647
Neo opical Ich hyology, 7(4):647-658, 2009
Copy igh © 2009 Sociedade B asilei a de Ic iologia
S a is ical dis ibu ion models o mig a o y ish in Jacuí basin, Sou h B azil
Thaís P. Al es and Nelson F. Fon ou a
The aim o he p esen s udy was o iden i y he dis ibu ion pa e ns o mig a o y ishes in he Jacuí i e basin (Rio G ande
do Sul, Sou h B azil), p oposing a s a is ical model o p esumed dis ibu ion based on geomo phologic en i onmen al da a.
Th ough maps o occu ence p obabili y, we hope o con ibu e o decisions ega ding basin en i onmen al managemen . The
analyzed species we e: Salminus b asiliensis (dou ado), Lepo inus ob usidens (pia a), P ochilodus linea us (g uma ã) and
Pimelodus pin ado (pin ado). Samples we e made h ough in e iews wi h ishe men and local inhabi an s, co e ing he main
channel and ibu a ies o he i e s Jacuí, Taqua i-An as, Vacacaí, Vacacaí-Mi im, Pa do, Pa dinho, Sinos, and Caí. The
sampling p og am esul ed in 204 in e iews, being 187 conside ed as alid in 155 di e en sampling poin s. The p obabili y o
mig a o y ish occu ence was adjus ed h ough he LOGIT ou ine o he Id isi Andes So wa e: P = e(b0 + b1 . al i ude + b2 . basin a ea) .
(1 + e(b0 + b1 . al i ude + b2 . basin a ea))-1, whe e P is he occu ence p obabili y o he species (0-1) and b0, b1 and b2 a e he equa ion
pa ame e s. Model accu acy, o es ima ing p esence, anged om 82% o 93%. Pimelodus pin ado was ci ed o occu in 121
poin s among he 155 sampled (78.06%), P ochilodus linea us in 72 (46.45%), L. ob usidens in 62 (40.00%) and S. b asiliensis
in 58 (37.42%). Equa ion pa ame e s we e es ima ed (± s anda d e o ) as ollow: S. b asiliensis: b0 = -2.8762 ± 0.2597; b1 = -
1.3028 ± 0.0332; b2 = 1.1487 ± 0.0301; L. ob usidens: b0 = -0.8364 ± 0.2213; b1 = -1.5564 ± 0.0462; b2 = 0.9947 ± 0.0206; P ochilodus
linea us: b0 = 0; b1 = -1.3067 ± 0.0544; b2 = 0.8128 ± 0.0177; Pimelodus pin ado: b0 = -0.9487 ± 0.3688; b1 = -0.8269 ± 0.0496; b2 =
0.9255 ± 0.0304.
O obje i o do p esen e es udo oi iden i ica o pad ão de dis ibuição de peixes mig ado es da bacia hid og á ica do io Jacuí
(Rio G ande do Sul, Sul do B asil), p opondo um modelo ma emá ico de dis ibuição p esumida baseado em pa âme os
ambien ais geomo ológicos. A a és de mapas de p obabilidade de oco ência, espe a-se con ibui pa a a omada de decisões
elacionadas ao ge enciamen o des a bacia hid og á ica. As espécies analisadas o am: Salminus b asiliensis (dou ado),
Lepo inus ob usidens (pia a), P ochilodus linea us (g uma ã) e Pimelodus pin ado (pin ado). As amos as o am ob idas a
pa i de en e is as com pescado es e mo ado es locais, pe co endo-se a calha p incipal dos ios Jacuí, Taqua i-An as,
Vacacaí, Vacacaí-Mi im, Pa do, Pa dinho, Sinos e Caí. O p og ama de amos agens esul ou em 204 en e is as, sendo 187
conside adas como álidas em 155 pon os di e enciados. A p obabilidade de oco ência de peixes mig ado es oi ajus ada
