jbi_1317 1835..1854

ORIGINAL

ARTICLE

The biogeography of lower

Mesoamerican freshwater ﬁshes

Scott A. Smith

1,2

* and Eldredge Bermingham

1,2

Smithsonian Tropical Research Institute,

Panama City, Republic of Panama and

Department of Biology, McGill University,

Montreal, Quebec, Canada

*Correspondence: Scott A. Smith, Department

of Biology, McGill University, 1205 Avenue

Docteur Penﬁeld, Montreal, Quebec, Canada

H3A 1B1.

E-mail: [email protected]

ABSTRACT

Aim This paper examines the importance of regional processes in determining

the patterns of distribution and diversity of lower Mesoamerican freshwater

ﬁshes.

Location We focused our analyses on the lower Mesoamerican region, which we

deﬁne to include all the rivers of Panama and Costa Rica. The geographic

boundaries are the Colombian Choco to the south and Lake Nicaragua to the north.

Methods We described the biogeographical provinces of lower Mesoamerica

(LMA) using presence/absence data of primary and secondary LMA freshwater

ﬁshes. We conducted subsequent analyses at the spatial resolution of the

biogeographical provinces and described patterns of community composition,

species richness, endemism, range size, and the permeability of dispersal barriers

between biogeographical provinces.

Results This study represents the ﬁrst attempt since that of W. A. Bussing in

1976 to investigate the biogeographical regions of Mesoamerica, and our analyses

demonstrate increased regional complexity in biodiversity patterns relative to

previous studies. Changes in community composition across LMA clearly

highlight the importance of both extrinsic geological processes and intrinsic

biological differences among freshwater ﬁsh species in shaping the dispersal and

diversiﬁcation histories of the LMA freshwater ﬁsh fauna. The inﬂuence of

biology and geology is also exempliﬁed by patterns of endemism and turnover

between biogeographical provinces, which suggests that the relative importance of

regional speciation and dispersal varies spatially across the LMA landscape.

Finally, it would seem to follow that secondary freshwater ﬁshes will have larger

range sizes than primary ﬁshes as a result of the increased salinity tolerance

posited for the former group, and thus the increased probability of dispersal along

coastlines. We did not, however, ﬁnd a signiﬁcant difference between the average

range size of primary and secondary freshwater ﬁshes, indicating that the putative

differences in physiological tolerance to seawater between the two groups are not

reﬂected in their distribution patterns at the scale of LMA. The geometric

distribution of range size of LMA freshwater ﬁshes suggests that dispersal of both

primary and secondary freshwater ﬁshes along coastlines must be infrequent.

Main conclusion The observation that regional processes exerted a strong

inﬂuence on the assembly and maintenance of LMA freshwater ﬁsh communities

has important consequences for both theory and conservation. We suggest that

large-scale biogeographical analyses are required to illuminate the backdrop upon

which local interactions play themselves out, supporting a top-down approach to

the study of biological diversity. Our results also identify areas of high

conservation priority, providing a baseline for informing conservation strategies

for freshwater ﬁshes in LMA. We conclude by calling for conservation planning

and action that acknowledges the importance that regional processes play in

Journal of Biogeography (J. Biogeogr.) (2005) 32, 1835–1854

ª 2005 Blackwell Publishing Ltd www.blackwellpublishing.com/jbi doi:10.1111/j.1365-2699.2005.01317.x

1835

INTRODUCTION

Identiﬁcation of the relative contributions of local and regional

processes in determining patterns of species richness is essential

to understanding the mechanisms underlying patterns of global

species richness (Huston, 1999; Gaston, 2000). Traditionally,

ecologists have focused on the importance of local, deterministic

processes such as competition, predation and mutualism in

shaping patterns of species richness and community composi-

tion (Ricklefs, 1987). More recently, ecologists have emphasized

the role that regional processes, such as dispersal, speciation and

extinction, play in the assembly and maintenance of biological

communities (Ricklefs, 1987, 2004; Ricklefs & Schluter, 1993;

Gaston, 2000, 2003; Mora et al., 2003; Smith et al., 2004; Smith

& Shurin, in press). However, it has seldom been possible to

directly evaluate the importance of regional processes in

determining patterns of distribution and diversity for entire

continental faunas because molecular phylogeographic analyses

have only recently begun to illuminate historical patterns of

colonization and community assembly (Bermingham & Avise,

1986; Losos, 1992; Qian & Ricklefs, 1999, 2000; Losos & Schluter,

2000; Parra-Olea & Wake, 2001; Ricklefs & Bermingham 2001;

Mueller et al., 2004), and reliable presence/absence data at the

continental scale remain scarce for most taxa found in tropical

and subtropical regions of the world (but see Unmack, 2001).

Lower Mesoamerica (LMA) provides an interesting backdrop

against which to study the importance of regional processes. The

Pliocene completion of the Isthmus of Panama connected North

and South America, permitting a massive exchange of ﬂora and

fauna between the two continents, an event termed the Great

American Biotic Interchange (Marshall et al., 1979; Stehli &

Webb, 1985). The emerging isthmus enabled the initial

colonization of Mesoamerica by freshwater ﬁshes from putative

source populations in north-western Colombia (Miller, 1966;

Myers, 1966; Bussing, 1985; Bermingham & Martin, 1998).

However, the embryonic isthmus provided differential coloni-

zation opportunities for these ﬁshes (Miller, 1966; Myers, 1966;

Bussing, 1976, 1985; Bermingham & Martin, 1998; Martin &

Bermingham, 1998; Perdices et al., 2002). Secondary freshwater

ﬁshes such as cichlids and Rivulus colonized Mesoamerica

between 18 and 15 Ma (Martin & Bermingham, 1998; see also

Murphy & Collier, 1996), approximately 10 Ma earlier than did

primary freshwater ﬁshes (Bermingham & Martin, 1998),

according to molecular clock estimates.

Two important facts suggest that the study of patterns of

distribution and diversity of LMA freshwater ﬁshes will permit

inferences regarding the role that regional processes played in the

assembly and maintenance of freshwater ﬁsh communities in

this region. First, although there has been controversy regarding

the timing of the arrival of primary freshwater ﬁshes in

Mesoamerica (Miller, 1966; Myers, 1966; Bussing, 1985),

molecular systematic research indicates that the evolutionary

assembly of the LMA freshwater ﬁsh fauna is recent

(Bermingham & Martin, 1998; Perdices et al., 2002), suggesting

that the historical record of colonization is largely intact because

it is not likely to have been overwritten by multiple rounds of

dispersal. Second, drainage basins function as terrestrial islands

that create repeated patterns of subdivision of populations and

communities within discrete boundaries. These patterns arise

largely because of the dispersal constraints on obligate (primary

and secondary) freshwater ﬁshes. Primary freshwater ﬁshes are

hypothesized to be physiologically intolerant of saline condi-

tions and thus rarely cross marine barriers (Myers, 1938, 1966;

Unmack, 2001). The dispersal of primary freshwater ﬁshes is

thought to be entirely dependent on the formation of

direct connections between drainage basins arising from

physiographical changes of the landscape. Examples include

river anastomosis that occurs during periods of reduced sea

level coinciding with glacial maxima, as well as during the

high water of the rainy season, pulses of freshwater that connect

rivers along the coast during ﬂood events, and river capture

events that result from differential erosion across drainage

divides (Myers, 1938, 1966; Loftin, 1965; Miller, 1966; Berming-

ham & Martin, 1998; Unmack, 2001). On the other hand,

secondary freshwater ﬁshes are hypothesized to tolerate mod-

erate levels of salinity and thus may be able to disperse via saline

marine waters along coastlines, although the frequency of

such colonization events remains unknown. Recent colonization

of LMA by obligate freshwater ﬁshes coupled with the disper-

sal limitation of these ﬁshes therefore provides an interesting

system in which to study the importance of regional processes in

determining present-day patterns of distribution and diversity.

A ﬁrst step towards studying the importance of regional

processes is the description of biogeographical provinces.

Biogeographical provinces represent geographical regions of

relatively homogeneous faunal composition, and their descrip-

tion permits analysis of the processes responsible for turnover

across the landscape, in terms of both species richness and

community composition. We describe the biogeographical

provinces of LMA freshwater ﬁshes and test whether the

biogeographical provinces represent areas of shared history that

extend beyond the similarity arising simply from the spatial

arrangement of the rivers and the source of colonists. In turn, we

evaluate patterns of community composition, species richness

and endemism at the scale of biogeographical provinces in order

to document physiological and regional inﬂuences on the

determining patterns of organismal diversity, and that incorporates these

processes in strategies to conserve remnant biological diversity.

Keywords

Beta diversity, biogeography, community assembly, dispersal, freshwater ﬁsh,

Mesoamerica, regional processes, species richness.

S. A. Smith and E. Bermingham

1836

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

dispersal and diversiﬁcation history of the Mesoamerican

freshwater ﬁsh fauna. Furthermore, we quantify the biological

turnover between biogeographical provinces in order to estimate

how the relative permeability of geographical barriers to

dispersal changes across the LMA landscape. Finally, we assess

the range-size distribution of the LMA freshwater ﬁshes by

determining the number of biogeographical provinces occupied

by each species, and we test whether hypothesized differences in

salinity tolerance between primary and secondary freshwater

ﬁshes have inﬂuenced the sizes of the geographical ranges of

these ﬁshes.

