Intraspecific colour preference in mate choice by female Apistogramma cacatuoides Hoedeman, 1951 (Teleostei: Perciformes: Cichlidae)

In mate-choice experiments females of the Neotropical cichlid Apistogramma cacatuoides unerringly differentiate between males of different populations and a red aquarium strain. Females use male colour traits to choose their mates. The preference of wild females for red-coloured males varies from 3 % to 69 % dependent on geographical origin, whereas aquarium-strain females prefer their own bright-coloured males in about 90 % of cases. The preferences of wild females show signifi cant differences depending on their origin and following a geographical gradient. The high degree of stability in assortative mating behaviour in female A. cacatuoides may indicate incipient speciation processes.

as well as domestic strains of variable colour are available.Hohl (2010a & b) presents background data on the origin of various aquarium-hobby colour morphs of A. cacatuoides.
In mate-choice experiments wild-caught females from the río Algodón (150 km to the north of Iquitos) prefer inconspicuous males of their own morph with high statistical confi dence instead of conspicuous males of a red-coloured aquarium strain (Römer & Beisenherz, 2005).This contradicts the widespread assumption of a general preference for mates with exaggerated traits as signals (Darwin, 1871;Bischoff et al., 1985;Bradbury & Andersson, 1987;Andersson, 1994;Houde, 1997).On the other hand, phylogenetic studies of different animal taxa indicate that secondary sexual characteristics can be lost due to environmental, social, and random factors (reviewed by Wiens, 2001).Hence mate-choice preference for conspicuous traits in different fi sh species may decrease (Forsgren, 1992;Endler & Houde, 1995;Godin & Briggs, 1996), reverse (Johnson & Basolo, 2003), or even disappear (Kingston et al., 2003) so that mating happens at random (Berglund, 1993;Breden & Stoner, 1987) and thus deviate from theoretical expectations.
The availability of wild specimens of different origin and coloration as well as of domestic strains offers an opportunity to monitor the way in which female A. cacatuoides choose their mates.
The importance of colour in communication within and between the sexes of Apistogramma species has been demonstrated by Römer (2000) and Rodrigues et al. (2009).Beisenherz & Römer (2005) provide data supporting the infl uence of colour in inter-and intraspecifi c mate choice.Ready et al. (2006) examined the specifi c status of allopatric Apistogramma populations with a similar habitus but differences in coloration.
Allopatric populations may practise assortative mating by colour and thus may have to be seen as different distinct biological species (Ready et al., 2006).We will test on A. cacatuoides whether a preference for inconspicuously-coloured males confi rms this general concept or is simply a matter of geographical variation.

