Is Diet Correlated with Feeding Morphology in Neotropical ... · The suckermouth armoured catfishes (Siluriformes: Loricariidae) are a taxonomically diverse group of benthic fish

Is Diet Correlated with Feeding Morphology in Neotropical Suckermouth Armoured Catfishes (Siluriformes:

Loricariidae)?

by

Stéphanie L. Lefebvre

A thesis submitted in conformity with the requirements for the degree of Masters of Science

Ecology and Evolutionary Biology University of Toronto

© Copyright by Stephanie L. Lefebvre 2014

ii

Is Diet Correlated with Feeding Morphology in Neotropical Suckermouth Armoured Catfishes (Siluriformes: Loricariidae)?

Stéphanie L. Lefebvre

Masters of Science

Ecology and Evolutionary Biology

University of Toronto

2014

Abstract

The suckermouth armoured catfishes (Siluriformes: Loricariidae) are a diverse group of

predominantly detritivorous fishes inhabiting rivers of South and Central America. Their distinct

jaws are specialized for scraping a wide range of substrate. Though it is hypothesized that

specialization of loricariid feeding morphology may have played a role in their diversification,

little is known about the ecological and evolutionary processes involved. The present study

quantifies variation in jaw functional traits and body morphology in a phylogenetic context.

Morphological variation was compared to species’ relative position in assemblage isotope space

(for both δ13

C and δ15

N) to test for correlations between diet and feeding morphology. Results

show that although jaw functional traits are decoupled from body morphology, both are

correlated with δ13

C values. Partitioning of terrestrial and aquatic resources can be explained by

both jaw functional traits and body morphology, however further partitioning of diet is only

attributed to the former.

iii

Acknowledgements

I would like to thank Nathan K. Lujan for providing isotope and jaw data, along with a wealth of

information about loricariids, and for organizing the provision of specimens from both the

Auburn University Museum Fish Collection (AUM) and the Academy of Natural Sciences of

Philadelphia (ANSP). I am deeply grateful to my supervisors, Hernán López-Fernández and

Nathan Lovejoy, for guidance, insight, and funding. My supervisory committee members, Jason

Weir and Don Jackson, for helpful discussion and guidance. The curators of the fish collections

at their respective institutions for allowing me to dissect their specimens – Jon Armbruster

(AUM), Mark Sabaj Pérez (ANSP), and Hernán López-Fernández (ROM). I am indebted to

Sarah Steele for being a great sounding board to my crazy ideas, for providing the outline to

Figure 4 (Methods, section 1.2.6), and for always being there for support. Finally, I’d like to

thank the López-Fernández lab members for all the discussions, insight, R code help, and laughs

along the way. You guys are awesome.

iv

Table of Contents

Acknowledgements ........................................................................................................................ iii

Table of Contents ........................................................................................................................... iv

List of Tables ................................................................................................................................. vi

List of Figures ............................................................................................................................... vii

List of Appendices ....................................................................................................................... viii

1.1 Introduction ......................................................................................................................... 1

1.2 Methods ............................................................................................................................... 4

1.2.1 Samples ................................................................................................................... 4

1.2.2 Jaw Dissection and Preparation .............................................................................. 4

1.2.3 Imaging ................................................................................................................... 5

1.2.4 Jaw Parameters ........................................................................................................ 6

1.2.5 Jaw Functional Traits .............................................................................................. 7

1.2.6 Body Morphology ................................................................................................... 8

1.2.7 Phylogenetic Corrections ...................................................................................... 10

1.2.8 Diet Information .................................................................................................... 12

1.2.8.1 Diet Categories ....................................................................................... 12

1.2.8.2 Stable Isotope Analysis .......................................................................... 13

1.3 Results ............................................................................................................................... 14

1.3.1 Significant Principal Components Test ................................................................ 14

1.3.2 Jaw Functional Traits Analyses ............................................................................ 15

1.3.3 Body Morphology Analysis .................................................................................. 18

1.3.4 Diet-Morphology Correlations .............................................................................. 21

v

1.4 Discussion ......................................................................................................................... 23

1.4.1 Jaw Functional Traits Analyses ............................................................................ 23

1.4.2 Body Morphology Analysis .................................................................................. 24

1.4.3 Diet Groups ........................................................................................................... 24

1.4.3.1 Algae Scrapers (Gray) ............................................................................ 24

1.4.3.2 Wood-Eaters (Purple) ............................................................................. 25

1.4.3.3 Insect and Mollusk Eaters (Pink) ........................................................... 26

1.4.3.4 Aufwuch Eaters (Yellow) ....................................................................... 27

1.4.4 Habitat Use ............................................................................................................ 28

1.4.5 Diet-Morphology Correlations .............................................................................. 29

1.4.5.1 Relationship Between δ13

C and Jaw Functional Traits .......................... 29

1.4.5.2 Relationship Between δ13

C and Body Morphology ............................... 30

1.4.5.3 Implications of Diet-Morphology Correlations in Loricariids ............... 30

1.5 Conclusions ....................................................................................................................... 31

References ..................................................................................................................................... 33

Appendix A ................................................................................................................................... 39

Appendix B ................................................................................................................................... 45

vi

List of Tables

Table 1. Parameters quantifying morphological variation in loricariid mandibles ........................ 6

Table 2. Morphological variables used to quantify aspects of body size and shape in loricariids. 9

Table 3. Diet information identified for a subset of loricariids in this study, based on references

to diet in literature. ........................................................................................................................ 12

Table 4. Species abbreviations used in all analyses.. .................................................................... 15

Table 5. Principal components analysis eigenvectors for jaw functional traits and body

morphology.. ................................................................................................................................. 16

Table 6. Results of PGLS regressions using centroid deviations of δ13

C and δ15

N against

residuals of the first two principal components of the jaw functional traits (f) and body

morphology (b) . ........................................................................................................................... 21

vii

List of Figures

Figure 1. Representative sample of lower jaws from Loricariidae examined in this study

showcasing morphological diversity in the family. ........................................................................ 2

Figure 2. Orientation of lower jaw ramus of Hypostomus hemiurus.. ............................................ 5

Figure 3. Mandible of Hypostomus macushi illustrating parameters used in this study, as defined

in Table 1. . .................................................................................................................................... 7

Figure 4. Visual representation of body measurements used to evaluate size and shape

differences in the present study.. ................................................................................................... 10

Figure 5. Multi-locus molecular phylogeny of a subset of loricariids, sequenced by Lujan et al.

(unpublished data).. ....................................................................................................................... 11

Figure 6. PC axis significance test for jaw functional traits analysis.. ......................................... 14

Figure 7. Graphic representation of the first two principal components of jaw functional traits of

a phylogenetically-corrected PCA, by means of morphospace (upper panel) and

phylomorphospace (lower panel).. ................................................................................................ 17

Figure 8. Graphic representation of the first two principal components of body morphology of a

phylogenetically-corrected PCA, by means of morphospace (upper panel) and

phylomorphospace (lower panel).. ................................................................................................ 20

Figure 9. Estimate of concentration of δ13C as a function of jaw functional traits.. ................... 22

Figure 10. Estimate of concentration of δ13

C as a function of body morphology.. ...................... 23

Figure 11. Despite their different body shapes, head morphology appears convergent in

representatives of Lamontichthys and Chaetostoma.. ................................................................... 25

viii

List of Appendices

Appendix A List of species prepared for this study, as well as catalogue numbers of the institutions from

which they came ........................................................................................................................... 39

Appendix B Species used in the stable isotope analyses.. ........................................................................ 45

1

1.1 Introduction

The suckermouth armoured catfishes (Siluriformes: Loricariidae) are a taxonomically

diverse group of benthic fish inhabiting rivers of South and Central America. With over 900

known species, loricariids are the most species-rich family of catfishes (Eschmeyer and Fricke,

2011). Loricariids are characterized by a dorsoventrally depressed body covered in ossified

dermal plates instead of scales, and their mouths are ventrally positioned as an oral disk. This

positioning of the mouth allows these fishes to attach to the substrate, which facilitates their

scraping mode of feeding. This suckermouth-scraping mode has evolved independently in

Neotropical Loricarioidea and African Mochokidae catfish from generalized benthic suction

feeders (Van Wassenbergh et al. 2008). Loricariid teeth are also ventrally oriented and their

flexibility is unique among vertebrates (Geerinckx et al. 2012). The upper jaws, which are

composed of two tightly linked premaxillae, are highly protrusible and move independently of

the lower jaws. Each ramus of the lower jaw is medially decoupled, allowing them to also move

independently from each other (and the upper jaw) while scraping (Geerinckx et al. 2009;

Adriaens et al. 2009).

Lujan and Armbruster (2012) measured aspects of jaw morphology and proposed a

biomechanical model to predict how morphological variation is linked with functional diversity

in loricariids. The present study includes 48 species (27 genera), and incorporates the functional

traits established by Lujan and Armbruster (2012) in order to explore loricariid jaw diversity in a

phylogenetic context. The phylogeny used is a subset of a newly established molecular

phylogeny of the Loricariidae in preparation by Lujan et al (unpublished data).

2

Figure 1. Representative sample of lower jaws from Loricariidae examined in this study

showcasing morphological diversity in the family. A) Baryancistrus xanthellus, B) Dekeyseria

scaphiryncha, C) Hypancistrus furunculus, D) Hypostomus macushi, E) Rineloricaria fallax, F)

Leporacanthicus triactis, G) Chaetostoma sp. Xingu, H) Lithoxus lithoides, I) Scobinancistrus

pariolispos, J) Sturisoma monopelte

In the present study, jaw morphological data were combined with dietary (isotope) data in

order to explore potential links between trophic niche and feeding morphology. A link between

diet and feeding morphology would be especially interesting in loricariids because despite their

diverse jaws, they predominantly feed on detritus (dead organic matter) or algae. Increasingly,

studies are linking these diet and jaw morphology in Neotropical fishes (Fugi et al. 2001;

Novakowski et al, 2004; López-Fernández et al. 2012; Montana and Winemiller, 2013), but

most studies focus on predators (piscivores or insectivores), for which prey items are discrete

and taxonomically distinguishable to the eye. Such diet discrimination is near impossible for

detritivores since their stomach contents are often homogenous and undiscernible (Hood et al.