METHODS

Data sources

We assembled a data base describing the presence and absence

of freshwater ﬁsh species in LMA drainage basins, a region

Figure 1 Drainage basins of lower Mesoamerica. 1, Rio San Juan; 2, Rio Tortuguero; 3, Rio Parismina; 4, Rio Matina; 5, Rio Sixaola; 6, Rio

San San; 7, Rio Changuinola; 8, International Cuenca no. 91; 9, Rio Uyama; 10, Quebrada Cilico Creek and Quebrada Marin; 11, Rio

Robalo; 12, Quebrada La Gloria; 13, Rio Guarumo; 14, Quebrada La Margarita; 15, Rio Guariviara; 16, Rio Cricamola; 17, Rio Canaveral; 18,

International Cuenca no. 95; 19, Rio Calovebora; 20, International Cuenca no. 99; 21, Rio Veraguas; 22, International Cuenca no. 103; 23,

Rio Cocle del Norte; 24, International Cuenca no. 107; 25, Rio Miguel de la Borda; 26, Rio Indio; 27, International Cuenca no. 113; 28, Rio

Chagres; 29, Rio Piedras; 30, Rio Cuanche; 31, Rio Cascajal; 32, Rio Claro; 33, Rio Pina Pina; 34, Rio Frio; 35, Rio Cuango; 36, Rio

Mandinga; 37, International Cuenca no. 121; 38, Rio Azucar; 39, Rio Playon Chico; 40, Rio Cuadi; 41, Rio Acla; 42, Rios of the Nicoya

Peninsula; 43, Rio Tempisque; 44, Rio Bebedero; 45, Rio Barranca; 46, Rio Tarcoles; 47, Rio Pirris; 48, Rio Terraba; 49, Rio Coto; 50, Rio

Palo Blanco; 51, Rio Chiriqui Viejo; 52, Rio Escarrea; 53, Rio Chico; 54, Rio Platanal; 55, Rio Chiriqui; 56, Rio Estero Salado; 57, Rio San

Juan; 58, Rio San Felix; 59, Rio Santiago; 60, Rio Tabasara; 61, Rio Bubi; 62, Rio San Pablo; 63, Rio Cate; 64, Quebrada Seca; 65, Rio Santa

Maria; 66, Rio San Pedro; 67, Rio Ponuga; 68, Rio Tebario; 69, Rio Pavo; 70, Rio Playita; 71, Rio Tonosi; 72, Rio Cana; 73, Rio Oria; 74,

International Cuenca no. 126; 75, Rio Guarare; 76, Rio Honda; 77, Rio La Villa; 78, Rio Parita; 79, Rio Cocle del Sur; 80, Rio Chorrera; 81,

Rio Estancia; 82, Rio Anton; 83, Rio Farallon; 84, Rio Chame; 85, Rio Sajalices; 86, Rio Capira; 87, Rio Caimito; 88, Rio Grande; 89, Rio Juan

Diaz; 90, Rio Cabra; 91, Rio Pacora; 92, Rio Bayano; 93, International Cuenca no. 150; 94, Rio Lara; 95, Rio Tuira; 96, Rio Iglesia; 97,

International Cuenca no. 160; 98, Rio Samba; 99, International Cuenca no. 164.

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1837

including the rivers of Panama and Costa Rica (Fig. 1). Our

data base was based on information of species distribution

contained in the NEODAT data base (http://www.neodat.org)

and the Smithsonian Tropical Research Institute (STRI)

Freshwater Fish Collection (Bermingham et al., 1997a). We

geo-referenced and veriﬁed each record’s merit using the

literature and the STRI Freshwater Fish Collection. Only

collection records for primary and secondary freshwater ﬁshes

were used in the analyses. No diadromous ﬁshes were included

in our analyses. A total of 18,366 records, representing the

distribution of 170 species, 72 genera and 23 families, were

included in our analyses.

Assumptions

Given the heterogeneity in collecting effort across rivers, we

documented many local absences of ﬁsh species that can be

attributed to insufﬁcient sampling effort exerted in particular

drainages. The fairly dense sampling of rivers across the isthmian

landscape, however, suggests that absence of a ﬁsh species at the

provincial level can be assumed to be real. Since all of our

analyses were undertaken at the provincial level, the inferences

that we make were not unduly inﬂuenced by absences arising

from insufﬁcient sampling effort in some drainage basins.

There are also some limitations in the taxonomic descrip-

tion of the fauna. For example, recent molecular systematic

research on the genus Roeboides (Bermingham & Martin, 1998)

uncovered phylogenetic variation that made geographical sense

of a bewildering distribution pattern based on the preceding

taxonomy and that led to the description of three new species

(Lucena, 2000). That said, we are conﬁdent that our

description of both taxa and geographical distributions is

largely correct because phylogenetic relationships of most

potentially problematic taxa have been clariﬁed by molecule-

based phylogeographic analyses (Bermingham & Martin, 1998;

Martin & Bermingham, 2000; Perdices et al., 2002). Table 1

provides a complete list of the freshwater ﬁsh species of LMA

as deﬁned by prevailing taxonomy and molecular phylogeog-

raphy, and their distribution in the LMA biogeographical

provinces.

Biogeographical provinces

Correspondence analysis was used to convert binary species

presence/absence data into continuous variables (Jackson &

Harvey, 1989; Hugueny & Leveque, 1994). This method of

ordination positions the objects (sites and species) in a space

that contains fewer dimensions than the original data set

(Legendre & Legendre, 1998). It thus simpliﬁes the recovery of

meaningful patterns in the ﬁrst ordination axes and defers

noise to later axes (Gauch, 1982). The noise eliminated by the

correspondence analysis is assumed to be non-informative

from a biogeographical perspective (Hugueny & Leveque,

1994). Correspondence analysis is preferred over other

methods of multivariate analysis because it is based on a

metric that does not include cases in which two species are

both absent from a given locality in its computation (Hugueny

& Leveque, 1994), and is thus strongly recommended for

ecological or biogeographical multivariate analyses that rely on

binomial data (Legendre & Legendre, 1998).

Next, Euclidian distances were used to compute the faunistic

distances between the rivers based on the coordinates

representing the ﬁrst three axes of the correspondence analysis.

An unweighted pair group method with arithmetic mean

(UPGMA) analysis produced a dendrogram depicting the

faunistic relationship between drainage basins for the LMA

freshwater ﬁsh fauna.

The faunistic relationship between rivers was also described

using Jaccard’s similarity coefﬁcient, a method that did not

involve the use of correspondence analysis to transform the

data initially. The UPGMA cluster analysis produced a

dendrogram describing the faunistic relationship between

drainage basins based on Jaccard’s similarity coefﬁcient (the

procedures described in the preceding paragraphs of this

section were implemented in MVSP v3.1, Provalis Research,

Montreal, Quebec).

We transformed the values of the UPGMA cluster matrices

(for clusters based on both Euclidean distances and Jaccard’s

similarity coefﬁcient) to matrices of cophenetic (ultrametic)

distances. We then subjectively tested the goodness-of-ﬁt of

our UPGMA clusters by calculating a cophenetic correlation

based on the cophenetic distance matrix and the original

distance matrix (Euclidean or Jaccard) that was used in the

UPGMA analysis. The cophenetic correlation measures the

extent to which the clustering results correspond to the

original resemblance matrix (Legendre & Legendre, 1998).

A cophenetic correlation coefﬁcient of > 0.9 represents a very

good ﬁt, whereas a coefﬁcient of 0.8 < r < 0.9 represents a

good ﬁt to the data (these analyses were performed in

NTSYSpc2.1, Exeter Software, New York, USA).

In addition, two partial Mantel tests were performed to test

the signiﬁcance of the faunistic relationships among drainage

basins described by Euclidean distances and Jaccard’s simi-

larity coefﬁcient (R Package v4.0). The data were ﬁrst

partitioned into two randomly deﬁned groups of species (y

and z). We used Euclidean distances (or Jaccard’s similarity

coefﬁcient) to produce two matrices that described the

faunistic similarity between the rivers for species group y

and species group z (x (y) and x (z)). The Nearest Features

extension of Arcview GIS (Jenness Enterprises, http://

www.jennessent.com) was used to calculate the geographical

distance between each pair of drainage basins, resulting in a

third matrix, the geographical distance matrix. A partial

Mantel test was then performed on the faunistic distance

matrices, using the geographical distance matrix to remove

the effect of the spatial positioning of the drainage basins, in

order to determine if the faunistic relationships depicted in

our dendrograms arose because of the spatial positioning of

the drainage basins alone.

Finally, we described the biogeographical provinces of LMA

by identifying convergent results between the UPGMA dendr-

ograms. In cases for which the dendrograms based on

S. A. Smith and E. Bermingham

1838

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

Table 1 The LMA freshwater ﬁsh species by their occurrence in the described biogeographical provinces. We assembled a data base

describing the presence and absence of freshwater ﬁsh species in LMA drainage basins, a region including the rivers of Panama and Costa

Rica. The biogeographical provinces in this table refer to the areas of faunal similarity described by our analyses. The Salinity tolerance

column refers to the hypothesized salinity tolerance of freshwater ﬁshes based on Myers (1949)

