ArticlePDF Available

Repeated invasions into the twilight zone: Evolutionary origins of a novel assemblage of fishes from deep Caribbean reefs

May 2016
Molecular Ecology 25(15)

May 2016
25(15)

DOI:10.1111/mec.13704

Authors:

Luke Tornabene

University of Washington Seattle

James van Tassell

American Museum of Natural History

D Ross Robertson

Smithsonian Tropical Research Institute

Carole C. Baldwin

Smithsonian Institution

Mesophotic and deeper reefs of the tropics are poorly known and underexplored ecosystems worldwide. Collectively referred to as the 'twilight zone,' depths below ~30-40 m are home to many species of reef fishes that are absent from shallower depths, including many undescribed and endemic species. We currently lack even a basic understanding of the diversity and evolutionary origins of fishes on tropical mesophotic reefs. Recent submersible collections in the Caribbean have provided new specimens that are enabling phylogenetic reconstructions that incorporate deep-reef representatives of tropical-fish genera. Here, we investigate evolutionary depth transitions in the family Gobiidae (gobies), the most diverse group of tropical marine fishes. Using divergence-time estimation coupled with stochastic character mapping to infer the timing of shallow-to-deep habitat transitions in gobies, we demonstrate at least four transitions from shallow to mesophotic depths. Habitat transitions occurred in two broad time periods (Miocene, Pliocene-Pleistocene), and may have been linked to the availability of underutilized niches, as well as the evolution of morphological/behavioral adaptations for life on deep reefs. Further, our analysis shows that at least three evolutionary lineages that invaded deep habitats subsequently underwent speciation, reflecting another unique mode of radiation within the Gobiidae. Lastly, we synthesize depth distributions for 95 species of Caribbean gobies, which reveal major bathymetric faunal breaks at the boundary between euphotic and mesophotic reefs. Ultimately our study is the first rigorous investigation into the origin of Caribbean deep-reef fishes, and provides a framework for future studies that utilize rare, deep-reef specimens. This article is protected by copyright. All rights reserved.

Deep-reef Caribbean gobies included in this study. (A) Pinnichthys aimoriensis Espirito Santo, Brazil; (B) Varicus cephalocellatus, Bonaire; (C) Varicus decorum, Curac ßao; (D) Varicus veliguttatus, Curac ßao; (E) Psilotris laurae, Bonaire; (F) Varicus sp. 1, Bonaire; (G) Coryphopterus curasub, Curac ßao; (H) Antilligobius nikkiae, Curac ßao; (I) Palatogobius sp. 1, Curac ßao; (J) Palatogobius grandoculus, Curac ßao; (K) Gobiidae sp. 1, Bonaire; (L) Gobiidae sp. 2, Curac ßao; (M) Gobiidae sp. 3, Curac ßao. Photographs by Hudson Pinheiro, Carole Baldwin, Ross Robertson and Pat Colin.

…

Deep-reef goby species included in this study

…

Ancestral habitat reconstruction for Gobiidae (showing relevant subclade only) from 1000 stochastic-mapping simulations on maximum-clade credibility tree from BEAST analysis. Red branches have 100% posterior probability of occupying deep-reef habitat; blue branches have 100% posterior probability of occupying shallow-water habitat.

…

Age estimates of habitat transitions for two possible evolutionary scenarios within the Nes subgroup. (A) Scenario wherein habitat transition occurred on branch leading to (Varicus + Pinnichthys ) + Psilotris clade; (B) and (C). Scenarios wherein independent habitat transitions occurred in the Varicus + Pinnichthys clade (B) and in the Psilotris clade (C), respectively. Note differences in Yaxis scale in each case.

…

Depth distributions for the two ecotypes of deep-reef Greater Caribbean gobies. *Uncertainty regarding a hovering and schooling vs. benthic and solitary lifestyle. **Only one specimen of 593 records of Priolepis hipoliti was collected below this range to 107 m.

…

Figures - uploaded by D Ross Robertson

Content may be subject to copyright.

Content uploaded by D Ross Robertson

Content may be subject to copyright.

Repeated invasions into the twilight zone: evolutionary

origins of a novel assemblage of ﬁshes from deep

Caribbean reefs

LUKE TORNABENE,* JAMES L. VAN TASSELL,†D. ROSS ROBERTSON‡and CAROLE C.

BALDWIN*

*National Museum of Natural History, Smithsonian Institution, PO Box 37012, Washington, DC 20013-7012, USA,

†Department of Ichthyology, American Museum of Natural History, Central Park West at 79th Street, New York,

NY 10024-5192, USA, ‡Smithsonian Tropical Research Institute, Balboa, Panama, Unit 9100, PO Box 0948, DPO AA 34002,USA

Abstract

Mesophotic and deeper reefs of the tropics are poorly known and underexplored

ecosystems worldwide. Collectively referred to as the ‘twilight zone’, depths below

~30–50 m are home to many species of reef ﬁshes that are absent from shallower

depths, including many undescribed and endemic species. We currently lack even a

basic understanding of the diversity and evolutionary origins of ﬁshes on tropical

mesophotic reefs. Recent submersible collections in the Caribbean have provided new

specimens that are enabling phylogenetic reconstructions that incorporate deep-reef

representatives of tropical ﬁsh genera. Here, we investigate evolutionary depth transi-

tions in the family Gobiidae (gobies), the most diverse group of tropical marine ﬁshes.

Using divergence-time estimation coupled with stochastic character mapping to infer

the timing of shallow-to-deep habitat transitions in gobies, we demonstrate at least

four transitions from shallow to mesophotic depths. Habitat transitions occurred in

two broad time periods (Miocene, Pliocene–Pleistocene), and may have been linked to

the availability of underutilized niches, as well as the evolution of morphological/be-

havioural adaptations for life on deep reefs. Further, our analysis shows that at least

three evolutionary lineages that invaded deep habitats subsequently underwent specia-

tion, reﬂecting another unique mode of radiation within the Gobiidae. Lastly, we syn-

thesize depth distributions for 95 species of Caribbean gobies, which reveal major

bathymetric faunal breaks at the boundary between euphotic and mesophotic reefs.

Ultimately, our study is the ﬁrst rigorous investigation into the origin of Caribbean

deep-reef ﬁshes and provides a framework for future studies that utilize rare, deep-reef

specimens.

Keywords: adaptive radiation, Deep Reef Observation Project, Gobiidae, mesophotic reefs, phy-

logenetic comparative methods, phylogeny, speciation

Received 29 October 2015; revision received 20 April 2016; accepted 3 May 2016

Introduction

Exploration of Caribbean deep-reef ﬁshes

Shallow reefs in the Caribbean have been extensively

surveyed over the last half-century, largely due to the

widespread availability of scuba diving technology

beginning in the 1950s. As a result, the diversity and

taxonomic structure of assemblages of reef ﬁshes on

Caribbean reefs living shallower than ~50 m are well

characterized (e.g. B€

ohlke & Chaplin 1993; Smith &

Knopf 1997; Floeter et al. 2008; Humann & Deloach

2014; Robertson & Van Tassell 2015). In contrast, much

more limited information is available on reef-ﬁsh

assemblages from the twilight zone—that is mesophotic

Correspondence: Luke Tornabene,

E-mail: luke.tornabene@gmail.com

Molecular Ecology (2016) doi: 10.1111/mec.13704

(~30–150 m, see Bongaerts et al. 2015) and deeper reefs

that extend down to 300 m or beyond in the same

region. Our current understanding of western Atlantic

ﬁsh communities from the twilight zone comes primar-

ily from a limited number of submersible observations

(Colin 1974, 1976 [depth range sampled =DRS 50–

305 m]; Nelson & Appeldoorn 1985 [DRS 50–915 m];

Parker & Ross 1986 [DRS 23–152 m]; Dennis & Bright

1988 [DRS 20–140 m]; Itzkowitz et al. 1991 [DRS 25–

100 m]; Ross & Quattrini 2007 [DRS 366–783 m]; unpub-

lished Johnson Sea-link Expeditions by Harbor Branch

Institute of Oceanography [DRS 20-915 m]), visual sur-

veys by remote-operated vehicles (Spieler et al. 2007

[DRS 70–100 m]; Bryan et al. 2013 [DRS 50–120 m]; Rosa

et al. 2015 [DRS 30–90 m]), trap or hook-and-line sur-

veys (Bunkley-Williams & Williams 2004) and surveys

from technical divers (Feitoza et al. 2005 [DRS 35–70];

Bejarano et al. 2011 [DRS 50–85 m], Bejarano et al. 2014

[DRS 30–70 m]; Garcia-Sais 2010 [DRS 15–50 m]; Garcia-

Sais et al. 2010 [DRS 30–50 m]; Pinheiro et al. 2015 [DRS

45–130 m]).

There are several challenges when trying to character-

ize a ﬁsh community at depth using the survey meth-

ods above. The majority of the deep-reef studies above

are strictly visual surveys, which often overlook the

small and cryptic but ecologically important, highly

speciose groups of reef ﬁshes, notably the gobies (Gobi-

idae), and blennies (Blenniidae, Chaenopsidae, Labriso-

midae), and other diminutive demersal ﬁshes such as

the basslets (Grammatidae), clingﬁshes (Gobiesocidae)

and some seabasses (Serranidae). Small cryptic ﬁshes

account for a dominant fraction of overall reef-ﬁsh

diversity and contribute substantially to coral-reef

trophodynamics (Ackerman & Bellwood 2000, 2002;

Depczynski & Bellwood 2003). Characterizing this com-

munity of cryptic species typically requires targeted col-

lections using ichthyocides (Robertson & Smith-Vaniz

2008). While this may be possible on deep reefs with

divers using closed-circuit rebreather technology (Pyle

1998, 1999, 2000), these methods are logistically chal-

lenging and require substantial training and experience

to avoid the numerous safety risks associated with deep

diving. Furthermore, bottom time and depth maxima

are still greatly limited, long decompression times are

involved, and rebreathers currently do not provide

access to reefs deeper than about 150 m. Ultimately, as

a result of these challenges, there have been few tar-

geted efforts to collect cryptic, deep-reef ﬁshes in the

wider Caribbean region. Collections made by the John-

son Sea Link submersible program in the 1970s and

1980s provided some valuable specimens, but tissue

samples that would facilitate molecular phylogenetic

analyses, population genetic analyses, divergence-time

estimation and other macro-evolutionary investigations

were not obtained. Thus, we currently lack both a clear

picture of the evolutionary origins of the majority of

deep-reef ﬁshes and a strong temporal understanding

of how and when deep-reef ﬁsh communities evolved

over geologic timescales.

Fish diversiﬁcation on deep reefs: gobies as a model

group

Deep-reef ﬁsh assemblages are more than just an under-

explored extension of the shallow-reef ecosystem.

Assemblages of benthic organisms that form ﬁsh habitat

differ considerably between shallow and deep reefs in

the Caribbean (Goreau & Goreau 1973; James & Gins-

burg 1979; Liddell & Ohlhorst 1988; Liddell et al. 1997;

Reed & Pomponi 1997). Speciﬁcally, the fore-reef slope

shows a marked drop in species richness of sclerac-

tinian corals and calcareous green algae with increasing

depth and changes in geomorphology (Goreau & Gor-

eau 1973; Liddell & Ohlhorst 1988) and coral endosym-

biont assemblages (Bongaerts et al. 2015). These changes

coincide with an increased diversity of sponges and an

increasing predominance of coralline algae with depth,

which in turn provide the foundation for a unique and

diverse ﬁsh community. Reef ﬁsh communities show

trends with depth similar to those for scleractinian cor-

als and calcareous green algae: species richness and

abundance decrease as depth increases, frequently cor-

relating with extent of live coral cover (Lukens 1981;

Nelson & Appeldoorn 1985; Dennis & Bright 1988;

Itzkowitz et al. 1991; but see Pinheiro et al. 2015). Given

that those reported trends are based largely on visual

surveys that, as mentioned, underestimate cryptic spe-

cies diversity, and relationships between reef ﬁshes and

depth should be re-analysed when sufﬁcient specimen-

based data are available. Regardless, many species com-

mon on shallow reefs are rare or absent at mesophotic

depths and beyond, with a signiﬁcant fraction of meso-

photic ﬁsh species belonging to a ‘true deep-reef

fauna’—that is a well-deﬁned group of deep-reef spe-

cialists that are absent from shallow communities (Colin

1974, 1976; Pyle 1999, 2000; Brokovich et al. 2008; Gar-

cia-Sais et al. 2010; Kahng et al. 2011; Bejarano et al.

2014). Much of the true deep-reef fauna consists of

undescribed taxa that potentially include locally ende-

mic species (Colin 1974, 1976; Pyle 1999, 2000).