2006; Lujan et al. 2011). This may account for the lack of research focusing on Loricariidae,

despite the fact that it is one of the most diverse families of Neotropical freshwater fish, second

only to Characidae (Reis et al, 2003). The few studies exploring diet-morphology correlations in

loricariids tend to focus on intestinal morphology (Kramer and Bryant, 1995; Delariva and

Agostinho, 2001; German et al. 2010), although limited studies of wood-eating loricariids have

linked jaw morphology to both gut contents and isotopically defined diet patterns in a non-

phylogenetically explicit manner (Armbruster 2003; Nogonaki et al. 2007; Lujan et al. 2011).

Dietary information for this study is based on stable isotope data from Lujan et al (2012).

Isotopes can give insight into dietary intake averages over time, and are especially useful in

detritivores, where the maceration and fast uptake of food results in traditional stomach content

3

analyses not being suitable for discerning individual food items in the gut. The majority of

loricariids’ diet consists of a combination of algae and detritus, with occasional ingestion of

macroscopic plant matter, insects and mollusks (Lujan et al. 2012).

Stable isotope analyses quantify the ratio between naturally occurring isotopes of an

element (typically C and N in diet studies). Since one isotope has more neutrons they differ in

atomic mass, with the heavier of the two isotopes being the most stable. For isotopic trophic

studies, the ratio of these isotopes is measured and compared with a standardized measurement

for that element – if the ratio of heavy to light element is higher than that of the standard, the

sample is said to be enriched (Ben-David and Flaherty, 2012). A ratio lower than the standard for

that element is considered depleted. In their analysis, Lujan et al. (2012) used the ratios between

13C and

12C, as well as the ratios between

15N and

14N to infer dietary discrimination among

sympatric loricariids.

δ13

C signatures in freshwater ecosystems vary largely in response to sources of dissolved

organic carbon. Values from algae may be upwards of 25‰ higher than those of dissolved

inorganic carbon, such as calcite, for example (Peterson and Fry, 1987). However, δ13

C values

do not vary significantly between trophic levels and can therefore be used to infer the relative

contribution of different primary producers in the ecosystem to a consumer’s biomass (Peterson

and Fry 1987; Ben-David and Flaherty, 2012). For example, ratios of δ13

C can help differentiate

between autochthonous (aquatic) and allochthonous (terrestrial) sources of carbon (Peterson and

Fry 1987), and previous studies used δ13

C values to determine that wood-eating loricariids

(Hypostomus cochliodon group, as well as species of Panaque and Panaqolus) have a different

δ13

C signal than those that strictly rely on autochthonous carbon sources (Nogonaki et al. 2007;

German and Miles, 2010; Lujan et al. 2012).

δ15

N signatures can be used to infer trophic level since 15

N is enriched along the chain of

consumers. Of the two naturally occurring forms of nitrogen, biochemical reactions responsible

for the assimilation amino acids and proteins prefer the heavier and rarer 15

N. For most

organisms, this enrichment of 15

N increases by 3 to 5‰ with each successive trophic level

(Minagawa and Wada 1984). It is important to note that as loricariids occupy the same trophic

level (German and Miles, 2010), the difference in values of δ15

N in this study are expected to be

minimal. However, these values will still be biologically relevant and may give insight into both

4

the proportion of protein in the diet (Kelly and del Rio, 2010), and niche partitioning of resources

(Layman et al. 2007).

In this study, I will first quantify variation in both jaw functional traits and body

morphology in a phylogenetic context. Second, I will test for correlations between jaw functional

traits and diet (based on assemblage standardized variation in δ13

C and δ15

N stable isotope data),

as well as body morphology and diet. Links between diet and either jaw function or body

morphology would suggest that these features can predict ecological traits. This, in turn, would

support an adaptive relationship between phenotype and feeding behavior in loricariids.

1.2 Methods

1.2.1 Samples

This study examined 166 specimens representing 48 species and 27 genera of loricariids.

Data on 66 specimens (14 species, 9 genera) were provided by Lujan and Armbruster (2012),

and novel data on 100 specimens (34 species, 18 genera) were gathered for this study. New

specimens examined were provided by the Auburn University Museum Fish Collection (n=85),

the Royal Ontario Museum (n=8), and the Academy of Natural Science of Philadelphia (n=7).

Only adult specimens were included in order to reduce confounding effects of allometric changes

during ontogeny on the dataset. See Appendix A for the list of specimens used in this study,

including geographical information regarding where each sample was collected.

1.2.2 Jaw Dissection and Preparation

For each specimen, both the right premaxilla and right mandible were dissected, and

individually treated following the Maceration and Staining of Jaw Bones protocol used by Lujan

and Armbruster (2012). This protocol aims to remove all soft tissue while keeping bones,

ligaments, and teeth intact, and prepares the jaws to be photographed. Following dissection, jaw

elements were dehydrated in individual vials containing 95% ethanol. After at least 24 hours,

ethanol was replaced with a solution of 1% KOH and 0.5% alizarin red. KOH macerates soft

tissue surrounding the bone, while alizarin red stains ossified structures to highlight details on

the surface of the bone. Jaws were kept in the KOH-alizarin solution for 3 days, after which they

were returned to 95% ethanol for 24 hours. Any remaining soft tissue was then easily removed

using forceps, and the newly exposed bones were left to air dry. This protocol was modified

5

slightly with either small or large jaws by adjusting time spent in each solution accordingly. The

smallest jaws were given special attention since too much time spent in 1.0 % KOH would

weaken the cartilage and ligaments that bind the dentary to the anguloarticular, and the jaw

would separate in two pieces.

1.2.3 Imaging

Digital images of each mandible were taken with a Nikon D100 digital camera mounted on a

Zeiss Stemi SV8 stereomicroscope using Camera Control Pro 2 software. Images were taken in

at least 2 of the 3 perspectives outlined by Lujan and Armbruster (2012), with most species being

photographed in all 3 perspectives (see Figure 2). These perspectives maximize homology and

ensure that all jaw regions could be observed in a standardized way using the angular complex as

a reference point (Lujan and Armbruster, 2012). The angular complex is parallel to the field of

view in both the ventral-horizontal and dorsal-horizontal perspectives, while the vertical-dorsal

perspective features the angular complex perpendicular to the field of view. Certain jaws have a

morphology that allows all of the linear measurements to be clearly seen in two perspectives,

while others are shaped in a way that requires all three perspectives in order to measure all of the

parameters in this study. Five linear measurements and one surface area measurement were taken

digitally in ImageJ (Rasband 1997), with each image scaled using the established distances of the

graph paper used as a background for each image. The premaxilla was neither photographed nor

included in analyses because homology and functional relevance of all structures could not be

ensured.

Figure 2. Orientation of lower jaw ramus of Hypostomus hemiurus. The orientation is shown in

each plane, ventral-horizontal (A), vertical-dorsal (B), and horizontal-dorsal (C), as outlined in Lujan and

Armbruster (2012).

6

1.2.4 Jaw Parameters

Six measurements were taken for each mandible in accordance with Lujan and

Armbruster (2012)’s protocol (see Table 1 and Figure 3). Of these, three are analogous to the

input and output lever arms that are standard measurements of teleost feeding biomechanics

(Westneat, 2004), while the remaining three are loricariid-specific and were established by Lujan

and Armbruster (2012). For clarity, these measurements will henceforth be referred to as

parameters throughout the text. These parameters are used for the calculation of jaw functional

traits, described below in section 1.2.5.

Table 1. Parameters quantifying morphological variation in loricariid mandibles. It should be

noted that the output lever arm is only measured once in traditional feeding biomechanics studies (Westneat 2004),

whereas it involves two components in loricariids due to the ventral orientation of their mandible.

Parameters Definition Measurement

Traditional (Westneat, 2004)

In Input lever arm

Distance between the center of the area of insertion of the adductor

mandibulae muscle and the anguloarticular condyle

Outprox Output lever arm

Distance between the anguloarticular condyle and the

tooth closest to the condyle

Outdist Output lever arm Distance between the

anguloarticular condyle and the tooth furthest to the condyle

Loricariid Specific (Lujan and Armbruster 2012)

TRL Tooth row length Distance of the proximal to the

distalmost tooth insertions

H1 Variation in height of the coronoid

arch relative to the distal-most tooth

Perpendicular distance between the coronoid arch and the line

formed by the Outdist measurement

AMarea Area of insertion of the adductor

mandibulae muscle Area of insertion of the adductor

mandibulae muscle

7

Figure 3. Mandible of Hypostomus macushi illustrating parameters used in this study, as

defined in Table 1. Measurements were taken in multiple perspectives, not necessarily the one

shown in this figure.

1.2.5 Jaw Functional Traits

The jaw parameters described above were used to calculate five functionally relevant

traits, developed by Lujan and Armbruster (2012). These traits quantify aspects of feeding

biomechanics, and give insight into the mechanical advantages of the jaw system.

1. Mechanical advantage at the distal-most tooth was calculated as the quotient of the input

lever (In) and the distance between the distalmost tooth and the anguloarticular condyle (Outdist).

A high value reveals a strength optimized jaw, while a low value reveals a speed-optimized jaw.

2. Mechanical advantage at the proximal tooth was calculated as the quotient of the input

lever (In) and the distance between the proximalmost tooth and the anguloarticular condyle

(Outprox). A high value favours strength, while a low value favours speed.

3. The combined measure of torque and distribution of force transmitted through the

mandible which reaches the substrate was calculated as the quotient of the perpendicular distance

between the coronoid arch and the line formed by the Outdist measurement (H1) and the tooth

row length (TRL). This measure quantifies the concentration of force that makes contact with the

8

substrate while scraping. A high ratio shows a concentration of force, while a low ratio shows

distributed force.

4. The maximum force entering the mandible per unit substrate contacted was measured as

the area of insertion of the adductor mandibulae muscle (AMarea) divided by the squared value

of the tooth row length (TRL). A high quotient favours concentrated force, while a low ratio

favours distributed force.

5. Lastly, an indicator of jaw protrusion, mechanical advantage, and variation in torque

across the tooth row was measured as the angle (∡ x) formed where the tooth row length (TRL)

meets the distance of the distalmost tooth and the anguloarticular condyle (Outdist) (see Figure 3).