Salinity

tolerance Family Species

San

Juan

Chiapas-

Nicaragua Bocas Chiriqui

Santa

Maria Chagres Tuira

1 Ageneiosidae Ageneiosus caucanus 00 0 0 0 0 1

1 Apteronotidae Apteronotus rostratus 00 0 0 0 0 1

1 Astroblepidae Astroblepus longiﬁlis 00 0 0 0 1 1

1 Astroblepidae Astroblepus trifasciatum 00 0 0 0 1 1

1 Auchenipteridae Parauchenipterus amblops 00 0 0 1 0 1

1 Callichthyidae Hoplosternum punctatum 00 0 0 0 1 1

1 Characidae Astyanax aeneus 11 1 1 1 1 1

1 Characidae Astyanax nasutus 10 0 0 0 0 0

1 Characidae Astyanax orthodus 00 1 0 0 0 0

1 Characidae Astyanax panamensis 00 0 0 0 1 1

1 Characidae Astyanax ruberrimus 00 0 0 1 1 1

1 Characidae Bramocharax bransfordii 10 0 0 0 0 0

1 Characidae Brycon argenteus 00 0 0 0 1 1

1 Characidae Brycon behreae 00 0 1 1 0 0

1 Characidae Brycon chagrensis 00 0 0 0 1 0

1 Characidae Brycon guatemalensis 10 1 0 0 0 0

1 Characidae Brycon obscurus 00 0 0 1 1 0

1 Characidae Brycon petrosus 00 0 0 0 1 1

1 Characidae Brycon sp. nov. ‘Acla’ 0 0 0 0 0 1 0

1 Characidae Brycon sp. nov. ‘Bocas’ 0 0 1 0 0 0 0

1 Characidae Brycon striatulus 00 0 0 0 0 1

1 Characidae Bryconamericus emperador 00 1 0 1 1 1

1 Characidae Bryconamericus ricae 00 1 0 0 0 0

1 Characidae Bryconamericus scleroparius 10 1 0 0 0 0

1 Characidae Bryconamericus terrabensis 00 0 1 0 0 0

1 Characidae Bryconamericus zeteki 00 0 0 1 0 0

1 Characidae Carlana eigenmanni 10 0 0 0 1 0

1 Characidae Characidium marshi 00 0 0 0 0 1

1 Characidae Compsura dialeptura 00 0 1 1 0 0

1 Characidae Compsura gorgonae 00 0 0 1 1 1

1 Characidae Compsura mitoptera 00 0 0 0 1 0

1 Characidae Compsura sp. nov. ‘Chiriqui’ 0 0 0 1 0 0 0

1 Characidae Creagrutus afﬁnis 00 0 0 0 0 1

1 Characidae Eretmobrycon bayano 00 0 0 0 0 1

1 Characidae Gephyrocharax atricaudata 00 0 0 0 1 1

1 Characidae Gephyrocharax chocoensis 00 0 0 0 1 0

1 Characidae Gephyrocharax intermedius 00 0 1 1 1 0

1 Characidae Gephyrocharax sp. nov. ‘Chiriqui’ 0 0 0 1 0 0 0

1 Characidae Hemibrycon dariensis 00 0 1 1 1 1

1 Characidae Hyphessobrycon panamensis 00 1 0 0 1 0

1 Characidae Hyphessobrycon savagei 00 0 1 0 0 0

1 Characidae Hyphessobrycon sp. nov. ‘Acla’ 0 0 0 0 0 1 0

1 Characidae Hyphessobrycon sp. nov. ‘Bayano’ 0 0 0 0 0 0 1

1 Characidae Hyphessobrycon sp. nov. ‘Bocas’ 0 0 1 0 0 0 0

1 Characidae Hyphessobrycon tortuguerae 10 0 0 0 0 0

1 Characidae Phenagoniates macrolepis 00 0 0 0 0 1

1 Characidae Pseudocheirodon afﬁnis 00 0 0 1 1 1

1 Characidae Pseudocheirodon terrabae 00 0 1 0 0 0

1 Characidae Pterobrycon myrnae 00 0 1 0 0 0

1 Characidae Roeboides bouchellei 11 0 1 0 0 0

1 Characidae Roeboides carti 00 0 0 0 1 0

1 Characidae Roeboides dayi 00 0 0 0 1 0

1 Characidae Roeboides guatemalensis 00 0 0 0 1 0

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1839

Table 1 continued

Salinity

tolerance Family Species

San

Juan

Chiapas-

Nicaragua Bocas Chiriqui

Santa

Maria Chagres Tuira

1 Characidae Roeboides ilseae 00 0 1 0 0 0

1 Characidae Roeboides occidentalis 00 0 1 1 0 1

1 Characidae Roeboides sp. nov. ‘Cocle del Norte’ 0 0 0 0 0 1 0

1 Ctenoluciidae Ctenolucius beani 00 0 0 1 0 1

1 Curimatidae Cyphocharax magdalenae 00 0 1 1 0 1

1 Erythrinidae Hoplias malabaricus 00 0 0 0 1 1

1 Erythrinidae Hoplias microlepis 00 0 1 1 1 1

1 Gasteropelecidae Gasteropelecus maculatus 00 0 0 0 0 1

1 Gymnotidae Gymnotus cylindricus 11 1 0 0 0 0

1 Gymnotidae Gymnotus maculosus 11 0 0 0 0 1

1 Hemiodontidae Saccodon dariensis 00 0 0 0 0 1

1 Lebiasinidae Piabucina boruca 00 0 1 0 0 0

1 Lebiasinidae Piabucina festae 00 0 0 0 0 1

1 Lebiasinidae Piabucina panamensis 00 0 0 0 1 1

1 Loricariidae Ancistrus chagresi 00 0 0 0 1 1

1 Loricariidae Ancistrus spinosus 00 0 0 1 0 1

1 Loricariidae Chaetostoma ﬁscheri 00 0 0 0 1 1

1 Loricariidae Crossoloricaria variegata 00 0 0 0 0 1

1 Loricariidae Dasyloricaria capetensis 00 0 0 0 0 1

1 Loricariidae Dasyloricaria tuyrensis 00 0 0 0 0 1

1 Loricariidae Hypostomus panamensis 00 0 1 1 1 1

1 Loricariidae Lasiancistrus planiceps 00 0 0 0 1 1

1 Loricariidae Leptoancistrus canensis 00 0 0 1 1 1

1 Loricariidae Rineloricaria altipinnis 00 0 0 0 0 1

1 Loricariidae Rineloricaria uracantha 00 0 1 0 1 1

1 Loricariidae Sturisoma panamense 00 0 0 1 0 1

1 Loricariidae Sturisomatichthys citurensis 00 0 0 0 0 1

1 Pimelodidae Imparales panamensis 00 0 0 1 1 1

1 Pimelodidae Nannorhamdia lineata 00 0 1 0 0 0

1 Pimelodidae Pimelodella chagresi 00 0 1 1 1 1

1 Pimelodidae Pimelodella sp. nov. ‘recent’ 0 0 0 0 1 0 1

1 Pimelodidae Pimelodus clarias 00 0 0 0 0 1

1 Pimelodidae Pseudopimelodus zungaro 00 0 0 0 0 1

1 Pimelodidae Rhamdia laticauda 11 1 1 1 1 0

1 Pimelodidae Rhamdia nicaraguensis 11 0 0 0 0 0

1 Pimelodidae Rhamdia quelen 11 1 1 1 1 1

1 Rhamphichthyidae Brachyhypopomus occidentalis 00 1 1 1 1 1

1 Sternopygidae Eigenmannia humboldtii 00 0 0 0 0 1

1 Sternopygidae Sternopygus macrurus 00 0 1 1 0 1

1 Trichomycteridae Trichomycterus striatus 00 0 1 1 1 1

2 Anablepidae Oxyzygonectes dovii 11 0 1 1 0 0

2 Aplocheilidae Rivulus birkhahni 00 1 0 0 0 0

2 Aplocheilidae Rivulus brunneus 00 0 0 0 1 1

2 Aplocheilidae Rivulus chucunaque 00 0 0 0 1 1

2 Aplocheilidae Rivulus frommi 00 0 0 1 1 0

2 Aplocheilidae Rivulus fuscolineatus 11 0 0 0 0 0

2 Aplocheilidae Rivulus hildebrandi 00 0 1 1 0 0

2 Aplocheilidae Rivulus isthmensis 11 1 0 0 0 0

2 Aplocheilidae Rivulus kuelpmanni 00 1 0 0 0 0

2 Aplocheilidae Rivulus montium 00 0 0 0 1 0

2 Aplocheilidae Rivulus rubripunctatus 00 1 0 0 0 0

2 Aplocheilidae Rivulus sp. nov. ‘Rio Cocle del Norte’ 0 0 0 0 0 1 0

2 Aplocheilidae Rivulus uroﬂammeus 00 0 1 0 0 0

2 Aplocheilidae Rivulus wassamanni 00 1 0 0 0 0

2 Aplocheilidae Rivulus weberi 00 0 0 0 1 1

2 Cichlidae Aequidens coeruleopunctatus 00 0 1 1 1 1

S. A. Smith and E. Bermingham

1840

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

Table 1 continued

Salinity

tolerance Family Species

San

Juan

Chiapas-

Nicaragua Bocas Chiriqui

Santa

Maria Chagres Tuira

2 Cichlidae Amphilophus calobrensis 00 0 0 0 0 1

2 Cichlidae Amphilophus citrinellus 10 0 0 0 0 0

2 Cichlidae Amphilophus lyonsi 00 0 1 0 0 0

2 Cichlidae Amphilophus umbriferum 00 0 0 0 0 1

2 Cichlidae Archocentrus centrarchus 10 0 0 0 0 0

2 Cichlidae Archocentrus myrnae 00 1 0 0 0 0

2 Cichlidae Archocentrus nanoluteus 00 1 0 0 0 0

2 Cichlidae Archocentrus nigrofasciatus 11 1 0 0 0 0

2 Cichlidae Archocentrus panamensis 00 0 0 0 1 1

2 Cichlidae Archocentrus sajica 00 0 1 0 0 0

2 Cichlidae Archocentrus septemfasciatus 10 0 0 0 0 0

2 Cichlidae Astatheros alfari 11 0 0 0 0 0

2 Cichlidae Astatheros altifrons 00 0 1 0 0 0

2 Cichlidae Astatheros bussingi 00 1 0 0 0 0

2 Cichlidae Astatheros calobrense 00 0 0 0 0 1

2 Cichlidae Astatheros diquis 00 0 1 0 0 0

2 Cichlidae Astatheros longimanus 11 0 0 0 0 0

2 Cichlidae Astatheros rhytisma 00 1 0 0 0 0

2 Cichlidae Astatheros rostratus 10 0 0 0 0 0

2 Cichlidae Cichlasoma atromaculatus 00 0 0 0 1 0

2 Cichlidae Geophagus crassilabris 00 0 0 0 1 1

2 Cichlidae Herotilapia multispinosa 11 0 0 0 0 0

2 Cichlidae Hypsophrys nicaraguensis 10 0 0 0 0 0

2 Cichlidae Neetroplus nematopus 10 0 0 0 0 0

2 Cichlidae Parachromis dovii 11 0 0 0 0 0

2 Cichlidae Parachromis loisellei 10 1 0 0 0 0

2 Cichlidae Parachromis managuensis 10 0 0 0 0 0

2 Cichlidae Theraps sieboldii 01 0 1 1 0 0

2 Cichlidae Theraps sp. nov. ‘Rio Viento’ 0 0 1 0 0 0 0

2 Cichlidae Theraps underwoodi 10 0 0 0 0 0

2 Cichlidae Vieja maculicauda 10 0 0 0 1 0

2 Cichlidae Vieja tuyrensis 00 0 0 0 0 1

2 Poeciliidae Alfaro cultratus 11 1 0 0 0 0

2 Poeciliidae Belonesox belizanus 10 0 0 0 0 0

2 Poeciliidae Brachyrhaphis cascajalensis 00 1 0 0 1 1

2 Poeciliidae Brachyrhaphis episcopi 00 0 0 1 1 1

2 Poeciliidae Brachyrhaphis holdridgei 10 0 0 0 0 0

2 Poeciliidae Brachyrhaphis olomina 11 0 0 0 0 0

2 Poeciliidae Brachyrhaphis parismina 10 0 0 0 0 0

2 Poeciliidae Brachyrhaphis rhabdophora 11 0 1 0 0 0

2 Poeciliidae Brachyrhaphis roseni 00 0 1 1 0 0

2 Poeciliidae Brachyrhaphis roswithae 00 0 0 1 1 0

2 Poeciliidae Brachyrhaphis terrabensis 00 0 1 0 0 0

2 Poeciliidae Gambusia nicaraguensis 10 0 0 0 0 0

2 Poeciliidae Neoheterandria cana 00 0 0 0 0 1

2 Poeciliidae Neoheterandria tridentiger 00 1 1 1 1 1

2 Poeciliidae Neoheterandria umbratilis 10 0 0 0 0 0

2 Poeciliidae Phallichthys amates 11 1 0 0 0 0

2 Poeciliidae Phallichthys quadripunctatus 00 1 0 0 0 0

2 Poeciliidae Phallichthys tico 10 0 0 0 0 0

2 Poeciliidae Poecilia caucana 00 0 0 0 0 1

2 Poeciliidae Poecilia gillii 11 1 1 1 1 1

2 Poeciliidae Poeciliopsis elongata 01 0 1 1 0 1

2 Poeciliidae Poeciliopsis paucimaculata 00 0 1 0 0 0

2 Poeciliidae Poeciliopsis retropinna 00 0 1 0 0 0

2 Poeciliidae Poeciliopsis sp. nov. ‘La Yeguada’ 0 0 0 0 1 0 0

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1841

Euclidean distances and Jaccard’s similarity coefﬁcient differed

in the identiﬁcation of drainage membership in biogeograph-

ical provinces, we examined the distribution of species’ ranges

across regions of high faunal turnover in order to resolve any

inconsistencies between dendrograms regarding the placement

of biogeographical boundaries. Moreover, we examined the

data for drainages that may have been inappropriately assigned

to a biogeographical province as a result of inadequate

sampling of the fauna of that particular drainage. In general,

when we deemed that drainages were positioned in an

inappropriate cluster owing to inadequate sampling, we placed

these drainages in a biogeographical province based on the

geographical location of the drainage.