Habitat partitioning by depth is thus an important

factor structuring reef ﬁsh communities, and in this

regard, the shallow-to-deep reef slopes of Caribbean

islands provide a habitat-rich ecological gradient for

ﬁshes to exploit. Bathymetric niche partitioning has

been observed within lineages of closely related coral

reef ﬁshes such as Greater Caribbean Liopropoma (sea-

basses) (Baldwin & Robertson 2014), Elacatinus (neon

2L. TORNABENE ET AL.

gobies) (Colin 1975, 2010), Coryphopterus (gobies) (Bald-

win & Robertson 2015), as well as NE Paciﬁc Sebastes

(rockﬁshes) (Ingram 2011), and Indo-Paciﬁc Eviota

(dwarfgobies) (Greenﬁeld & Randall 1999). Speciation

driven by depth partitioning may be common in many

other groups of reef ﬁshes and could contribute sub-

stantially to the origins of the ‘true deep-reef fauna’.

Speciﬁcally, deep reefs could provide a variety of

vacant niches that are available to species capable of

adapting to a low-light, cool, resource-limited environ-

ment, thus opening the door for adaptive radiations of

select lineages on deep reefs. We currently lack a clear

understanding not only of the temporal rate of shallow-

to-deep (or vice versa) evolutionary transitions, but also

the extent of speciation that occurred exclusively on

deep reefs following an invasion of a shallow-water lin-

eage. As a result, it is unclear whether the deep-reef

ﬁsh fauna comprises a conglomerate of species that lar-

gely represent single evolutionary offshoots of shallow-

water clades, or whether some lineages successfully

diversiﬁed at mesophotic depths.

For a variety of important reasons Caribbean gobies

represent an ideal group to use as a model exploring

these hypotheses. First, the Gobiidae comprise the most

diverse family of marine ﬁshes, and goby species exhi-

bit a remarkable array of morphological and beha-

vioural adaptations that have facilitated microhabitat

specialization and adaptive radiations in a variety of

marine and coastal freshwater habitats globally (R€

uber

et al. 2003; Yamada et al. 2009; Thacker & Roje 2011;

Zander 2011). Second, gobies are particularly well rep-

resented on coral reefs, where they comprise a major

component of overall diversity and occupy a broad

array of ecological niches (Ackerman & Bellwood 2000,

2002; Herler et al. 2011). In the western Atlantic, there

are nearly 100 species of shallow-water gobies that are

associated with reef habitats (Van Tassell 2011). Third,

gobies have the ability and propensity to speciate

rapidly in association with microhabitat partitioning

(e.g. R€

uber et al. 2003; Tornabene et al. 2013), rendering

them an ideal group for studying potential bathymetric

habitat partitioning in deep-reef ﬁshes. Lastly, gobies

on Greater Caribbean reefs represent multiple indepen-

dent phylogenetic lineages and thus provide an oppor-

tunity to explore the existence of patterns of parallel

evolution.

Deep Reef Observation Project: incorporating deep-reef

ﬁshes into molecular phylogenies

Recently, the Smithsonian Institution’s Deep Reef Obser-

vation Project (DROP) has added a wealth of new infor-

mation to the growing body of knowledge on Caribbean

deep-reef communities. Operating out of Substation

Curac

ßao (http://www.substation-Cuaracao.com), DROP

uses the manned submersible Curasub to explore deep

reefs to 300 m off the coast of Curac

ßao and nearby areas

in the southern Caribbean. Unlike many of the deep-reef

surveys mentioned above, DROP effectively captures

small and cryptic ﬁshes, some of which have been the

basis for descriptions of new species of deep-reef ﬁshes

(Baldwin & Robertson 2013, 2014, 2015; Baldwin &

Johnson 2014; Tornabene et al. 2016). For example, a sin-

gle dive to 198 m off the west coast of Curac

ßao in 2015

resulted in the capture of ~20 specimens of gobies rep-

resenting six undescribed species, in addition to many

other specimens of deep-reef ﬁshes and invertebrates.

Tissue samples from Curasub collections are now being

incorporated into larger molecular phylogenies to assess

the evolutionary origins of deep-reef taxa.

During a series of DROP-Curasub dives to depths of

300 m from 2011 to 2014, specimens of gobies were col-

lected that represent 13 species, most of them recently

described as new or awaiting description (Van Tassell

et al. 2012; Baldwin & Robertson 2015; Tornabene et al.

2016) (Fig. 1). Whole specimens plus tissue samples of

the specimens provide an excellent opportunity to

investigate the phylogenetic relationships of these deep-

reef species relative to their shallow-water goby rela-

tives. By integrating these new species into the most

comprehensive molecular phylogenetic hypotheses of

the Gobiidae to date (see Agorreta et al. 2013), we can

generate a clearer picture of the evolution of Caribbean

deep-reef gobies and investigate the existence and tim-

ing of parallel invasions between mesophotic and shal-

low reefs across independent lineages. Here, we use

multilocus sequence data to infer the phylogenetic posi-

tions of 13 species of deep-reef gobies from the western

Atlantic (Table 1) and 244 species of their relatives. The

resulting phylogeny is then used in combination with

relaxed molecular clock methods and ancestral charac-

ter-state inference to estimate the timing of speciation

events and of bathymetric habitat transitions in gobies,

and to assess whether lineages are diversifying follow-

ing a transition to deep reefs. In addition, we synthesize

depth distribution data for all Caribbean reef gobies to

reveal bathymetric faunal breaks associated with the

boundary between the euphotic and mesophotic zones,

and test the hypothesis that ﬁsh diversity decreases

with increasing depth. Lastly, we discuss the role of

niche partitioning in facilitating the existence of a taxo-

nomically and ecologically diverse assemblage of cryp-

tobenthic ﬁshes at mesophotic depths. This study

represents the ﬁrst attempt to examine the evolution of

multiple lineages of deep-reef ﬁshes in any tropical

region and may serve as a primer for future studies

investigating the evolutionary origins of the deep-reef

fauna in other taxa and regions of the world tropics.

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 3

(A)

(B)

(C)

(D)

(E)

(F)

(G)

(H)

(I)

(J)

(K)

(L)

(M)

Fig. 1 Deep-reef Caribbean gobies included in this study. (A) Pinnichthys aimoriensis Espirito Santo, Brazil; (B) Varicus cephalocellatus,

Bonaire; (C) Varicus decorum, Curac

ßao; (D) Varicus veliguttatus, Curac

ßao; (E) Psilotris laurae, Bonaire; (F) Varicus sp. 1, Bonaire; (G) Cor-

yphopterus curasub, Curac

ßao; (H) Antilligobius nikkiae, Curac

ßao; (I) Palatogobius sp. 1, Curac

ßao; (J) Palatogobius grandoculus, Curac

ßao; (K)

Gobiidae sp. 1, Bonaire; (L) Gobiidae sp. 2, Curac

ßao; (M) Gobiidae sp. 3, Curac

ßao. Photographs by Hudson Pinheiro, Carole Baldwin,

Ross Robertson and Pat Colin.

4L. TORNABENE ET AL.

Methods

Specimen collection and phylogenetic analysis

Collection methods for most of the deep-reef species in

this study follow that of Baldwin & Robertson (2014).

Speciﬁcally, 12 of the 13 deep-reef species used in this

study were captured by the Curasub submersible dur-

ing several dives from 2011 to 2014 (Fig. 1B–L) in Cur-

ßao and Bonaire. Each of the more than 130 Curasub

dives between 2011 and 2015 typically involved 3–6h

of time spent searching for ﬁsh below 60 m. During

dives, the ﬁsh anaesthetic quinaldine was pumped

from a reservoir to a tube connected to one hydraulic

arm, and a suction hose on the sub’s other hydraulic

arm collected ﬁshes and deposited them into an acrylic

container outside the submersible. Specimens were then

retrieved at the surface where they were photographed

and tissue sampled. Tissues were stored in saturated

salt–DMSO (dimethyl sulphoxide) buffer. In addition,

one other deep-reef species (Pinnichthys aimoriensis) was

collected by Hudson Pinheiro and Thiony Simon off

Espirito Santo, Brazil, using closed-circuit rebreathers

(Fig. 1A). Whole genomic DNA was extracted from tis-

sue samples using an automated phenol:chloroform

extraction protocol on the Autogenprep965 (Autogen,

Holliston, MA, USA). Four genes were sequenced for

phylogenetic analysis: mitochondrial cytochrome b

(cytb); the nuclear gene G protein-coupled receptor

(GPR85, aka SREB2); the nuclear gene recombination

activating gene subunit 1 (RAG1); and the nuclear gene

zinc ﬁnger 1 protein (zic1). We did not use the riboso-

mal regions from Agorreta et al. (2013) due to difﬁculty

with ampliﬁcation and questions regarding the homol-

ogy of poorly aligned regions. Primers for PCRs and

thermal proﬁles are identical to those used in Agorreta

et al. (2013).

New sequences from this study were combined with

a reduced data set from Agorreta et al. (2013), in which

we removed three problematic or ‘rogue’ taxa (Kraeme-

ria cunicularia,Schlindleria praematura,Schindleria pietsch-

manni) whose phylogenetic position could not be

resolved and inclusion in the analysis signiﬁcantly

reduced posterior probabilities across the entire tree. To

better capture the diversity of Caribbean reef gobies,

this alignment was then supplemented with cytb and

RAG1 sequences from several species from the genus

Lythrypnus, a shallow-reef group from the western

Atlantic and eastern Paciﬁc Oceans (sequenced by Max-

ﬁeld et al. 2012). We also included sequences for the

shallow-water members of the genus Coryphopterus (12

of 13 species, 92%), several additional shallow-water

species of Gobiosomatini and the Butidae species Kribia

nana to assist with molecular clock calibration (details

below). Sequences were assembled and aligned in Gen-

eious R6 (Biomatters Ltd., available at http://www.-

geneious.com). The total alignment had a length of

4395 bp (83% complete) and included 257 taxa, 35 more

taxa than the largest gobioid phylogeny to date (Agor-

reta et al. 2013). Of particular importance, our data set

is the most comprehensive in terms of western Atlantic

goby species: our matrix includes 50 of the 55 genera of

Caribbean gobies (91%), including 27 of the 30 genera

of the Gobiosomatini (92%), and 23 of the 25 genera of

non-Gobiosomatini gobies (90%). Partitioning scheme

and substitution model choice were chosen with PARTI-

TIONFINDER version 1.0.1 (Lanfear et al. 2012) using the

‘greedy’ algorithm and the Bayesian information crite-

rion. The ﬁnal alignment is available on Dryad

(doi:10.5061/dryad.c3k7s).

The concatenated partitioned data set was analysed

using Bayesian phylogenetic inference with MRBAYES ver-

sion 3.2.1 (Ronquist et al. 2012) on the CIPRES portal

version 3.3 (Miller et al. 2010). The analysis was

Table 1 Deep-reef goby species included in this study

Species Recorded depths (m) Method of capture Clade Habitat use

Pinnichthys aimoriensis 70 Rebreathers Nes subgroup Benthic, solitary

Psilotris laurae 113–137 Curasub Nes subgroup Benthic, solitary

Varicus cephalocellatus 114–159 Curasub Nes subgroup Benthic, solitary

Varicus decorum 99–251 Curasub Nes subgroup Benthic, solitary

Varicus veliguttatus 153–287 Curasub Nes subgroup Benthic, solitary

Varicus sp. 1 250 Curasub Nes subgroup Benthic, solitary

Antilligobius nikkiae 90–198 Curasub Microgobius group Hovering, schooling

Palatogobius grandoculus 224–276 Curasub Microgobius group Hovering, schooling

Palatogobius sp. 1 117–128 Curasub Microgobius group Hovering, schooling

Coryphopterus curasub 70–80 Curasub Coryphopterus Benthic, solitary

Gobiidae sp. 1 120–140 Curasub Priolepis lineage Hovering, schooling

Gobiidae sp. 2 248 Curasub Priolepis lineage Unknown

Gobiidae sp. 3 128–161 Curasub Priolepis lineage Hovering, schooling

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 5

replicated 4 times, each replicate consisting of two par-

allel MCMC runs (four chains each) for a length of

20 million generations. Convergence of parallel runs

was assessed by comparing plots of ln Lscores vs. gen-

eration in TRACER ver.1.5 (Rambaut & Drummond 2007),

the standard deviation of split frequency statistic and

the Potential Scale Reduction Factor statistic from MR-

BAYES, and by visually inspecting the resulting topolo-

gies of maximum-clade credibility (MCC) trees from

each run. We assessed adequate mixing of MCMC

chains by examining estimated sample sizes (ESS) from

each run in TRACER.

Divergence-time estimation and ancestral habitat

inference

We used an uncorrelated, lognormal, relaxed-clock

model to infer divergence times using the program

BEAST 2.2.3 (Bouckaert et al. 2014), as our data did not ﬁt

a strict molecular clock (likelihood ratio test, chi-square

distribution, P>0.05). We used the following three cali-

bration points. First, following Thacker (2014), we set

the crown age of all gobioid ﬁshes (root age of the tree)

to a minimum of 52 Myr with a very soft upper bound

(exponential distribution, mean =10, offset =52), based

on the oldest known fossil gobioid, an Eocene otolith

(Bajpai & Kapur 2004; Gierl et al. 2013). The most recent

common ancestor of K. nana and Bostrychus zonatus was

set to a minimum of 23 Myr with a soft upper maxi-

mum (lognormal distribution, mean =1.5, sigma =0.8,

offset =23), based on a well-preserved whole body fos-

sil of an upper Oligocene Butidae species that was

determined to be the sister taxa of Kribia (Gierl et al.