Small angles show minimal protrusion of the jaw, low strength and high torque, while bigger

angles favour the opposite with maximal protrusion of the jaw, maximized strength and low

torque.

1.2.6 Body Morphology

Body morphology was measured in this analysis using seven exterior linear measures

representing body size and shape in loricariids (Table 2, Figure 4). Data for Chaetostoma breve,

C. microps, Etsaputu relictum, Panaqolus nocturnus, and Panaque nigrolineatus were provided

by Nathan Lujan (unpublished data from Lujan et al. 2012).

9

Table 2. Morphological variables used to quantify aspects of body size and shape in loricariids.

1 It should be noted that due to the anatomy of their jaw structure, loricariid mouths cannot be closed (Geerinckx et

al. 2010), and so the measurement for mouth length may have been influenced by the degree which their mouths

were naturally open in a resting position based on the morphology of the dentary when the measurements were

taken.

Variable Measurement Distance between

lanmarks in Figure 4

Body Length

Standard Length Distance from the anterior margin of the snout to the base of the

caudal fin

1-2

Snout Length Distance from the anterior margin of the snout to the nares

1-12

Head Length Distance from the anterior margin of the snout to the posterior margin of the supraoccipital

1-4

Mouth Length (Lip Diameter)1 Distance from the anteriormost margin of the premaxilla to the posteriormost part of the lower

lip

12-13

Body Depth

Head Depth Vertical distance from ventrum to supraoccipital

4-5

Body Depth at Anal-Fin Insertion Vertical distance from ventrum to dorsalmost part of the body at the

anal-fin insertion

6-7

Body Depth at Caudal Peduncle Vertical distance at posterior margin of adipose fin insertion

8-9

Body Width

Body Width Body width at cleithrum 10-11

10

Figure 4. Visual representation of body measurements used to evaluate size and shape

differences in the present study. A) Lateral view, B) dorsal view, C) ventral view. Loricariid

body outlines provided by Sarah Steele.

1.2.7 Phylogenetic Corrections

All phylogenetic corrections were based on a subset of a newly developed multi-locus

molecular phylogeny of loricariids (Figure 5) based on one mitochondrial gene (Cytb) and two

nuclear genes (RAG1 and RAG2) (Lujan et al, unpublished data). This phylogeny was made

ultrametric in Mesquite (Maddison and Maddison 2011), with total branch lengths from the root

of the tree to any given tip being set to 1 since the phylogeny is not yet time-calibrated. Branches

were pruned to correspond with taxa available for the jaw functional traits and body

morphometric analyses. It should be noted that this phylogeny has uncovered the tribe

Hypostomini as being placed within Ancistrini, rendering the latter paraphyletic. However, since

this phylogeny is not yet published and only contains a small subsample of the total diversity that

will be included in the upcoming phylogeny, the relationships herein may not reflect the true

evolutionary history of the group. Therefore, in this study, I will continue to refer to

Hypostomini as being a separate tribe, as established by Armbruster (2004).

11

Figure 5. Multi-locus molecular phylogeny of a subset of loricariids, sequenced by Lujan et

al. (unpublished data). Only taxa included in the jaw functional traits analysis were included in

the tree. Subtribes identified are Loricarini (green), Hypostomini (orange), and Ancistrini (blue).

The tree was further pruned to correspond with taxa available for the body morphology analysis.

Phylogenetic correction ensures that values used in the analysis are statistically

independent and evenly distributed by removing the effect of evolutionary relationships among

taxa (Felsenstein 1985). In order to quantify variation in jaw functional traits and body

morphology in loricariids, all measurements were log-transformed to increase normality of the

dataset. These log-transformed values were then size-corrected by performing a regression

against Standard Length (which was also log-transformed). The residuals of this regression

(Revell 2009) were used in phylogenetically-corrected Principal Components Analyses (PCAs),

using the R packages “ape” (Paradis et al. 2004) and “phytools” (Revell 2012). Because of

highly divergent body sizes and shapes between the subfamilies Loricariinae (Loricarini) and

12

Hypostominae (Hypostomini + Ancistrini) in the dataset, all variables in both subfamilies were

size corrected separately, and then combined into a new matrix for the PCAs. Phylomorphospace

plots (Sidlauskas 2008) were created from each PCA using the “phytools” package in R (Revell

2012). These plots show the evolutionary relationships between occupied regions of

morphospace, and allow for an examination of relative rates of evolution in the morphological

traits based on distances between taxa.

1.2.8 Diet Information

1.2.8.1 Diet Categories

Though diet information is sparse in loricariids, several studies have described the

feeding ecology of certain species. As such, four diet categories have been identified, and will be

compared throughout the study (Table 3). These food groups (guilds) are based on identification

of macroscopic particles, and do not include fish for which diet was identified as either detritus

or of unknown origin.

Table 3. Diet information identified for a subset of loricariids in this study, based on references to diet in the

literature. The four diet categories include wood eaters (purple), algae scrapers (gray), insect and mollusk eaters

(pink) and aufwuch eaters (yellow).

1Aufwuch consists of small crustaceans, larval insects and pieces of algae attached to rocks and substrate.

13

Diet Species Within Group References to Diet in Literature

Wood

Panaqolus gnomus P. nocturnus Panaque nigrolineatus Hypostomus macushi H. pyrineusi H. taphorni

Schaefer and Stewart 1993; Nelson et al. 1999; Armbruster 2003, 2004; Lujan et al. 2011; Lujan et al.

2012

Algae

Chaetostoma breve C. microps C. stroumpoulos C. sp. Xingu Lamonthichthys sp.

Hood et al. 2005; Lujan et al. 2011

Insects + Mollusks

Leporacanthicus galaxias L. heterodon L. triactis Scobinancistrus pariolispos

Burgess 1994

Aufwuchs1

Hypancistrus contradens H. delibittera H. furunculus H. lunaorum Lithoxus lithoides

Horeau et al. 1998; Armbruster et al. 2011

1.2.8.2 Stable Isotope Analysis

For dietary comparisons, isotope values for δ13

C and δ15

N were provided by Lujan

(unpublished data from Lujan et al. 2012). These isotope values for syntopic loricariid

assemblages were standardized to a centroid value for each locality. Distance from the

assemblage centroid to each taxon in the assemblage is the centroid deviation (Lujan et al. 2012).

See Appendix B for list of species used, and the mean centroid deviation values for each δ13

C

and δ15

N. The mean centroid deviation of samples from a particular species was calculated for

each locality, and this value was used to calculate a mean centroid deviation for each species.

This was done so that uneven sample sizes for each locality did not influence a species’ centroid

deviation value. Although isotope data did not come from the same specimens as the

morphological data, this should not be an issue since it is expected that interspecific variation

should be greater than intraspecific variation (Lujan et al. 2012)

Phylogenetically Generalized Least Squares (PGLS) (Grafen 1989; 1992) was used on

the mean centroid deviation for each species separately, for each δ13

C and δ15

N and the residuals

of the first two principal components (PCs) of the PCAs for both the jaw functional traits and

body morphology datasets. This analysis was a modified Generalized Least Squares regression

14

model using a phylogenetic tree to account for the non-independence of taxa (Felstenstein 1985;

Grafen 1989), and assuming that the traits evolved under Brownian motion. The “nlme” package

in R was used to run these analyses (Penheiro et al. 2013).

1.3 Results

1.3.1 Significant Principal Components Test

To determine which (PC) axes were significant and retainable for further analysis,

eigenvalues of the observed data were compared to simulated eigenvalues using the “picante”

package in R (Kembel et al. 2010). Measurements of jaw functional traits and body morphology

were randomized within the dataset, and eigenvalues of the randomized matrices were calculated

separately along with the mean of these values. This permutation process was completed 1000

times for each dataset to create a distribution of possible eigenvalue means assuming random

distribution of the data. The mean of the observed value was compared to the distribution of

simulated data, with the expectation that critical axes would have a mean score higher than that

of the simulated data. For both jaw functional traits and body morphology, the permutation test

determined that only PC1 and PC2 were significant (see Figure 6 for jaw functional traits test –

data not shown for body morphology test).

Figure 6. PC axis significance test for jaw functional traits analysis. Randomized eigenvalue

means (red) were created and compared with actual eigenvalues obtained in the PCA (black).

15

Critical axes have a mean score higher than that of the simulated data. In this case, only the first

two axes can be interpreted in the analysis.

1.3.2 Jaw Functional Traits Analyses

In the analysis of five variables of jaw functional traits for 48 species, PC1 and PC2

together represented nearly 74% of the total variation in this dataset (Figure 7, Table 4).

Variation along PC1 was explained by mechanical advantage measures to proximal and distal

teeth (see section 1.2.5 – functional variables 1 and 2). Low ratios of mechanical advantage were

associated with negative scores while high ratios of mechanical advantage were associated with

positive scores. As such, the functional interpretation is that PC1 reflects a trade-off between jaw

speed (low ratios) and jaw strength (high ratios). Similarly, variation along PC2 is explained by

the maximum force entering the mandible per unit of substrate contacted (AM/ (TRL^2)) (see

section 1.2.5 – functional variable 3). Low ratios are associated with negative scores while high

ratios are associated with positive scores. Therefore, the functional interpretation is that PC2

reflects a gradient of distributed force across the tooth row length (low values) and concentrated

force (high values).

Table 4. Species abbreviations used in all analyses. Colours represent the three tribes included in the analysis,

with Loricarini (green), Hypostomini (orange), and Ancistrini (white).