Community composition, species richness, endemism,

and range size

We calculated the relative community composition of bioge-

ographical provinces by dividing the species richness of each

freshwater ﬁsh family by the total number of species present in

each biogeographical province, providing an estimate of the

relative contribution of each family to the observed species

richness for the province. Endemic species were deﬁned as

species whose global distributions are limited to one biogeo-

graphical province in the LMA region. The number of endemic

species for each biogeographical province was divided by the

species richness of the province to obtain a measure of the

relative degree of endemism of each biogeographical province.

The range size of LMA freshwater ﬁshes was calculated by

summing the number of biogeographical provinces occupied

by each species.

Beta-diversity

The relative permeability of dispersal barriers between bioge-

ographical provinces was estimated by the speciﬁc covariance

of occurrence between provinces. The speciﬁc covariance of

occurrence (Bell, 2003) provides a standard method of

expressing Whittaker’s (1975) concept of beta-diversity and

is deﬁned as

CovðX

; X

Þ¼

 n

NðN  1Þ

;

where n

is the number of species that occur at both sites

(provinces in this article), n

is the number of species that

occur at site j but not at site k, n

is the number of species that

occur at site k but not at site j, n

is the number of species

that do not occur at either of the sites, and N is the number of

species in the survey (in this case, 170). Larger values of

speciﬁc covariance between sites indicate greater faunal

similarity, or decreased beta-diversity or turnover between

the two sites.

RESULTS AND DISCUSSION

Biogeographical provinces

The high degree of faunal turnover between the Rio Tuira in

eastern Panama and the Rio Atrato in north-western Colombia

(Loftin, 1965; Bermingham et al., 1997a) suggests that the

boundary between the countries, which follows the eastern

limit of the LMA cordillera extending from the Caribbean to

the Paciﬁc coastline at the Colombian–Panama border

(Fig. 1), is also an important biogeographical barrier. Further-

more, north-western Colombia has a history and ichthyofauna

that predate the rise of the Isthmus of Panama (Coates et al.,

2004). At the other geographical extreme of LMA, a marine

corridor existed in the region of Lake Nicaragua and the Rio

San Juan well into the Pliocene (Coates & Obando, 1996),

separating LMA from nuclear Mesoamerica near the current

political boundary between Costa Rica and Nicaragua. Miller

(1966) and Bussing (1976), however, hypothesized that the

biogeographical provinces of both the Atlantic and Paciﬁc

slopes of Costa Rica continued northwards into Nicaragua (the

San Juan and Chiapas-Nicaragua provinces, respectively),

suggesting that the political border between Costa Rica and

Nicaragua may not represent the northern terminus of the

biogeographical provinces described herein.

We identiﬁed seven biogeographical provinces (Fig. 2),

primarily based on convergent results of the UPGMA dendr-

ograms (Figs 3 & 4). For the large majority of cases, both

dendrograms yielded very similar results with respect to

faunistic relationships between drainage basins and between

areas. In addition, our cophenetic correlation analyses for

UPGMA clusters based on Jaccard’s similarity coefﬁcient

(r ¼ 0.85) and Euclidean distances (r ¼ 0.83) indicate that

our UPGMA clusters represent a good ﬁt to our data.

Nevertheless, several inconsistencies between the dendro-

grams exist. First, the dendrogram based on Euclidean

distances clustered several drainage basins from the Tuira

and Chagres provinces together. This depiction of the faunistic

Table 1 continued

Salinity

tolerance Family Species

San

Juan

Chiapas-

Nicaragua Bocas Chiriqui

Santa

Maria Chagres Tuira

2 Poeciliidae Poeciliopsis turrubarensis 01 0 1 1 0 1

2 Poeciliidae Priapichthys annectens 11 1 0 0 0 0

2 Poeciliidae Priapichthys darienensis 00 0 0 1 0 1

2 Poeciliidae Priapichthys panamensis 01 0 1 0 0 0

2 Synbranchidae Synbranchus marmoratus 11 1 1 1 1 1

S. A. Smith and E. Bermingham

1842

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

relationships between drainages differed from that based on

Jaccard’s similarity coefﬁcient (Fig. 4), which placed the

drainage basins of the Tuira and Chagres provinces in separate

clusters. The clusters based on Euclidean distances (Fig. 3)

reﬂect the faunal similarity of these rivers owing to the

substantial number of shared species between the Tuira and

Chagres provinces. In addition, the substantial number of

species whose distribution is limited to the extremes of eastern

Paciﬁc Panama (the Tuira province) probably contributed to

the relative dissimilarity among drainages within the Tuira

province. We conclude, however, that the drainage basins on

the Caribbean and Paciﬁc slopes of eastern Panama represent

discrete areas of faunal similarity and therefore qualify as

separate biogeographical provinces (Fig. 4).

Furthermore, the dendrograms (Figs 3 & 4) identiﬁed

different drainages separating the Santa Maria and Chiriqui

provinces. This inconsistency clearly arises as a result of

inadequate sampling of the Rio Playita, Rio Pavo, Rio Oria and

Rio Cana drainage basins in the transitional area between

provinces, as well as the natural decline in species richness

along the Azuero Peninsula owing to a peninsular effect. Thus,

we placed these drainages in biogeographical provinces based

on the drainages’ geographical locations. Finally, in cases for

which the two dendrograms yielded differing descriptions of

the faunal relationships between drainage basins and between

areas, we also considered the distribution of species’ ranges

across province boundaries to settle inconsistencies between

dendrograms. For example, based on the eight species whose

westernmost distributions are located in the Rio Santa Maria,

and the six species whose easternmost distributions are

situated in the Rio Tabasara and the Rio San Pedro (three in

each), we concluded that the boundary between the biogeo-

graphical provinces of the Chiriqui and Santa Maria provinces

should be placed along the eastern edge of the Rio San Pablo

and the Rio San Pedro drainages (Fig. 4). The entire region,

however, clearly represents an area of broad faunal turnover

between the Santa Maria and the Chiriqui provinces.

Hugueny & Leveque (1994) point out that, because the

dispersal of freshwater ﬁshes often depends on direct

connections between neighbouring rivers, the mean faunistic

distance between drainage basins may be positively correlated

with the geographical distance between drainages. Thus, it is

critical to remove the variation that results from a drainage’s

geographical position. We therefore analysed our data using

partial Mantel tests in order to remove the effect of the spatial

positioning of the drainage basins, and found a signiﬁcant

correlation between the faunistic distances described in matrix

x (y) and matrix x (z). We conclude that the faunistic

relationships did not arise solely as a result of the spatial

arrangement of the drainage basins (Table 2). Rather, the

results of the partial Mantel tests suggest that the clusters that

we used to deﬁne the biogeographical provinces identiﬁed

groups of drainages with shared biological history. Similar

processes acting in the described biogeographical provinces

have given rise to the observed faunal similarity among

drainages within a province, and, therefore, patterns within the

provinces can be used to illuminate the mechanisms respon-

sible for the distribution and diversity of freshwater ﬁshes at

the landscape scale of LMA.

Previous attempts to describe the biogeographical provinces

of the freshwater ﬁsh fauna of Mesoamerica led to the

description of only four regions – the Chiapas-Nicaragua,

Usumacinta, San Juan, and Isthmian provinces (Miller, 1966;

Bussing, 1976). The reduced biogeographical resolution of

these studies results from differences in the knowledge

regarding the distribution of Mesoamerican ﬁshes, and from

the larger spatial scale at which their analyses were performed.

For example, our results are based on more complete

Figure 2 The biogeographical provinces of

LMA. The faunistic relationships between

LMA drainage basins were inferred based on

concordance between methodologies and

summarized to describe the biogeographical

provinces of LMA freshwater ﬁshes depicted

in this ﬁgure.

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1843

3.6 3 2.4 1.8

UPGMA

Euclidean

1.2 0.6 0

Rio Robalo

Rio Guarumo

Rio Guariviara

Rio Canaveral

Rio San San

Rio Calovebora

Rio Changuinola

Rio Tortuguero

Rio San Juan CR

Rio Parismina

Rio Matina

Rio Tempisque

Rio Bebedero

Rio Tarcoles

Rio Barranca

Rio Terraba

Rio Pirris

Rio San Juan

Rio Palo Blanco

Rio Escarrea

Rio Platanal

Rio Chiriqui

Rio Coto

Rio Chiriqui Viejo

Rio Chico

Rio Tebario

Rio Tonosi

Rio Farallon

Rio La Villa

Rio Santa Maria

Rio Chorrera

Rio Chame

Rio Parita

Rio Anton

Rio Pavo

Rio Oria

Rio Playita

Rio Cana

Rio Tabasara

Rio San Felix

Rio Estero Salado

Rio San Pedro

Rio San Pablo

Rio Cate

IC#126

Rio Lara

Rio Iglesia

Rio Tuira

Rio Sambu

Rio Bayano

Rio Cocle del Sur

Rio Sajalices

Rio Grande

Rio Capira

Rio Pacora

Rio Juan Diaz

Rio Caimito

Rio Cabra

Rio Mandinga

Rio Chagres

Rio Azucar

Rio Cascajal

Rio Cuadi

Rio Acla

Rio Cuango

Rio Cocle del Norte

Rio Pina Pina

Rio Miguel de la Borda

Rio Indio

Rio Playon Chico

IC#121

Rios of Nicoya Peninsula

Rio Cricamola

Rio Sixaola

Rio Uyama

Q. Marin

Q. La Margarita

Q. La Gloria

Q. Cilico Creek

IC#91

Figure 3 Dendogram depicting the faunal relationships among LMA drainage basins as deﬁned by a UPGMA cluster analysis based on the measure of Euclidean distances. The colours adjacent

to the names of the drainages denote their respective biogeographical provinces (we used the same colour scheme as in Fig. 2). The biogeographical provinces (denoted by their colours in

parentheses) are as follows: San Juan (yellow), Chiapas-Nicaragua (green), Bocas (orange), Chiriqui (maroon), Chagres (red), Santa Maria (grey) and Tuira (teal).

