to be detrimental. Therefore, it is possible that the dragonfish and plunderfish 

species developed multiple genetic mutations affecting their cellular membrane and 

cytoskeleton, which led to the abnormally large area and roundness observed in some dragonfish 

species and the off-centered nuclei I observed in several dragonfish and plunderfish species. 

When genetic defects cause RBCs to develop abnormal cellular shape (such as large area, large 

roundness, or being very ovular) these mutations often occur in the cell membrane, leading to 



 

59 
 

weaker membranes more inclined to the cell breaking and releasing hemoglobin into the blood 

(hemolysis) (Barcellini et al., 2011). Furthermore, the bigger and the more round a RBC is, the 

more difficult it is for these cells to fit through capillaries and efficiently transport blood, often 

leading to malformed cells being squeezed and bursting in the process (Barcellini et al., 2011). 

When cells undergo hemolysis and rupture in the bloodstream, hemoglobin is then released and 

freely floats into the blood plasma (Barcellini et al., 2011). Under normal conditions in RBCs, 

hemoglobin is a heterotetramer made up of two alpha- and two beta-globin chains, but when it is 

dissolved directly in the blood plasma it dissociates into heterodimers, which are toxic for 

organisms (Vallelian et al., 2022). The dimers can be translocated across tissue barriers and 

negatively impact critical organs and muscles, eventually leading to vasoconstriction and renal 

injury (Vallelian et al., 2022). Thus, the aforementioned mutations may have accumulated and 

gotten worse in the icefish ancestor until RBC were completely incapable of traveling throughout 

the body to deliver oxygen without bursting and poisoning the organism. Under these 

circumstances, if RBC were incapable of holding hemoglobin due to instability in the RBC cell 

membrane then it may have been better to completely get rid of the hemoglobin genes rather than 

experience the constant toxicity of the blood upon hemolysis. Under this hypothesis, the best 

option for survival would have been for icefish to completely rid themselves of the hemoglobin 

genes rather than trying to fix them, which is evolutionarily unlikely. Furthermore, if the genes 

were “broken” before the loss of RBC then the heterodimers would likely not have formed in the 

first place, which means there would have been no toxicity and thus no need for the icefish to 

lose their RBCs. Thus, the RBC maturation process was likely altered to the point of being 

detrimental and unfunctional due to genetic mutations, which triggered the arrest of hemoglobin 

genes in the icefish ancestor. This maladaptive phenotype of stopping RBC maturation and 



 

60 
 

eliminating hemoglobin genes, which requires many energetically unfavorable physiological 

compensatory adaptations, may have been the only way for icefish to survive. Likely, the only 

reason that icefish remained alive in the evolution of this maladaptive phenotype is because of 

the high oxygen saturation of the Antarctic waters and their extremely limited amount of niche 

competition (Sidell & O’Brien, 2006); without these unique circumstances the species 

presumably would have gone extinct.  

Since I have found connections between the cellular morphology characteristic of the 

human diseases hereditary elliptocytosis and spherocytosis and the cellular abnormalities arising 

in the dragonfish species, a future direction of research to pursue is a deeper analysis of the 

genetic similarities underlying those diseases in humans and the genes found in icefish. In taking 

a closer look at the genetic similarities between the cellular abnormalities more research may 

lead to further classification of the genetic mutations likely arising in lineages phylogenetically 

close to icefishes and leading to the eventual arrest of RBC production. Furthermore, icefishes 

are an evolutionary mutant model that could be highly beneficial for observing and 

characterizing anemias in humans, possibly leading to better therapeutics in the future if more 

information on the genetics involved are obtained. A second future direction to pursue based on 

my research would be taking a closer look at the RBC maturation process in Antarctic species. 

By investigating the RBC maturation process more closely, research may reveal when anomalies 

arise, which could provide more information on the loss of RBC in icefishes and potentially the 

timing involved. Overall, due to the fascinating blood morphology of icefishes, looking more 

closely at their phylogenetically related species may provide even more information into the 

strange evolution of these Antarctic fish and ultimately lead to discoveries relevant to human 

diseases as well. 



 

61 
 

Bibliography 

Barber, D.L. et al. (1981) The blood cells of the Antarctic icefish Chaenocephalus aceratus 
Lönnberg: light and electron microscopic observations. J. Fish Biol. 19, 11–28 

Beck, E. A., Healey, H. M., Small, C. M., Currey, M. C., Desvignes, T., Cresko, W. A., & 
Postlethwait, J. H. (2022). Advancing human disease research with fish evolutionary 
mutant models. Trends in genetics:TIG, 38(1), 22–44. 
https://doi.org/10.1016/j.tig.2021.07.002 

Beers, J. M., Borley, K. A., & Sidell, B. D. (2010). Relationship among circulating hemoglobin, 
nitric oxide synthase activities and angiogenic poise in red- and white-blooded Antarctic 
notothenioid fishes. Comparative Biochemistry and Physiology Part A: Molecular & 
Integrative Physiology, 156(4), 422–429. https://doi.org/10.1016/j.cbpa.2010.03.027 

Bista, I., Wood, J. M. D., Desvignes, T., McCarthy, S. A., Matschiner, M., Ning, Z., Tracey, A., 
Torrance, J., Sims, Y., Chow, W., Smith, M., Oliver, K., Haggerty, L., Salzburger, W., 
Postlethwait, J. H., Howe, K., Clark, M. S., Detrich, H. W. 3rd., Cheng S.-H. C., Miska, 
E,