Rapid growth, early sexual maturation and high reproductive investment are traits typical of organisms inhabiting temporary and unpredictable habitats [1, 2]. We documented extremely early maturation in two species of African annual fish. Notably, Nothobranchius do not compromise early sexual maturation with modification of its developmental trajectory and displayed all the morphological features typical of its evolutionary lineage .
Current published records of early maturity in a vertebrate are for a miniature pedomorphic goby, Schindleria sp. (23 days at 15 mm) . Among non-pedomorphic vertebrates, female Alston’s brown mouse (Scotinomys teguina) achieve maturity at 28 to 39 days in the wild , females of some lab strains of house mouse (Mus musculus) mature at 23 days (although males mature much later, at the age of approximately 6 weeks) , a pygmy goby (Eviota sigillata) at 34 days , and the chameleon F. labordi at 2 months . In other vertebrate taxa, sexual maturation typically takes considerably longer.
We demonstrated that N. kadleci can mature after 17 days at a size of 31/37 mm (female/male) and N. furzeri after 18 days at 32/38 mm (female/male). This finding is much earlier than previous reports for the GRZ laboratory strain of N. furzeri of four weeks . Previous reports of early maturation in N. furzeri are based on anecdotal general statements ; here we provide a quantitative analysis. Further, the GRZ strain has been bred in captivity since 1969, is highly inbred  and likely artificially selected for traits facilitating captive breeding. In contrast, our study fish were non-inbred F2 generations of wild populations with naturally evolved life history trade-offs. Our study was designed to mimic natural conditions through ad libitum feeding, reducing fish density over time, and using the largest individuals to estimate rates of growth and development. In both N. furzeri and N. kadleci intense male-male competition favors high survival of the largest males and males of comparable size are frequently encountered in the wild [8, 22]. The only age estimates obtained in the wild indicate that all individuals older than 27 days were sexually mature, with no younger fish included in that study .
All study populations of Nothobranchius demonstrated rapid growth, especially in the first 28 days. They also showed rapid deceleration of growth, with negligible growth after the age of 60 days (Figure 2). Maximum growth rates were recorded during their second week of life, reaching up to 2.15 to 2.72 mm (17 to 23.4% of TL) day-1 in the largest males. This period corresponded with a decrease in fish density, increase in water temperature (from 25 to 27.5°C), and weaning to a nutritionally richer diet, which are all features that match temporal changes in nature. After 14 days, fish growth rates decreased (Figure 2) as fish started to allocate resources to reproduction. Growth rates of N. furzeri and N. kadleci are unusually high among killifish (Cyprinodontiformes). For example, the maximum growth rate of juvenile Austrolebias viarius (>1 month old), a Neotropical annual killifish, achieved 0.70 mm (2.3% of TL) day-1 at 25°C in captivity and 0.66 mm day-1 in the wild (17 to 23°C) . The growth rates of N. furzeri and N. kadleci are also unique within the genus [24, 25]. For example, in Nothobranchius korthausae, a congeneric species of comparable size at hatching but considerably longer-lived (57 and 81 months for mean and maximum lifespan), fish at an age of 2 to 4 weeks are only 25% of the size of N. furzeri and N. kadleci. Further, N. korthausae approach asymptotic length at the age of 40 weeks, compared to age of 10 to 12 weeks in N. furzeri and N. kadleci.
The growth rates in the wild are comparable and may be even higher. A wild population of N. furzeri (NF220, in close proximity to the study population NF222), with a known maximum age of 11 weeks (estimated from the fact that a logger recorded watering of the habitat 78 days prior to fish collection, see  for more details) attained 71.7 ± 2.1 mm (maximum 76 mm) and 60.0 ± 1.0 mm (maximum 63 mm) for males and females, respectively. The growth rates in the wild are variable and affected by fish density and prey availability . However, the presented growth rates and maturation schedule are within the natural range of the studied species and favorable habitat conditions can likely promote even faster growth.
High adult mortality risk experienced by Nothobranchius is predicted to select for increased allocation to reproduction [1, 2]. A notable outcome of the study was the high fecundity of experimental females. Egg production markedly increased between 21 and 42 days, when females still allocated a large proportion of resources to growth, but remained stable and high afterwards. Note that an increase in egg production coincided with change in experimental conditions, though we consider the effect of increasing body mass more important than a decrease in population density as food was delivered ad libitum. The increase in daily egg production when environmental conditions were temporarily enhanced (fresh food, partial water exchange) was reported in Neotropical annual fish, Austrolebias nigripinnis. In general, egg production of A. nigripinnis is markedly lower than in Nothobranchius[11, 12, 24], with the number of eggs rarely exceeding 20 per day (typically 0 to 12 eggs) .
Fertilization success gradually increased from 21 to 63 days (Table 1). Maximum recorded fecundity was 583 eggs laid during 2.5 hours by N. kadleci female (Pungwe population) (age 63 days, 56.9 mm). Nothobranchius normally lay 5 to 50 eggs daily and each egg is laid singly . In our setting, such fecundity was typical for females until 21 to 42 days. Once growth rate decelerated, females laid up to several hundreds of eggs, demonstrating the ability of N. furzeri and N. kadleci to elevate their energetic allocation to reproduction if conditions are favorable. Environmental conditions in Nothobranchius habitats vary temporally and spatially [8, 26], and fecundities presented here are likely close to the upper limit of the reaction norm (that is, plastic phenotype expression of a genotype across environmental conditions) expressed under conditions of low population density and high food availability.
In this study we found a proportion of embryos to develop within 15 days at 25.0 ± 0.2°C and Valenzano et al.  provided an estimate of only 12 days for N. furzeri at 26°C. This results in a plausible minimum generation time of one month in both species. This interval is long compared to invertebrate models; D. melanogaster (10 days) and C. elegans (4 days), but considerably shorter than lab mice (70 to 80 days) . The calculated generation time is applicable to laboratory conditions; fish follow a natural annual cycle in the wild and produce one generation each year . Likewise, the other model species have much longer generation times sensu Caswell (2001)  in the wild. We also note that higher temperatures (at least 28°C) for embryo incubation are possible and should result in even shorter embryo development and, hence, generation time, though this has not been tested. Finally, we did not follow hatched fish and did not estimate the proportion of fish with insufficient swimming capacities (‘belly sliders’). They frequently occur, at least in captive conditions, including our study populations and are unable to reproduce normally. Consequently, only a study following rapidly developing fish over more generations would unequivocally demonstrate the ability of N. kadleci and N. furzeri to produce a large number of generations in rapid succession.
A further important outcome of our tests of laboratory protocols is that over 50% of embryos skipped all diapauses and developed within a short period (30 days), along with a good survival rate (73% for embryos) for embryos developing on a damp peat moss. This extraordinary short generation time and high embryo survival in the lab makes them ideal model species for aging research .
The natural generation time of wild Nothobranchius is strictly determined by filling and desiccation of their habitats and thus strongly linked to the alternation of rainy and dry seasons [5, 26]. However, we believe that the ability to complete a generation within one month is adaptive. Both studied species live at a periphery of Nothobranchius distribution [5, 30] where rainfall is unpredictable and erratic. A pool with fish may occasionally dry out and be watered again within a single rainy season. Under these circumstances the embryos with fast development may give rise to the second generation of fish that would populate a secondary pool.