S. A. Smith and E. Bermingham

1844

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

0.04 0.2 0.36 0.52

UPGMA

Jaccard’s Coefficient

0.68 0.84 1

Rio Tortuguero

Rio San Juan CR

Rio Parismina

Rio Matina

Rio Tempisque

Rio Tarcoles

Rio Bebedero

Rio Robalo

Rio Guarumo

Rio Cricamola

Rio Sixaola

Rio Changuinola

Q. Cilico Creek

Rio Uyama

IC#91

Rio Pina Pina

Rio Sajalices

Rio Playita

Rio Pavo

Rio Oria

Rio Cana

Rio Terraba

Rio Pirris

Rio San Juan

Rio San Felix

Rio Palo Blanco

Rio Tabasara

Rio San Pedro

San Pablo

Rio Estero Salado

Rio Chiriqui Viejo

Rio Coto

Rio Chiriqui

Rio Escarrea

Rio Platanal

Rio Chico

Rio Chorrera

Rio Tebario

Rio Cate

Rio Tonosi

Rio Farallon

Rio Chame

Rio La Villa

Rio Parita

Rio Santa Maria

Rio Cocle del Sur

Rio Anton

IC#126

Rio Sambu

Rio Lara

Rio Iglesia

Rio Pacora

Rio Juan Diaz

Rio Grande

Rio Capira

Rio Caimito

Rio Cabra

Rio Tuira

Rio Bayano

Rio Miguel de la Borda

Rio Indio

Rio Cocle del Norte

Rio Chagres

Rio Cuango

Rio Cuadi

Rio Cascajal

Rio Mandinga

Rio Playon Chico

Rio Azucar

Rio Acla

IC#121

Rio San San

Q. La Gloria

Rio Calovebora

Rio Barranca

Rios of Nicoya Peninsula

Q. La Margarita

Q. Marin

Rio Guariviara

Rio Canaveral

Figure 4 Dendogram depicting the faunal relationships among LMA drainage basins as deﬁned by a UPGMA cluster analysis based on Jaccard’s similarity coefﬁcient. The colours adjacent to the

names of the drainages denote their respective biogeographical provinces (we used the same colour scheme as in Fig. 2). The biogeographical provinces (denoted by their colours in parentheses)

are as follows: San Juan (yellow), Chiapas-Nicaragua (green), Bocas (orange), Chiriqui (maroon), Chagres (red), Santa Maria (grey) and Tuira (teal).

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1845

geographical sampling, and clearly demonstrate signiﬁcant

faunal turnover across the Isthmian province described by

Miller (1966) and Bussing (1976). We deﬁne four discrete

biogeographical provinces in this region: Chiriqui, Santa

Maria, Tuira and Chagres (Fig. 2). In addition, Bussing

(1976) locates the southern boundary of the San Juan province

east of the Rio Calovebora, a result not supported by our data,

which establish a faunal break between the Rio Matina and the

Rio Sixaola drainage basins. Our analysis identiﬁes a new

biogeographical province in the region of Bocas del Toro

encompassing the drainage basins between the Rio Calovebora

and the Rio Sixaola (Bocas province), with the San Juan

province to the north, probably extending to the Prinzapoloka

drainage basin of Nicaragua, as previously asserted by Bussing

(1976). The divide between the Rio Tarcoles and the Rio Pirris

forms the biogeographical province of Chiapas-Nicaragua to

the north, and the newly named Chiriqui province to the

south. This boundary corresponds to that described by Miller

(1966) and Bussing (1976) between their Chiapas-Nicaragua

and Isthmian provinces.

Community composition and species richness of LMA

biogeographical provinces

Changes in community composition and to a lesser extent in

species richness are striking as one travels east to west across

the LMA landscape, reﬂecting changes in the relative

contribution of primary versus secondary freshwater ﬁshes.

Species representing families of the secondary freshwater

ﬁshes Cichlidae and Poeciliidae have undergone substantial

diversiﬁcation and predominate in nuclear Mesoamerica, and

this is reﬂected in the biogeographical provinces of north-

western LMA (the San Juan, Chiapas-Nicaragua, and Bocas

provinces, Fig. 5), whereas primary freshwater ﬁshes are a

considerably more prominent feature of south-eastern LMA

communities. For example, Cichlidae and Poeciliidae consti-

tute more than 60% of the San Juan fauna, whereas

Characidae constitute only 18% of the total species richness

of this province and Loricariidae are completely absent. In

contrast, primary freshwater ﬁshes of the Characidae family,

and to a lesser extent of the Loricariidae family, contribute a

much larger percentage of the total species richness of the

south-eastern biogeographical provinces of LMA (the Chir-

iqui, Santa Maria, Chagres and Tuira provinces, Fig. 5). This

trend is most clearly exempliﬁed in the Chagres province,

where Characidae and Loricariidae make up approximately

50% of the fauna, while Poeciliidae and Cichlidae represent

only 17% of the total number of species present. However,

changes in community composition across the LMA land-

scape are not mirrored by changes in the species richness of

the biogeographical provinces. Species richness is highest at

the eastern extreme of LMA, but does not decline smoothly

as one travels west across the LMA, reﬂecting the counter-

acting contribution of the diversiﬁcation of the Cichlidae and

Poeciliidae in north-western lower and nuclear Mesoamerica.

This pattern is most clearly demonstrated by the increase in

species richness in the San Juan and Chiriqui provinces

(Fig. 6).

Figure 5 The relative contributions of

selected LMA freshwater ﬁsh families to the

species richness of the LMA biogeographical

provinces.

Table 2 Results from two partial Mantel tests that were used to

evaluate the signiﬁcance of the faunistic relationships among LMA

drainage basins established by UPGMA cluster analyses based

on Euclidean distances and Jaccard’s similarity coefﬁcient. Three

matrices were used for each of the two partial Mantel tests that

were performed. Each partial Mantel test used a matrix that des-

cribed the geographic distances between drainages to remove the

effects of the spatial positioning of the drainages, and two matrices

that described the faunistic distance/similarity between rivers

based on the two sets of randomly deﬁned species lists

Measure of faunal

similarity/distance r

No. of

permutations P-value

Euclidean distance 0.447 9999 < 0.0005

Jaccard’s similarity

coefﬁcient

0.609 9999 < 0.0005

S. A. Smith and E. Bermingham

1846

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

Regional processes and biological differences among fresh-

water ﬁsh families played a central role in giving rise to the

geographical variation in community composition across the

LMA landscape. The emergence of the Isthmus of Panama over

a period of 15 million years beginning in the mid-Miocene,

coupled with episodes of sea-level regression, set the stage for

the colonization of Mesoamerica by primary freshwater ﬁshes

(Miller, 1966; Myers, 1966; Bussing, 1976, 1985; Bermingham

& Martin, 1998; Perdices et al., 2002), and probably by

secondary freshwater ﬁshes as well (Murphy & Collier, 1996;

Martin & Bermingham, 1998). In particular, several periods of

geological development of the LMA landscape appear to have

played important roles in determining the success of LMA

freshwater ﬁsh colonists.

In the mid-Miocene, the region of present-day LMA was

made up of a series of islands; marine connections between the

Paciﬁc and Caribbean were commonplace, and Central and

South America were separated by an ocean barrier of abyssal-

bathyal depths (> 2000 m) in the region of present-day eastern

Panama and the Colombian Choco (Coates & Obando, 1996).

There is no evidence that primary freshwater ﬁshes colonized

Mesoamerica during this time, but molecular analyses based

on an mtDNA cytochrome b molecular clock suggest that

heroine cichlids (Martin & Bermingham, 1998) and Rivulus

(Murphy & Collier, 1996) arrived approximately 18–15 Ma

(1–1.2% sequence divergence per Myr, Bermingham et al.,

1997b). How they did so is unknown. However, our

presumption is that colonization was facilitated by some

physiological tolerance of seawater.

There are many examples of salinity tolerance in cichlids;

for example, the earliest fossil records of cichlids in South

America in the Miocene strongly suggest that cichlids

migrated from Africa to South America across the South

Atlantic via the South Equatorial Current (Murray, 2001),

providing biogeographical evidence that cichlids are capable

of crossing marine barriers. Furthermore, it is important to

note when considering the dispersal of freshwater ﬁshes

across salt-water barriers that salinity levels of seawater may

vary (Matthews, 1998). Lindsey & McPhail (1986) noted that

temporary freshwater or brackish-water bridges can be

formed in the sea during periods of high runoff, which

may facilitate dispersal of freshwater taxa across oceanic gaps

separating freshwater environments. It is therefore possible

that salinity tolerance coupled with large ﬂooding events

associated with the northward discharge of the proto-Amazon

in the region of the Magdalena river and Maracaibo basin

(Lundberg et al., 1998; Perdices et al., 2002) provided the

means for the Miocene colonization of Mesoamerica by

secondary freshwater ﬁshes, an event that pre-dated the ﬁnal

uplift of the Isthmus of Panama. The hypothesis of chance

colonization during the Miocene is supported by molecular

analyses: the phylogenies for heroine cichlids and Rivulus

suggest that the colonization of Mesoamerica occurred once

or no more than several times in each clade approximately

18–15 Ma (Martin & Bermingham, 1998).

Molecular analyses of extant taxa suggest that modern

lineages of freshwater ﬁshes did not colonize Mesoamerica

prior to the mid-Miocene (Bermingham & Martin, 1998;

Martin & Bermingham, 1998), and the fossil record does not

suggest otherwise. Moreover, none of the North American

primary freshwater ﬁsh families successfully colonized Meso-

america south of Guatemala, and only the salt-tolerant

Neararctic gars have managed to colonize as far south as the

Great Lakes basin of Nicaragua and Costa Rica (Miller, 1966).

Multiple lines of evidence therefore suggest that the streams

and rivers of the LMA landscape very probably contained

highly depauperate primary and secondary freshwater ﬁsh

communities. Thus, we postulate that early LMA freshwater

ﬁsh colonists encountered many empty ecological niches.

Conditions during the mid-Miocene in nuclear Central

America fostered an adaptive radiation among the Cichlidae

and Poeciliidae in order to ﬁll the ‘ostariophysan vacuum’ that

existed in Mesoamerica at the time (Myers, 1966; Martin &

Bermingham, 1998). This early radiation explains the relatively

large contribution of Cichlidae and Poeciliidae to the species

richness of northern LMA.

The continuing uplift of the Panamanian Isthmus, com-

bined with the dramatically reduced sea level (Haq et al., 1987)

at the close of the Miocene, may have resulted in the ephemeral

emergence of the Isthmus (Savin & Douglas, 1985) perhaps

providing the earliest opportunity for the colonization of LMA

by primary freshwater ﬁshes (Bermingham & Martin, 1998).

The Pliocene high sea-level stand would certainly have

inundated the nascent Isthmus, causing extirpation and

allopatric separation in central and eastern Panama. Local

extinction events probably facilitated the subsequent coloniza-

tion of freshwater ﬁshes in this region, which may partially

explain the relative richness of primary freshwater ﬁshes in this

region. Unlike the initial colonization of Mesoamerica by

Figure 6 Species richness of the biogeographical provinces of

LMA.