2013). Lastly, the most recent common ancestor of the

Atlantic and Paciﬁc species of Gobulus was set to a min-

imum age of 3 Myr with a soft upper maximum

(gamma distribution, a=1.9, b=3.0, offset =2.5),

based on the ﬁnal closure of the Isthmus of Panama,

which marks the latest point in which this split could

have occurred. While the date of the ﬁnal closure of the

Isthmus of Panama is still contentious (e.g. Bacon et al.

2015a,b; Lessios 2015; Marko et al. 2015; Montes et al.

2015), the combination of a hard minimum and soft

upper maximum in our calibration point accommodates

ages from multiple hypotheses and allows for ﬂexibility

in the inference of the split between Gobulus species. In

addition, despite uncertainty in the identiﬁcation of

gobioid otoliths, we chose to use the c. 52-million-year-

old Eocene otolith fossil of Bajpai and Kapur (2004) for

the root gobioid calibration point rather than the oldest

articulated skeleton (c. 44 Ma; Gaudant 1996), as this

older fossil agrees better with the considerably older

age estimates of the origin of gobioids derived from

other phylogenetic analyses that included wider

sampling, albeit without fossil gobioid calibration points

(Near et al. 2012; Betancur-R et al. 2013). To remedy the

problem of the BEAST analysis not converging, we used a

ﬁxed topology for the analysis. To do this, we con-

verted our MCC tree from MRBAYES to an ultrametric tree

using the penalized likelihood approach (Sanderson

2002) with the above calibration points, and inputted

this as our starting tree, ﬁxing topology by disabling

topology-sampling MCMC operators in the BEAST analy-

sis. For the BEAST analysis, we partitioned our data set

by gene and assigned each partition a HKY +G substi-

tution model, assigned different clock models for the

cyt b and the group of three nuclear genes, and imple-

mented a Yule speciation prior with parameters condi-

tional on the root age of the tree. The analyses were run

four times, each with 30 million generations. We com-

pared plots of posterior probability vs. generations for

each analysis in TRACER, ensured adequate ESS (>200),

and visually inspected divergence times of the four

MCC trees to see that each analysis recovered similar

results.

To infer the frequency, dates and topological location

of habitat transitions between shallow-water and deep

reefs on our phylogeny, we used stochastic character

mapping of discrete traits via SIMMAP (Bollback 2006) in

R(R Core Team 2015), through the function make.sim-

map (‘PHYTOOLS’ package; Revell 2012). Species in our

phylogeny were coded as being deep-reef species if

they are known exclusively from depths at or below

60 m. Species with depth ranges that extend no deeper

than 60 m were coded as shallow-water species. Very

few species of Caribbean reef gobies (four species of

Coryphopterus, four species of Elacatinus and Priolepis

hipoliti) that are primarily shallow-reef species have

been recorded (rarely) in upper-mesophotic depths.

More than 130 Curasub dives of 3- to 6-h duration/dive

spanning 5 years have yielded only one specimen of

P. hipoliti from 107 m and only three Coryphopterus

venezuelae from 65 to 69 m. None of the other 49 spe-

cies of shallow-reef gobies from Curac

ßao were

observed or collected below ~60 m. Lastly, species in

the genus Bollmannia frequently extend into mesophotic

depths, but these species are not reef specialists and

instead occur in association with burrows over ﬂat

mud bottoms that are largely homogeneous with

respect to depth. Because we are primarily interested

in transitions into deep-reef habitats that are ecologi-

cally distinct from shallow-water habitats, we coded

Bollmannia as shallow-water species for the purpose of

the SIMMAP analysis.

The SIMMAP analysis assumed a Markov model with a

symmetrical rate matrix for character-state changes,

where changes between-character states in either direc-

tion were considered equally probable a priori. We ran

6L. TORNABENE ET AL.

the SIMMAP analysis on both a single MCC tree from our

BEAST analysis, as well as a subset of 100 random trees

from the post-burn-in posterior distribution of trees

from the BEAST analysis. The subset of post-burn-in trees

was analysed to accommodate divergence-time uncer-

tainty in our estimation of the timing of habitat transi-

tions. For the analysis of the single MCC tree, we used

10 000 MCMC iterations to sample the posterior distri-

bution of possible shallow/deep transition rates. For

computational efﬁciency, in the analysis of 100 post-

burn-in trees, we used a single maximum-likelihood

point estimate for estimating the transition rate matrix

rather than using MCMC. Each SIMMAP analysis

included 100 character-mapping simulations per tree

(100 trees 9100 simulations =10 000 total simulations

for our analysis of 100 post-burn-in trees). For the SIM-

MAP analysis of our single MCC BEAST tree, the results of

all simulations were visually summarized on the tree

using the function density.map (‘PHYTOOLS’ package; Rev-

ell 2012). We also extracted the estimated dates of habi-

tat transitions for each unique lineage of deep-reef

gobies from our SIMMAP analysis of post-burn-in trees.

These estimates of transition times were plotted as his-

tograms, and we compared the timing of habitat transi-

tions across groups using ANOVA to determine whether

independent groups of gobies were simultaneously

transitioning between shallow and deep habitats.

Depth distributions of Caribbean reef gobies

Rather than sampling in both shallow and deep water

that would enable the identiﬁcation of any clear-cut

break point between euphotic (shallow) and mesopho-

tic (deep) faunas, studies of mesophotic ﬁshes typi-

cally involve sampling only below some arbitrary

depth between ~30 and 60 m. Hence, to provide a less

subjective estimate of the depth zone at which there

might be a changeover between the euphotic and

mesophotic components of the Greater Caribbean reef-

goby fauna, we examined the known depth distribu-

tions of all members of that fauna. Using those data,

we assessed whether and at which depth there is any

peak in the relative abundance of depth-range end-

points (both depth maxima and minima) that would

indicate a changeover in faunal composition (see Roy

et al. 1998).

To determine depth distributions of gobies, we col-

lected records for the 95 species that are associated with

reefs in the western Atlantic. To mitigate biases from

incomplete data, we limited our depth data to species

from the western Atlantic Ocean where our groups of

interest occur, and where the information on goby

depth distributions and diversity is considerably more

complete. Unlike most western Atlantic gobies, for

which collections are abundant and identiﬁcations of

species have been veriﬁed by the authors, a great num-

ber of Indo-Paciﬁc gobiid taxa are poorly known and

virtually no depth-range data are available—especially

for deep-reef taxa. Depth data used here were gathered

from databases based only on physical collections rather

than visual observations, given the inherent difﬁculty of

identifying many species of gobies in the ﬁeld. Speciﬁ-

cally, we used data from Robertson & Van Tassell

(2015), supplemented with museum records from the

American Museum of Natural History online database

and from FISHNET2 (www.ﬁshnet2.com), which queries

73 natural history collections from around the world.

To highlight bathymetric breaks, we divided the

euphotic–mesophotic zone into 10-m intervals and cal-

culated the total number of species that were present in

each depth interval. We then reﬁned this count by tabu-

lating the number of species with depth minima or

maxima in each interval. Lastly, to remove the effect of

uneven species richness across depth intervals and

identify potential bathymetric faunal breaks, we divided

the total number of species with depth endpoints (mini-

mum or maximum) in each interval by the total number

of species present in that interval, giving us a measure

of the relative number of species that start or end in the

given 10-m depth interval. For species present at 250–

300 m, we refrained from designating a speciﬁc depth

maximum due to infrequent collections and the sam-

pling bias associated with the maximum depth reach-

able by the Curasub (300 m).

Results

Phylogenetic analysis

Thirteen deep-reef species were recovered in three

monophyletic clades and two single-species lineages

within the Gobiidae (Fig. 2). The majority of the deep-

reef species belong to the tribe Gobiosomatini. Within

that tribe, three species in two genera, Antilligobius and

Palatogobius, were recovered within the Microgobius

group (sensu Birdsong et al. 1988; Van Tassell et al.

2012), and the two genera were recovered as sister taxa

with strong support. The remaining six deep-reef spe-

cies of gobiosomatins belong to the Nes subgroup (sensu

Van Tassell et al. 2012; Tornabene et al. 2016), including

four species of Varicus (V. cephalocellatus,V. decorum,

V. veliguttatus, and the undescribed Varicus sp. 1;

Fig. 1B–D,F), one Psilotris (P. laurae; Fig. 1E) and one

Pinnichthys (P. aimoriensis; Fig. 1A). The mesophotic

species Coryphopterus curasub was recovered as sister to

a clade containing Coryphopterus dicrus,Coryphopterus

glaucofraenum,Coryphopterus tortugae and Coryphopterus

venezuelae.Coryphopterus was recovered as sister to

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 7

Lophogobius, with the western Atlantic Coryphopterus

kuna falling outside those taxa. The remaining three

deep-reef species (Gobiidae sp. 1, sp. 2 and sp. 3;

Fig. 1K–M) represent a new genus and form a single

clade within the Priolepis lineage (sensu Agorreta et al.

2013), which also contains the New World genus

Lythrypnus, the Indo-Paciﬁc genera Trimma and Feia,

and the Indo-Paciﬁc/Atlantic genus Priolepis (Fig. 2).

Divergence-time estimation and mapping of habitat

transitions

Our SIMMAP analysis (Fig. 3) indicates that there were

between 4 and 6 (mean 5.2) transitions between shal-

low-water and deep-reef habitats in our study group.

The number of habitat transitions on the tree differed

depending on the likelihood estimation of the

Fig. 2 Bayesian maximum-clade credibility trees of selected Gobiidae. (A) Phylogram from MRBAYES, node labels are posterior proba-

bilities. (B) Time-calibrated phylogeny from BEAST, node labels and bars are median divergence-time estimates with 95% highest pos-

terior density intervals. Clade names follow that of Agorreta et al. (2013) and Van Tassell et al. (2012). Western Atlantic species are

highlighted in grey. Stars denote deep-reef species.

8L. TORNABENE ET AL.

Fig. 3 Ancestral habitat reconstruction for Gobiidae (showing relevant subclade only) from 1000 stochastic-mapping simulations on

maximum-clade credibility tree from BEAST analysis. Red branches have 100% posterior probability of occupying deep-reef habitat;

blue branches have 100% posterior probability of occupying shallow-water habitat.

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 9

transition rates and the branch lengths of individual

tree analysed. Most habitat transitions were from shal-

low water to mesophotic habitats (mean 4.49 transi-

tions per simulation), while deep-to-shallow habitat

transitions were far less common (mean 0.73 transi-

tions per simulation). The only potential case of deep-

to-shallow habitat transition occurred within the Nes

subgroup, as a reversal within an otherwise deep-

water clade. In this group, the SIMMAP analysis inferred

two different scenarios for the evolution of deep-reef

species (Figs 3 and 6); (i) a shallow-to-deep transition

occurred on the branch leading to the clade containing

Varicus,Psilotris and Pinnichthys, followed by a deep-

to-shallow reversal in the ancestor of Psilotris celsa and

Psilotris kaufmani; or (ii) two shallow-to-deep transi-

tions, one in the ancestor of the Varicus +Pinnichthys

clade and the other in P. laurae. The latter scenario is

more probable as it occurred in approximately twice

the number of simulations as the former scenario

(Fig. 4).

From a familywide perspective, our SIMMAP analysis

indicates that shallow-to-deep transitions occurred pri-

marily during two distinct phases (Fig. 4). The ﬁrst

phase occurred between 10 and 30 Ma and involved

two lineages, the Antilligobius +Palatogobius clade and a

clade of three undescribed species in the Priolepis lin-

eage. The second phase, which involved at least the

divergence of Coryphopterus curasub from the shallow

members of Coryphopterus, occurred within the last

10 Myr. Depending on the SIMMAP scenario, members of

the Nes subgroup may have transitioned to deep reefs

during both the early and late phases (if P. laurae transi-

tioned to deep reefs independently of Varicus +Pin-

nichthys), or exclusively during the early phase (if the

transition occurred on the branch leading to Vari-

cus +Pinnichthys +Psilotris). In the latter scenario, the

timing of the single deep-to-shallow reversal in Psilotris

is consistently estimated within the last 10 Myr (mean

4.71 Ma). Although a familywide perspective indicates

two major pulses of shallow-to-deep transitions over

time, as evident by the strongly bimodal distribution in

Fig. 4, the mean estimates for the dates of habitat transi-

tions from a clade-by-clade perspective are signiﬁcantly

different from one another (ANOVA,P<0.001; pairwise

Tukey’s HSD P<0.001; Figs 5 and 6), indicating that

each transition could arguably be considered temporally

distinct.