Materials and methods
The dwarf cichlid Apistogramma cacatuoides was used for our standardised dichotomous mate-choice experiments.A detailed description of the cichlid genus Apistogramma and the investigated species A. cacatuoides has been given elsewhere (Kullander, 1980;Römer, 2000Römer, , 2006)).
Females and males of A. cacatuoides are sexually dimorphic and dichromatic.Adult males are signifi cantly larger than females.Additionally they have longer lappets in the dorsal fi n and a lyrate caudal fi n (Fig. 1).Courting and breeding female Apistogramma show a yellowish body with contrasting black markings (Fig. 2).
The diagrams show the correlation between female preference for red-coloured males (red) and the proportion of wild males with red-coloured ornamentation on the caudal fi n.
A. cacatuoides prefer richly structured areas with scattered driftwood and branches and a thick leaf layer where they can hide and establish their breeding territories (see Römer, 2000).Females of most Apistogramma species attach their eggs to the upper interior surface of cave-like shelters and perform most of the brood care (for further information see Römer, 2000).
In experiments for intraspecifi c mate choice, wildcaught males and their fi rst and second generation (F 1 , F 2 ) descendants from random spawns with wild-caught females were tested against males from a red-coloured aquarium strain (fi g. 3).Males of this captive strain are much more colourful than wild males and exhibit large red or yellow dots or speckles on their dorsal and caudal fi ns (Fig. in Römer, 2000, p. 118).We used wild-stock A. cacatuoides from different importations and locations, obtained by commercial collectors from four different collecting sites.Wild stock I originated from the río Algodón (150 km to the north of Iquitos) (Römer & Beisenherz, 2005).Two further wild stocks were collected near Iquitos: wild stock II derived from the río Itaya 20 km northwest of Iquitos, and wild stock III from the río Nanay 3.6 km south -west of Iquitos; specimens of wild stock IV were captured in small tributaries of the río Huallaga about 500 km west of Iquitos.
All fi sh were kept in the laboratory in water with matching parameters.Temperature, pH, and water hard ness were measured electronically.The water was puri fi ed by fi ltration and regular weekly water changes.All fi sh were fed ad libitum with Artemia nauplii.
In mate-choice experiments females of A. cacatuoides were offered two males as potential reproductive partners in experimental tanks (60 × 30 × 30 cm) which were divided into three unequal subunits by two separating glass dividers siliconed in at the left-and right-hand ends (compare Beisenherz et al., 2006;Römer & Beisenherz, 2005).Any infl uence of scent on fi sh behaviour during the experiments could thus be excluded.Our experimental set-ups also excluded acoustic information to a certain extent, as permanent external and internal noise was generated by air blowers and bubbling water.We therefore postulate that fi lms, without lid) were placed in the female's section of each experimental tank to act as caves, one close to each of the lateral compartments housing the males.
To exclude habituation or general preference by females for a particular end of a tank during the experiments, males of wild stocks and the domestic strain were randomly placed into the left-or right-hand compartments of the experimental tanks.To exclude size acoustic information, which may be of importance in mate choice in cichlid fi shes (Slabbekoorn et al., 2010), probably had no decisive infl uence on the experimental results.One female was placed in the larger, central compartment of each aquarium (27 litres, 30 × 30 × 30 cm), and a male in each of the smaller lateral compartments (13.5 litres, 15 × 30 × 30 cm).Two plastic canisters (the type used for photographic  red aquarium strain.In addition, 74 females of the red aquarium strain were tested using males of their own strain and wild stock I.
Differences in preferences between wild-stock and aquarium-stock females, as well as between the wild populations, were tested using Chi-Square and Fisher´s exact tests.Repeated female choice was analysed by McNemar´s Chi-Square for dependent samples.
The frequency of colour ornamentation in male fi nnage in each wild stock was noted to permit checking for correlation with preferences for red-coloured aquarium-strain males.The coeffi cient of determination (R 2 ) and Spearman's rank correlation coeffi cient were used to estimate probability of future results.All analyses were processed using the program STATISTICA Version 6.0 (StatSoft, Tulsa, USA).