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1847

secondary freshwater ﬁshes, where the successful colonization

of nuclear Central America was determined by hypothesized

differences in salinity tolerance among freshwater ﬁsh families,

the success of subsequent colonization events would have been

primarily determined by the ﬁshes’ abilities to disperse across

the landscape as well as by the conditions (distribution and

diversity of ﬁshes) of the putative source populations in north-

western South America. The species richness of Characidae and

Loricariidae in north-western Colombia far surpasses the

species richness of other families of freshwater ﬁshes (Table 3).

These conditions clearly favoured colonization and subsequent

allopatric diversiﬁcation of the Characidae, and to a lesser

extent of the Loricariidae, in eastern and central Panama,

explaining the high species richness of primary freshwater

ﬁshes in south-eastern LMA.

The ﬁnal closure of the Isthmus was completed approxi-

mately 3.5–3.1 Ma, serving to establish a permanent connec-

tion between north-western Colombia and eastern Panama,

which in turn permitted the Great American Biotic Inter-

change across the LMA landbridge (Marshall et al., 1979; Stehli

& Webb, 1985; Coates et al., 1992; Coates & Obando, 1996).

Subsequent geological evolution of the LMA landscape

resulted in the isolation of certain areas, which had important

consequences for the dispersal and diversiﬁcation of the fauna

in these regions (Bermingham & Martin, 1998). For example,

the rise of the Central Cordillera was an important vicariant

event, which separated the faunas of the Caribbean and Paciﬁc

slopes of LMA (3–2 Ma, Collins et al., 1995). Moreover,

dispersal events of freshwater ﬁshes were increasingly limited

in their spatial extent as a result of the geological development

of the LMA landscape (Bermingham & Martin, 1998). Thus, it

is likely that, owing to geological processes, only the ﬁrst

colonization event of primary freshwater ﬁshes reached the

northern drainage basins of Costa Rica (Bermingham &

Martin, 1998), leading to the disparity of the relative richness

of primary and secondary ﬁshes across LMA.

Dispersal and speciation following the colonization of

LMA

The relative importance of dispersal and speciation events

in shaping patterns of distribution and diversity of LMA

freshwater ﬁshes following initial colonization varies across the

spatial extent of LMA. Fifty-eight percent of the LMA

ichthyofauna is endemic to LMA, suggesting that regional

speciation has clearly played an important role in shaping

patterns of diversity of the LMA freshwater ﬁsh fauna.

Nevertheless, it is apparent that dispersal is a persistent aspect

of the LMA biota.

Adjacent provinces along the same slope have exchanged

faunas with increased frequency relative to provinces separated

by the continental divide. Cross-cordillera turnover values of

adjacent provinces were more pronounced than those of

adjacent provinces along the same slope in 12 out of 18 cases

(Table 4). This contrast in faunal turnover is probably the

result of two general geological mechanisms: (1) river capture;

and (2) river anastomosis across low-gradient palaeoland-

scapes. Both of these geological mechanisms certainly have

larger impacts on adjacent drainages on the same slope for two

principal reasons. First, drainage captures across a continental

divide function to increase faunal similarity only to the extent

that the ﬁsh species are likely to inhabit the captured reaches of

the river. As Hildebrand (1938) noted, LMA rivers become

increasingly depauperate with increasing elevation, thus

precluding a large number of ﬁsh species from taking part in

cross-Cordillera dispersal. Furthermore, Bishop (1995) indi-

cated that the drainage rearrangements necessary for the

dispersal of primary freshwater ﬁshes across the landscape are

rarer than biogeographers had previously speculated. In sum,

river capture may rarely cause cross-cordillera faunal exchange

Table 4 Speciﬁc covariation among the

biogeographical provinces of LMA. The spe-

ciﬁc covariation metric is an estimate of beta-

diversity, where larger values of the speciﬁc

covariation indicate reduced species turnover

between biogeographical provinces. We used

the presence/absence of freshwater ﬁsh in the

described biogeographical provinces to

calculate these values

San Juan

Chiapas-

Nicaragua Bocas Chiriqui Santa Maria Chagres Tuira

San Juan X

Chiapas-

Nicaragua

0.090 X

Bocas 0.032 X

Chiriqui )0.023 0.029 )0.011 X

Santa Maria )0.002 0.079 X

Chagres )0.007 0.064 X

Tuira 0.065 0.073 X

Table 3 Contribution of several freshwater ﬁsh families to the

overall species richness of the Rio Atrato and Rio Magdalena

drainage basins in north-western Colombia. We estimated the

percentage contribution of several obligate freshwater ﬁsh families

to the total species richness of the Rio Atrato and Magdalena

drainage basins based on data contained in the NEODAT data base

(http://www.neodat.org)

Freshwater ﬁsh family Salinity tolerance Percentage contribution

Characidae 1 35.2

Loricariidae 1 11.1

Aplocheilidae 2 2.7

Cichlidae 2 6.0

Poeciliidae 2 2.4

Synbranchidae 2 0.3

S. A. Smith and E. Bermingham

1848

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

of any magnitude. Second, river anastomosis obviously has no

impact on the faunal similarity of rivers on opposite sides of a

continental divide, but ﬁgures importantly in the facilitation of

faunal exchange among adjacent rivers along the same slope,

particularly at reduced sea level in regions where the gradient

of the continental slope is small.

An obvious exception to the general lack of cross-Cordillera

faunal exchange regards the biogeographical provinces of San

Juan and Chiapas-Nicaragua, marked by the lowest recorded

value of species turnover (Table 4). This region is character-

ized by relatively low topography, and the traverse from the

Rio Tempisque drainage basin (Paciﬁc) across the Isthmus to

the Rio San Carlos (part of the Rio San Juan drainage basin,

Caribbean) does not rise above 45 m in altitude (Coates &

Obando, 1996). The dispersal of freshwater ﬁshes has

obviously not been as strongly impeded between the San Juan

and Chiapas-Nicaragua provinces in comparison with other

areas in LMA where the Cordillera reaches much higher

elevations and represents a more formidable barrier to ﬁsh

dispersal. The faunal similarity between these provinces may

also have been promoted by local extinction events, and

subsequent re-colonization, associated with marine inunda-

tions of the southern Chiapas-Nicaragua province during

interglacial periods (Haq et al., 1987). Extinction can also be

invoked to give a partial explanation of the low species richness

and absence of endemic ﬁshes in the Chiapas-Nicaragua

province.

The Paciﬁc slope of Panama, owing to the large number of

shared species between biogeographical provinces, provides

one of the clearest examples of facilitated dispersal among

rivers along the same slope. 16 of the 25 species shared between

the Chiriqui and the Santa Maria provinces are also shared

with the Tuira province. The low degree of faunal turnover

between Paciﬁc slope drainages, coupled with the pattern of

increasing faunal turnover from east to west, has several

possible explanations. It is probable that the nascent Isthmus

was characterized by relatively little topographic heterogeneity

and thus by relatively large drainage basins with very similar

faunas among Paciﬁc coast biogeographical provinces. As the

topographic complexity of the Isthmus increased, opportun-

ities for river anastomosis and faunal exchange between some

biogeographical provinces are likely to have decreased. For

example, a coastal mountain chain bisecting the Santa Maria

and Chiriqui biogeographical provinces developed in the

region of the Sona peninsula. This peninsula extends close to

the continental edge and thus there would have been little or

no opportunity for rivers at the adjacent margins of the

Chiriqui and Santa Maria provinces to anastomose across the

exposed continental ﬂoodplain. Our analysis suggests, not

surprisingly, that both the geographical description of bioge-

ographical provinces, and the porosity of biogeographical

barriers separating them probably changed over time. Fur-

thermore, the relatively high endemicity of Chiriqui province

(30%) is probably a result of its increasing isolation over time.

This view of landscape change over time, and the increasing

evolutionary independence of the biogeographical provinces

along the Paciﬁc slope, is supported by molecule-based

phylogeographic analysis of widespread taxa across the

LMA landscape (Bermingham & Martin, 1998; Perdices et al.,

2002).

Our results combined with molecular analyses (Bermingham

& Martin, 1998) permit a strong inference that dispersal events

resulting from river anastomosis occurring during periods of

reduced sea level have had a large impact on the faunal

similarity of the Santa Maria and Tuira provinces. Thirty

species are shared by the Santa Maria and Tuira provinces, and

the faunal turnover between these provinces is relatively low

(Table 4). The increased likelihood of dispersal events in this

region can in part be attributed to the low gradient of the

continental shelf in the Bay of Panama (Loftin, 1965;

Bermingham & Martin, 1998). The gradual decline of the

continental shelf of the Bay of Panama combined with periods

of reduced sea level during glacial maxima would have greatly

facilitated ﬁsh dispersal through anastomizing lowland streams

and swamps extending from the Rio Tuira to the streams of the

Azuero Peninsula (Loftin, 1965; Bermingham & Martin, 1998).

We modelled the distribution of streams in the Bay of Panama

during the last glacial maxima using GRASS in order to

examine how periods of reduced sea level during the last glacial

maxima might have inﬂuenced the connectivity of present-day

drainages. Our GIS modelling analysis (Fig. 7) suggests that

many of the rivers that empty into the Bay of Panama were

connected during the last glacial maxima; however, the spatial

resolution of the data is not sufﬁcient for us to present the

palaeodrainage patterns with precision. Nevertheless, it is clear

that geological processes facilitating dispersal between drai-

nages in this area have had a large inﬂuence on the patterns of

distribution and diversity observed in the Santa Maria and

Tuira provinces. Our ﬁndings complement those of Bishop

(1995), who suggested that the frequency of river captures is

rarer than previously hypothesized by biogeographers. Our

results indicate that river anastomosis during glacial maxima

may occur more frequently and have a signiﬁcantly larger

impact on patterns of distribution and genetic divergence of

freshwater ﬁshes than the process of river capture.

The pan-Paciﬁc dispersal corridor of central Panama stands

in contrast to the relative isolation of Bocas province from

adjacent provinces on the Caribbean slope (Table 4). Speciation

has clearly had a large impact on patterns of distribution and

diversity in this region: 42% of the fauna is endemic to the

province (Fig. 8). Our analyses indicate that very strong

dispersal barriers exist between the Bocas province and the

ichthyological provinces along the same slope (the San Juan and

Chagres provinces). The faunal assembly of the Bocas province

may contain species (e.g. Brachyhypopomus; see Bermingham &

Martin, 1998) derived from an early colonization event 7–4 Ma,

followed by isolation from the rest of LMA approximately

3 Ma. The relative degree of isolation of the Bocas province

from its Caribbean slope neighbours can be partly attributed

to the gradient of the continental shelf in this area, which

probably acted to reduce the frequency and extent of river

anastomosis in this region relative to central Paciﬁc Panama.

