DNA barcoding LT cichlids of the rocky shores

Breman FC1, Van Steenberge M2, Jordaens K1 & Snoeks J2

1 Royal Museum for Central Africa 2 Royal Museum for Central Africa and Catholic University of Leuven

DNA barcoding of LT cichlids of the rocky shores

• Introduction to the topic • Methods

– Collection – Specimens – Methods

• Specimens • Data and library setup • Sequence based • Species based

– Analysis • Sequence based identification • Species based identification

– Results • Sequence based • Species based • Species complexes • Example of a complex and a newly described species

– Testing OTU´s

• Discussion • Questions?

Tropheus duboisi

Introduction

• Fish DNA barcoding successful so far

• Most species can be identified

• Different methods can be used

Introduction

• Habitat – rocky shores of Lake

Tanganyika – Alkaline environment

(average pH 8,4) – Highly diverse habitats – Highly specialized

• Breeding strategies – Mouth brooders – Shell brooders – Substrate brooders

• Feeding strategies – Predation – Algae scraping

Introduction

• 200+ recognized species of cichlids

– Dozens remain to be described

– >95% endemic

• Model organisms for evolution and speciation

• Economic importance with aquarium enthusiasts

• 75 non cichlid fishes are also present in LT

Lake Tanganyika

Collection sites

• 676 km from N-S and average 50 km across

• Average depth 570m (max 1470m)

• 4 countries

– Tanzania

– Congo

– Burundi

– Zambia

Collection sites

• 15 sites

• 3 expeditions (1992, 1995, 2010)

• 1000s of specimens

Lepdiolamprologus elongatus

Tropheus brichardi

Lobochilotus labiatus

Gnatohochromis pfefferi

Methods Specimens

• Covering 11 tribii and 37 genera

• Library A) 78 (98 with singletons) OTUs

• Library B) 70 (91 with singletons) OTUs

• Library C) 52 (66 with singletons) OTUs (11 complexes)

Methods Data setup

• Three groups

– A) all taxonomic, behavioural and distributional knowledge (published and unpublished) were used to assign a name to a specimen

– B) only currently recognized species used for assigning a name to a specimen

– C) groups with known difficulties in evolutionary history (hybridization, incomplete lineage sorting) and taxonomy are grouped in species clusters

Methods Sequence based

– BM/BCM method

– Software compares each sequence to all the others and the chosen threshold for the dataset

– Returns a statement on the sequence/specimen with regard to threshold and presence of same/related species

– Returns a success percentage in terms of sequences

– Influenced by dataset properties

Methods Species based

– Uses the sequences assigned in the sequence based method, but now classified according to threshold per species

– Species identifiable or not

– Returns success percentage in terms of species

Analysis Species based BM/BCM

• Categories

– True Negative (TN) Best match above threshold and allospecific

– True Positive (TP) Best match is same species below threshold

– False Positive (FP) Closest match is below threshold but is an allospecific

– False Negative (FN) Intraspecific distances are above threshold and conspecific

Threshold

• Fixed or based on dataset, in our case threshold was determined via an R script

• Obtained via two curves of intra and inter specific distances

• The optimum is then chosen as threshold

FN FP

Analysis Species based using a NJ tree

• Using a NJ tree

• Counting species in distinct clusters

Results Sequence based

(A)morphospecies (B)accepted names (C) species complexes

optimal threshold 1.91% 2.37% 2.20%

Sequences: 398 398 398

Correct id's according to "Best Match": 276 (69.34%) 320 (80.4%) 352 (88.44%)

Ambiguous according to "Best Match": 80 (20.1%) 36 (9.04%) 21 (5.27%)

Incorrect id's according to "Best Match": 42 (10.55%) 42 (10.55%) 25 (6.28%)

Correct id's according to "Best Close Match": 274 (68.84%) 318 (79.89%) 349 (87.68%)

Ambiguous according to "Best Close Match": 80 (20.1%) 36 (9.04%) 21 (5.27%)

Incorrect id's according to "Best Close Match": 31 (7.78%) 34 (8.54%) 17 (4.27%)

Sequences without any match closer than threshold 13 (3.26%) 10 (2.51%) 11 (2.76%)

Results Species based

Library # species TP FP TN FN percentage ID

success A 78 55 14 5 4 0.71 B 70 46 21 2 1 0.59 C 52 40 3 6 3 0.51

Relative ID error Precision Overall ID error Accuracy A 0.20 0.80 0.23 0.77

B 0.31 0.69 0.31 0.69 C 0.07 0.93 0.12 0.88

% success NJ

0.74 0.74 0.81

-Marginal improvement when species complexes in success %, -but an increase in accuracy and precision

Results Species complexes

Results Species complex and new species

• Examples of species complexes with unresolved clusters

• Eretmodus cyanostictus specimens have recently been described as Eretmodus marksmithi

Results • Identification of putative new species with

DNA barcodes

putative species based on expert opinion

# sequences in dataset

intrasp. dist. < intersp dist. to nearest other species (y/n)

dist. to nearest other species

average generic distance +/- se

putative species y/n (difference > generic average +/- se)

Chalinochromis spbifrenatus 1 NA 2.33 1.84 +/- 0.42 y Ectodus cfdescampsi 3 y 1.7 1.45 +/- 0.33 n Neolamprologus cfpetricola 1 NA 0.61 5.70 +/- 0.55 n Neolamprologus speseki 3 y 1.07 5.70 +/- 0.55 n Petrochromis cfmacrognathus 2 y 0.15 2.45 +/- 0.36 n Petrochromis spephippiumsouth 2 y 1.39 2.45 +/- 0.36 n Petrochromis sppolyodonelongate 3 y 0.76 2.45 +/- 0.36 n Petrochromis sppolyodonhigh 2 n 0.15 2.45 +/- 0.36 n Tropheus cfannectens 8 y 1.54 1.55 +/- 0.26 n Tropheus ikola 3 n 0 1.55 +/- 0.26 n Tropheus mpimbwe 51 n 0 1.55 +/- 0.26 n

Discussion

• LT cichlids well studied

– Genetically and morphologically

• They are examples of ongoing speciation

– mtDNA evolution is slower than morphological evolution

• Hybridisation is common

Discussion

• Incomplete taxonomy • Unbalanced dataset, influence of sequence

composition • Success percentages not high compared to other

fish groups, however still ok for a complex group such as this.

• Distance method not very useful for detecting potential new species

• Single sequences cannot be evaluated with either method

Questions?

Thank you for your attention

感谢您的关注

Foto credits: Royal Museum for Central Africa Maarten Van Steenberge

Dimitri Geelhand de Merxem