Results
Different preferences for red-versus inconspicuously-coloured males were detected in our experiments (Fig 4,Tab. 1).Whereas 97 % of females of wild population I preferred inconspicuously-coloured males of preferences (here taken as total length) as a choicemodifying factor, only males of similar size were offered to a female.Females were randomly placed in the larger, central compartments of the experimental aquaria at least one day after males had been placed in the smaller lateral compartments.
Displaying males were visible to females in the adjacent tank compartments.Females chose a male by spawning with him; this provides irrefutable evidence of their mate-choice behaviour (for detailed description see Beisenherz & Römer (2006)).
During the experiments all but two females laid their eggs in one of the "caves" next to the glass pane close to one of the displaying males.Therefore a choice was considered made as soon as a female had spawned close to one of the available males.The two exceptions split their clutches between the two caves adjacent to each of the available males.
After spawning, females were transferred from the experimental tanks to other aquaria where they were kept for several days before being retested with different males.
67 females of wild stock I, 44 of wild stock II, 34 of wild stock III, and 13 of wild stock IV were tested in mate-choice experiments.The females chose between males of their own population and males of the their own population, about 90 % of red aquariumstock females chose males of their own strain (see also Römer & Beisenherz, 2005).Females of wild populations II, III, and IV showed a preference for redcoloured males in about 33 %, 51 %, and 69 % cases, respectively (Tab.1).
Individual repeatability of female colour choice was measured to exclude random mate choices.Preference for inconspicuous wild or red-coloured domestic-strain males was repeated in about 84 % of cases in wild stock II, and about 86 % in wild stock III (Tab.2).This indicates high but not entirely consistent mate-choice behaviour.
In both populations preference for own inconspicuous males was reduced during the repeat experiments, but the differences were not signifi cant (Tab.2); how-  Wild females of different geographical populations showed different degrees of preference for conspicuous red-coloured males; however, only a few wild males in the populations we used in our experiments exhibited red fi n ornamentation.The results of experiments with two A. cacatuoides populations were consistent on repetition.
While females of wild population I almost exclusively chose their own inconspicuous males, those of populations II, III, and IV showed higher degrees of preference for red-coloured males.
The sexes have different investments in reproductive activities.Females usually invest more energy in direct reproduction than do males, as, for example, production of eggs is generally more costly than production of sperm (Andersson, 1994;Jennions & Petrie, 1997;Pomiankowski, 1987).Thus females need to be more selective in their choice of a potential mate.They probably use secondary sexual traits indicating the genetic quality of males, for example colour, size, etc. (Kodric-Brown & Brown, 1984;Andersson, 1994;Jennions & Petrie, 1997;Brooks & Endler, 2001;Stein & Uy, 2006).
Furthermore, the criteria for mate choice may be dependent on geographically determined differences in the environment, which may also infl uence the direction and intensity of sexual selection (Endler & Houde, 1995;Houde, 1997;Jennions & Petrie, 1997).Thus the different degrees of preference for red-coloured males between populations tested in our experiments may be explained by differences in selecting ecofactors, determined by geographical distance.
Different studies indicate that sexually-selected secondary sexual characteristics can be lost (reviewed in Wiens, 2001).In many animal taxa there is a discernible discrepancy between the frequency of female preferences and conspicuous male traits.This indicates that environmental, sociobiological, and random factors may infl uence the appearance of sexually-selected secondary sexual characteristics differently in females and males (reviewed in Wiens, 2001).ever, this may simply refl ect the comparatively small data set.
Females of the two populations tested increased their preference for red-coloured males by a corresponding amount.In these choice-repetition experiments the preference for red males increased to about 57 % in females of population II, and was about the same as previously at 58 % in females of population III.
There was weak (but not signifi cant) correlation between female choice of brightly-coloured males and frequency of red ornamentation on the caudal fi n of own-population males (Fig. 5).