u ilizando-se a o ina LOGIT do so wa e Id isi Andes: P = e(b0 + b1 . al i ude + b2 . á ea de bacia) . (1 + e(b0 + b1 . al i ude + b2 . á ea de bacia))-1; onde
P é a p obabilidade de oco ência da espécie (0-1) e b0, b1 e b2 são os pa âme os da equação. Pimelodus pin ado oi ci ado
como p esen e em 121 pon os den e os 155 amos ados (78.06%), P ochilodus linea us em 72 (46.45%), L. ob usidens em 62
(40.00%), e S. b asiliensis em 58 pon os (37.42%). A p ecisão do modelo, pa a a p esença es imada, icou en e 82% e 93%. Os
pa âme os es imados da equação são desc i os a segui : S. b asiliensis: b0 = -2.8762 ± 0.2597; b1 = -1.3028 ± 0.0332; b2 = 1.1487
± 0.0301; L. ob usidens: b0 = -0.8364 ± 0.2213; b1 = -1.5564 ± 0.0462; b2 = 0.9947 ± 0.0206; P ochilodus linea us: b0 = 0; b1 = -1.3067
± 0.0544; b2 = 0.8128 ± 0.0177; Pimelodus pin ado: b0 = -0.9487 ± 0.3688; b1 = -0.8269 ± 0.0496; b2 = 0.9255 ± 0.0304.
Key wo ds: Salminus b asiliensis, Lepo inus ob usidens, P ochilodus linea us, Pimelodus pin ado, Gene alized Linea Models.
Faculdade de Biociências, Pon i ícia Uni e sidade Ca ólica do Rio G ande do Sul (PUCRS). A . Ipi anga, 6681, Caixa Pos al 1429, 90619-
900 Po o Aleg e, RS, B azil. [email p o ec ed] (TPA), [email p o ec ed] (NFF)
In oduc ion
Mo e han 15% o he Neo opical ish auna is composed
by mig a o y species (Ca ols eld e al., 2004), which may need
o co e se e al kilome e s o s imula e gonadal ma u a ion
(Godoy, 1987). The wa e shed a ea o main aining mos o
hese species could each 40.000 km2 (Godoy, 1987). Spawning
use o happen ups eam he adul eeding a eas, and s eam
cu en ca ies he eggs and la ae o a eas whe e hey will
de elop. In hese si es, ju eniles will eed and g ow un il
hey become big enough o join he main s ock (Pi che &
Ha , 1982).
Malaba ba (1989) p esen s a lis o species o eshwa e
ish o Pa os Lagoon d ainage, which includes he Jacuí Ri e .
Following Agos inho & Júlio (1999) and Vazzole e al. (1997),
he mig a o y species o Jacuí basin a e: Salminus b asiliensis
S a is ical dis ibu ion models o mig a o y ish
648
espec , p edic ing species occu ence h ough a modeling
app oach based on geog aphical in o ma ion sys em (GIS)
ep esen s a use ul me hodological ool (Pea ce & Fe ie ,
2000b).
Gene alized linea models (GLM), implemen ed wi hin a GIS
a e especially use ul as i can be applied wi h many ypes o
p edic o s (con inuous, bina y, quali a i e, and o dinal), and
he dis ibu ion o he species should comp ise jus p esence
and absence da a (Sya inilia & Tsuyuki, 2008), in o ma ion
a ailable h ough in e iews. These models a e hen applied o
ex apola e he p obabili y o occu ence o species ac oss he
en i e egion o in e es (Pea ce & Fe ie , 2000b).
Al hough se e al ecological ac o s a e known o
in luence ish dis ibu ion pa e ns, access o such da a is
no easy o e en accu a e a la ge a eas. In a la ge egional
scale, he use o simple ac o s de i ed om GIS, as al i ude
and basin a ea, seems o be enough o unde s anding
dis ibu ional pa e ns along he i e s (Ma hews e al., 1992;
Pouilly e al., 2006).
