Nothobranchius fishes require very special habitat conditions. They cannot survive through more than one generation in pools that do not have the necessary habitat conditions, primarily an appropriate substrate for egg survival and development. Perhaps the most comprehensive account on habitat requirements of Nothobranchius is by Watters [⁸] who, based on field observations, experimental data and published studies, discusses the conditions that prevail in Nothobranchius habitats and the factors that control their distribution.

Nothobranchius fishes have been present in East Africa for many million years and during that time, geomorphological evolution of the landscape, with accompanying changes in drainage patterns, have been very effective in spreading Nothobranchius species to all sites that have the conditions necessary for their survival over the long term [⁸,⁹]. In other words, if Nothobranchius are capable of existing as a viable population in any particular seasonal pool then natural processes have ensured that they are already there. Brian Watters examined more than a thousand Nothobranchius habitats [¹⁰,¹¹], as well as pools that do not host Nothobranchius fishes. In all cases, the reasons for the presence or absence of Nothobranchius were primarily determined by the nature of the substrate. Without the particular type of substrate (alluvial vertisols) no eggs can survive the dry season.

During the course of an extensive project in the Kruger National Park (KNP) in South Africa, numerous sites in the northern part of the park, to which Nothobranchius had been translocated, were examined [¹⁰,¹¹]. Two species of Nothobranchius (N. rachovii, N. orthonotus) occur naturally at only two localities in the park and, for conservation purposes, a decision was made in the mid-1970 s to translocate specimens of these two species to various seasonal pools in a different part of the park. These efforts were carried out during the period 1975-1985 and involved 10 different sites and 15 translocation events. The translocations involved large numbers of fish that were transported at considerable expense (by helicopter in many cases). None of these translocations were successful over even a relatively short term and only in a few cases did the fish survive through a full seasonal cycle. The investigations showed conclusively that the failure of these translocations was due primarily to an unsuitable substrate at the translocation sites that could not sustain the viability of eggs deposited therein by the fish. The only reason that in a few rare cases the translocated populations were able to survive through a seasonal cycle was because the translocations occurred during a particularly wet period. When normal conditions returned the populations became extinct [¹⁰].

Further, indirect evidence for the low success of Nothobranchius translocation comes from the original translocations by Vanderplank [¹²] who initiated introductions of N. taeniopygus from the vicinity of Old Shinyanga to other regions in Tanzania (Dodoma area), Uganda, Kenya and Swaziland. Numerous sampling efforts have been made in many of these areas over decades since these events, but no evidence for the survival of the introduced N. taeniopygus has ever been found [¹³].

The most important factor that makes the translocation of Nothobranchius fishes to new habitats pointless is that any seasonal pool within the overall range of distribution of Nothobranchius that is capable of sustaining a population of these fishes will already host one or more species of Nothobranchius. It makes no sense to introduce a new, non-native species (or population) into a pool that already hosts other species/populations of the same genus. Research carried out on the factors that control the distribution of Nothobranchius species over time [⁹,¹⁴] provides evidence for this as well as the vast field observational experience of the authors who regularly conduct field studies of Nothobranchius fishes in the wild.

Tanzania, a potential target of suggested introductions [⁷], is rich in Nothobranchius fishes and they are present in almost every temporary body of water with suitable environmental conditions [¹⁵,¹⁶]. Unfortunately, it appears that Matias and Adrias have not made a serious effort to research this and to determine what the natural distribution and population density of these fishes are before suggesting a scheme that will have no greater effect on the mosquito population than existing populations of Nothobranchius already have.

In summary, introductions to sites presently lacking Nothobranchius fishes will very likely have the same negative results (i.e. no population establishment in new habitats) as previous attempts.