Discussion
In our experiments female A. cacatuoides chose their mates on the basis of shape and colour pattern through a glass pane acting as a physical barrier.Thus other signals than visual for mate choice were not available to the females.They laid their eggs next to the preferred male.This experimental design excludes direct male interaction -especially aggression -which may  on closer examination may turn out to be different genotypic clusters or biological species (Ready et al., 2006).
Earlier experiments by Römer & Beisenherz (2005) have shown that female Apistogramma of different species are able clearly to distinguish between conspecifi c and heterospecifi c mates.Interestingly they make mistakes regarding potential mates only when potential mates are males of allopatric species.This indicates that mate choice by females may be the key factor in preventing hybridisation and for stabilising genotypic clusters in the wild (Kocher, 2004;Stelkens et al., 2008;van der Sluijs et al., 2008a), even if hybridisation has been shown to be a possible mechanism triggering speciation in some African cichlids (Egger et al., 2007;Rüber et al., 2001;Salzburger et al., 2002;Schilthuizen et al., 2004;Seehausen, 2004;Seehausen et al., 2007;Smith et al., 2003;Stauffer et al., 1996;Stelkens et al., 2009;van der Sluijs et al., 2008b).In our case the clear geographical cline seen in the distribution of males with coloured fi n ornamentation and in female assortative mating may indicate the beginning of speciation in Apistogramma cacatuoides.
Depending on morphological and/or environmental factors female mate choice may lead to alternative reproductive strategies (Alonzo & Sinervo, 2001;Barlow, 1973Barlow, , 2000;;Brooks & Endler, 2001;Hurtado-Gonzales et al., 2010;Kingston et al., 2003;Rios-Cardenas et al., 2007).For instance, heavy predation on conspicuous males may reduce female preference for conspicuous males in a given population.If susceptibility to visual predators reduces the advantages of mating with conspicuous males, this may indirectly lead to reduction of conspicuousness in males (Houde & Endler, 1990).Alternative mating strategies may be triggered by this type of interaction between female mating preferences and susceptibility to predation within populations (Bond & Kamil, 2006;Gamble et al., 2003;Hurtado-Gonzales et al., 2010).A secondary effect of this interaction may be the stable occurrence of certain secondary sexual traits and colour morphs in given fi sh populations (Hurtado-Gonzales et al., 2010).Hurtado-Gonzales et al. (2010) and Lindholm et al. (2004) found the same frequency patterns of such traits and morphs across the geographical distribution of the cyprinodont fi sh species Poecilia parae.
The situation is completely different in the polychromatic cichlid species Apistogramma cacatuoides: in this species the occurrence of certain colour morphs, especially that involving red blotches in the caudal and the soft portion of the dorsal fi n, apparently follows a geographical cline (Fig. 5).In general there is a lack of systematically collected fi eld data on variation in male coloration in Apistogramma populations (Römer, 2000;Römer & Beisenherz, 2005;Hohl, 2010), but we have taken data relating to our test populations from numerous imported specimens and have interviewed Peruvian and US-American collectors (cf.Römer & Beisenherz, 2005).
Relating this data to female choice in our experimental fi sh, differences in preference apparently follow a geographical cline (Fig. 5).Nevertheless, the correlation between female choice and occurrence of fi n ornamentation in wild populations is still weak owing to the small sample size.
Mate choice based on colour is important in cichlid fi sh evolution.Many cichlid radiations, at least in part, involve the evolution through mate choice of secondary sexual characteristics based on colour (Barlow, 2000;Dominey, 1984), in other words female colour preference is responsible for the development and maintenance of colour morphs through mate choice (Salzburger et al., 2006).
Various species of the genus Apistogramma are widespread throughout the entire neotropical region east of the Andes and are often found sympatric (Kullander, 1980(Kullander, , 1986;;Römer, 2000;2006).Many Apisto gramma species exist in various colour morphs, and

Fig. 1 .
Fig. 1.Apistogramma cacatuoides: typical adult wild male without any red markings on the unpaired fi ns, collected in the vicinity of Iquitós, Peru.

Fig. 3 .
Fig. 3. Apistogramma cacatuoides: typical adult male of the red aquarium strain used in the experiments, with clearly expressed red dots and ocelli in the unpaired fi ns.
behaviour (cf.Beisenherz & Römer (2006) andReady et al. (2006)).Our results suggest that different strains and populations of A. cacatuoides have different preferences in relation to male colour.

Fig. 5 .
Fig. 5. Origin of Apistogramma cacatuoides populations I -IV in the Peruvian Amazon basin (Loreto): red dot: population I; green dot: population II; blue dot: population III; yellow dot: population IV.Red: amount of females choosing outgroup male; light blue: amount of males showing fi n ornamentation.
Tab. 1. Intraspecifi c mate-choice experiments in female Apistogramma cacatuoides.Results for wild populations I -IV and a red aquarium strain (V).Signifi cant differences between female preferences analysed by Chi 2 (with Yate's correction) and Fisher's exact test (if frequency of choices lower than 6) marked *) (= signifi cant) or **) (= not signifi cant).
Tab. 2. Repeatability of female mate choice in Apistogramma cacatuoides of wild populations II and III.Differences were analysed by Chi 2 test (with Yate's correction) and Fisher's exact test (if frequency of choices lower than 6).Comparison of repeatability of choices between population II and III shows no difference (p = 0.526) (Chi 2 and Fisher's exact test).Analyses of fi rst and repeated choices by McNemar's Chi 2 marked *) (= signifi cant difference) or **) (= no signifi cance).Fig.4.Intraspecifi c colour preference of female Apistogramma cacatuoides.Results of female mate choice in laboratory experiments using females of four wild populations (I -IV) and an aquarium strain (V).Differences between groups II and III and between groups III and IV are not signifi cant (Chi 2 -test and Fisher's exact test).