An.m Ancistrus macropthalmus Lm.sp Lamonthichthys sp. Pk.s Peckoltia sabaji

An.r Ancistrus ranunculus La.s Lasiancistrus schomburgkii Pk.v Peckoltia vittata

Ba.b Baryancistrus beggini La.t Lasiancistrus tentaculatus Psc.l Pseudacanthicus leopardus

Ba.x Baryancistrus xanthellus Le.g Leporacanthicus galaxias Psa.s Pseudancistrus sidereus

Ch.b Chaetostoma breve Le.h Leporacanthicus heterodon Psl.a Pseudolithoxus anthrax

Ch.m Chaetostoma microps Le.t Leporacanthicus triactis Psl.d Pseudolithoxus dumus

Ch.s Chaetostoma stroumpoulos Li.l Lithoxus lithoides Psl.t Pseudolithoxus tigris

Ch.x Chaetostoma sp. Xingu Ne.b Neblinichthys brevibacchium Ri.f Rineloricaria fallax

De.s Dekeyseria scaphiryncha Ol.p Oligancistrus punctatissimus Sc.p Scobinancistrus pariolispos

Et.r Etsaputu relictum Pqo.a Panaqolus albomaculatus Sp.L Spectracanthicus L020

Fa.n Farlowella amazona Pqo.g Panaqolus gnomus St.m Sturisoma monopelte

He.s Hemiancistrus subviridis Pqo.n Panaqolus nocturnus Hyo.m Hypostomus macushi

Hya.c Hypancistrus contradens Pqe.n Panaque nigrolineatus Hyo.p Hypostomus pyrineusi

Hya.d Hypancistrus delibittera Par.n Parancistrus nudiventris Hyo.t Hypostomus taphorni

Hya.f Hypancistrus furunculus Pk.ba Peckoltia bachi Hyo.r Hypostomus rhantos

Hya.l Hypancistrus lunaorum Pk.b Peckoltia braueri Hyo.n Hypostomus niceforoi

Fishes categorized as feeding on wood, insects, algae or aufwuch occupy different areas

of morphospace (Figure 7), indicating that jaw functional traits may influence diet. The guilds

16

are associated with the outermost edges of the defined morphospace, whereas fish that were

classified as either of unknown diet or detritivorous -which do not have a colour associated with

them- have smaller ranges in jaw functional traits.

The four Chaetostoma species (Ancistrini) included in the analysis are convergent in

morphospace with Lamontichthys sp. (Loricarini), despite the fact that these fish are from

different tribes. Both Chaetostoma and Lamontichthys are algae scrapers (Hood et al. 2005;

Lujan et al. 2011), and are shaded in gray in the PCA (Figure 7). Both these groups display long

tooth row lengths and large adductor-mandibulae insertion areas (data not shown). These fish

load positively on PC1, which indicates they have jaws optimized for strength. As such,

Lamontichthys sp. is far from the cluster of other members of Loricarini in the analysis.

At the opposite end of PC1, Leporacanthicus heterodon and Scobinancistrus pariolispos,

have speed-optimized jaws. They also load positively on PC2, which indicates concentrated

force. Their diets consist primarily of mollusks and insects (Burgess 1994), and they, along with

other species of the genus Leporacanthicus are shaded in pink.

Shaded in yellow are fish whose diet primarily consists of aufwuchs (Horeau et al. 1998;

Armbruster et al. 2011). They include species of Hypancistrus and Lithoxus, which are both in

the tribe Ancistrini, and are characterized by distributed force along the length of the tooth row,

as well as optimized speed.

The Hypostomini occupy a unique area of morphospace characterized by concentrated

force (high loading on PC2). The phylomorphospace reveals a distinct split between the wood-

eating Hypostomus cochliodon group (as defined by Armbruster, 2003), and the non-wood eating

Hypostomini, represented in this study by H. rhantos and H. niceforoi. The H. cochliodon group

is convergent with the other wood-eating loricariids (of the tribe Ancistrini), and are shaded

purple in the PCA. Interestingly, Peckoltia bachi is also found among the wood-eating loricariids

in this analysis, despite not being a wood eater (Armbruster 2008).

17

Figure 7. Graphic representation of the first two principal components of jaw functional

traits of a phylogenetically-corrected PCA, by means of morphospace (upper panel) and

phylomorphospace (lower panel). PCA analysis included five size corrected variables for the

mean values of 48 species of loricariids. Symbol colours indicate the three tribes included in the

analysis, Loricarini (green), Hypostomini (orange), and Ancistrini (blue). Shaded ploygons in the

PCA indicate to primary diet category: wood (purple), aufwuchs (yellow), algae (gray), and

insects/mollusks (pink). See Table 5 for loadings of each PC axis, and Table 4 for full names of

species’ abbreviations.

18

Table 5. Principal components analysis eigenvectors for jaw functional traits and body

morphology. All analyses were calculated using a phylogenetically-corrected PCA.

Jaw Functional PCA eigenvectors Body Morphology PCA eigenvectors

PC1 PC2 PC1 PC2 Eigenvalue 2.288 1.389 Eigenvalue 3.059 2.606 Cumulative percent variance explained

45.77 73.56 Cumulative percent variance explained

38.24 70.82

AM/(TRL^2) -0.317 0.876 Head Length -0.776 0.399 Mechanical Advantage -D 0.661 0.577 Mouth Length -0.545 0.676 Mechanical Advantage -P 0.871 0.299 Head Depth -0.776 -0.361 H1/TRL -0.790 0.445 Cleithral Width -0.682 -0.060 Tooth Row Angle -0.606 0.021 Snout Length -0.818 0.325

Depth at Anal Fin -0.509 -0.816 Depth at Caudal Peduncle -0.382 -0.787

1.3.3 Body Morphology Analysis

The morphospace defined by the eight body morphology variables for 39 species (in

which PC1 and PC2 represent a combined 71% of the total variation in the dataset), appears

more dispersed than that of jaw functional traits (Figure 8). Some sister taxa occupy vastly

different areas of morphospace, and there appears to be an accelerated rate of morphological

change along these branches which is made evident in the phylomorphospace plot (Figure 8,

lower panel). For example, species in the genus Chaetostoma are highly dispersed in

morphospace. In constrast, species of the genus Hypostomus have a well-defined morphospace.

This pattern contrasts with that of the jaw functional traits, where intrageneric clustering and

more conserved morphology are evident for nearly all genera. PC1shows a gradient in head

shape with Head Length, Head Depth and Snout Length loading strongly along this axis (Table

5). On one end of PC1 are long and deep heads and on the other are short and flat heads.

Variation along PC2 is explained primarily by variation of body depth, with one end describing

deeper bodies and the other describing shallower bodies.

In contrast to the jaw functional traits analysis (Figure 7), the body morphology analysis

does not reveal segregation of guilds in morphospace. Rather, there is overlap between the diet

polygons (wood and algae, as well as insects and mollusks and aufwuchs), and these guilds

occupy vast areas or morphospace, especially along PC1 in the case of the wood-eaters (purple).

19

When comparing intrageneric variation in morphospace, the majority of congeners have a

similar position along PC2 (body depth), but are divergent along PC1 (head depth). This pattern

is consistent when sister taxa across all three tribes are examined. Both Ancistrus species, as well

as all three Peckoltia species (tribe Ancistrini) are prime examples. An exception to this trend is

Baryancistrus, which differs very little along PC1, but B. beggini has a much deeper body

relative to B. xanthellus.

The Hypostomini appear to have similar body morphologies, with deep bodies and short,

flat heads. Contrary to the jaw functional traits, this tribe is not convergent with other wood-

eating loricariids in body shape (gray), as both ancistrin wood-eaters (Panaqolus nocturnus and

Panaque nigrolineatus) have much deeper and longer heads than the hypostomin wood-eaters

(Hypostomus pyrineusi, H.taphorni, and H. macushi).

Chaetostoma breve and Panaque nigrolineatus appear to be convergent on a body plan

defined by deep bodies and heads, despite having different overall maximum body sizes (C.

breve has a maximum size of 30cm, while P. nigrolineatus can grow upwards of 43cm (Fisch-

Muller, 2003). They also have different diets, with the former eating algae and the latter feeding

on wood.

The genus Leporacanthicus and Scobinancistrus pariolispos (pink) also have similar

body sizes and shapes. They are the only members of identified guilds that appear to cluster in

the body morphology PCA. Despite their specialized jaws, their body morphology falls near the

center of family morphospace, indicating that their proportions are average among loricariids.

It is interesting to note the relatively high divergence of Rineloricaria fallax from other

loricarines, and its position within morphospace shared by deep-bodied loricariids. This was due

to the separate size correction (see Methods) between the Loricarinae (Loricarini) and

Hypostominae (Hypostomini + Ancistrini). Rineloricaria fallax is more robust than its

counterparts, which are markedly thin, narrow, and elongate, and this led to the low score on

PC2 despite the fact that it does not appear to resemble the surrounding deep-bodied fish. An

attempt to resolve this issue was to size correct R. fallax with the more robust Hypostominae, but

this caused R. fallax to become an outlier in the analysis (data not shown), and so the original

residuals were kept. However, any functional interpretation along PC2 should be made with

caution.

20

Figure 8. Graphic representation of the first two principal components of body morphology

of a phylogenetically-corrected PCA, by means of morphospace (upper panel) and

phylomorphospace (lower panel). PCA analysis included eight size corrected variables for the

mean values of 39 species of Loricariids. Symbol colours indicate the three tribes included in the

analysis: Loricarini (green), Hypostomini (orange), and Ancistrini (blue). Shaded ploygons in the

PCA indicate to primary diet category: wood (purple), aufwuchs (yellow), algae (gray), and

21

insects/mollusks (pink). See Table 5 for loadings of each PC axis, and Table 4 for full names of

species’ abbreviations.

1.3.4 Diet-Morphology Correlations

Phylogenetic Generalized Least Squares (PGLS) analyses for jaw functional traits

included mean values for 38 taxa, while the body morphology analysis included 31 taxa. These

analyses revealed a significant correlation (p=0.0239) between jaw functional traits and centroid

deviation values of δ13

C, as well as a significant correlation (p=0.0299) between body

morphology and centroid deviation values of δ13

C (Table 6).

A regression for the significant results of jaw functional traits was also run excluding

PC1 in order examine to what extent correlation could be attributed to PC2 alone, rather than a

combination of PC axes. Results did not show any change in significance, and the AIC was

similar (data not shown) meaning that PC1 did not add important information to the model.

Therefore only PC2 will be considered in the interpretation of the jaw functional traits results.

For jaw functional traits, the value of the regression slope is negative (-2.2994), indicating that

the centroid deviation value of δ13

C is inversely proportionate with force concentration in jaws

(Figure 9).

