Development and life cycle Migration and reproduction Multiple line of evidence seems to indicate that reproduction occurs in the brackish or marine environment. Hogan and Nicholson (1987) were the first to establish that the optimal sperm motility for salinity ranged between 25 ‰ and 32 ‰ in individuals from the Fiji Islands. Another study by Henderson (2010) indicated optimal sperm motility occurs at 36 ‰ salinity for males caught in Queensland. These differences could reflect either regional or experimental variability. In captivity, only individuals maintained in freshwater before receiving hormonal injections and then being moved to high salinity tanks (30 ‰) were capable of breeding (Hutchinson et al., 2009). While spawning behaviour has yet to be observed in the wild, Hogan and Nicholson (1987) suggested that individuals converge to the tidal zone for breeding along coastal and offshore reefs. Additionally, breeding appears to be performed in a short-time frame (Lewis and Hogan, 1987; Henderson, 2010). The spawning period seems to occur during full moon and new moon phases (Hogan and Nicholson, 1987; Hutchinson, 2009; Henderson, 2010) of the rainy season (ER, Anamparéla, Feutry, Marsden, Valade, 2015). The increase in flow velocity and a favourable moon phase could have a determinant synergic effect favouring the migration into preferential breeding habitat. Analysis of hormonal concentrations during the different gonad maturation stages have shown that the reproduction period could last for months, from October / November to March / April in Australia (Hutchinson et al., 2008; Henderson, 2010). Additionally, Hutchinson et al. (2009) reported females bearing different oocyte sizes, thus suggesting females K. rupestris perhaps have the ability to spawn multiple times each season. This is possibly an adaptation to the high variability and unpredictability of environmental conditions faced in insular systems. Kuhlia rupestris compensates for his short larval phase by having a long breading season, ensuring its survival via low and episodic recruitments staggered over time rather than one big recruitment event. On Réunion Island, surveys of larval recruitment in the mouth of the two largest rivers indicate that the reproduction period could last up to 10 months, from November / December to August / September (Lagarde et al., 2012). Sexual maturity The study of sexual maturity requires a high number of mature specimens. Thankfully, for several decades now, sacrifices are no longer required and have been replaced by non-lethal sampling. Biopsies, allowing the measure of hormonal concentrations, are done using a cannula inserted into the urogenital aperture. However, in K. rupestris, this method has only produced satisfactory results for females (Henderson, Hutchinson, 2015, pers. comm.). For the males, it is still difficult to obtain their semen (by massaging them) even if they are at the appropriate maturity stage (ER, Henderson, 2015). Lewis and Hogan (1987) were the first to report a minimum size of 17 cm SL for “ running ripe ” males, whereas the minimum size for nearly ripe female was 21 cm SL. More recently, in Australia, Henderson (2010) has collected some blood samples and made ovarian biopsies on a quarterly basis in order to determinate reproduction activity. This work showed that ripe females were larger than 23.4 cm SL as opposed to 17 - 18 cm SL for ripe males. In Japan, however, the smallest mature male observed was 12 cm SL and more than half of the males mature at 17 cm SL (Oka and Tachihara, 2017, pers. comm.), therefore suggesting regional variability. Furthermore, the age for sexual maturity ranges between 3 and 7 years (Gelineau et al., 2015). In Réunion Island it has been estimated that individuals reach sexual maturity after 5 years (Gelineau et al., 2015), whereas in Queensland, in natural environment, it is ranged between 4 and 6 years (Henderson, 2010). Trophic ecology Kuhlia rupestris seems to have a fairly diverse diet. The analysis of the diet of 50 specimens from 3 different locations in northern Australia by Pusey et al. (2004) revealed that the main food source was terrestrial invertebrates, aquatic ones coming second (Fig. 2). Ants, spiders, grasshoppers or even frogs and smaller vertebrates were found in the stomachs of large individuals. Leave debris and fruits were also identified, but quite surprisingly, very few fish appeared in the stomach content. For young individuals, Trichopteran and Ephemeroptera larvae as well as adult stages formed the bulk of the food. Note that a breakthrough towards successful breeding in captivity was the consumption of small rotifers by K. rupestris larvae (Hutchison et al., 2009 b). Given the wide distribution of K. rupestris, additional studies with consideration of local habitat and seasonality would probably help define its diet more precisely across its range. According to experts, this species feeds on different fish species (Sicyopterus sp. fry, young Anguilla spp.) and macrocrustaceans (Atyidae and Macrobrachium spp.) (ER, Marquet, Pusey, Lagarde, 2015). International experts consider K. rupestris an omnivorous and opportunistic species (ER, Marquet, Anamparéla, 2015). It is qualified as visual predator by some researchers (ER, Pusey, Feutry, 2015) having the ability to catch floating and deriving preys such as insects fallen into the water. Habitat Morphodynamical data Kuhlia rupestris habitat is found between the lower and the middle course of the river