Finally, our results indicate three instances in which

evolutionary lineages have subsequently diversiﬁed

after transitioning to deep reefs (Nes subgroup, Priolepis

lineage, Antilligobius +Palatogobius lineage). Only in the

most recent shallow-to-deep transitions (Coryphopterus

curasub and perhaps P. laurae) do deep-reef species

appear to represent single evolutionary offshoots. The

one possible deep-to-shallow transition led to subse-

quent diversiﬁcation of shallow-water Psilotris (P. kauf-

mani and P. celsa).

Depth distributions of Caribbean reef gobies

Depth distribution data for our 13 species of deep-reef

gobies indicate very little depth range overlap between

deep and shallow members of the Nes group, Microgo-

bius group and Priolepis lineage (Fig. 7). Data for Car-

ibbean reef gobies as a whole indicate that species

richness decreases as depth increases (Fig. 8A). Most

species (67) have the upper limit of their depth distri-

butions between 0 and 10 m, and very few species

have the lower limit of their depth distributions below

70 m (Fig. 8B). The depth intervals that contain the

most species’ depth maxima are 21–30 m (15 species)

and 51–60 m (12 species; Fig. 8B). Two peaks in the

relative abundance of depth-range endpoints are evi-

dent in Fig. 8C: at 0–10 m and between 51 and 70 m.

Deeper peaks than the latter could well represent sam-

pling artefacts due to small numbers of species pre-

sent.

Fig. 4 Age estimates of habitat transitions across Greater Carib-

bean members of the family Gobiidae from SIMMAP simulations

on 100 trees from BEAST analysis.

10 L. TORNABENE ET AL.

Discussion

Eco-evolution of deep-reef species in the Microgobius

group (Gobiosomatini)

Antilligobius and Palatogobius, sister taxa in our phy-

logeny (Fig. 2), share several ecological characteristics

that are unique within the Microgobius group. In addi-

tion to occurring at greater depths, Antilligobius and

Palatogobius are the only group members typically

found hovering in the water column over hard sub-

strates. In contrast, most species of other genera in that

group (Bollmannia,Akko,Microgobius and Parrella) are

strictly associated with soft mud and ﬁne sediments,

and many are known to use burrows as substrate

refuges. Antilligobius and Palatogobius were commonly

observed during Curasub dives hovering up to 50 cm

above rocky-reef slopes or in front of small caves or

Fig. 5 Age estimates of habitat transi-

tions for selected clades of Greater Carib-

bean gobies. (A) Priolepis lineage

shallow-to-deep transition; (B) Microgob-

ius group shallow-to-deep transition; (C)

Coryphopterus (Lophogobius group) shal-

low-to-deep transition. Note differences

in Y-axis scales.

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 11

crevices along vertical rocky walls in Curac

ßao, often in

groups of tens to scores of individuals, and did not

retreat to benthic shelters when disturbed. Such near-

bottom schooling behaviour is rare in the Microgobius

group and in gobies in general. When hovering in

schools, Antilligobius feeds on copepods and other zoo-

plankton (F. Young & D. Marine, personal communica-

tion). Similarly, Palatogobius species may also be

planktivorous. An increase in the percentage of plank-

tivorous ﬁshes with increasing depth along a reef slope

has been documented in the Indo-Paciﬁc (Thresher &

Colin 1986; Brokovich et al. 2008), and schooling cou-

pled to planktivory may be selectively advantageous

features for ﬁsh lineages expanding onto deep

reefs. Antilligobius and Palatogobius display some

morphological features that may be adaptive for a

hovering lifestyle at mesophotic depths. Although the

gas bladder is absent or variably developed in many

gobiid groups (McCune & Carlson 2004), it is present in

Antilligobius and Palatogobius sp. 1, and a modiﬁcation

of the haemal arch in these species (see Van Tassell

et al. 2012: Fig. 8B) allows the well-developed gas blad-

der to extend throughout the elongate abdominal cavity

and into the expanded haemal arch. The cleared and

stained specimens identiﬁed as Palatogobius paradoxus

examined by Van Tassell et al. (2012) had neither an

expanded haemal arch nor a gas bladder. However,

radiographs of the holotype of P. paradoxus and recently

available photographs of what appears to be P. para-

doxus or another species of Palatogobius from Curac

ßao

Fig. 6 Age estimates of habitat transitions

for two possible evolutionary scenarios

within the Nes subgroup. (A) Scenario

wherein habitat transition occurred on

branch leading to (Varicus +Pin-

nichthys)+Psilotris clade; (B) and (C). Sce-

narios wherein independent habitat

transitions occurred in the Varicus +Pin-

nichthys clade (B) and in the Psilotris clade

(C), respectively. Note differences in Y-

axis scale in each case.

12 L. TORNABENE ET AL.

do show an expanded haemal arch and well-developed

gas bladder. Currently, P. paradoxus is known from

shallower depths (20–80 m, Greenﬁeld 2002) than either

Palatogobius grandoculus or Palatogobius sp. 1 (see

Table 1; Fig. 7). A well-developed gas bladder may pro-

vide an energy-efﬁcient means of controlling buoyancy,

particularly in the resource-limited environment of deep

reefs (Kahng et al. 2011). Further, such a bladder may

also serve to amplify sound for enhancing communica-

tion in low-light habitats. Sound production is an

important method of communication in gobies, and sev-

eral species are known to produce sounds for both sex-

ual and aggressive signalling (e.g. Mok 1981; Tavolga

et al. 1981; Torricelli et al. 1990; Lugli et al. 1995; Mala-

vasi et al. 2008; Horvati

cet al. 2015). An expanded hae-

mal arch has evolved independently in several other

groups of hovering gobiids, including the pantropical

genus Ptereleotris (Birdsong et al. 1988) and some species

of the Indo-central Paciﬁc genus Trimma (Winterbottom

1984; Winterbottom et al. 2007), where it was hypothe-

sized to serve a similar function with regards to allow-

ing expansion of the gas bladder. Although some Indo-

Paciﬁc representatives of Ptereleotris and Trimma (an

exclusively Indo-Paciﬁc genus) have been observed at

mesophotic depths and form small groups or schools,

and those species were not available for inclusion in

our phylogeny.

Our SIMMAP analysis on the sample of post-burn-in

BEAST trees indicates that the transition from shallow

water to deep reefs in the Antilligobius +Palatogobius

clade occurred c. 10–30 Ma (mean 19.2 Ma; Fig. 5B).

This overlaps broadly with the deep-reef transition in a

clade of three new species from the Priolepis lineage

(Fig. 5A), as well as with the earliest estimates of the

deep-reef transition in the Nes subgroup (Fig. 6A). The

three species in the Priolepis group also display hover-

ing behaviour, and at least one of them (Gobiidae sp. 3)

has been observed in schools of hundreds of individu-

als at Curac

ßao. In one instance, the Curasub collected

both Palatogobius sp. 1 and Gobiidae sp. 3 from a single

mixed school of scores of ﬁshes that included individu-

als of both species.

The Antilligobius +Palatogobius clade represents a

clear example of a lineage diversifying on deep reefs.

After the initial shallow-to-deep transition at the base of

this clade, the split between Antilligobius and

Fig. 7 Depth distributions for the two

ecotypes of deep-reef Greater Caribbean

gobies. *Uncertainty regarding a hover-

ing and schooling vs. benthic and soli-

tary lifestyle. **Only one specimen of

593 records of Priolepis hipoliti was col-

lected below this range to 107 m.

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 13

Palatogobius was estimated at 13.7 Ma [95% highest pos-

terior density (HPD) =8.3–21.2 Ma], followed by the

divergence between the two Palatogobius species

c. 7.9 Ma (95% HPD =4.1–12.8 Ma). In addition to these

two speciation events, Gilbert (1971, 1977) noted that

there may be an additional undescribed species of

deep-water Palatogobius from Panama, indicating

another possible speciation event at depth. With the

exception of the Curasub specimens, the few specimens

of Palatogobius in collections have come from trawls or

dredges primarily over ﬂat bottoms. In contrast, obser-

vations from Curasub dives indicate that some species

in this genus are abundant and conspicuous at the junc-

tion of sand and rock substrata with rocky reefs. The

exception is P. paradoxus, which is known from shal-

lower water and lives on unconsolidated bottoms. If the

deep-living species in this genus indeed prefer more

complex substrates that are intentionally avoided by

benthic sampling gear, the genus may contain many

more species than those currently known simply due to

a sampling bias. In contrast, Antilligobius nikkiae has

been observed or collected from Belize, Bahamas, Cur-

ßao, Cuba, Mexico and Puerto Rico, and all morpho-

logical evidence suggests the presence of a single

species throughout the Caribbean. However, this has

yet to be tested with DNA sequence data. Nonetheless,

the Antilligobius +Palatogobius clade as a whole presents

another example of a lineage of Gobiosomatini that has

invaded a novel niche and subsequently diversiﬁed,

providing additional support that the Gobiosomatini is

one of the best known cases of adaptive radiation in the

marine environment (R€

uber et al. 2003).

Eco-evolution of deep-reef ﬁshes within the Nes

subgroup (Gobiosomatini)

The SIMMAP analysis suggests that in the Nes clade (Fig. 2)

there was either a single shallow-to-deep transition fol-

lowed by a deep-to-shallow transition in Psilotris

(Fig. 6A) or that there were two independent shallow-to-

Fig. 8 Depth distribution data for 95

species of Greater Caribbean reef gobies.

(A) Total number of species present in

each depth interval; (B) total number of

species that have their maximum or min-

imum depth within each 10-m interval;

depth interval that have their minimum

or maximum depth within that interval.

Dotted lines indicate the inferred transi-

tional zone between the shallow-reef and

deep-reef faunas. Numbers above select

bars indicate the total number of species

present in that interval.

14 L. TORNABENE ET AL.

deep transitions (Fig. 6B,C). The latter scenario appeared

in approximately twice the number of SIMMAP simulations

as the former, and hence was statistically more probable.

Pending additional investigation, we consider both sce-

narios equally plausible at this point.

With the exception of a single species, Nes longus,

which typically lives in burrows with Alpheus snapping

shrimp, shallow-water species in this group are consid-

ered to be strictly cryptobenthic, occurring around vari-

ous rock or calcareous substrates in the littoral or

sublittoral zone. Many species occupy tiny interstitial

spaces on or around coral and rocky reefs. This habitat

specialization may have coincided with several morpho-

logical adaptations, including a reduction in body size

(standard length <20 mm in many species), loss of pel-

vic ﬁn fusion, reduction in sensory head pores and

canals, reduction in squamation and cryptic coloration

(Findley 1983). The deep-reef members of the Nes clade

have retained many of these characters, but differ eco-

logically in some regard. Most deep-reef species in the

Nes subgroup are larger and more robust than their

shallow-water relatives and have bright yellow body

coloration; several were collected while perching on

open sand and rubble substrates. The phenotypic and

behavioural differences between shallow and deep-reef

species could well reﬂect different selective forces

between the two habitats. Total ﬁsh abundance and

diversity generally are lower on deep reefs in compar-

ison with shallow reefs (Brokovich et al. 2008; Rosa et al.

2015), which may result in fewer predators (both in

abundance and diversity) and less competition for space

and resources. A greater availability of niche options in

deep-reef habitats than shallow habitats, where goby

diversity and abundance is much higher (e.g. Brokovich

et al. 2008), may explain the strong bias towards shal-

low-to-deep transitions over deep-to-shallow transitions

(4.4 vs. 0.7) among Caribbean gobies. Factors such as

reduced competition and predation on deeper reefs

could relax the evolutionary constraints on body size

and reduce the selective advantages of an extreme cryp-

tobenthic lifestyle. In this regard, the invasion of deep

reefs by Nes-clade gobies from shallow-water habitats

may have created a speciation opportunity by enabling

the use of deep-reef niches unoccupied by other goby

species. This ‘ecological release’ may have enabled the

subsequent speciation events in the deep-reef lineages

of this clade, facilitating a radiation into an otherwise

species-depauperate ecosystem from the mid-Miocene

and throughout the Pliocene. There may be additional

speciation events at depth in the Nes subgroup not

reported here. Several deep-water species in the Nes

clade were not sampled in our study (e.g. Pinnichthys

prolata,P. bilix,P. atrimela,Varicus benthonis,V. vespa,

V. bucca,V. marilynae), although recent combined

molecular and morphological analyses strongly support

that these species are members of the deep-reef clades

identiﬁed here (Tornabene et al. 2016). Thus, it is proba-

bly that the deep-reef radiation within the Nes subgroup

is even more substantial than our phylogeny depicts.