The p esen s udy aims o iden i y he longi udinal
his o ical dis ibu ion pa e n o long dis ance mig a o y ish
in he hyd og aphic basin o he Jacuí Ri e , which al eady
has an ins alled hyd oelec ic p oduc ion o 1.428,674 MW,
and a po en ial, s ill no ins alled one, o 271,004 MW. Apa
om p esen ing dis ibu ion maps o long dis ance mig a o y
ish, p obabilis ic dis ibu ion models o each species will
also be p oposed, using al i ude and basin a ea as main
p esence p edic o s.
Ma e ial and Me hods
Eigh ield expedi ions we e ca ied ou , co e ing he mos
impo an i e s d aining o he Jacuí basin: Taqua i-An as,
Vacacaí/Vacacaí-Mi im, Pa do/Pa dinho, Sinos and Caí, as hei
main ibu a ies. The samplings we e pe o med be ween
Feb ua y and July o 2007. Expedi ions we e ca ied ou making
use o he oad ne wo k and i s c ossing poin s wi h he i e s,
esul ing in 155 di e en sampling poin s (Fig. 1).
A each sampling si e, local ishe men and oldes esiden s
we e sea ched and in e iewed. Wi h he aid o a boa d
con aining a pic u e o in e es species, each in e iewed
pe son was asked abou hei knowledge o he ishes, and
hen, hei p esence o absence in ha speci ic i e segmen
(p esen o his o ical). In case o mo e han one in e iew a
he same sampling si e, only he mos equen in o ma ion o
consensus was egis e ed. The p ojec o alized 204
in e iews, 187 egis e ed as use ul in 155 di e en poin s.
Also, a su ey in he main B azilian ish collec ions was
done by using he SIBIP/NEODAT III da abase (MCP, MNRJ,
MZUSP, UFRGS); besides li e a u e in o ma ion (Pe y &
Schulz, 2006). This su ey was ex ended h ough he analysis
o En i onmen al Impac Repo s (EIA-RIMA) p oduced o
he licensing powe -dams (UHE) in he Jacuí and Taqua i-
An as basins (Dona F ancisca; 14 de Julho, Cas o Al es,
Mon e Cla o).
Da a analysis and ca og aphic p oduc s we e pe o med
(Cu ie , 1816) (dou ado); P ochilodus linea us (Valenciennes,
1836) (g uma ã); Lepo inus ob usidens (Valenciennes, 1836)
(pia a); and Pimelodus pin ado Azpelicue a, Lundbe g &
Lou ei o, 2008 (pin ado). The spo ed ca ish o he Pa os basin
was known un il ecen ly as Pimelodus macula us Lacépede,
1803, a species wi h wide dis ibu ion in neo opics and maybe
a species complex. A e Azpelicue a, Lundbe g & Lou ei o
(2008) he Pa os spo ed ca ish was e iewed and desc ibed
as Pimelodus pin ado, wi h ew in o ma ion conce ning
mig a o y o ep oduc i e biology bu maybe wi h simila
biological pa e ns as Pimelodus macula us.
Ne e heless, including Pimelodus pin ado as a mig a o y
ish species is s ill con o e sial. B aun (2005) sugges s ha
he species seems o ep oduce in li le channels, ibu a ies o
Pa os Lagoon. This in o ma ion is ein o ced by he occu ence
o Pimelodus pin ado in Ba os lagoon, which ecei es wa e
om only small c eeks o i s o de , sugges ing no dependence
o la ge scale mig a ion o ep oduc i e pu pose. Howe e ,
conside ing he impo ance o his species o a isanal ishe y,
Pimelodus pin ado will be included among he compulso y
mig an species o he Jacuí Ri e basin.