Most Nothobranchius species readily feed on mosquito larvae, including larval Culex spp. [¹⁵] and mosquito larvae are used as the most common diet for captive Nothobranchius [¹⁶]. This finding has been quantitatively confirmed by Matias and Adrias [¹]. Studies on the diet of Nothobranchius in the wild are rare and Matias and Adrias cite a work which found undetermined mosquito larvae as main food item for a Nothobranchius species from Somalia. We suppose that the exact reference in the manuscript is incorrect as it suggests that the statement refers to findings from an experimental study concerning the resistance of egg chorion to external chemical damage [¹⁷], and we deduce that this information comes from a WHO (World Health Organization) report by Wildekamp [⁴] on N. microlepis or previous WHO assignment reports [¹⁸,¹⁹] describing the diet of Nothobranchius patrizii (then determined as N. palmqvisti). However, Matias and Adrias selectively used only a part of the study that supports their results. The report by Wildekamp [⁴] indeed describes the frequent presence of mosquito larvae in N. microlepis stomachs. This report also contains data showing that N. microlepis had a clear preference for small planktonic crustaceans and their nauplii. The consumed mosquito larvae, mentioned in the report [⁴] were taken during experimental trials when mosquito larvae were the only food offered. Nothobranchius microlepis may be ecologically distinct from N. guentheri used by Matias and Adrias since it was discovered later that N. microlepis have specially adapted gill rakers to prey on small food items [²⁰]. Mosquito larvae were frequently found in the diet of N. cyaneus (= N. jubbi) but they also frequently preyed on larger sized planktonic crustaceans and Coryxa nymphs (Hemiptera), and the conclusion that Nothobranchius have a preference for mosquito larvae is certainly not supported by [⁴] despite showing that mosquito larvae were frequently consumed (together with some other prey types) by some species.

Another recent study examined the diet of three sympatric species of Nothobranchius in southern Mozambique and found that mosquito larvae were extremely rare items in the diet of all three Nothobranchius species studied [²¹], while crustaceans (in N. furzeri and N. rachovii, two species with body size and morphology comparable to N. guentheri) and coarse insect larvae such as Odonata, Ephemeroptera and Coleoptera (in N. orthonotus, a larger species) were the primary prey. We acknowledge the possibility that mosquito larvae may have already been eradicated if they constituted the preferred food and hence were not encountered during diet analysis, but the conservative interpretation is that mosquito larvae may not be as important component of the Nothobranchius diet in the wild, especially considering that the natural habitats of annual fishes and mosquito larvae are not identical.

Yet another pitfall with regard to the ability of Nothobranchius to eradicate mosquito larvae in the wild is that very substantial numbers of larvae would survive in the wide margins of natural pools overgrown by thick grass vegetation [¹⁴] in water too shallow for Nothobranchius fishes to reach. Mosquito larvae tend to be concentrated in the marginal zones of water bodies [²²] and anopheline larvae typically reside among vegetation with limited motion and hence do not attract visual predators (such as Nothobranchius) [²³]. This sedentary behaviour is in contrast to active swimming by culicine larvae used in the experiments by Matias and Adrias [¹]. Even if Nothobranchius fishes were able to eradicate mosquito larvae in temporary habitats in which they are capable of surviving, their special habitat requirements prevent them from inhabiting many typical habitats of mosquito larvae, such as almost any type of receptacle capable of holding even very small amounts of rainwater (including waste items such as bottles, cans and any water containers in households) or cattle hoofprints (very typical feature of African countryside with a human settlement). These small bodies of water are significant habitats for mosquito larvae, but are impossible to stock with predators because of their abundance and small size [²⁴]. This points to the serious logistical impracticality of the proposed scheme.

Countries of sub-Saharan Africa are richly endowed with many species of Nothobranchius and related genera of annual fishes. For example, the coastal region of Tanzania alone includes more than 20 species of Nothobranchius. The assertion that N. guentheri is native to Tanzania [¹] is correct but very imprecise, since that species is endemic to Zanzibar Island and does not occur on the African mainland [¹⁵]. As pointed out by Matias and Adrias [¹], there are indeed localities with more than a single Nothobranchius species [⁴,¹⁴,¹⁵], but this does not diminish the potential threat of hybridization between indigenous and introduced species. There are numerous accounts of introgressive hybridization between native and introduced congeneric species in many organisms [²⁵], including killifishes [²⁶]. For example, hybridization between a non-native species of pupfish (Cyprinodon variegatus) and its native congener not only resulted in extinction of the native species via hybridization but also facilitated a further expansion of non-native pupfish and its hybrid swarm [²⁷]. Within the genus Nothobranchius alone, Reichard and Polačik [²⁸] experimentally showed that there were almost no reproductive isolating barriers between an island and mainland populations of N. korthausae despite their significant morphological and genetic differentiation leading some authors to assign the mainland populations as a separate species [²⁹]. The risk of hybridization is especially high in taxa where sexual selection plays an important role in reproductive success [³⁰] as is the case in Nothobranchius with brightly coloured males [³¹]. Native species (or populations) may show a preference for partners from non-native populations/species [³²,³³], which promotes rapid extinction via extensive hybridization [³⁴].