Eco-evolution of deep-reef ﬁshes in the Priolepis and

Lophogobius lineages

Three deep-reef species (Gobiidae sp. 1–3 in Fig. 2)

were recovered in the Priolepis lineage—an exception-

ally diverse clade consisting primarily of species that

live on shallow reefs (Fig. 2). There are scant records of

gobies from this group in the Indo-Paciﬁc region from

depths that extend marginally, or in some cases consid-

erably, into the mesophotic zone. None of these Indo-

Paciﬁc species were available for inclusion in this study,

as very few gobies have been collected from tropical

Indo-Paciﬁc deep reefs. The mean estimate for the tim-

ing of the shallow-to-deep habitat transition in the Pri-

olepis lineage is 20.4 Ma (standard deviation =7.0 Myr;

Fig. 5A). However, the age of this lineage is probably to

change with increased taxon sampling. We are rela-

tively conﬁdent that the putative new genus (Gobiidae

sp. 1–3) belongs in the Priolepis lineage (posterior proba-

bility =1.0), but the position within this group is uncer-

tain and may be clariﬁed with the addition of more

taxa. Our phylogenetic sampling of Indo-Paciﬁc gobies

is limited, and it is probably that the true sister group

to the new genus is missing from our study. Increased

sampling leading to the inclusion of a more closely

related sister group to this clade would have two

impacts on the inferred age of the shallow-to-deep tran-

sition: (i) the age of the new genus, and thus the age of

the habitat transition, will be shifted towards the pre-

sent, as the addition of a sister group will shorten the

internal branch leading to the most recent common

ancestor of Gobiidae sp. 1, sp. 2 and sp. 3; (ii) adding

more taxa to the Priolepis group may cause the width of

our age estimates to become narrower for this group (as

well as for the corresponding habitat transition), due to

additional information being available on mutations

occurring along internal branches within the Priolepis

lineage. These branches are currently very long and

have ages that are difﬁcult to estimate precisely, as indi-

cated by the wide age range in Fig. 5A. Currently, the

age estimates for the habitat transition in the Priolepis

group overlap broadly with that of the Antilligob-

ius +Palatogobius clade (Fig. 5B), and this overlap could

become more precise with the addition of closely

related species within the Priolepis lineage. Little is

known about the ecology of Gobiidae sp. 1 and sp. 2, as

both are known from only a limited number of speci-

mens; however, Gobiidae sp. 3 is ecologically similar to

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 15

Antilligobius and Palatogobius in hovering above the sub-

strate in small schools.

Coryphopterus curasub is the only known deep-reef

member of the Lophogobius lineage and is one of the

most recent invasions of deep reefs among the Greater

Caribbean gobies studied here. The mean estimate for

the timing of the shallow-to-deep transition for C. cura-

sub is 5.49 Ma (standard deviation =3.42 Myr; Fig. 5C).

Species of Coryphopterus can be divided into two general

ecological groups based on their behaviour. Coryphop-

terus personatus and Coryphopterus hyalinus are the small-

est species in the genus, and both form large

aggregations that hover in the water column within

~50 cm above the substrate where they feed on zoo-

plankton. They are frequently the most numerically

abundant species of gobies on shallow reefs in the Car-

ibbean (Greenﬁeld & Johnson 1999; Dominici-Arose-

mena & Wolff 2005). The remaining species in the

genus, including the deep-reef C. curasub, are benthic

species that typically perch on sand, rock or coral sub-

strates, solitarily or in loose aggregations of a few indi-

viduals, and move about in contact with the substrate.

It seems paradoxical that planktivory, hovering and

schooling are all common and potentially adaptive

behaviours in deep-reef gobies, including several of

those studied here, yet the species of Coryphopterus that

display these behaviour are not found below 50–72 m

depth, and the only ‘true’ mesophotic Coryphopterus col-

lected to date is a demersal species found on sand and

rubble (Baldwin & Robertson 2015; Fig. 6). This paradox

may be related to the timing of diversiﬁcation events

and the sequence of habitat transitions across gobies as

a whole, coupled with the temporal availability of

niches. The mean age of the genus Coryphopterus is esti-

mated at 13.36 Ma, and the origin of the clade contain-

ing the two hovering species is estimated at 2.54 Ma.

Our analysis of the timing of habitat transitions indi-

cates that by the middle-to-late Miocene (10–15 Ma),

deep reefs were already inhabited by the three hovering

planktivorous species from the Microgobius group, plus

another three from the Priolepis lineage. Ecologically,

these mesophotic species and the shallow hovering spe-

cies of Coryphopterus are very similar, and could well

occupy similar microhabitat and trophic niches on reefs.

The presence of large hovering groups of Antilligobius,

Palatogobius and Gobiidae sp. 3 on deep reefs could

have restricted the ecologically similar hovering Cory-

phopterus species to shallow depths. Both the absence of

hovering Coryphopterus on deep reefs and the restriction

of C. curasub to the upper edge of the mesophotic zone

may also be due, in part, to these relatively young spe-

cies not having sufﬁcient time to develop a complex set

of habitat-speciﬁc adaptations for exploiting the meso-

photic zone on Caribbean reefs.

Bathymetric trends and oceanographic considerations

Our depth data for all Caribbean gobies highlight two

major trends. First, as observed in reef-building corals

and reef ﬁshes in general, goby species richness

decreases as depth increases (Fig. 8A). The most promi-

nent bathymetric break in Caribbean reef-goby distribu-

tions occurs between 50 and 70 m (Figs 7 and 8).

Among them, 14 species (61%) that occur between 51

and 60 m and 10 species (63%) that occur between 61

and 70 m have depth ranges that either start or end in

this zone (Fig. 8B,C). These high frequencies of depth

minima and maxima in the 50–70 m zone suggest that

it represents the transitional zone between the shallow-

reef community and the true deep-reef fauna, at least in

the case of this taxon and biogeographic region.

The exact depths of this transitional zone and of spe-

cies depth ranges in general should be interpreted cau-

tiously. The 50–70 m transitional zone may also be

somewhat biased by historical collection patterns. The

recreational scuba limit is ~40 m, and while there is a

history of a handful of ichthyologists diving slightly

below this limit to collect and observe ﬁshes in the Car-

ibbean, it is probably that this limit explains the known

depth maxima for some species. However, in over 130

submersible dives spanning 5 years at Curac

ßao, there

have been no observations or collections of shallow-

water gobies at depths greater than 70 m, despite their

abundance and diversity (49 species) on shallow reefs

there. Thus, historical sampling bias alone cannot

account for the 50–70 m transitional zone observed

here. Beyond any potential sampling artefacts, the maxi-

mum depth of a species is probably determined by a

complex combination of factors such as temperature,

light and substrate—all of which may not correlate with

depth in a uniform manner across a species’ entire geo-

graphic range. Nonetheless, the preliminary bathymetric

patterns shown here (Figs 7 and 8) reinforce conclu-

sions of prior studies on reef ﬁshes below 50 m, which,

although limited in number and scope, collectively

point to the existence of a unique, truly deep-reef fauna

comprising species that are rare or absent from shallow

reefs. Brokovich et al. (2008) noted that the bathymetric

fauna breaks in reef ﬁshes correlated with a decrease in

branching corals with increasing depth. The overall

drop in coral cover may explain why some species of

gobies that are dependent on reef-building corals as

adults are absent from mesophotic reefs (e.g. some Ela-

catinus spp., Coryphopterus lipernes). On the other hand,

Srinivasan (2003) found that for several species of coral-

reef ﬁshes, there were signiﬁcant differences in recruit-

ment at different depths despite similar substrates

across depths. This suggests that species depth prefer-

ences may be driven in part by other factors beyond

16 L. TORNABENE ET AL.

substrate that could impact ﬁshes at early life history

stages, including abiotic factors such as light and tem-

perature. We recorded water temperature for 1 year

(September 2012 to August 2013) at 30-m increments

along the deep-reef slope (0–250 m) at Substation Cur-

ßao and found signiﬁcantly different seasonal trends at

euphotic vs. mesophotic depths. From September to

December, depths 50 m and shallower routinely

reached 29 °C, whereas depths from 75 to 250 m ranged

from 15 to 25 °C. In late December, water temperatures

at 50 m and shallower dropped sharply from 29 to

25 °C, whereas temperatures at deeper depths (180–

250 m) remained constant or even increased by as much

as 2 °C (75–220 m). While these data represent only a

single year at a single locality, they illustrate the possi-

bility that signiﬁcant seasonal differences in tempera-

ture with depth may be one of several mechanisms

driving the bathymetric faunal break observed in gobies

and other reef ﬁshes.

Glacioeustatic sea-level ﬂuctuations are one possible

mechanism that could initiate habitat transitions from

shallow to deep reefs. After dramatic drops in sea levels

during glacial cycles, subsequent rises in sea level

would cause shallow-reef fauna to move with the

changing seas, ultimately abandoning the areas occu-

pied during low sea level stands in favour of warmer

shallow water. Similarly, increases in sea surface tem-

perature alone (in the absence of corresponding sea-

level ﬂuctuations) may cause species to initially move

deeper in the water column to avoid thermal stress,

and subsequently vacate deep-reef habitats when tem-

peratures cool. In both scenarios, abandoned deep-reef

habitats would then represent underutilized niches

available to species that are able to stay deep and adapt

to the new deep-reef conditions. Thus, under both sce-

narios we would expect to see shallow-to-deep habitat

transitions immediately following periods where sea

level fell dramatically and subsequently rose in a short

time period, or where temperatures rose dramatically

and subsequently dropped. When our estimates of habi-

tat transition times are plotted against sea-level ﬂuctua-

tions and global temperatures (Fig. S1, Supporting

information), our age estimates of deep-reef invasions

overlap broadly with dramatic sea level drops at 30, 10

and 0–4 Ma, and also with abbreviated warming-to-

cooling climate cycles at 26–28 and 13–18 Ma. However,

the wide ranges in our estimates of transition times

obscure any obvious causative relationship between

deep-reef invasions, sea level and global climate.

In summary, the recent resurgence in the reconnais-

sance of Caribbean deep reefs, driven in large part by

DROP, is further illuminating a true, deep-reef, Carib-

bean ﬁsh fauna and has enabled the discovery of

numerous new species (Van Tassell et al. 2012; Baldwin

& Robertson 2013, 2014, 2015; Baldwin & Johnson 2014;

Tornabene et al. 2016). The deep-reef gobies investi-

gated here are a major component of this deep-reef ﬁsh

fauna, which collectively represent ﬁshes that are eco-

logically divergent from their shallow-water counter-

parts. The deep-reef Caribbean goby fauna formed via

several independent invasions from shallow reefs dur-

ing the Miocene and Pliocene. While our estimates of

the dates of each of these invasions are signiﬁcantly dif-

ferent from each other, the habitat transitions can be

broadly grouped into two main phases (Figs 3 and 4),

each of which was followed by subsequent adaptive

radiations on deep reefs. The initial phase occurred

c. 10–30 Ma and involved independent, near-simulta-

neous shallow-to-deep transitions in the Priolepis lineage

and the Microgobius group. Both of these lineages

underwent subsequent diversiﬁcations as hovering,

school-forming gobies on deep reefs. The second broad

period of deep-reef radiation occurred within the last

10 Myr and included predominately demersal species

that inhabit sand-rubble substrates—C. curasub and,

possibly, the Nes clade, although the ancestors of the

latter may instead have invaded during the ﬁrst phase.

It should be noted that while our taxon sampling of

Caribbean gobies is robust (91% of Caribbean genera;

92% of Gobiosomatini genera; 90% of non-Gobiosoma-

tini genera), our sampling across the entire Gobiidae is

much more limited. A more comprehensive sampling

across the gobiid phylogeny could affect the inferred

phylogenetic position and divergence-time estimates of

the deep-reef species that have Indo-Paciﬁc relatives

(i.e. the Priolepis lineage). In addition, continued recon-

naissance of deep reefs both within the Caribbean and

worldwide may reveal more undescribed mesophotic

gobies, which could result in several phylogenetic pat-

terns. New deep-reef species could potentially be (i)

imbedded in our deep-reef clades, resulting in no

change in our hypotheses of shallow-to-deep transi-

tions; (ii) new deep-reef species could be resolved as a

basal member of the deep-reef clades here, which could

push estimates of the shallow-to-deep transitions

towards the past; (iii) new deep-reef species could rep-

resent independent shallow-to-deep transitions beyond

those identiﬁed in this study and could require addi-

tional hypotheses regarding the mechanisms driving

habitat shifts. Given the well-known ability for gobies

to rapidly exploit novel microhabitats and diversify via

niche partitioning (e.g. R€

uber et al. 2003; Yamada et al.

2009; Tornabene et al. 2013), it should come as no sur-

prise that multiple independent lineages of gobies have

successfully radiated into deep reefs throughout the last

30 Myr and contributed substantially to the formation

of a true deep-reef fauna in the Caribbean. As explo-

ration of the twilight zone continues, future studies will

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 17

investigate whether other major deep-reef lineages dis-

play patterns of ecological and taxonomic diversiﬁca-

tion that mirror those in gobies, lending insights into

the evolutionary origins of the deep-reef community as

a whole.