A i e is a complex sys em, wi h con inuous changing
s uc u e om ups eam o lowe a eas. The species ichness
inc eases downwa ds (Reyes-Ga ilán e al., 1996) and in
highe al i udes bo h physical and chemical condi ions a e
mo e s ess ul, wi h a lowe numbe o species showing
physiological and ecological capaci ies o su i e in such
places (Ma hews & S y on-Júnio , 1981).
As a gene al ule, he dis ibu ional limi o a species is
d i en by se e al ac o s, which combine o in e ac o c ea e
he pa e ns ound in na u e (Hall e al., 1992). The al i ude,
o example, seems o be b oadly ela ed o hyd odynamics
and mo phology o a i e . Wi h he combined ac o s such
as wa e empe a u e, conduc i i y, pH, cu en speed,
decli i y, and he p esence o pools and wa e alls, al i ude is
capable o c ea ing an en i onmen al g adien ha exe s
in luence on he species ichness in each egion (Pouilly e
al., 2006). A aújo & Teje ina-Ga o (2009), ound ha pH, wa e
eloci y, channel wid h and wa e empe a u e s ongly
a ec ed ish assemblages in uppe Pa aná ibu a ies. Using
geomo phologic da a, Súa ez & Pe e e-Júnio (2007)
sugges ed al i ude as he mos impo an s uc u ing ac o
o ish communi y in he Igua emi Ri e basin. Apa om
al i ude, he wa e olume d ained o a speci ic poin seems
o exe a la ge in luence on he ish dis ibu ion pa e ns
(Ga u i, 1988). When he al i ude dec eases, he e is an
inc ease in he wid h and deep o he i e , allowing he su i al
o bigge ish species (Vázquez e al., 1979).
B azil has app oxima ely 4360 dams, no aking in o accoun
homemade ba ie s and/o non- egis e ed ones (Sil ei a &
C uz, 2005). Cu en ly, 90% o he B azilian elec ic powe
p oduc ion is gene a ed by hyd oelec ic u bines (Pe e e-
Júnio e al., 2002). Hyd oelec ic plan s ha e been in e e ing
in ish mig a ion all a ound he wo ld. Knowing he longi udinal
dis ibu ion o mig a o y ishes is an impo an ool o suppo
decisions ela ed o dams planning in a egional scale. In his
T. P. Al es & N. F. Fon ou a 649
Fig. 1. Sampling si es (n = 155) co e ing he main channel and ibu a ies o he i e s Jacuí, Taqua i-An as, Vacacaí, Vacacaí-
Mi im, Pa do, Pa dinho, Sinos and Caí. Hyd og aphic map gene a ed h ough a digi al ele a ion model (DEM, ada al i ude in
pixel o 92.6 pe 92.6 m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo, 2006).
using Id isi Andes so wa e (Cla ck Labs, 2006) and a digi al
ele a ion model (DEM, ada al i ude in pixel o 92.6 pe 92.6
m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo,
2006).
The hyd og aphic ma ix was ob ained h ough ou basic
s eps: (1) DEM homogeniza ion by applying he Fil e Min
3x3 pixels; (2) de ining pa hways wi h mono onically
dec easing a i udes h ough Pi Remo al ool; (3) by
applying he Runo ool o es ima e he ups eam ca chmen
a ea o each pixel, co ec ing pixel numbe o he eal a ea
(km²) by mul iplying pixel alues by 0.00857476; (4)
con e ing each pixel wi h ca chmen a ea smalle hen 10 o
ze o and bigge hen 10 in o one, and hen c ea ing a
hyd og aphic ma ix wi h wa e shed bigge hen 10 km².
Hyd og aphic ma ix wi h ish p esence, o each species,
was c ea ed h ough Pa hway ool, using eco ded p esence
(GPS coo dina es) as Ta ge Image and he DEM as Cos
Su ace. Mask o da a analysis was ob ained by using also
he Pa hway ool, wi h ups eam sampled coo dina es as
Ta ge Image and he DEM as Cos Su ace.