Results show that wood-eaters have the greatest centroid deviation of the guilds in the

analysis. This is consistent with their relatively greater ingestion and assimilation of terrestrial

(allochthonous) sources of carbon (i.e. trees), whereas the more δ13

C depleted values of the three

other guilds are consistent with their relatively greater ingestion and assimilation of carbon from

aquatic sources.

Table 6. Results of PGLS regressions using centroid deviations of δ13

C and δ15

N against

residuals of the first two principal components of the jaw functional traits (f) and body

morphology (b) PCAs. Significant results (p< 0.05) are indicated in orange.

22

Model Resid. Std. Error Value Std. Error t-value p-value

C~PC2f + PC1f 3.382 Intercept -0.3865 1.7303 -0.2233 0.8246 PC2f -2.2994 0.9737 -2.3615 0.0239 PC1f 0.1848 0.8118 0.2277 0.8212 N~PC2f + PC1f 1.649 Intercept 0.0229 0.9331 0.0246 0.9805 PC2f -0.3079 0.5251 -0.5864 0.5613 PC1f -0.3228 0.4378 -0.7373 0.4658 C~PC2b + PC1b 3.377 Intercept -0.2319 1.7522 -0.1323 0.8957 PC2b 1.0272 1.1692 0.8786 0.3874 PC1b -2.6275 1.1465 -2.2917 0.0299 N~PC2b + PC1b 1.868 Intercept -0.0778 0.9693 -0.0802 0.9366 PC2b 0.0564 0.6467 0.0873 0.9311 PC1b 0.5215 0.6343 0.8223 0.1481

Figure 9. Estimate of concentration of δ13C as a function of jaw functional traits. PGLS

model-predicted regression shows that jaws optimized for force concentration are inversely

proportionate with δ13C intake (t-value = -2.36; p-value=0.0239). See table 6 for all results of

the analysis.

Similarly, PC1 was found to be significant (rather than a combined effect of both PC1

and PC2) for the relationship between body morphology and δ13

C. Interpretation of results will

therefore only include PC1. The value of the regression slope is negative (-2.6275), indicating

23

that the centroid deviation value of δ13

C is inversely proportionate with force concentration in

jaws (Figure 10).

Figure 10. Estimate of concentration of δ13

C as a function of body morphology. PGLS

model-predicted regression shows that short heads and snouts, coupled with flat heads are

inversely proportionate with δ13

C intake (t-value = -2.29; p-value=0.0299). See table 6 for all

results of the analysis.

1.4 Discussion

The results of the jaw functional traits analysis show a clear separation of the dietary

guilds in morphospace, whereas these guilds overlap in the body morphology analysis. This will

be explored in the upcoming section, and will be complemented by a look at the results for each

dietary guild in relation to morphospace. Next the relationship between body morphology and

habitat use will be discussed, and finally, the implications of the diet-morphology correlations

uncovered in the PGLS regressions will be explored and discussed in regards to feeding

adaptation in loricariids.

1.4.1 Jaw Functional Traits Analyses

The PCA analysis for jaw functional traits (Figure 7) reveals that taxa are partitioned in

morphospace based on their known dietary specializations, as indicated by the separation of

24

dietary guilds in morphospace. A functional gradient is evident between speed-optimized jaws

and strength-optimized jaws (PC1), as well as between jaws with force either concentrated or

distributed (PC2).

PC1 corresponds to a functional gradient with invertivores with speed-optimized jaws at

one end with (including Leporacanthicus, Scobinancistrus, Hypancistrus, and Lithoxus, and

algae scrapers with strong jaws on the other (Chaetostoma and Lamontichthysis). In the middle

along PC1, we notice the wood-eaters (purple) which seem to have jaws optimized for strength,

albeit to a lesser extent than the invertivores. The dietary gradient is also apparent on PC2.

Loricariids requiring concentrated force to pry invertebrates from their shells (pink), or to scrape

pieces of wood from the substrate (purple) load positively on PC2, whereas the aufwuch-eaters

(yellow) require distributed force, presumably to remove the larva off rocks (Grillet and Barrera

1997).

1.4.2 Body Morphology Analysis

The body morphology analysis does not reveal segregation of guilds (coloured groups) in

morphospace. Rather, there is overlap between guilds, indicating that taxa with different body

shapes may be eating similar diets. The overlap between guilds also suggests that diet may not be

closely related to body morphology, especially along PC1 (which represents head shape).

1.4.3 Diet Groups

A closer look at the results for each dietary guild in relation to morphospace may provide

insight into association of diet and feeding morphology in loricariids.

1.4.3.1 Algae Scrapers (Gray)

Chaetostoma and Lamontichthys appear to converge on similar jaw functional traits

(Figure 7). This was not expected since they are from different tribes. However, dietary

information reveals that both are primarily algae scrapers, which suggests that this specialized

mode of feeding exerts a constraint on morphospace. Interestingly, these genera are found

sympatrically in swift piedmont rivers of the Andes and the Guiana Shield (Lujan and

Armbruster 2011), which indicates that they are convergent on the same resources despite living

in the same environment.

25

Although Lamontichthys sp. was not included in the body morphology analysis due to

unavailable data (see Methods), it is interesting to note that this fish does not appear at all similar

to Chaetostoma from an external body shape perspective – it has a much more depressed and

narrow body, as is characteristic of most Loricarini (Paixao and Toledo-Piza, 2009). However,

despite the overall body differences, the heads look quite similar (Figure 11), which coupled with

their diet of algae, suggests that there may be a link between diet and morphology in these fish.

Figure 11. Despite their different body shapes, head morphology appears convergent in

representatives of Lamontichthys and Chaetostoma. A) Lamontichthys sp., b) Chaetostoma

microps. Photographs are not to scale, and the body and mouth pictures are from different

individuals of the same species. Individual images provided by Nathan Lujan.

1.4.3.2 Wood-Eaters (Purple)

The analysis of jaw functional traits revealed that the Hypostomini occupy a unique area

of morphospace, and the phylomorphospace analysis revealed a distinct split between the H.

cochliodon group and the non-wood eating Hypostomini, represented by Hypostomus rhantos

and H. nocturnus (Figure 7). The H. cochliodon group is convergent with the other wood-eating

loricariids (Schaefer and Stewart 1993; Nelson et al. 1999), and occupy an area of morphospace

characterized by concentrated force in jaw function. Convergence within concentrated force

morphospace of wood-eaters suggests considerable force is needed to process submerged wood

whereas distributed force may limit loricariids to scraping biofilm on the surface of the wood.

26

The body morphology analysis shows great variation along PC1 for the wood eaters

(Figure 8). This indicates that although Hypostomus and Panaque/Panaquolus have very similar

jaw functional characteristics, these are independent of body shape and size. P. bathyphilus has a

long and deep head, whereas Hypostomus have short and flat heads. P. nocturnus appears to have

an intermediate head shape in comparison to the other wood-eaters. However, the wood-eaters

have similar body depths, as they are loading closely along PC2. These results suggest that head

morphology is not linked to jaw functionality in these fish.

Peckoltia bachi, is found among the wood-eating loricariids in the both jaw functional

traits and the body morphology analyses. Though little is known about P. bachi’s diet, it has

been known to live among submerged twigs, and possesses muscular adaptations to grasp twigs

in its habitat (Armbruster 2008). It may be possible that P.bachi’s jaws are morphologically and

functionally similar in order to attach to woody substrates, perhaps even removing wood from

branches in search for food. It would be interesting to explore potential convergence of P. bachi

body morphology; however, this could not be verified in this study since exterior morphometrics

were not available for this species. Interestingly, Peckoltia was not recovered as monophyletic in

the molecular phylogeny (Figure 5), as a result of P. bachi not grouping with other Peckoltia.

Initial inclusion of this species in the genus Peckoltia could be due to the convergence of

characters used in previous morphological analyses which do not accurately reflect the

evolutionary history of this species. In light of this and the results of both jaw functional traits

and the body morphology analyses, a taxonomic revision of P. bachi could be warranted.

1.4.3.3 Insect and Mollusk Eaters (Pink)

Based on the analysis of jaw functional traits, Leporacanthicus heterodon and

Scobinancistrus pariolispos are shown to have converged on speed-optimized jaws. In addition,

both are also very positively loaded on PC2, which indicates concentrated force (Figure 7).

Species in the genus Leporacanthicus and S. pariolispos are characterized as having few very

long premaxillary teeth. These teeth, which are upwards of three times the length of the dentary

teeth in species of Leporacanthicus, are hypothesized to be used for mining insects from wood,

as well as extracting snails from their shells (Burgess 1994). It is thus not surprising that their

jaws are optimized for speed and concentrated force, both useful in catching and extracting prey

from the substrate.

27

Despite not including measurements from the premaxilla in the present study, the

specialization of this feeding mode is captured in measurements from the mandible. The dentary

must have also converged to meet the demands of invertivory, and we can see evidence of this in

the jaw functional traits PCA (Figure 7), where Leporacanthicus and S. pariolispos occupy a

unique area of morphospace.

The insect and mollusc eaters are the only dietary group to have similar body

morphologies (Figure 8). They are clustered in morphospace, and have relatively average sized

heads and body depths. These results suggest that body morphology and jaw function may be

linked in this group of loricariids.

1.4.3.4 Aufwuch Eaters (Yellow)

Aufwuch eaters are characterized by their speed-optimized jaws with force distributed

along the tooth row length. They occupy a unique area of morphospace in regards to their jaw

functional traits. Distributed force would allow these fish to effectively scrape larvae,

crustaceans and algae from rocks, where most of their prey is found (Armbruster et al. 2011).

The body morphology analysis reveals that these fish are very dispersed throughout

morphospace, along both PC1 and PC2 (Figure 8). This appears to be in part due to

Hypancistrus’ conserved body morphology which differs significantly from that of Lithoxus

lithoides’ body morphology. Hypancistrus have much deeper bodies, as well as longer and

deeper heads, and their morphospace overlaps that of the other invertivorous group (shaded in

pink).

L. lithoides appears to be at the edge of morphospace in all three analyses. From a body

shape perspective, it has the most depressed body of all loricariids in the analysis. In addition,

the head and snout are also exceptionally dorso-ventrally depressed compared to most loricariids

included in the analysis. These traits are consistent with its habitat requirements, as L. lithoides is

found under rocks in small rivers and streams (Armbruster 1998).