Acknowledgements

We thank Barbara Brown, Bruce Brandt, Barry Brown, Cristina

Castillo, Frank Pezold, Diane Pitassy, Hudson Pinheiro, Adriaan

Schrier, Thiony Simon, Barbara van Bebber and Jeffrey T. Wil-

liams for their various contributions to this study. Liam Revell

provided helpful information on SIMMAP analyses. This study

was funded in part by the American Museum of Natural History

Lerner Gray Award and the Smithsonian Institution Peter Buck

Fellowship to LT. Funding for the Smithsonian Institution’s

Deep Reef Observation Project was provided internally by the

Consortium for Understanding and Sustaining a Biodiverse Pla-

net to CCB, the Competitive Grants for the Promotion of Science

Program to CCB and DRR, the Herbert R. and Evelyn Axelrod

Endowment Fund for Systematic Ichthyology to CCB, and exter-

nally by National Geographic Society’s Committee for Research

and Exploration to CCB (Grant #9102-12). This study is OHF/

CSA/SC contribution #26.

References

Ackerman JL, Bellwood DR (2000) Reef ﬁsh assemblages: a re-

evaluation using enclosed rotenone stations. Marine Ecology

Progress Series,206, 227–237.

Ackerman JL, Bellwood DR (2002) Comparative efﬁciency of

clove oil and rotenone for sampling tropical reef ﬁsh assem-

blages. Journal of Fish Biology,60, 893–901.

Agorreta A, San Mauro D, Schliewen U et al. (2013) Molecular

phylogenetics of Gobioidei and phylogenetic placement of

European gobies. Molecular Phylogenetics and Evolution,69,

619–633.

Bacon CD, Silvestro D, Jaramillo C, Smith BT, Chakrabarty P,

Antonelli A (2015a) Biological evidence supports an early and

complex emergence of the Isthmus of Panama. Proceedings of

the National Academy of Sciences of the USA,112, 6110–6115.

Bacon CD, Silvestro D, Jaramillo C, Smith BT, Chakrabarty P,

Antonelli A (2015b) Reply to Lessios and Marko et al.: Early

and progressive migration across the Isthmus of Panama is

robust to missing data and biases. Proceedings of the National

Academy of Sciences of the USA,112, E5767–E5768.

Bajpai S, Kapur VV (2004) Oldest known gobiids from Vastan

Lignite Mine (early Eocoene), Surat District, Gujarat. Current

Science,87, 433–435.

Baldwin CC, Johnson GD (2014) Connectivity across the Carib-

bean Sea: DNA barcoding and morphology unite an enigmatic

ﬁsh larva from the Florida straits with a new species of sea

bass from deep reefs off Curac

ßao. PLoS ONE,9, e97661.

Baldwin CC, Robertson DR (2013) A new Haptoclinus blenny

(Teleostei, Labrisomidae) from deep reefs off Curac

ßao, south-

ern Caribbean, with comments on relationships of the genus.

ZooKeys,306,71–81.

Baldwin CC, Robertson DR (2014) A new Liopropoma sea bass

(Serranidae, Epinephelinae, Liopropomini) from deep reefs off

Curac

ßao, southern Caribbean, with comments on depth distri-

butions of western Atlantic liopropomins. ZooKeys,409,71–92.

Baldwin CC, Robertson DR (2015) A new, mesophotic Cory-

phopterus goby (Teleostei, Gobiidae) from the southern Carib-

bean, with comments on relationships and depth

distributions within the genus. ZooKeys,513, 123–142.

Bejarano I, Nemeth M, Appeldoorn RS (2011) Use of mixed-gas

rebreathers to access ﬁsh assemblages in mesophotic coral

ecosystems off La Parguera shelf -edge, Puerto Rico. Gulf and

Caribbean Fisheries Institute,63, 130–133.

Bejarano I, Appeldoorn RS, Nemeth M (2014) Fishes associated

with mesophotic coral ecosystems in La Parguera, Puerto

Rico. Coral Reefs,33, 313–328.

Betancur-R R, Broughton RE, Wiley EO et al. (2013) The tree of

life and a new classiﬁcation of bony ﬁshes. PLoS Currents

Tree of Life, doi: 10.1371/currents.tol.53ba26640df0ccaee75b-

b165c8c26288.Abstract.

Birdsong RS, Murdy EO, Pezold F (1988) A study of the verte-

bral column and median ﬁn osteology in gobioid ﬁshes with

comments on gobioid relationships. Bulletin of Marine Science,

42, 174–214.

B€

ohlke JE, Chaplin CCG (1993) Fishes of the Bahamas and Adja-

cent Tropical Waters. University of Texas Press, Austin, Texas.

Bollback JP (2006) SIMMAP: stochastic character mapping of

discrete traits on phylogenies. BMC Bioinformatics,7, 88.

Bongaerts P, Frade PR, Hay KB et al. (2015) Deep down on a

Caribbean reef: lower mesophotic depths harbor a specialized

coral-endosymbiont community. Scientiﬁc Reports,5, 7652.

Bouckaert R, Heled J, K€

uhnert D et al. (2014) BEAST 2: a soft-

ware platform for Bayesian evolutionary analysis. PLoS Com-

putational Biology,10, e1003537.

Brokovich E, Einbinder S, Shashar N, Kiﬂawi M, Kark S (2008)

Descending to the twilight-zone: changes in coral reef ﬁsh

assemblages along a depth gradient down to 65 m. Marine

Ecology Progress Series,371, 253–262.

Bryan DR, Kilfoyle K, Gilmore RG, Spieler RE (2013) Charac-

terization of the mesophotic reef ﬁsh community in south

Florida, USA. Journal of Applied Ichthyology,29, 108–117.

Bunkley-Williams L, Williams EH (2004) New locality, depth,

and size records and species character modiﬁcations of some

Caribbean deep-reef/shallow slope ﬁshes and a new host

and locality record for the Chimaera cestodarian. Caribbean

Journal of Science,40,88–119.

Colin PL (1974) Observation and collection of deep-reef ﬁshes

off the coasts of Jamaica and British Honduras (Belize). Mar-

ine Biology,24,29–38.

Colin PL (1975) The Neon Gobies. Comparative Biology of the

Gobies of the Genus Gobiosoma, Subgenus Elacatinus (Pisces:

Gobiidae) in the Tropical Western Atlantic. T.F.H. Publications,

Neptune City, New Jersey.

Colin PL (1976) Observations of deep-reef ﬁshes in the Tongue-

of-the-ocean, Bahamas. Bulletin of Marine Science,26, 603–605.

Colin PL (2010) Fishes as living tracers of connectivity in the trop-

ical western North Atlantic: I. Distribution of the neon gobies,

genus Elacatinus (Pisces: Gobiidae). Zootaxa,2370,36–52.

Dennis GD, Bright TJ (1988) Reef ﬁsh assemblages on hard

banks in the Northwestern Gulf of Mexico. Bulletin of Marine

Science,43, 280–307.

Depczynski M, Bellwood DR (2003) The role of cryptobenthic

reef ﬁshes in coral reef trophodynamics. Marine Ecology Pro-

gress Series,256, 183–191.

18 L. TORNABENE ET AL.

Dominici-Arosemena A, Wolff M (2005) Reef ﬁsh community

structure in Bocas del Toro (Caribbean, Panama): gradients

in habitat complexity and exposure. Caribbean Journal of

Science,4, 613–637.

Feitoza BM, Rosa RS, Rocha LA (2005) Ecology and zoogeogra-

phy of deep-reef ﬁshes in northeastern Brazil. Bulletin of Mar-

ine Science,76, 725–742.

Findley LT (1983) A revision of the Eastern Paciﬁc species of the

gobiid genus Chriolepis (Teleostei: Gobioidei). Unpublished PhD

Dissertation, University of Arizona, Arizona, USA.

Floeter SR, Rocha LA, Roberston DR et al. (2008) Atlantic reef

ﬁsh biogeography and evolution. Journal of Biogeography,35,

22–47.

Garcia-Sais JR (2010) Reef habitats and associated sessile-

benthic and ﬁsh assemblages across a euphotic-mesophotic

depth gradient in Isla Desecheo, Puerto Rico. Coral Reefs,29,

277–288.

Garcia-Sais JR, Castro-Gomez RL, Sabater-Clavell J et al. (2010)

Mesophotic benthic habitats and associated marine communities at

Abrir La Sierra, Puerto Rico. Final Report NOAA Coral Grant,

FNA07NMF4410117.

Gaudant J (1996) Find of the oldest known Gobioids (Teleos-

tean ﬁshes) in the marine Lower Lutetian of Catalonia

(Spain). Comptes Rendus de l’Aacademie des Sciences de la Terre

et des Planetes,332,71–76.

Gierl C, Reichenbacher B, Gaudant J, Erpenbeck D, Pharisat A

(2013) An extraordinary gobioid ﬁsh fossil from Southern

France. PLoS ONE,8, e64117.

Gilbert CR (1971) Two new genera and species of western Atlan-

tic gobiid ﬁshes with vomerine teeth. Copeia,1971,27–38.

Gilbert CR (1977) The gobiid ﬁsh Palatogobius paradoxus in the

Northern Gulf of Mexico. Northeast Gulf Science,1,48–51.

Goreau T, Goreau N (1973) The ecology of Jamaican coral reefs.

II. Geomorphology, zonation and sedimentary phases. Bul-

letin of Marine Science,23, 399–464.

Greenﬁeld DW (2002) Palatogobius grandoculus: a new deep-

water Western Caribbean goby (Teleostei: Gobiidae). Copeia,

2002, 716–718.

Greenﬁeld DW, Johnson RK (1999) Assemblage structure and

habitat associations of Western Caribbean Gobies (Teleostei:

Gobiidae). Copeia,199, 251–266.

Greenﬁeld DW, Randall JE (1999) Two new Eviota species from

the Hawaiian Islands (Teleostei: Gobiidae). Copeia,1999, 439–

446.

Haq BU, Hardenbol J, Vail PR (1987) Chronology of ﬂuctuating

sea levels since the Triassic. Science,235, 1156–1167.

Herler J, Munday PL, Hernaman V (2011) Gobies on coral

reefs. In: The Biology of Gobies (eds Patzner RA, Van Tassel

JL, Kovacic M, Kapoor BG), pp. 493–529. Science Publishers,

Enﬁeld, New Hampshire.

Horvati

c S, Cavraro F, Zanella D, Malavasi S (2015) Sound pro-

duction in the Ponto-Caspian goby Neogobius ﬂuviatilis and

acoustic afﬁnities within the Gobius lineage: implications for

phylogeny. Biological Journal of the Linnean Society, 564–573.

doi: 10.1111/bij.12696.

Humann P, Deloach N (2014) Reef Fish Identiﬁcation, Florida Car-

ibbean Bahamas, 4th edn. New World Publications Inc, Jack-

sonville, Florida.

Ingram T (2011) Speciation along a depth gradient in a marine

adaptive radiation. Proceedings of the Royal Society of London

B: Biological Sciences,278, 613–618.

Itzkowitz M, Haley M, Otis C, Evers D (1991) A reconnaissance

of the deeper Jamaican coral reef ﬁsh communities. Northeast

Gulf Science,12,25–34.

James NP, Ginsburg RN (1979) The Seaward Margin of the Belize

Barrier and Atoll Reefs. Blackwell Publishing, Oxford.

Kahng SE, Garcia-Sais JR, Spalding HL et al. (2011) Community

ecology of mesophotic coral reef ecosystems. Coral Reefs,29,

255–275.

Lanfear R, Calcott B, Ho SYW, Guindon S (2012) PartitionFin-

der: combined selection of partitioning schemes and substi-

tution models for phylogenetic analyses. Molecular Biology

and Evolution,29, 1695–1701.

Lessios HA (2015) Appearance of an early closure of the Isth-

mus of Panama is the product of biased inclusion of data in

the metaanalysis. Proceedings of the National Academy of

Sciences of the USA,112, E5765.

Liddell WD, Ohlhorst SL (1988) Hard substrata community

patterns, 1–120 m, North. Palaios,3, 413–423.

Liddell WD, Avery WE, Ohlhorst SL (1997) Patterns of benthic

community structure, 10–250 m, the Bahamas. Proceedings of

the 8th International Coral Reef Symposium,1, 437–442.

Lugli M, Pavan G, Torricelli P, Bobbio L (1995) Spawning

vocalizations in male freshwater gobiids (Pisces, Gobiidae).

Environmental Biology of Fishes,43, 219–231.

Lukens RR (1981) Observations of deep-reef ichthyofauna from

the Bahama and Cayman Islands with notes on relative

abundance and depth distribution. Gulf Research Reports,7,

79–81.

Malavasi S, Collatuzzo S, Torricelli P (2008) Interspeciﬁc varia-

tion of acoustic signals in Mediterranean gobies (Perciformes,

Gobiidae): comparative analysis and evolutionary outlook.

Biological Journal of the Linnean Society,93, 763–778.

Marko PB, Eytan RI, Knowlton N (2015) Do large molecular

sequence divergences imply an early closure of the Isthmus

of Panama? Proceedings of the National Academy of Sciences of

the USA,112, E5766.