The occu ence p obabili y o each mig a o y ish species
h oughou he hyd og aphic ma ix was es ima ed using he
S a is ical dis ibu ion models o mig a o y ish
650
Mul inomial Logis ic Reg ession ou ine (Logis ic eg, adjus
by maximum likelihood h ough he New on-Raphson
algo i hm): P (X = 1) = e(b0 + b1 . al i ude + b2 . basin a ea) . (1 + e(b0 + b1 . al i ude
+ b2 . basin a ea))-1, whe e P (X = 1) is he occu ence p obabili y o
he species and b0, b1 and b2 a e he equa ion pa ame e s. The
measu ed a iables, he measu emen scale and da a sou ce
is desc ibed in Table 1.
Da a analysis was pe o med in wo s eps. Fi s , he
p obabili y model o each species was es ima ed using da a
om 175 in e iews om 150 di e en poin s, including
li e a u e, EIA-RIMA and museum da a. A esidual map was
gene a ed showing di e ences among obse ed p esence
(0/1) and p edic ed p esence p obabili y. Sampling poin s wi h
esiduals la ge hen 0.5 we e e isi ed o new in e iews.
Du ing his new ield ip, addi ional 29 in e iews and i e new
poin s we e in oduced in o de o iginal da a ma ix o new
model p ocessing. Adhe ence be ween in o med dis ibu ion
and p esumed dis ibu ion we e es ima ed as he pe cen age o
he i e segmen s wi h p esence/absence in o med as co ec ,
acco ding o he p esumed dis ibu ion model (conside ed as
p esumed p esence p obabili ies highe han 0.5 and as
p esumed absence p obabili ies lowe han 0.5).
Conside ing ha he Logis ic eg unc ion o he Id isi
Andes so wa e does no es ima e he s anda d e o s o he
adjus ed pa ame e s, 10 independen adjus men s we e ca ied
ou o each species h ough andom selec ion o 10% o he
pixels om de sample ma ix. The s anda d e o s we e
calcula ed as he s anda d de ia ion o he di e en es ima es
o each pa ame e , ob aining he signi icance ho ough he
Wald s a is ic, he a io be ween he es ima ed pa ame e and
i s s anda d de ia ion, calcula ing P h ough he Z dis ibu ion
(Tabachnick & Fidell, 1996).
in 58 (37.42%).
Is shown in Fig. 2 he cumula i e equency o in o med
occu ence as a unc ion o al i ude and ca chmen a ea.
Resul s we e simila o all he species bu Pimelodus pin ado.
As iden i ied, P ochilodus linea us, L. ob usidens and S.
b asiliensis ha e an al i udinal h eshold o abou 280-300 m
whe eas Pimelodus pin ado was eco ded in al i udes
exceeding 500 m (one in o med occu ence a 680 m).
Occu ences, conce ning basin a ea, we e eco ded o all
species in i e segmen s wi h ca chmen s o e 20 km².
The es ima ed pa ame e s o he p obabili y model o each
species a e shown in Table 2. Table 3 p esen s he adhe ence
be ween in o med dis ibu ion and p esumed dis ibu ion.
Fo each analyzed species, Figs. 3 o 6 show (a)
hyd og aphic map o he in o med occu ence h ough he
in e iews, as well as he egis e ed occu ence poin s in
li e a u e and museums; (b) p obabili y map o he p esumed
occu ence h ough he adjus ed logis ic model o he whole
hyd og aphic basin wi h a ca chmen a ea la ge han 10 km2;
Fig. 2. In e se cumula i e equency o in o med occu ence
o mig a o y ish o Jacuí basin acco ding o an ups eam
g adien o al i ude (m) and basin ca chmen (km²).
Table 1. Measu ed a iables, measu emen scale and da a
sou ce o p obabili y dis ibu ion models adjus men o
mig a o y ishes a Jacuí basin (B azil).