The results of the analyses do not suggest that jaw functional traits and body morphology

is linked in these insectivorous fish.

28

1.4.4 Habitat Use

The gradient along PC1 of the body morphology PCA (Figure 8) appears to give insight

into habitat use, specifically along a water-velocity gradient. This gradient has been reported in

other Neotropical species (Casatti and Castro 2006), although it seems to be to a lesser extent in

loricariids. As such, we would expect to find that species with long and deep heads (notably

Chaetostoma and Panaque) would be found in fast moving waters as their robustness would be

advantageous against the current. This is indeed the case for both Chaetostoma which are found

in fast-moving rocky waters (Salcedo 2006), and Panaque which live in fast-moving Amazonian

headwaters (Nelson et al. 1999). At the opposite end of PC1, Lithoxus and Pseudolithoxus dwell

under rocks in rivulets (Armbruster 1998; Lujan and Brindelli 2011), which can be described as

slower moving water. In addition, Hypostomus, which is found intermediately on PC1, has been

known to live in shallow pools and lakes which have a slow current (Armbruster 1998).

However, some species may be found in rivers with swift flow (Burgess 1989), which could

explain its deep body.

Mouth Length is loading positively on PC2 of the body morphology analysis, which is a

gradient of body depth. Since mouth length is a measurement of oral disk diameter, it may give a

hint as to suction surface. As expected, loricariids with dorso-ventrally flattened bodies have

longer mouth lengths (Table 5), potentially to attach themselves in fast moving currents.

Body size and shape are known to be important for accessing food sources in pelagic fish.

However, since loricariids are limited to the bottoms of lakes and rivers, their body shape may be

more associated with their habitats and the substrates on which they live rather than their specific

food sources. For example, Lithoxus lithoides has a dorso-ventrally compressed body, which

allows it to live under rocks (Armbruster 1998). However, in certain taxa, there appears to be a

disconnect between body size, shape, and the food resources consumed. For example, despite

Chaetostoma and Lamonthichthys’ very different body shapes (Figure 11), they have similar

diets, and both were collected from the Marañon River in northern Peru (Lujan and Armbruster

2012). It should be noted that detailed habitat and dispersal data is lacking in loricariids, and this

severely limits inferences that can be made about how body size and shape may be important for

food acquisition or niche differentiations. Further studies should therefore incorporate

29

information about habitat use and co-existence in loricariids in order to test if fish with similar

body sizes and shapes share ecological niches.

1.4.5 Diet-Morphology Correlations

1.4.5.1 Relationship Between δ13C and Jaw Functional Traits

The PGLS regression revealed a significant correlation (p=0.0239) between jaw

functional traits and the concentration of δ13

C. Since only PC2 of jaw functional traits is

significantly correlated with diet, it is the only axis of variation that can be interpret from the

analysis. The value of the regression slope is negative (-2.2994), indicating concentration of δ13

C

is inversely proportionate with force concentration in jaws (Figure 9).

Studies have shown that wood-eating loricariids have a different δ13

C signal than those

that strictly rely on autochthonous carbon sources (Nogonaki et al. 2007; German and Miles,

2010; Lujan et al. 2012). These findings are consistent with those of this study, as we can clearly

see that there is no overlap between the wood-eaters (purple) and the known non-wood eating

loricariids. For the most part, the wood eaters have a bigger centroid deviation value than the

other guilds. This is not surprising since the mean centroid deviation calculated for each locality

sampled would have been influenced by a high proportion of autochthonous carbon sources.

Because allochthonous carbon sources have a different δ13

C signal, it is expected that the

deviation from the mean centroid would be greater. It is interesting to note that certain fish not

associated with a diet guild have similar centroid deviations for δ13

C to wood-eaters. It can

therefore be hypothesized that Hypostomus rhantos, H. niceforoi, Panaqolus albomaculatus,

Rineloricaria fallaa, Pseudolithoxus dumus and Baryancistrus beggini also have diets consisting

of allochthonous carbon sources.

The aufwuch eaters, algae scrapers and insect/mollusk eaters have overlapping centroid

deviation values for δ13

C, which could indicate that they obtain their food from enriched

autochtonous sources of carbon. Certain taxa that were considered either generalists, detritivores,

or of unknown dietary composition have similar centroid deviation values to the pink, yellow

and gray groups, indicating that they too probably consume a diet rich in aquatic sources.

30

Conversely, taxa with high centroid deviation values are likely consuming terrestrial carbon

sources since their values are similar to those of the wood-eaters.

1.4.5.2 Relationship Between δ13C and Body Morphology

A significant correlation (p=0.0299) was uncovered between body morphology and

concentration δ13

C (Table 6, Figure 10). PC1 was the only axis found to be significant (rather

than a combined effect of both PC1 and PC2). Interpretation of results is therefore limited to

only include PC1. The value of the regression slope is negative (-2.6275), indicating that the

centroid deviation value of δ13

C is proportionate with head length (Figure 10), since taxa with

shorter heads have smaller centroid deviations of δ13

C.

The same trend is observed as with the correlation between δ13

C and jaw functional traits,

wherein the wood-eaters have a higher centroid deviation than the other groups. The position of

the insect/mollusk eaters in the regression is striking since they are completely contained within

the same space as the aufwuch eaters, indicating that they have overlapping body morphology

along PC1 as well as overlapping centroid deviation values for δ13

C. It is obvious that the only

axis that separates these two groups is that of PC2 for jaw functional traits.

1.4.5.3 Implications of Diet-Morphology Correlations in Loricariids

Morphological variation with regards to jaw functional traits and body morphology were

quantified and tested independently for correlations with diet, by means of species’ relative

position in assemblage isotope space (for both δ13

C and δ15

N) and diet categories (Table 3).

Despite the expectation that jaw functional traits would be linked to body morphology and that

both could be used to predict diet, the results of this study only partially support this. Results

showed that jaw functional traits do not appear to be linked to body morphology in Loricariidae.

Jaw functional traits were correlated with δ13

C, an indication of the carbon source (aquatic versus

terrestrial). Jaw functional traits were not correlated with δ15

N, a general indication of

proportions of protein in the diet. Partitioning of morphospace of jaw functional traits (Figure 7)

indicate a link between jaw morphology and diet categories (Table 3). Convergence of jaw

functional traits among taxa exploiting similar resources indicated that jaw functional traits may

be used to predict diet categories across the family. This study does not explicitly test whether

functional traits associated with the mandible are specialized for particular purposes (e.g., wood

31

eating), however the divergence of jaw functional traits along the δ13

C gradient suggest there

may be a difference in performance capability to exploit different carbon sources.

Body morphology was also correlated with δ13

C, which may be explained by habitat use

(water-velocity gradient) of various body shapes and the potential differences in the proportion

of autochthonous versus allochthonous resources among different habitat types. Due to the

current lack of detailed distribution and habitat data for taxa included in this study, only general

statements regarding this association can be made. However, further analyses of habitat-

morphology associations may shed light on the processes driving the pattern of the δ13

C-body

morphology correlation observed in this study.

Questions of diet-morphology relationships addressed in this study are only the first steps

in understanding the ecology of this group of fishes in an evolutionary context. Loricariidae is

the second most taxonomically diverse lineage in the Neotropics (Reis et al. 2003) and it was

qualitatively described as having relatively high ecological diversity (Lujan et al. 2012). This

study supports the initial observations of Lujan et al. (2012) by quantifying functional diversity

with regards to feeding ecology. It has yet to be tested whether this functional diversity can be

attributed to the taxonomic richness of Loricariidae, or whether there is an adaptive component

that has accelerated the rates of morphological and species diversification in this group.

1.5 Conclusions

Despite being taxonomically diverse, loricariids show relatively little dietary diversity

and share the same trophic level – they are predominantly detritivores in Neotropical systems. It

has been hypothesized that they partition their food resources on a nutrient gradient, (Lujan et al.

2011; Hood et al. 2005; Nogogaki et al. 2007), yet little research has shown an explicit link

between diet and feeding morphology in loricariids. The present study was the first to quantify

the diversity of jaw functional traits in loricariids in an explicit phylogenetic context, and has

demonstrated that these fish are not only taxonomically diverse, but their jaws are also

functionally diverse. Recorded dietary information for certain taxa indicates that their jaws are

specialized for their diets (wood-eating, invertivory, algae-scraping). Body morphology was also

quantified in a phylogenetic context, and revealed that jaw functional traits are decoupled from

32

body morphology. Loricariids with different body sizes and shapes converge on similar jaw

functional traits if they have similar diets.

Both jaw functional traits and body morphology show correlations with δ13

C isotope data.

Jaw functional traits were inversely correlated with centroid deviation values of δ13

C, and body

morphology was also inversely correlated with δ13

C values. These results are the first quantified

links between diet and feeding morphology in loricariids. This suggests that both jaw function

and body morphology can predict ecological traits, which, in turn, suggests an adaptive

relationship.

This study is evidence that more research is warranted to fully understand the relationship

between diet and morphology in loricariids, and perhaps begins to grasp how this family of fish

has become so diverse.

33

References

Adriaens D, Geerinckx T, Vlassenbroeck J, Van Hoorebeke L, Herrel A. 2009. Extensive jaw

mobility in suckermouth armoured catfishes (Loricariidae): A morphological and

kinematic analysis of substrate scraping mode of feeding. Physiological and Biochemical

Zoology, 82(1)51-62.

Armbruster JW. 1998. Modifications of the Digestive Tract for Holding Air in Loricariid and

Scoloplacid Catfishes. Copeia, 3:663-675.

Armbruster JW. 2002. Hypancistrus inspector, a new species of suckermouth armored catfish

(Loricariidae: Ancistrinae). Copeia, 2002(1):86-92.

Armbruster JW. 2003. The species of the Hypostomus cochliodon group (Siluriformes:

Loricariidae). Zootaxa, 249:1-60.

Armbruster JW. 2004. Phylogenetic relationships of the suckermouth armoured catfishes

(Loricariidae) with emphasis on Hypostominae and the Ancistrinae. Zoological Journal

of the Linnean Society, 141:1-80.

Armbruster JW. 2008. The genus Peckoltia with the description of two new species and a

reanalysis of the phylogeny of the genera of the Hypostominae (Siluriformes:

Loricariidae). Zootaxa, 1822:1-76.