The second potential threat to native Nothobranchius species arises from interspecific competition [²⁵]. The ecological niche of most Nothobranchius fishes is similar [¹⁵,¹⁶,²¹] and, at present, there is serious lack of information on how Nothobranchius communities (presence/absence of species and their relative abundance) are shaped. Despite the fact that seemingly ecologically similar species may co-occur syntopically [¹⁴], in most habitats with multiple sympatric Nothobranchius species, the species present do appear to be ecologically separated [¹⁵].

Finally, there is an important risk of introducing diseases [³⁵]. Introduced fish may transmit diseases contracted during captive breeding into the natural environment, with devastating consequences on natural populations [³⁶]. Infections of serious pathogens such as the microporidian Glugea sp. are common in Nothobranchius cultures [³⁷].

In summary, we believe that transport of Nothobranchius fish and eggs outside their current range is irresponsible and may have significant effects on the distribution of native congeners as well as other organisms in aquatic communities.

We believe that the Matias and Adrias initiative [¹] is driven by the necessity to develop a new approach to combat infectious insect-borne diseases. The authors, both affiliated with the Poseidon Science Foundation, a non-profit branch of Poseidon Sciences biotechnological company with a large portfolio of insect (mosquito) control products [³⁸], stated that there were no competing interests and hence we assume that the study constitutes an attempt to generously contribute to a mosquito control programme without any prospect of commercial or other interests. We welcome this attempt, but urge the authors to adhere to the mission targets of the Poseidon Sciences Foundation that include protection and preservation of the aquatic environment as their first aim [⁷].

We think that the recommendations given in the paper [¹] pose a potentially very significant threat to natural communities in temporary freshwater pools across the African continent and adjacent islands. We are pleased that authors are aware that the introduction of N. guentheri to non-native areas within Tanzania and elsewhere, is inappropriate. A solution offered by Matis and Adrias is that another species with wider distribution, such as N. melanospilus, may be used instead. While this may represent a step forward from the original plans, it should be pointed out that: (1) N. melanospilus populations are variable and have a strong phylogeographical pattern, which is common to all Nothobranchius species studied to date [³⁹], and hence even translocation within its wide range is not recommended; (2) N. melanospilus will probably hybridize with related and currently allopatric/parapatric species such as N. makondorum or N. lucius; (3) N. melanospilus may pose a risk to other native Nothobranchius via competition (including its indirect effects [²⁵]); (4) experimental data obtained for N. guentheri may not be transferable to other species of the genus [⁴⁰].

The study by Matias and Adrias misquotes the results of previous studies (see an example above on mosquito larvae as the dominant natural prey) and includes some incorrect statements. For example, the statement that juvenile Nothobranchius do not feed for the first three days is highly inaccurate since the juveniles must start feeding within a few hours after hatching (certainly within less than a day) in order to survive [¹⁶]. Another important fact is that the photograph published in the Matias and Adrias paper as N. guentheri, the study species, actually depicts a different Nothobranchius species (an undescribed N. sp. aff. kirki from Malawi), that is morphologically very different to N. guentheri and likely not even closely related. It is therefore unclear whether N. sp. aff. kirki or N. guentheri were used for the experiments. Also a map depicting the distribution of African annual fishes (Figure 8 in [¹]) is largely incorrect, even if other African genera with potentially annual species (Fundulopanchax, Callopanchax, and Raddaella) are included.

In conclusion we reiterate that we are aware of serious problems with mosquito-borne diseases in tropical Africa and elsewhere, and we encourage responsible means of biological control of parasite vectors. There are several cases where native fish species were found to be effective predators of mosquito larvae [e.g. [⁴¹,⁴²]] but there are also many cases where the ecological balance has been catastrophically affected [e.g. [²⁵,⁴³,⁴⁴]]. We believe that any serious attempt to advocate particular fish species as a biocontrol agent needs to be based on sound scientific attempts to assess both the ability of species to control the parasite vector or pest, and the potential impact on natural ecosystems [²³]. Otherwise, there is a significant risk of repeating mistakes made in the past decades, such as the introduction of Gambusia holbrooki into many countries with catastrophic consequences for native species [⁴³,⁴⁴] and with no effect on mosquito populations [⁴⁵,⁴⁶].