Va iable Measu emen Scale Da a O igin
P esence (pe specie
o mig a o y ish) Bina y:
p esen (1); absen (0)
P esence in ca alogues
Reco ds in li e a u e
In e iews
Al i ude Me ic (na u al loga i hm
o he al i ude in me e s) Digi al Ele a ion Model
Ca chmen a ea Me ic (na u al loga i hm
o he basin a ea in km2) Digi al Ele a ion Model
Resul s
Sea ch in a ailable in o ma ion (museums, li e a u e, EIA-
RIMA) esul ed in jus i e occu ences o S. b asiliensis, 13
o P ochilodus linea us, 16 o L. ob usidens and 26 o
Pimelodus pin ado. Pimelodus pin ado was p esen in 121
poin s among he 155 sampled (78.06%), P ochilodus linea us
in 72 (46.45%), L. ob usidens in 62 (40.00%) and S. b asiliensis
T. P. Al es & N. F. Fon ou a 651
and (c) dis ibu ion map o esiduals, highligh ing poin s in
which he di e ence be ween in o med occu ence and
es ima ed p obabili y was highe han 0.3 (33%).
Conside ing he e ec s o dams as ba ie s o ish
mig a o y mo emen s, Fig. 7 p esen s he cumula i e
dis ibu ion o la ge and small powe dams o he Jacuí basin
(n = 24, da a om Na ional Elec ic Powe Agency - ANEL)
acco ding o an al i udinal g adien . As could be iden i ied,
a ound 50% o egis e ed dams a e p esen abo e he
al i udinal h eshold o 280-300 m. Figu e 8 p esen s he
geog aphical dis ibu ion o powe dams (yellow spo s) in
he Jacuí basin acco ding o al i udinal ca ego ies o unde
300 m ( ed i e segmen s) and o e 300 m (g een i e
segmen s). Inspec ing he spa ial dis ibu ion, i becomes clea
ha mos o he cons uc ed powe dams a e loca ed abo e
o nea he al i udinal h eshold o 280-300 m o uppe
mig a o y ish dis ibu ion, sugges ing ha mos ope a ing
powe dams p esen s li le e ec as ba ie s o ups eam
mo emen s o mig a o y ish.
Discussion
In he las wen y yea s, he use o mul i a ia e s a is ic in
he modeling o species dis ibu ion has inc eased h ough
he applica ion o a la ge a ie y o echniques. Pa icula ly,
Table 2. A e age, S anda d De ia ion, Wald and signi icance
o he coe icien s (In e cep , b0; al i ude, b1; and basin a ea,
b2) es ima ed om en independen adjus men s o he logis ic
eg ession pe o med using he so wa e Id isi Andes o each
o he ou mig a o y ish species in Jacuí basin (B azil). S.D.
= S anda d De ia ion.
Table 3. Mean accu acy pe cen age (adhe ence) o he
p esumed occu ence model o each one o he ou mig a o y
ish species in Jacuí basin (B azil).
Reg ession Coe icien s
Species In e cep Al i ude (m) Basin A ea (km2)
A e age -2.8762 -1.3028 1.1487
S.D. 0.2597 0.0332 0.0301
Wald 11.0735 39.2022 38.1045
S. b asiliensis
p < 0.0001 < 0.0001 < 0.0001
A e age -0.3176 -1.3067 0.8128
S.D. 0.2731 0.0544 0.0177
Wald 1.1628 23.9999 45.9075
P ochilodus linea us
p 0.2719 < 0.0001 < 0.0001
A e age -0.8364 -1.5564 0.9947
S.D. 0.2213 0.0462 0.0206
Wald 3.7789 33.6909 48.1894
L. ob usidens
p 0.0036 < 0.0001 < 0.0001
A e age -0.9487 -0.8269 0.9255
S.D. 0.3688 0.0496 0.0304
Wald 2.5722 16.6855 30.4644
Pimelodus pin ado
p 0.0278 < 0.0001 < 0.0001
S. b asiliensis P. linea us L. ob usidens P. pin ado
P esence adhe ence (%) 83.2 82.0 82.9 93.3
Absence adhe ence (%) 84.2 81.3 85.0 48.9
he eg ession models ha e been b oadly used o p edic
dis ibu ion, abundance and habi a p e e ence o species
(B i o e al., 1999). Howe e , when his in o ma ion is joined
wi h Geog aphic In o ma ion Sys ems (GIS), i is possible o
inco po a e ecological ac o s di ec ly ob ained om emo e
images o p edic ion models. Also, indi ec g adien s
cons i u ed by a iables ha do no show di ec physiological
ac ion, as al i ude, decli i y and opog aphic posi ion, a e
easie o be ob ained han en i onmen al pa ame e s o di ec
g adien , such empe a u e o pH, which o en in oduce
spa ial unce ain ies due o he lack o da a and in e pola ion
bias (Guisan e al., 1999).