Armbruster JW, Lujan NK, Taphorn DC. 2007. Four new Hypancistrus (Siluriformes:

Loricariidae) from Amazonas, Venezuela. Copeia, 2007(1): 62-79.

Ben-David M, and Flaherty EA. 2012. Stable isotopes in mammalian research: a beginner’s

guide. Journal of Mammology, 93(2):312-328.

Burgess, WE. 1989. An atlas of freshwater and marine catfishes, a preliminary survey of the

Siluriformes. T.F.H. Publications, Neptune City, New Jersey. 784 pp

Burgess WE. 1994. Scobinancistrus aureatus, a new species of loricariid catfish from the Rio

Xingu (Loricariidae: Ancistrinae). Tropical Fish Hobbyist, 43:236-242.

34

Casatti L, and Castro RM. 2006. Testing the ecomorphological hypothesis in a headwater riffles

fish assemblage of the rio São Francisco, southeastern Brazil. Neotropical Ichthyology,

4(2): 203-214.

Delariva RL, and Agostinho AA. 2001. Relationship between morphology and diets of six

neotropical loricariids. Journal of Fish Biology, 58:832-847.

Ershmeyer WN., Fricke R. 2011. Catalogue of Fishes electronic version.

http://research.calacademy.org/research/ichthyology/catalog/fishcatmain.asp

Felsenstein J. 1985. Phylogenies and the Comparative Method. The American Naturalist, 125(1):

1-15.

Fisch-Muller, S. 2003. Loricariidae-Ancistrinae (Armored catfishes). p. 373-400. In R.E. Reis,

S.O. Kullander and C.J. Ferraris, Jr. (eds.) Checklist of the Freshwater Fishes of South

and Central America. Porto Alegre: EDIPUCRS, Brasil

Fugi, R, Agostinho AA, & Hahn NS. 2001. Trophic morphology of five benthic-feeding fish

species of a tropical floodplain. Revista Brasileira de Biologia, 61(1), 27-33.

Geerinckx T, Herrel A, Adriaens D. 2010. Suckermouth Armored Catfish Resolve the Paradox

of Simultaneous Respiration and Suction Attachment: A Kinematic Study of

Pterygoplichthys disjunctivus. Journal of Experimental Zoology, 315: 121-131.

Geerinckx T, Huysseune A, Boone M, Claeys M, Couvreur M, De Kegel B, Mast P, Van Hoorebeke L, Verbeken K, Adriaens D. 2012. Soft Dentin Results in Unique Flexible Teeth in Scraping Catfishes. Physiological and Biochemical Zoology, 85(5): 481-490.

German DP, Neuberger DT, Callahan MN, Lizardo NR, Evans DH. 2010. Feast to famine: The effects of food quality and quantity on the gut structure and function of a detritivorous catfish (Teleostei: Loricariidae), Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology, 155(3): 281-293.

German D, Miles RD, 2010. Stable carbon and nitrogen incorporation in blood and fin tissue of the catfish Pterygoplichthys disjunctivus (Siluriformes, Loricariidae). Environmental Biology of Fishes, 89:117–133.

35

Grafen A. 1989. The phylogenetic regression. Philosophical Transactions of the Royal Society Series B, 326, 119–157.

Grafen A. 1992. The uniqueness of the phylogenetic regression. Journal of Theoretical Biology, 156(4):405-423.

Grillet ME and Barrera R. 1997. Spatial and temporal abundance, substrate partitioning and

species co-occurrence in a guild of Neotropical blackflies (Diptera:Simuliidae).

Hydobiologia. 345:197-208.

Hood JM, Vanni MJ, Flecker AS. 2005. Nutrient recycling by two phosphorus rich grazing

catfish: the potential for phosphorus-limitation offish growth. Oecologia, 146:247-257.

Horeau V, Cerdan P, Champeau A, Richard S. 1998. Importance of aquatic invertebrates in the

diet of rapids-dwelling fish in the Sinnamary River, French Guiana. Journal of Tropical

Ecology, 14: 851-864.

Kelly LJ, del Rio CM. 2010. The fate of carbon in growing fish: an experimental study of

isotopic routing. Physiological and Biochemical Zoology, 83:473–480.

Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, Blomberg SP,

Webb CO. 2010. Picante: R tools for integrating phylogenies and ecology. Bioinformatics

26:1463-1464.

Kramer DL, Bryant MJ. 1995. Intestine length in the fishes of a tropical stream: 2. Relationships

to diet — the long and short of a convoluted issue. Environmental Biology of Fishes,

42(2):129-141.

Layman CA, Arrington DA, Montaña CG, Post DM. 2007. Can stable isotope ratios provide for

community-wide measures of trophic structure? Ecology, 88:42–48.

López-Fernández H., Winemiller KO., Montaña C., Honeycutt R. 2012. Diet-Morphology

Correlations in the Radiation of South American Geophagine Cichlids (Perciformes:

Cichlidae: Cichlinae). Plos One, 7(4): e33997. doi:10.1371/journal.pone.0033997.

Lujan NK, and Armbruster JW. 2011. The Guiana Shield. In: Historical Biogeography of

Neotropical Freshwater Fishes. James S. Albert and Roberto E. Reis (Eds.). Chapter 13:

211-242. 388 pp. University of California Press. ISBN-10: 0520268687. ISBN-13: 978-

0520268685.

36

Lujan NK, and Armbruster JW. 2012. Morphological and Functional Diversity of the Mandible

in Suckermouth Armored Catfishes (Siluriformes: Loricariidae). Journal of Morphology,

273: 24-39.

Lujan NK and Birindelli JO. 2011. A new distinctively banded species of Pseudolithoxus

(Siluriformes: Loricariidae) from the upper Orinoco River. Zootaxa, 2941:38-46.

Lujan NK, German DP, Winemiller KO. 2011. Do wood-grazing fishes partition their niche?:

morphological and isotopic evidence for trophic segregation in Neotropical Loricariidae.

Functional Ecology, 25:1327-1338.

Lujan NK, Winemiller KO, Armbruster JW. 2012. Trophic diversity in the evolution and

community assembly of loricariid catfishes. BMC Evolutionary Biology. 12:124

Maddison, WP. and Maddison DR. 2011. Mesquite: a modular system for evolutionary analysis.

Version 2.75 http://mesquiteproject.org

Minagawa M, Wada E. 1984. Step­wise enrichment of 15N along food chains: Further evidence

and the relation between δ15N and animal age. Geochimica et Cosmochimica Acta,

48(5):1135-1140.

Montana CG, Winemiller KO. 2013. Evolutionary convergence in Neotropical cichlids and

Nearctic centrarchids: evidence from morphology, diet, and stable isotope analysis.

Biological Journal of the Linnean Society, 109(1):146-164.

Nelson JA., Wubah DA., Whitmer ME., Johson EA., Stewart DJ. 1999. Wood-eating catfishes of

the genus Panaque: gut microflora and cellulolytic enzyme activities. Journal of Fish

Biology, 54:1069-1082.

Nonogaki H, Nelson JA, Patterson WP. 2007. Dietary histories of herbivorous loricariid

catfishes: evidence from δ13C values of otoliths. Environmental Biology of Fishes, 78:

13-21.

Novakowski GC, Fugi R, Hahn NS. 2004. Diet and dental development of three species of

Roeboides (Characiformes: Characidae). Neotropical Ichthyology, 2(3), 157-162

37

Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R

language. Bioinformatics 20: 289-290

Paixao, AD and Toledo-Piza M. 2009. Systematics of Lamontichthys Miranda-Ribeiro

(Siluriformes: Loricariidae), with the description of two new species. Neotropical

Ichthyology, 7(4):519-568.

Peterson BJ, Fry B. 1987. Stable Isotopes in ecosystem studies. Annual Review of Ecology and

Systematics, 18(1):293-320.

Pinheiro J, Bates D, DebRoy S, Sarkar D and the R Development Core Team (2013). nlme:

Linear and Nonlinear Mixed Effects Models. R package version 3.1-113.

Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA,

http://imagej.nih.gov/ij/, 1997-2012

Reis R, Kullander S, Ferraris C, Jr. 2003. Check list of the freshwater fishes of South and Central

America. Porto Alegre: Pontificia Universidade Catolica do Rio Grande do Sul. 729p.

Revell LJ. 2009. Size-correction and principal components for interspecific comparative studies.

Evolution. 63(12):3258-3268.

Revell, L J. (2012) phytools: An R package for phylogenetic comparative biology (and other

things). Methods in Ecology and Evolution, 3 217-223

Salcedo NJ. New Species of Chaetostoma (Siluriformes: Loricariidae) from Central Peru.

Copeia, 2006(1): 60-67.

Schaefer SA, and Lauder GV. 1986. Historical Transformation of Functional Design

Evolutionary Morphology of Feeding Mechanisms in Loricarioid Catfishes. Systematic

Zoology, 35(4): 489-508.

Schaefer SA and Lauder GV. 1996. Testing hystorical hypotheses of morphological change:

biomechanical decoupling in loricarioid catfishes. Evolution. 50(4):1661-1675.

38

Schaefer SA and Stewart DJ. 1993. Systematics of the Panaque dentex species group

(Siluriformes: Loricariidae), wood-eating armored catfishes from tropical South America.

Ichthyological Exploration of Freshwaters, 4:309-342.

Sidlauskas B. 2008. Continuous and arrested morphological diversification in sister clades of

characiform fishes: a phylomorphospace approach. Evolution, 62(12):3135-3156.

Van Wassenbergh, S., Herrel, A., Adriaens, D. & Aerts, P. 2007. No trade-off between biting

and suction feeding performance in clariid catfishes. Journal of Experimental Biology, 210: 27 –

36.

Van Wassenbergh S., Lieben T., Herrel A., Huysentruyt F., Geerinckx T., Adriaens D., Aerts P.

(2008) Kinematics of benthic suction feeding in Callichthyidae and Mochokidae, with

functional implications for the evolution of food scraping in catfishes. Journal of

Experimental Biology, 212: 116-126.

Westneat MW. 2004. Evolution of levers and linkages in the feeding mechanisms of fishes.