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1849

Distribution of species ranges and dispersal limitation

The distribution of range size for the freshwater ﬁshes of LMA

is geometric, whereby many more species have small ranges

than large ranges (Fig. 9). This pattern is mirrored by the

distribution of species ranges as a function of the number of

drainage basins occupied (data not presented). Hugueny (1990)

reported a similar geometric distribution for the range size of

Nilo-Sudanian freshwater ﬁshes. Gaston (1994, 1996) has

argued that the principal determinants of range-size distribu-

tion are habitat availability, habitat generalism, breadth of

environmental tolerance and dispersal ability. The pattern of

the geometric distribution of range size, however, has also been

successfully reproduced by neutral community models in which

the demographic properties of individuals are the same and

dispersal gives rise to many of the macroecological patterns

reported in the ecological literature (Bell, 2001; Hubbell, 2001).

Dispersal limitation, whether it is as a manifestation of neutral

Figure 8 Patterns of endemism in the LMA biogeographical

provinces. The number of endemic species for each biogeo-

graphical province was divided by the species richness of the

province to obtain a measure of the relative degree of endemism of

each biogeographical province. The number above each of the bars

indicates the number of endemic species located in each province.

Figure 9 The distribution of range size (as measured by the

number of biogeographical provinces occupied) for the freshwater

ﬁshes of LMA.

Figure 7 Bathymetric streams of LMA during the last interglacial (sea level was set at ) 110 m from present sea level). The bathymetric

streams were modelled using the r. function of GRASS.

S. A. Smith and E. Bermingham

1850

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

processes, or a function of the breadth of a species’ environ-

mental tolerances and the corresponding distribution of habitat

across the landscape, or some combination thereof, is probably

responsible for the observed patterns of range-size distribution

of LMA freshwater ﬁshes.

Given dispersal limitation, it would seem to follow that

secondary freshwater ﬁshes would have larger range sizes than

primary ﬁshes as a result of the increased salinity tolerance

posited for the former group, and thus an increased probability

of dispersal along coastlines. However, the average range sizes

of primary and secondary freshwater ﬁshes are not signiﬁcantly

different (t-test, P ¼ 0.54), suggesting that these ﬁshes experi-

ence similar constraints on dispersal. The range-size distribu-

tion of both primary and secondary freshwater ﬁshes follows a

geometric curve similar to that observed for the combined plot

of species ranges across the LMA biogeographical provinces

(Fig. 9). Many more species of both primary and secondary

freshwater ﬁshes therefore have smaller ranges than larger

ranges. The small average range size of both primary and

secondary freshwater ﬁshes indicates that dispersal along

coastlines must be a fairly infrequent event and that the

differential salinity tolerance hypothesized to distinguish

primary freshwater ﬁshes from secondary freshwater ﬁshes is

not an important determinant of range size. Hugueny (1990)

also rejected the hypothesis that secondary freshwater ﬁsh

species have larger ranges and are less dispersal-limited than

primary ﬁshes for the Nilo-Sudanian freshwater ﬁsh fauna.

These results call into question the salinity tolerance of

primary and secondary freshwater ﬁshes, which has never been

experimentally veriﬁed. However, there is no escaping the

observation that the distribution of the vast majority of primary

freshwater ﬁshes is limited to the freshwaters of continents and

continental islands, whereas the distribution of secondary

freshwater ﬁshes includes many islands that have not had

continental connections during the Cenozoic (West Indian

islands, Madagascar, the Seychelles, etc., Myers, 1949). Resolu-

tion of this apparent paradox would suggest that, whereas ﬁshes

of some freshwater ﬁsh families encountered in Mesoamerica,

for example the Characidae and the Pimelodidae, are unlikely

ever to cross marine barriers, others such as the Cichlidae and

Poeciliidae are physiologically capable of doing so, but do so

only rarely. The hypothesis of rare marine dispersal posits a high

probability of monophyly in groups that have crossed marine

barriers to colonize new landscapes. This prediction appears to

be largely met by the Heroini (Cichlidae) found in Meso-

america and the Greater Antilles (Martin & Bermingham, 1998;

G. A. Concheiro et al. , unpubl. data). Additional phylogenetic

analysis as well as sophisticated physiological experimentation of

the salinity tolerance and adaptive potential of Neotropical

freshwater ﬁshes are required to determine how variable this trait

is within and between families. More detailed analysis of Myers’s

(1949) intelligent and provocative deduction regarding salinity

tolerance differences in freshwater ﬁshes based on the biogeo-

graphical distribution of freshwater ﬁsh families is warranted,

given the enormous importance of his ideas for the interpret-

ation of ﬁsh dispersal probability, and, in turn, Earth history.

CONCLUSION

Our study clearly indicates that regional processes had an

important inﬂuence on the assembly of the LMA freshwater

ﬁsh fauna and that, in particular, the process of dispersal

probably plays an ongoing role in the maintenance of LMA

freshwater ﬁsh assemblages. This observation has important

ramiﬁcations for both theory and conservation. Our results

suggest that the study of regional processes is essential to

explaining patterns of biological diversity. Biogeographical

analyses of regional faunas at large spatial scales coupled with

phylogeographic analyses that elucidate regional colonization

history will certainly provide further insight into the processes

that determine patterns of biodiversity. Biogeographical

studies paint the backdrop upon which local interactions play

themselves out, permitting inference regarding the relative

importance of local and regional processes in shaping the

diversity and structure of ecological communities. This

observation lends support to the importance of a top-down

approach to the study of species richness and diversity (Tonn

et al., 1990; Ricklefs & Schluter, 1993; Whittaker et al., 2001).

Freshwater faunas are among the most threatened taxa on

the planet (Ricciardi & Rasmussen, 1999). However, conser-

vation efforts in areas such as LMA have suffered from a

scarcity of large-scale descriptions of biodiversity patterns

capable of informing conservation decisions. Our analyses

identify areas of high faunal endemism and species richness of

the LMA freshwater ﬁsh fauna. These areas are of high

conservation value and provide a baseline for informing

conservation strategies in LMA. Similar studies of other

organisms in this area and across the globe will provide

important information for conservation planners and policy

makers alike.

Our results also emphasize the importance of designing

conservation strategies that permit the continued inﬂuence of

regional processes on patterns of local diversity. The import-

ance of dispersal in maintaining species richness over ecolog-

ical and geologic time-scales via rescue effects (Brown &

Kodric-Brown, 1977) suggests that the maintenance of

dispersal corridors between areas of suitable habitat both

within and among drainage basins should be an important

conservation priority. Drainage basins that are increasingly

isolated are more susceptible to local extirpation (or extinction

if the species is endemic to the drainage basin) and less likely to

receive colonizers (Olden et al., 2001). We conclude, therefore,

by calling for conservation planning and action that acknow-

ledges the importance that regional processes play in deter-

mining patterns of biological diversity and that incorporates

these processes in our strategies to conserve remnant bio-

diversity.

ACKNOWLEDGEMENTS

We would especially like to thank Gustavo A. Concheiro P. for

his help with GRASS- and GIS-related analyses presented in

this manuscript. We would also like to thank the following

Lower Mesoamerican ﬁsh biogeography

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1851

people, who provided invaluable assistance in the preparation

of the data presented herein: Vijay Aswani, Grettehun Grajales,

Rigoberto Gonzalez, Oris Sanjur, as well as other members of

the Bell and Bermingham laboratories for their support. This

study would not have been possible without the development

of the STRI freshwater ﬁsh collection, and the major contri-

butions to the collection made by Heidi Banford, Andrew

Martin and Anabel Perdices. Bernard Hugueny provided

insightful statistical advice as well as comments on a previous

draft of this manuscript. Leslie G. Payne kindly read an early

version of this manuscript, and Bob McDowall provided sound

editorial advice. S.A.S. was supported by the Canadian

government through a Natural Sciences and Engineering

Research Council of Canada PGS A scholarship. We would

also like to thank the Smithsonian Tropical Research Institute

for ﬁnancial and logistical support. This article is a contribu-

tion of the McGill-STRI Neotropical Environment Option,

and represents partial fulﬁlment of the requirements of the

MSc degree for that program.

REFERENCES

Bell, G. (2001) Neutral macroecology. Science, 293, 2413–2418.

Bell, G. (2003) The interpretation of biological surveys. Pro-

ceedings of the Royal Society of London Series B, Biological

Sciences, 270, 2531–2542.

Bermingham, E. & Avise, J.C. (1986) Molecular zoogeography

of freshwater ﬁshes in the south-eastern United States.

Genetics, 113, 939–965.

Bermingham, E. & Martin, A.P. (1998) Comparative mtDNA

phylogeography of neotropical freshwater ﬁshes: testing

shared history to infer the evolutionary landscape of lower

Central America. Molecular Ecology, 7, 499–517.

Bermingham, E., Branford, H., Martin, A.P. & Aswani, V.

(1997a) STRI freshwater ﬁsh collection. Neotropical ﬁsh

collections (ed. by L. Malabarba), pp. 37–38. Museu de

Ciencias e Tecnologia, PUCRS, Puerto Alegre.

Bermingham, E., McCafferty, S. & Martin, A.P. (1997b) Fish

biogeography and molecular clocks: perspectives from the

Panamanian Isthmus. Molecular systematics of ﬁshes (ed. by

T. Kocher and C. Stepien), pp. 113–128. Academic Press,

San Diego.

Bishop, P. (1995) Drainage rearrangement by river capture,

beheading and diversion. Progress in Physical Geography, 19,

449–473.

Brown, J.H. & Kodric-Brown, A. 1977. Turnover rates in

insular biogeography: effect of immigration on extinction.

Ecology, 58, 445–449.

Bussing, W.A. (1976) Geographic distribution of the San Juan

ichthyofauna of Central America with remarks on its origin

and ecology. Investigations of the ichthyofauna of Nicaraguan

lakes (ed. by T.B. Thorson), pp. 157–175. University of

Nebraska, Lincoln.

Bussing, W.A. (1985) Patterns of the distribution of the Cen-

tral American ichthyofauna. The Great American Biotic

Interchange (ed. by F.G. Stehli and S.D. Webb), pp. 453–473.

Plenum Press, New York.

Coates, A.G. & Obando, J.A. (1996) The geologic evolution of

the Central American Isthmus. Evolution and environment in

tropical America (ed. by J.B.C. Jackson, A.F. Budd and A.G.

Coates), pp. 21–57. University of Chicago Press, Chicago.

Coates, A.G., Jackson, J.B., Collins, L.S., Cronin, T.M.,

Dovosett, H.J., Bybell, L.M., Jung, P. & Obando, J.A. (1992)

Closure of the Isthmus of Panama: the near-shore marine

record of Costa Rica and western Panama. Geological Society

of America Bulletin, 104, 814–828.