Al hough in he p esen wo k we ha e op ed o model
dis ibu ion based jus in wo pa ame e s o indi ec g adien ,
al i ude and ca chmen a ea, he p edic ed dis ibu ion, when
con on ed wi h he in o med dis ibu ion, always p oduced
an adhe ence a io highe han 80%. The combined e ec o
bo h a iables sugges ha al i ude is he main ac o limi ing
mig a o y ish dis ibu ion o a eas comp ising he Se a Ge al
(al i udinal a eas) whe eas ca chmen a ea limi s dis ibu ion
in he plains o he Cen al Dep ession o Rio G ande do Sul.
This esul sugges s ha e en a ela i ely small se o
geomo phologic a iables a e able o show an impo an
p edic i e powe , maybe due o colinea i y e ec s wi h
associa ed di ec pa ame e , such as wa e empe a u e, i e
dimension (deep, wid h), cu en speed and u bidi y.
Ne e heless, e en conside ing adhe ence o e han 80%
as sa is ac o y, he esul ed e o should no be neglec ed.
Among he ac o s con ibu ing o he obse ed esidues a e:
(1) in o ma ion misma ch; (2) de iciency in de e mining he exac
poin o species disappea ance in o a i e sys em; (3) in luence
o o he en i onmen al ac o s, such as ba ie downs eam;
and (4) he in insic a iabili y ela ed o any na u al phenomena.
Al hough we ha e ied o minimize he p obabili y o
mis aken in o ma ion, h ough a esample p ocess o
disc epan in e iews ( alida ion), his p ocess was limi ed
o i e segmen s in which esiduals (bo h posi i e and
nega i e) we e highe han 0.5 o a leas wo species, due o
he la ge geog aphical a ea unde analysis. Also, new
in e iews esul ed mos ly in o con i ma ion o p e ious da a
and no signi ican change was ob ained.
On he o he hand, he me hodological app oach used in
his wo k, s a is ic ope a ion o images ins ead o di ec nume ic
da a, p esen s ad an ages and disad an ages ha need
close examina ion. The main ad an age consis s in inc easing
he a ailable da a se . Pea ce & Fe ie (2000b) iden i ied a
sha p inc ease in p edic i e pe o mance o GLM as sample
size is inc eased om 50 o 250 si es. By applying con en ional
s a is ic so wa e, each poin is an independen sample, wi h
measu ed a ibu es. In he p esen wo k, we ha e 155 di e en
poin s sp ead in he Jacuí basin. Howe e , by using
Geog aphic In o ma ion Sys em (GIS) and s a is ic ope a ion
o images, addi ional in o ma ion could be inco po a ed. When
he p esence o a species is egis e ed in adjacen poin s o a
i e , i is easonable o suppose ha he species do also
occu s along he whole segmen be ween hese wo poin s.