Integrative and Comparative Biology. 44:378-389

39

Appendix A

List of species prepared for this study, as well as catalogue numbers of the institutions from which they came.

Geographical information includes GPS coodinates from the sampled locality. All fish were collected in Brazil

(BR), Guyana (G), or Venezuela (V).

Inst. Catalog # Species Country Latitude Longitude

AUM 42105 Ancistrus macropthalmus V 3.96731 -67.2535

AUM 42105 Ancistrus macropthalmus V 3.96731 -67.2535

AUM 42105 Ancistrus macropthalmus V 3.96731 -67.2535

AUM 42105 Ancistrus macropthalmus V 3.96731 -67.2535

ANSP 199612 Ancistrus ranunculus BR 3.39040 -52.2243

AUM 54990 Baryancistrus beggini V 3.88218 -67.0136

AUM 54990 Baryancistrus beggini V 3.88218 -67.0136

AUM 54990 Baryancistrus beggini V 3.88218 -67.0136

AUM 54990 Baryancistrus beggini V 3.88218 -67.0136

ANSP 193015 Baryancistrus xanthellus BR 3.50355 -52.4408

ANSP 199686 Chaetostoma sp. Xingu BR 3.27209 -52.5534

ANSP 199686 Chaetostoma sp. Xingu BR 3.27209 -52.5534

AUM 54474 Dekeyseria scaphiryncha V 4.02372 -66.9773

AUM 54474 Dekeyseria scaphiryncha V 4.02372 -66.9773

AUM 54270 Dekeyseria scaphiryncha V 4.72977 -67.8191

40

AUM 54270 Dekeyseria scaphiryncha V 4.72977 -67.8191

AUM 53524 Hemiancistrus subviridis V 4.38418 -67.7747

AUM 54989 Hemiancistrus subviridis V 3.88218 -67.0136

AUM 54989 Hemiancistrus subviridis V 3.88218 -67.0136

AUM 54989 Hemiancistrus subviridis V 3.88218 -67.0136

AUM 54993 Hypancistrus contradens V 3.88218 -67.0136

AUM 54993 Hypancistrus contradens V 3.88218 -67.0136

AUM 53528 Hypancistrus delibittera V 4.38418 -67.7747

AUM 53528 Hypancistrus delibittera V 4.38418 -67.7747

AUM 58530 Hypancistrus furunculus - - -

AUM 58530 Hypancistrus furunculus - - -

AUM 58530 Hypancistrus furunculus - - -

AUM 58530 Hypancistrus furunculus - - -

AUM 54470 Hypancistrus lunaorum V 4.07870 -66.8585

AUM 39854 Hypancistrus lunaorum V 4.08042 -66.8651

AUM 42163 Hypancistrus lunaorum V 3.10036 -66.4628

AUM 42163 Hypancistrus lunaorum V 3.10036 -66.4628

AUM 48692 Hypostomus hemiurus G 3.50207 -59.0339

ROM 87216 Hypostomus hemiurus G 5.15091 -58.1952

ROM 87216 Hypostomus hemiurus G 5.15091 -58.1952

41

ROM 87216 Hypostomus hemiurus G 5.15091 -58.1952

AUM 48768 Hypostomus macushi G 3.8950 -59.2937

AUM 49801 Hypostomus macushi G 3.8950 -59.2937

AUM 48160 Hypostomus macushi G 3.66467 -59.3428

AUM 53885 Hypostomus rhantos V 5.34648 -66.0333

AUM 53675 Hypostomus rhantos V 4.75346 -66.3726

AUM 54306 Hypostomus rhantos V 4.29481 -66.2889

AUM 48172 Hypostomus taphorni G 4.18268 -59.0638

ROM 86837 Hypostomus taphorni G 4.90681 -58.2447

ROM 86837 Hypostomus taphorni G 4.90681 -58.2447

ROM 86837 Hypostomus taphorni G 4.90681 -58.2447

AUM 39278 Lasiancistrus tentaculatus V 5.34637 -66.0335

AUM 39278 Lasiancistrus tentaculatus V 5.34637 -66.0335

AUM 39278 Lasiancistrus tentaculatus V 5.34637 -66.0335

AUM 39278 Lasiancistrus tentaculatus V 5.34637 -66.0335

AUM 39279 Leporacanthicus galaxias V 5.34637 -66.0335

AUM 39279 Leporacanthicus galaxias V 5.34637 -66.0335

AUM 42144 Leporacanthicus galaxias V 4.05736 -66.9326

AUM 42144 Leporacanthicus galaxias V 4.05736 -66.9326

ANSP 193009 Leporacanthicus heterodon BR 3.50355 -52.4408

42

AUM 56909 Leporacanthicus triactis V 5.38683 -66.1159

AUM 56909 Leporacanthicus triactis V 5.38683 -66.1159

AUM 56909 Leporacanthicus triactis V 5.38683 -66.1159

AUM 58516 Leporacanthicus triactis V 5.38683 -66.1159

AUM 39040 Lithoxus lithoides G 2.22654 -58.2938

AUM 39040 Lithoxus lithoides G 2.22654 -58.2938

AUM 39040 Lithoxus lithoides G 2.22654 -58.2938

AUM 39001 Lithoxus lithoides G 2.22654 -58.2938

ROM 93530 Neblinichthys brevibacchium G 5.48681 -60.7777

ROM 93534 Neblinichthys brevibacchium G 5.48681 -60.7777

ANSP 199539 Oligancistrus punctatissimus 'short snout' BR 3.20568 -52.1119

ANSP 193072 Parancistrus nudiventris BR 3.22557 -51.4411

AUM 48547 Peckoltia braueri G 2.95963 -59.9625

AUM 48547 Peckoltia braueri G 2.95963 -59.9625

AUM 48547 Peckoltia braueri G 2.95963 -59.9625

AUM 48547 Peckoltia braueri G 2.95963 -59.9625

AUM 39588 Peckoltia sabaji V 5.44198 -66.1507

AUM 39588 Peckoltia sabaji V 5.44198 -66.1507

AUM 39588 Peckoltia sabaji V 5.44198 -66.1507

AUM 39588 Peckoltia sabaji V 5.44198 -66.1507

43

AUM 39313 Peckoltia vittata V 5.42863 -66.1362

AUM 39313 Peckoltia vittata V 5.42863 -66.1362

AUM 39313 Peckoltia vittata V 5.42863 -66.1362

AUM 39313 Peckoltia vittata V 5.42863 -66.1362

AUM 44440 Pseudacanthicus leopardus G 3.91755 -59.1002

AUM 43443 Pseudancistrus sidereus V 2.36280 -66.5648

AUM 43443 Pseudancistrus sidereus V 2.36280 -66.5648

AUM 43443 Pseudancistrus sidereus V 2.36280 -66.5648

AUM 39246 Pseudolithoxus anthrax V 4.08042 -66.8651

AUM 39246 Pseudolithoxus anthrax V 4.08042 -66.8651

AUM 39246 Pseudolithoxus anthrax V 4.08042 -66.8651

AUM 39246 Pseudolithoxus anthrax V 4.08042 -66.8651

AUM 42118 Pseudolithoxus dumus V 3.28998 -66.6000

AUM 42118 Pseudolithoxus dumus V 3.28998 -66.6000

AUM 42118 Pseudolithoxus dumus V 3.28998 -66.6000

AUM 42118 Pseudolithoxus dumus V 3.28998 -66.6000

AUM 42110 Pseudolithoxus tigris V 3.96731 -67.2535

AUM 42110 Pseudolithoxus tigris V 3.96731 -67.2535

AUM 42110 Pseudolithoxus tigris V 3.96731 -67.2535

AUM 42110 Pseudolithoxus tigris V 3.96731 -67.2535

44

ANSP 199622 Scobinancistrus pariolispos BR 3.39040 -52.2243

ANSP 193077 Spectracanthicus L020 'bola blanca' BR 3.22557 -51.4411

45

Appendix B

Species used in the stable isotope analyses. Mean centroid deviation for a species was calculated as the mean of the

mean of each centroid deviation for that species at a given locality.

Species Mean centroid deviation (δ13C) Mean centroid deviation δ15N)

Ancistrus macrophthalmus -1.1241 0.1994

Baryancistrus beggini 0.5517 0.2425

Baryancistrus xanthellus -2.9246 0.7764

Chaetostoma breve -1.6536 0.4203

Chaetostoma microps -1.6082 -0.0524

Chaetostoma stroumpoulos 0.0975 0.0803

Dekeyseria scaphirhyncha 0.0219 -2.2046

Etsaputu relictum -0.6993 0.6747

Farlowella nattereri -0.414 -0.7258

Hemiancistrus subviridis -0.0219 0.6926

Hypancistrus contradens -2.2276 0.8413

Hypancistrus furunculus -0.3681 0.2303

Hypancistrus lunaorum -1.401 0.7272

Hypostomus macushi 2.9494 1.6004

Hypostomus niceforoi 0.8039 -0.367

Hypostomus pyrineusi 0.4748 -0.1336

Hypostomus rhantos 1.8828 -1.2128

Hypostomus taphorni 0.3751 -0.7367

46

Lamontichthys sp -2.2577 0.1555

Lasiancistrus schomburgkii -0.5223 -0.1429

Lasiancistrus tentaculatus -1.45 -0.87

Leporacanthicus galaxias -1.1188 0.9339

Leporacanthicus triactis -2.2826 0.9852

Lithoxus lithoides 0.3889 1.1112

Oligancistrus punctatissimus 0.0518 0.6212

Panaqolus albomaculatus 1.2416 0.3424

Panaqolus gnomus 0.4444 -0.1858

Panaqolus nocturnus -0.2341 -0.4424

Panaque bathyphilus 0.8088 0.0532

Peckoltia braueri -1.8378 1.269

Pseudacanthicus leopardus -0.68 1.9229

Pseudancistrus sidereus 0.1102 0.2475

Pseudolithoxus anthrax -0.057 0.2845

Pseudolithoxus dumus 1.5667 0.3256

Pseudolithoxus tigris -0.4127 0.6909

Rineloricaria fallax 1.5462 -0.2841

Scobinancistrus pariolispos -1.4668 0.9447

Sturisoma monopelte -1.3319 -0.0056

47