Maxﬁeld JM, Van Tassel JL, St Mary CM, Joyeux JC, Crow KD

(2012) Extreme gender ﬂexibility: using a phylogenetic

framework to infer the evolution of variation in sex alloca-

tion, phylogeography, and speciation in a genus of bidirec-

tional sex changing ﬁshes (Lythrypnus, Gobiidae). Molecular

Phylogenetics and Evolution,64, 416–427.

McCune AR, Carlson RL (2004) Twenty ways to lose your

bladder: common natural mutants in zebraﬁsh and wide-

spread convergence of swim bladder loss among teleost

ﬁshes. Evolution & Development,6, 246–259.

Miller MA, Pfeiffer W, Schwartz T (2010) Creating the CIPRES

science gateway for inference of large phylogenetic trees. In 2010

Gateway Computing Environments Workshop, GCE 2010.

Mok HK (1981) Sound production in the Naked Goby, Gobio-

soma bosci (Pisces, Gobiidae)—a preliminary study. In: Hear-

ing and Sounds Communication in Fishes (eds Tavolga WN,

Popper AN, Fay RR), pp. 447–456. Springer-Verlag, New

York, New York.

Montes C, Cardona A, Jaramillo C et al. (2015) Middle Miocene

closure of the Central American Seaway. Science,348, 226–

229.

Near TJ, Eytan RI, Dornburg A et al. (2012) Resolution of ray-

ﬁnned ﬁsh phylogeny and timing of diversiﬁcation. Proceed-

ings of the National Academy of Sciences of the USA,109,

13698–13703.

EVOLUTIONARY ORIGINS OF TWILIGHT ZONE FISHES 19

Nelson WR, Appeldoorn RS (1985) Cruise Report R/V Seward

Johnson. A Submersible Survey of the Continental Slope of Puerto

Rico and the U.S. Virgin Islands. Report submitted to NOAA,

NMFS, SEFC, Mississippi Laboratories. University of Puerto

Rico, Department of Marine Sciences, p 76.

Parker RO, Ross SW (1986) Observing reef ﬁshes from sub-

mersibles off North Carolina. Northeast Gulf Science,8,

31–49.

Pinheiro HT, Goodbody-Gringley G, Jessup ME, Shepherd B,

Chequer AD, Rocha LA (2015) Upper and lower mesophotic

coral reef ﬁsh communities evaluated by underwater visual

censuses in two Caribbean locations. Coral Reefs,35, 139–151.

Pyle RL (1998) Use of advanced mixed-gas diving technology

to explore the coral reef “Twilight Zone”. In: Ocean Pulse: A

Critical Diagnosis (eds Tanacredi JT, Loret J), pp. 71–88. Ple-

num Press, New York, New York.

Pyle RL (1999) Mixed-gas, closed-circuit rebreather use for

identiﬁcation of new reef ﬁsh species from 200–500 fsw. In:

Assessment and Feasibility of Technical Diving Operations for

Scientiﬁc Exploration (eds Hamilton RW, Pence DF, Kesling

DE), pp. 53–65. American Academy of Underwater Sciences,

Nahant, MA.

Pyle RL (2000) Assessing undiscovered ﬁsh biodiversity on

deep coral reefs using advanced self-contained diving tech-

nology. Marine Technology Society Journal,34,82–91.

R Core Team (2015) R: A Language and Environment for Statisti-

cal Computing. R Foundation for Statistical Computing,

Vienna, Austria. Available from http://www.R-project.org/.

Rambaut A, Drummond AJ (2007) Tracer version 1.5. Available

from http://tree.bio.ed.ac.uk/software/tracer/.

Reed JK, Pomponi SA (1997) Biodiversity and distribution of

deep and shallow water sponges in the Bahamas. Proceedings

of the 8th International Coral Reef Symposium,2, 1387–1392.

Revell LJ (2012) phytools: an R package for phylogenetic com-

parative biology (and other things). Methods in Ecology and

Evolution,3, 217–223.

Robertson DR, Smith-Vaniz WF (2008) Rotenone: an essential

but demonized tool for assessing marine ﬁsh diversity. BioS-

cience,58, 165–170.

Robertson DR, Van Tassell JL (2015) Shoreﬁshes of the Greater

Caribbean: Online Information System, Version 1.0. Smithsonian

Tropical Research Institute, Balboa, Panam

Ronquist F, Teslenko M, Van Der Mark P et al. (2012) Mrbayes

3.2: efﬁcient Bayesian phylogenetic inference and model

choice across a large model space. Systematic Biology,61,

539–542.

Rosa M, Laves A, Medeiros D et al. (2015) Mesophotic reef ﬁsh

assemblages of the remote St. Peter and St. Paul’s Archipe-

lago, Mid-Atlantic Ridge, Brazil. Coral Reefs,35, 113–123.

Ross SW, Quattrini AM (2007) The ﬁsh fauna associated with

deep coral banks off the southeastern United States. Deep-

Sea Research Part I: Oceanographic Research Papers,54, 975–

1007.

Roy K, Jablonksi D, Valentine JW, Rosenberg G (1998) Marine

latitudinal diversity gradients: tests of causal hypotheses.

Proceedings of the National Academy of Sciences of the USA,95,

3699–3702.

R€

uber L, Van Tassell JL, Zardoya R (2003) Rapid speciation

and ecological divergence in the American seven pined gob-

ies (Gobiidae: Gobiosomatini) inferred from a molecular phy-

logeny. Evolution,57, 1584–1598.

Sanderson MJ (2002) Estimating absolute rates of molecular

evolution and divergence times: a penalized likeli-

hood approach. Molecular Biology and Evolution,19, 101–

109.

Smith C, Knopf A (1997) National Audubon Society Field Guide to

Tropical Marine Fishes of the Caribbean, the Gulf of Mexico, Flor-

ida, the Bahamas, and Bermuda. University of Texas Press,

Austin, Texas.

Spieler R, Bryan D, Kilfoyle K (2007) Reef Fishes of

Broward County Natural Substrate and Vessel-reefs in

50–120 Meter Water Depth: Final Report, FWC Grant

FWC-05017. 88p.

Srinivasan M (2003) Depth distributions of coral reef ﬁshes: the

inﬂuence of microhabitat structure, settlement, and post-set-

tlement processes. Population Ecology,137,76–84.

Tavolga WN, Popper AN, Fay RR (eds) (1981) Hearing and

Sound Communication in Fishes. Springer-Verlag, New York,

New York.

Thacker CE (2014) Species and shape diversiﬁcation are inver-

sely correlated among gobies and cardinalﬁshes. Organisms

Diversity and Evolution,14, 419–436.

Thacker CE, Roje DM (2011) Phylogeny of Gobiidae and identi-

ﬁcation of gobiid lineages. Systematics and Biodiversity,9,

329–347.

Thresher RE, Colin PL (1986) Trophic structure, diversity

and abundance of ﬁshes of the deep reef (30–300 m) at

Enewetak, Marshall Islands. Bulletin of Marine Science,38,

253–272.

Tornabene L, Ahmadia GN, Berumen ML et al. (2013) Evolu-

tion of microhabitat association and morphology in a

diverse group of cryptobenthic coral reef ﬁshes (Teleostei:

Gobiidae: Eviota). Molecular Phylogenetics and Evolution,66,

391–400.

Tornabene L, Van Tassell JL, Gilmore RG, Robertson DR,

Young F, Baldwin CC (2016) Molecular phylogeny, analysis

of character evolution, and submersible collections enable a

new classiﬁcation for a diverse group of gobies (Teleostei:

Gobiidae: Nes subgroup), including nine new species and

four new genera. Zoological Journal of the Linnean Society, doi:

10.1111/zoj.12394.

Torricelli P, Lugli M, Pavan G (1990) Analysis of sounds pro-

duced by male Padogobius martensi (Pisces, Gobiidae) and

factors affecting their structural properties. Bioacoustics,2,

261–275.

Van Tassell JL (2011) Gobiiformes of the Americas. In: The Biol-

ogy of Gobies (eds Patzner RA, Van Tassell JL, Kovacic M,

Kapoor BG), pp. 139–176. Science Publishers, Enﬁeld, New

Hampshire.

Van Tassell JL, Tornabene L, Collin PL (2012) Review of the

western Atlantic species of Bollmannia (Teleostei: Gobiidae:

Gobiosomatini) with the description of a new allied genus

and species. Aqua, International Journal of Ichthyology,18,61–

94.

Winterbottom R (1984) A review of the gobiid ﬁsh genus

Trimma from the Chagos Archipelago, central Indian Ocean,

with description of seven new species. Canadian Journal of

Zoology,62, 695–715.

Winterbottom R, Zur M, History N, Museum RO, Park Q

(2007) Three new species of genus Trimma from Palau, Wes-

tern Paciﬁc (Percomorpha: Gobiidae). Aqua, International Jour-

nal of Ichthyology,13,1–25.

20 L. TORNABENE ET AL.

2024 Schwarzhans & Aguilera- Otoliths of the Gobiidae from the Neogene of tropical America

Article

Full-text available

Mar 2024

Otoliths are common and diverse in the Neogene of tropical America. Following previous studies of Neogene tropical American otoliths of the lanternfishes (Myctophidae), marine catfishes (Ariidae), croakers (Sciaenidae), and cuskeels (Ophidiiformes), we describe here the otoliths of the gobies (Gobiidae). The Gobiidae represent the richest marine fish family, with more than 2000 species worldwide and about 250 in America. In the fossil record too they are the species richest family in the Neogene of tropical America. We have investigated otoliths sampled from Ecuador, Pacific and Atlantic Panama, Atlantic Costa Rica, Dominican Republic, Venezuela, and Trinidad, ranging in age from late Early Miocene (late Burdigalian) to late Early Pleistocene (Calabrian). Most of the studied material originates from the collection expeditions of the Panama Paleontology Project (PPP). Our study represents the first comprehensive record of fossil gobies from America, and we recognize 107 species, of which 51 are new to science, 35 are in open nomenclature, and 19 represent species that also live in the region today. Previously, only two fossil otolithbased goby species have been described from the Neogene of tropical America. The dominant gobies in the fossil record of the region are from the Gobiosomatini, particularly of genera living over soft bottoms or in deeper water such as Bollmannia, Microgobius, Antilligobius, and Palatogobius. Another purpose of our study is to provide a first comprehensive account of otoliths of the extant Gobiidae of America, which we consider necessary for an adequate identification and interpretation of the Neogene otoliths. We studied otoliths of 130 extant American gobiid species and figured 106 of them for comparison. We also present a morphological analysis and characterization of the extant otoliths as a basis for the identification of fossil otoliths. Problems that commonly arise with the identification of fossil otoliths and specifically of fossil goby otoliths are addressed and discussed. A comparison of the history of the Gobiidae in tropical America reveals a high percentage of shared species between the Pacific and the Atlantic basins during the Late Miocene (Tortonian and Messinian) from at least 11 to 6 Ma. A recording gap on the Pacific side across the Pliocene allows a comparison again only in the late Early Pleistocene (Calabrian, 1.8 to 0.78 Ma), which shows a complete lack of shared species. These observations support the effective closure of the former Central American Seaway and emersion of the Isthmus of Panama in the intervening time. Groups that today only exist in the East Pacific were also identified in the Miocene and Pliocene of the West Atlantic, and there is also at least one instance of a genus now restricted to the West Atlantic having occurred in the East Pacific as late as the Pleistocene. The evolution of gobies in tropical America and the implications thereof are extensively discussed. Furthermore, observations of fossil gobies in the region are discussed in respect to paleoenvironmental indications and paleobiogeographic aspects.

Assembly rules of coral reef fish communities along the depth gradient

Article

Mar 2023
CURR BIOL

Coral reefs are home to some of the most studied ecological assemblages on the planet. However, differences in large-scale assembly rules have never been studied using empirical quantitative data stratified along the depth gradient of reefs. Consequently, little is known about the small- and regional-scale effects of depth on coral reef assemblages. Using a large dataset of underwater surveys, we observed that the influence of classic biogeographic drivers on the species richness of coral reef fishes changes significantly with depth, shaping distinct assemblages governed by different rules in mesophotic coral ecosystems. We show that a general pattern of decreased taxonomic and functional richness of reef fish assemblages with depth results from convergent filtering of species composition and trophic strategies on deeper reefs across ocean basins and that at smaller scales deep-reef communities are less influenced by regional factors than shallower reefs.