S a is ical dis ibu ion models o mig a o y ish
652
Fig. 3. Salminus b asiliensis. (a) Hyd og aphic map o p esumed occu ence (blue) h ough in e iews (yellow spo s),
li e a u e and museums (pu ple spo s) o e sampled hyd og aphy ( ed). (b) P obabili y map o he p esumed occu ence
h ough he adjus ed logis ic model o he whole hyd og aphic basin wi h a ca chmen a ea highe han 10 km2. (c) Dis ibu ion
map o esiduals applied o sampled hyd og aphic basin. Red segmen s ep esen nega i e esiduals less hen -0.34, indica ing
p edic ed p esence bu wi h no egis e ed occu ence. Yellow segmen s ep esen posi i e esiduals g ea e hen 0.34, indica ing
egis e ed occu ence do no p edic ed by he model. Hyd og aphic maps gene a ed h ough a digi al ele a ion model (DEM,
ada al i ude in pixel o 92.6 pe 92.6 m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo, 2006).
T. P. Al es & N. F. Fon ou a 653
Fig. 4. P ochilodus linea us. (a) Hyd og aphic map o p esumed occu ence (blue) h ough in e iews (yellow spo s), li e a u e
and museums (pu ple spo s) o e sampled hyd og aphy ( ed). (b) P obabili y map o he p esumed occu ence h ough he
adjus ed logis ic model o he whole hyd og aphic basin wi h a ca chmen a ea highe han 10 km2. (c) Dis ibu ion map o
esiduals applied o sampled hyd og aphic basin. Red segmen s ep esen nega i e esiduals less hen -0.34, indica ing
p edic ed p esence bu wi h no egis e ed occu ence. Yellow segmen s ep esen posi i e esiduals g ea e hen 0.34, indica ing
egis e ed occu ence do no p edic ed by he model. Hyd og aphic maps gene a ed h ough a digi al ele a ion model (DEM,
ada al i ude in pixel o 92.6 pe 92.6 m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo, 2006).
S a is ical dis ibu ion models o mig a o y ish
654
Fig. 5. Lepo inus ob usidens. (a) Hyd og aphic map o p esumed occu ence (blue) h ough in e iews (yellow spo s),
li e a u e and museums (pu ple spo s) o e sampled hyd og aphy ( ed). (b) P obabili y map o he p esumed occu ence
h ough he adjus ed logis ic model o he whole hyd og aphic basin wi h a ca chmen a ea highe han 10 km2. (c) Dis ibu ion
map o esiduals applied o sampled hyd og aphic basin. Red segmen s ep esen nega i e esiduals less hen -0.34, indica ing
p edic ed p esence bu wi h no egis e ed occu ence. Yellow segmen s ep esen posi i e esiduals g ea e hen 0.34, indica ing
egis e ed occu ence do no p edic ed by he model. Hyd og aphic maps gene a ed h ough a digi al ele a ion model (DEM,
ada al i ude in pixel o 92.6 pe 92.6 m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo, 2006).
T. P. Al es & N. F. Fon ou a 655
Fig. 6. Pimelodus pin ado. (a) Hyd og aphic map o p esumed occu ence (blue) h ough in e iews (yellow spo s), li e a u e
and museums (pu ple spo s) o e sampled hyd og aphy ( ed). (b) P obabili y map o he p esumed occu ence h ough he
adjus ed logis ic model o he whole hyd og aphic basin wi h a ca chmen a ea highe han 10 km2. (c) Dis ibu ion map o
esiduals applied o sampled hyd og aphic basin. Red segmen s ep esen nega i e esiduals less hen -0.34, indica ing
p edic ed p esence bu wi h no egis e ed occu ence. Yellow segmen s ep esen posi i e esiduals g ea e hen 0.34, indica ing
egis e ed occu ence do no p edic ed by he model. Hyd og aphic maps gene a ed h ough a digi al ele a ion model (DEM,
ada al i ude in pixel o 92.6 pe 92.6 m) o he B azilian o icial e e ence sys em (SAD 69; Labgeo, 2006).