Coates, A.G., Aubry, M.P., Berggren, W.A., Collins, L.S. &

Kunk, M. (2003) Early neogene history of the

Central American arc from Bocas del Toro, western Panama.

Geological Society of America Bulletin, 115, 271–287.

Coates, A.G., Collins, L.S., Aubry, M.P. & Berggren, W.A.

(2004) The geology of the Darien, Panama, and the late

Miocene-Pliocene collision of the Panama arc with north-

western South America. Geological Society of America Bul-

letin, 116, 1327–1344.

Collins, L.C., Coates, A.G. & Obando, J.A. (1995) Timing and

rates of emergence of the Limon and Bocas del Toro Basins.

Geologic and tectonic development of the Caribbean Plate

boundary in southern Central America (ed. by P. Mann), pp.

263–289. Geological Society of America Special Paper 295,

Boulder.

Gaston, K.J. (1994) Rarity. Chapman and Hall, London.

Gaston, K.J. (1996) Species range-size distributions: patterns,

mechanisms and implications. Trends in Ecology and Evo-

lution, 11, 197–201.

Gaston, K.J. (2000) Global patterns in biodiversity. Nature,

405, 220–227.

Gaston, K.J. (2003) Ecology: the how and why of biodiversity.

Nature, 421, 900–901.

Gauch, H.G., Jr (1982) Noise reduction by eigenvector ordi-

nations. Ecology, 63, 1643–1649.

Haq, B.U., Hardenbol, J. & Vail, P.R. (1987) The chronology of

ﬂuctuating sea level since the Triassic. Science, 235, 1156–1167.

Hildebrand, S.F. (1938) A new catalogue of the fresh-water

ﬁshes of Panama. Chicago Field Museum of Natural History,

XXII, 217–359.

Hubbell, S.P. (2001) The uniﬁed neutral theory of biodiversity

and biogeography. Princeton University Press, Princeton.

Hugueny, B. (1990) Geographic range of west African fresh-

water ﬁshes: role of biological characteristics and stochastic

processes. Acta Oecologica, 11, 351–375.

Hugueny, B. & Leveque, C. (1994) Freshwater ﬁsh zoogeo-

graphy in west Africa: faunal similarities between river

basins. Environmental Biology of Fishes, 39, 365–380.

Huston, M.A. (1999) Local processes and regional patterns:

appropriate scales for understanding variation in the

diversity of plants and animals. Oikos, 86, 393–401.

Jackson, D.A. & Harvey, H.H. (1989) Biogeographic associa-

tions in ﬁsh assemblages: local versus regional processes.

Ecology, 70, 1472–1484.

S. A. Smith and E. Bermingham

1852

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

Legendre, P. & Legendre, L. (1998) Numerical ecology. Elsevier,

New York.

Lindsey, C.C. & McPhail, J.D. (1986) Zoogeography of ﬁshes

of the Yukon and Mackenzie basins. The zoogeography of

North American freshwater ﬁshes (ed. by C.H. Hocutt and

E.O. Wiley), pp. 639–674. John Wiley and Sons, New York.

Loftin, H.G. (1965) The geographical distribution of freshwater

ﬁshes in Panama. Florida State University, Florida.

Losos, J.B. (1992) The evolution of convergent structure in

Caribbean Anolis communities. Systematic Biology, 41, 403–

420.

Losos, J.B. & Schluter, D. (2000) Analysis of an evolutionary

species–area relationship. Nature, 408, 847–850.

Lucena, C. (2000) Revisao taxonomica e ﬁlogenia des especies

transandinas do genero Roeboides guther (teleostei: Ostar-

iophysi: Characiforme). Comun. Mus. Cienc. Tecnol. PUCRS

ser. zool., 13, 3–63.

Lundberg, J.G., Marshall, L.G., Guerrero, J., Horton, B.,

Malabra, M.C.L.S., Wesselingh, F. (1998) The stage for Neo-

tropical diversiﬁcation: a history of tropical South American

rivers. Phylogeny and classiﬁcation of Neotropical ﬁshes (ed. by

L.R. Malabra, R.E. Rreis, R.P. Vari, Z.M. Lucena and C.A.S.

Lucena), pp. 13–48. Edipuers, Porto Alegre, Brazil.

Marshall, L., Butler, R.F., Drake, R.E., Curtis, G.A. & Tedforth,

R.H. (1979) Calibration of the Great American Interchange.

Science, 204, 272–279.

Martin, A.P. & Bermingham, E. (1998) Systematics and

evolution of lower Central American cichlids inferred

from analysis of cytochrome b gene sequences. Molecular

Phylogenetics and Evolution, 9, 192–203.

Martin, A.P. & Bermingham, E. (2000) Regional endemism

and cryptic species revealed by molecular and morphologi-

cal analysis of a widespread species of Neotropical catﬁsh.

Proceedings of the Royal Society of London Series B, Biological

Sciences, 267, 1135–1141.

Matthews, W.J. (1998) Patterns in freshwater ﬁsh ecology.

Chapman & Hall, New York.

Miller, R.R. (1966) Geographical distribution of Central

American freshwater ﬁshes. Copeia, 4, 773–802.

Mora, C., Chittaro, P.M., Sale, P.F., Kritzer, J.P. & Ludsin, S.A.

(2003) Patterns and processes in reef ﬁsh diversity. Nature,

421, 933–936.

Mueller, R.L., Macey, J.R., Jaekel, M., Wake, D.B. & Boore, J.L.

(2004) Morphological homoplasy, life history evolution,

and historical biogeography of plethodontid salamanders

inferred from complete mitochondrial genomes. Proceedings

of the National Academy of Sciences of the USA, 101, 13,820–

13,825.

Murphy, W.J. & Collier, G.E. (1996) Phylogenetic relationships

within the Aplocheiloid ﬁsh genus Rivulus (Cyprinodonti-

formes, Rivulidae): implications for Caribbean and Central

American Biogeography. Molecular Biology and Evolution,

13, 642–649.

Murray, A. (2001) The fossil record and biogeography of the

Cichlidae (Actinopterygii: Labroidei). Biological Journal of

the Linnean Society, 74, 517–532.

Myers, G.S. (1938) Fresh-water ﬁshes and West Indian zoo-

geography. Annual Report Smithsonian Institution, 1937,

339–364.

Myers, G.S. (1949) Salt-tolerance of fresh-water ﬁsh groups in

relation to zoogeographical problems. Bijdr. Dierk., 28, 315–

322.

Myers, G.S. (1966) Derivation of the freshwater ﬁsh fauna of

Central America. Copeia, 4, 766–773.

Olden, J.D., Jackson, D.A. & Peres-Neto, P.R. (2001) Spatial

isolation of ﬁsh communities in drainage lakes. Oecologia,

127, 572–585.

Parra-Olea, G. & Wake, D.B. (2001) Extreme morphological

and ecological homoplasy in tropical salamanders. Pro-

ceedings of the National Academy of Sciences of the USA, 98,

7888–7891.

Perdices, A., Bermingham, E., Montilla, A. & Doadrio, I.

(2002) Evolutionary history of the genus Rhamdia (Tele-

ostei: Pimelodidae) in Central America. Molecular Phyloge-

netics and Evolution, 25, 172–189.

Qian, H. & Ricklefs, R.E. (1999) A comparison of the taxo-

nomic richness of vascular plants in China and the United

States. The American Naturalist, 154, 160–181.

Qian, H. & Ricklefs, R.E. (2000) Large-scale processes and the

Asian bias in species diversity of temperate plants. Nature,

407, 180–182.

Ricciardi, A. & Rasmussen, J.B. (1999) Extinction rates of

North American freshwater fauna. Conservation Biology, 13,

1220–1222.

Ricklefs, R.E. (1987) Community diversity: relative roles of

local and regional processes. Science, 235, 167–171.

Ricklefs, R.E. (2004) A comprehensive framework for global

patterns of biodiversity. Ecology Letters, 7, 1–15.

Ricklefs, R.E. & Bermingham, E. (2001) Non-equilibrium

diversity dynamics of the Lesser Antillean avifauna. Science,

294, 1522–24.

Ricklefs, R.E. & Schluter, D. (1993) Species diversity in ecolog-

ical communities: historical and geographical perspectives.

University of Chicago Press, Chicago.

Savin, S.M. & Douglas, R.G. (1985) Sea level, climate, and the

Central American land bridge. The Great American Biotic

Interchange (ed. by F.G. Stehli and S.D. Webb), pp. 302–322.

Plenum Press, New York.

Smith, S.A. & Shurin, J.B. (in press) Room for one more?

Evidence for invasibility and saturation in ecological commu-

nities. Conceptual ecology and invasions biology: reciprocal

approaches to nature (ed. by M. Cadotte, S. McMahon and

T. Fukami), Springer.

Smith, S.A., Bell, G. & Bermingham, E. (2004) Cross-Cordil-

lera exchange mediated by the Panama Canal increased

the species richness of local freshwater ﬁsh communities.

Proceedings of the Royal Society of London Series B, Biological

Sciences, 271, 1889–1896.

Stehli, F.G. & Webb, S.D. (1985) The Great American Biotic

Interchange. Plenum Press, New York.

Tonn, W.M., Magnuson, J.J., Rask, M. & Toivonen, J. (1990)

Intercontinental comparison of small-lake ﬁsh assemblages:

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd 1853

Lower Mesoamerican ﬁsh biogeography

the balance between local and regional processes.

The American Naturalist, 136, 345–375.

Unmack, P.J. (2001) Biogeography of Australian freshwater

ﬁshes. Journal of Biogeography, 28, 1053–1089.

Whittaker, R.H. (1975) Communities and ecosystems . Mac-

millan, New York.

Whittaker, R.J., Willis, J. & Field, R. (2001) Scale and species

richness: towards a general, hierarchical theory of species

diversity. Journal of Biogeography, 28, 453–470.

BIOSKETCHES

Scott A. Smith is currently pursing his research interests in a PhD program in the Department of Biology at McGill University,

where he is studying the nature of invasion and its impacts on community structure, as well as the effects of human modiﬁcation of

the environment on the functioning of ecological systems. The research for this paper was conducted during the completion of

Scott’s Master’s degree under the auspices of the Neotropical Environment Option (NEO), and was carried out jointly at the

Department of Biology of McGill University and the Smithsonian Tropical Research Institute.

Eldredge Bermingham is a senior staff scientist and Deputy Director of the Smithsonian Tropical Research Institute where he

pursues his research interests in the historical biogeography and molecular systematics of neotropical ﬁshes, Caribbean island birds,

as well as a suite of other taxa.

Editor: R. M. McDowall

1854

Journal of Biogeography 32, 1835–1854, ª 2005 Blackwell Publishing Ltd

S. A. Smith and E. Bermingham