Description of the First Species of Polylepion (Teleostei: Labridae) from the Atlantic Ocean with Analysis of Evolutionary Relationships of the New Species

Article

Full-text available

Apr 2023

The Habitat Persistence Hypothesis: a new perspective on the distribution of coral-reef organisms

Article

Full-text available

Dec 2022

Depth distributions of benthic and pelagic species highlight the potential of mesophotic and deep habitats to serve as marine refugia

Article

Full-text available

Nov 2022

Shallow marine ecosystems are threatened by human impacts. The prospect that deeper environments could serve as refugia is promising but remains uncertain, as it has been evaluated primarily in benthic taxa and on small spatial scales. We use a dataset of all biota from the Gulf of Mexico to evaluate the degree to which species’ occurrences extend across shallow (0-30 m), mesophotic (>30-150 m), and deep (>150 m) environments. High percentages of species inhabit multiple depth zones: 70% of benthic and 81% of pelagic species occurring in the mesophotic zone are also found in the shallow zone, and 45% of benthic and 80% of pelagic species found in the shallow zone extend into mesophotic depths. After accounting for confounding effects using logistic regression, pelagic species have significantly higher odds of extending between depth zones than benthic species. Analyses of species in adjacent 20 m depth bands reveal taxonomically distinct communities of benthic species from 0 to 100 m and pelagic species from 0 to 200 m. Peaks in probability distributions show that pelagic species generally have broader depth ranges than benthic species. Depth ranges that overlap between deeper and shallower zones show that movement of individuals or propagules in one or both directions between these habitats is either occurring now or has occurred in the past. Over the long term, deeper habitats can allow species to escape regional extirpation, and can contribute to the resilience and persistence of coastal ecosystems in the face of increased disturbance in shallow water. This is particularly true when pelagic species are considered.

A global synthesis of the current knowledge on the taxonomic and geographic distribution of major coral diseases

Article

Full-text available

Apr 2022

Although knowledge on the diseases affecting corals has been accumulating exponentially since the 2000s, even more effort is required to summarize and guide further investigation. Here, we used the Web of Science database to review 226 studies published, between 2000 and 2020, to identify the major geographic and taxonomic gaps in the literature, and propose future directions for the study of coral diseases. We classified the studies according to the ocean, ecoregion, coral species, disease types, approach (e.g., observational or experimental), and depth. In total, 22 types of diseases were reported for 165 coral species. Acropora spp. was the most studied taxa with 12 types of diseases and 8.2% of the records. Black band, white plague, white syndromes, skeletal eroding, dark spot, and yellow band were the six most common diseases, accounting together for 76.8% of the records. As expected, most studies were conducted in the Caribbean and Indo-Pacific (34.0% and 28.7%, respectively), but only in 44 of the 141 global ecoregions that harbour corals. Observational approaches were the most frequent (75.6% of the records), while experimental approaches accounted for 19.9% and were mainly done on Acropora. The vast majority of studies (∼98%) were performed in shallow waters (<30 m depth). We conclude that over the past two decades, coral diseases have been assessed on a very small fraction of coral species, in very few locations around the globe, and at a limited range of their depth distribution. While monitoring bleaching is mandatory for reef ecology and conservation, the ecoepidemiology of coral diseases deserves more space in the research agenda of reef ecosystems.

Characterization of fish assemblages in mesophotic reefs of Cuba

Article

Full-text available

Jul 2021
B MAR SCI

Although mesophotic coral reef ecosystems (MCEs) are widespread and diverse, they remain largely unexplored. The Caribbean MCEs are among the most studied in the world, but detailed information regarding Cuban MCEs is limited. In this study we present the first analyses of the ichthyofauna of the Cuban MCEs based on data from a research cruise in June 2017. Forty-two sites were sampled around the Archipelago of Cuba using a Remotely Operated Vehicle at depths between 30 and 180 m. Overall, 151 species were observed, including 13 threatened species. Differences in ichthyofauna were observed between marine protected areas (MPAs) and unprotected areas, with greater abundances and species richness inside national parks. Invasive lionfish had greater abundances on the north coast, while the south coast had greater abundances and richness of threatened and commercially valuable species. A detailed analysis (Generalized Lineal Model) showed effects of depth, protection level, reef type (seamount or fringing reef), and region on abundance (as fish observed per minute), but no interaction between them. A transition was observed in the fish assemblages between the 30–60 m and 60–100 m depth ranges. Species abundance and richness decreased with increasing depth. The most represented families of commercial importance (Serranidae, Lutjanidae, and Carangidae) were well represented in the upper, middle, and lower areas of the MCEs, but dominated in deeper areas. The percentage of herbivores in the fish assemblage decreased with increasing depth, contrary to what was observed with piscivores and invertebrate feeders/piscivores. Twenty-seven functional groups were identified, with cryptic micro-invertivores and cryptic planktivores being the most abundant, while roving macro-invertivores and roving piscivores were the most functionally redundant. This research constitutes a valuable contribution to the study and management of ichthyofauna in Cuba and highlights the effectiveness of MPAs in the conservation of fish biodiversity.

Phylogenetic Classification of Living and Fossil Ray-Finned Fishes (Actinopterygii)

Article

Apr 2024

Two new species of Varicus from Caribbean deep reefs, with comments on the related genus Pinnichthys (Teleostei, Gobiidae, Gobiosomatini, Nes subgroup)

Article

Full-text available

Sep 2023

Tropical deep reefs (~40–300 m) are diverse ecosystems that serve as habitats for diverse communities of reef-associated fishes. Deep-reef fish communities are taxonomically and ecologically distinct from those on shallow reefs, but like those on shallow reefs, they are home to a species-rich assemblage of small, cryptobenthic reef fishes, including many species from the family Gobiidae (gobies). Here we describe two new species of deep-reef gobies, Varicus prometheus sp. nov. and V. roatanensis sp. nov. , that were collected using the submersible Idabel from rariphotic reefs off the island of Roatan (Honduras) in the Caribbean. The new species are the 11 th and 12 th species of the genus Varicus , and their placement in the genus is supported by morphological data and molecular phylogenetic analyses. Additionally, we also collected new specimens of the closely-related genus and species Pinnichthys aimoriensis during submersible collections off the islands of Bonaire and St. Eustatius (Netherland Antilles) and included them in this study to expand the current description of that species and document its range extension from Brazil into the Caribbean. Collectively, the two new species of Varicus and new records of P. aimoriensis add to our growing knowledge of cryptobenthic fish diversity on deep reefs of the Caribbean.

A new lineage of deep-reef gobies from the Caribbean, including two new species and one new genus (Teleostei: Gobiidae: Gobiosomatini)

Article

Apr 2022

Fish communities on tropical deep reefs are dominated by species that belong to families primarily composed of shallow-water species. Collections of deep-reef fishes via submersibles have allowed us to include these deep-reef species in molecular phylogenies, providing insights into the timing and frequency of invasions from shallow to deep reefs. Here we provide evidence of a new deep-reef invasion in the tribe Gobiosomatini in the family Gobiidae (gobies). We describe two new species, one of which belongs to a new genus, and incorporate these taxa into a time-calibrated molecular phylogeny of Gobiosomatini to show that, collectively, these two genera represent a previously unreported independent invasion on to deep reefs that occurred approximately 20–30 million years ago. These new taxa are readily distinguished from related genera and species by a combination of live coloration, pelvic-fin morphology, meristic characters, head-pore patterns and other osteological characters. We discuss the relevance of these two new species to the systematics of the tribe Gobiosomatini and include a comparison to all known genera in the tribe.

Observing Reef Fishes from Submersibles Off North Carolina

Article

Full-text available

Mar 1986

Molecular phylogeny, analysis of character evolution, and submersible collections enable a new classification of a diverse group of gobies (Teleostei: Gobiidae: Nes subgroup), including nine new species and four new genera

Article

Full-text available

May 2016

The Nes subgroup of the Gobiosomatini (Teleostei: Gobiiformes: Gobiidae) is an ecologically diverse clade of fishes endemic to the western Atlantic and tropical eastern Pacific Oceans. It has been suggested that morphological characters in gobies tend to evolve via reduction and loss associated with miniaturization, and this, coupled with parallel evolution of adaptations to similar microhabitats, may lead to homoplasy and ultimately obscure our ability to discern phylogenetic relationships using morphological characters alone. This may be particularly true for the Nes subgroup of gobies, where several genera that are diagnosed by ‘reductive characters’ have been shown to be polyphyletic. Here we present the most comprehensive phylogeny to date of the Nes subgroup using mitochondrial and nuclear sequence data. We then evaluate the congruence between the distribution of morphological characters and our molecular tree using maximum likelihood ancestral state reconstruction and test for phylogenetic signal in characters using Pagel’s λ tree transformations (Pagel 1999). Our results indicate that all of the characters previously used to diagnose genera of the Nes subgroup display some degree of homoplasy with respect to our molecular tree however, many characters display considerable phylogenetic signal and thus may be useful in diagnosing genera when used in combination with other characters. We present a new classification for the group in which all genera are monophyletic and in most cases diagnosed by combinations of morphological characters. The new classification includes 9 new species described here, many of which were collected from rarely sampled deep Caribbean reefs using manned submersibles. Lastly, we provide a key to all 38 named species of the Nes subgroup along with photographs and illustrations to aid in identification.

A Reconnaissance of the Deeper Jamaican Coral Reef Fish Communities

Article

Nov 1991

The Gobiid Fish Palatogobius paradoxus in the Northern Gulf of Mexico

Article

Jun 1977

Carter R. Gilbert

Hearing and Sound Communication in Fishes

Book

Jan 1981

This volume is a compilation of the papers presented at a meeting that took place in April 1980 at the Mote Marine Laboratory, Sarasota, Florida. The meeting and this volume are outgrowths of two earlier international meetings on marine bio-acoustics that occurred in 1963 and 1966 (Tavolga 1964, 1967). The first meeting took place at the Lerner Marine Laboratory of the American Museum of Natural History, while the second meeting was at the American Museum itself, and was under the sponsorship of the Department of Animal Behavior. It is apparent that these two volumes have had immense impact on the current study of marine bio-acoustics, and particularly on fish audition. In a preliminary conference in Sarasota in 1979 we decided that it was time for another such meeting, to bring together as many as possible of the investigators interested in fish acoustics in order to assess the current state of our knowledge and predict directions for research for the next several years. Such a meeting appeared par ticularly timely, since over the past four or five years there have been many new studies that have provided new empirical and theoretical work on basic mechanisms of fish audition. Furthermore, it became evident, as we made up preliminary lists of possible participants, that few of the currently active workers were in the field back in 1966. In fact, of the current participants, only Drs.

R: A Language and Environment for Statistical Computing

Book

Jan 2015

Core R Team

SIMMAP: stochastic character mapping of discrete traits on phylogenies

Article

J.P. Bollback

Sound Production in the Naked Goby, Gobiosoma bosci (Pisces, Gobiidae)—A Preliminary Study

Chapter

Jan 1981

Hin-Kiu Mok

To date, Bathygobius soporator, Gobius jozo, and Neogobius melanostomus are the reported soniferous species within the large ubiquitous perciform family, Gobiidae. The breeding males of B. soporator emit low pitched grunts (frequency, 100–150 Hz, non-harmonic; duration, 150–350 msec) as a communicatory signal associated with pre-spawning behavior (Tavolga 1958). Similar hydrodynamic sounds were noted in G. jozo and N. melanostomus by Kinger and Protasov et al., respectively (in Tavolga 1976). While observing the nest-guarding behavior of a male Gobiosoma bosci (naked goby) in a 15-gallon aquarium with six conspecifics (3 females and 3 males; total body length ranges from 28 to 40 mm), a clicking sound produced by this male was recorded. Succeeding additional observations and recordings were thus made. The recording equipment consisted of a hydrophone (Chesapeake Instrument Co., SH-102) and a recorder (Sony TC158SD). The present paper summarizes the results of a preliminary study on the acoustic behavior of this species.

Use of Advanced Mixed-Gas Diving Technology to Explore the Coral Reef “Twilight Zone”

Chapter

Jan 1998

Richard Lawrence Pyle

Marine environments harbor an enormously diverse array of organisms. S.A. Earle (1991) wrote, “Although fewer species appear to inhabit the ocean than the land, based on numbers of species currently known, consideration solely of numbers of species can be a misleading measure of diversity. At higher taxonomic levels (Class, Phylum, and various subdivisions of these broad categories), marine ecosystems have a significantly higher degree of genetic diversity.... [I]f the presence of individuals representing various broad categories of life are given somewhat greater weight than the splintery ends of diversity known as species, then the greatest diversity of life is unquestionably in the sea.”

Find of the oldest known Gobioids (Teleostean fishes) in the marine Lower Lutetian of Catalonia (Spain)

Article

Jan 1996

J. Gaudant

The oldest presently known articulated skeletons of Gobioids are reported from the marine Lower Lutetian of the surroundings of Borredâ (North Catalonia, Spain). Their discovery confirms that of an isolated otolith, recently described in the Lutetian of India.

Repeated invasions into the twilight zone: Evolutionary origins of a novel assemblage of fishes from deep Caribbean reefs

Abstract and Figures

Recommended publications

BIOGEOGRAPHY OF CARIBBEAN OLIGOCENE AND MIOCENE DECAPODS (THALASSINIDEA: BRACHYURA)

Mitochondrial DNA and Color Pattern Variation in Three Western Atlantic Halichoeres (Labridae), with...

Genetic Relationships of the Color Morphs of the Serranid Fish Hypoplectrus unicolor

The mode of operation of Antillean fish traps and the relationships between ingress, escapement, cat...