and is most often stopped by the first significant obstacle, generally a waterfall. In some rivers, especially in steep tropical islands, this can mean only a few hundred metres of suitable habitat. Exceptional hydrological conditions may help K. rupestris reaching higher parts of the rivers than they normally would. In the absence of obstacles, K. rupestris may be encountered far inland. For example, on the Onilahy River, in Madagascar, K. rupestris occurs at up to 900 m of altitude and as far as 250 km from the sea and (Loiselle and Stiassny, 2007). In Australia, some young individuals were found 50 km from the river mouth at an altitude of 50 m (Pusey et al., 2004). In Réunion Island, K. rupestris is found from 0 to 450 m of altitude (Antemi et al., 2011 - 2013). Microhabitat and preference curves Kuhlia rupestris seems to have a distinct preference for clear water rivers with rocky substrates (Lewis and Hogan, 1987; Hutchinson et al., 2008) presenting diverse profiles. Two different studies, both in the Pacific region, investigated K. rupestris microhabitat preferences. It appears that K. rupestris preferentially lives in water bodies deeper than 60 cm and tends to avoid those shallower than 30 cm (Keith et al., 2009). Its preference for deep water was also confirmed in Réunion Island where subaquatic observations were carried out (Gelineau and Saget, 2016). These observations also revealed that K. rupestris seems to prefer to stay near the bottom or in the water column rather than close to the surface. Kuhlia rupestris occurs in a large spectrum of sediments (Pusey et al., 2004; Keith et al., 2009) although they tend to prefer substrates dominated by hard materials such as boulders and slabs. During the adult stage, several experts report that adults prefer deep-water environments, especially those with shelter provided by boulders, dead wood or roots system under the river bank (ER, Marquet, Marsden, Hutchinson, Valade, 2015) and it is considered that this species prefers a flow velocity lower than 0.2 and 0.3 m. s – 1, generally avoiding those ranged between 0.8 and 1 m. s – 1 (Keith et al., 2009). In terms of river flow, this species is quite tolerant, occurring in rivers with flows ranging from less than 1 up to several hundred of m 3. s – 1. In other words, they can be found in small creeks as well as in large rivers draining water from vast watersheds. In the tropics, this often means the potential for big flood events with very high-water flows. Some degree of sexual segregation has been reported in Australia (Lewis and Hogan, 1987; Henderson, 2010). Females tend to move further upstream, whereas males inhabit mid to lower reaches of the river systems. Lewis and Hogan (1987) also found that the sex ratio was approximately 10 females for 1 male and only males were found in estuaries. More recently, a sex ratio of 4 females for 1 male was reported for the Wyuna Creek population in Queensland (Henderson, 2010). Environmental tolerances Kuhlia rupestris probably has an affinity for well-oxygenated waterbodies. Indeed, in the South-West Pacific Ocean it has been collected in environments, which generally exhibit mean dissolved oxygen values close to 7 mg. l – 1 (Pusey et al., 2004) and around 8 mg. l – 1 in Réunion Island (Gelineau and Saget, 2016). It is however important to note that a study in Queensland revealed that K. rupestris could be tolerant to low dissolved oxygen concentrations (<4 mg. l – 1), at least temporarily (Hogan and Graham, 1984). This species has wide temperature and acidity tolerances. It survives to winters in Okinawa Island, southern Japan, where the lowest temperature is about 13 ° C (K. Maeda, unpubl. data) and is resistant to average temperatures in excess of 25 ° C (ER, Boseto, Henderson, Valade, 2015). Kuhlia rupestris has been observed in acid water with a pH as low as 4.5 (Pusey et al., 2004) and in basic water with a pH as high as 9 (ER, Raynaud, 2015). Population dynamic Fry and juvenile growth Very little data available regarding the early growth of K. rupestris. Neither its embryonic development duration nor its incubation period has been studied in the wild. In Australia, during the first attempts of reproduction in captivity, fertilised eggs size were about 600 μm and larvae hatched within 12 to 15 hours after spawning. Two days old larvae measured about 2.4 mm (Hutchinson 2008). Recent studies on the larval dispersal and migratory flows in Réunion Island revealed that K. rupestris larvae recruiting in freshwater at sizes of about 18 - 30 mm SL (Richarson et al., 2010). In Japan, K. rupestris juveniles recruit into streams at about 20 mm SL mainly from October to February (Oka and Tachihara, 2017, pers. comm.). In Australia, the first size classes observed in freshwater are about 19 to 25 mm SL (Hutchinson, 2008; Henderson, 2010) and display a bimodal size distribution indicative of two major peaks of recruitment each season. In the Indian Ocean, and particularly in Réunion Island, the peak of juvenile recruitment in freshwater is observed in January-February coinciding with the new moon and the full moon phases during the rainy season (Lagarde et al., 2012). There is also a high variability in the size of the youngest individuals, which probably reflects the extent of the breeding season. The timing of recruitment could further increase this size heterogeneity. Indeed, it was suggested that a recruitment occurring in wet and hot season, with a likely high trophic potential, could enhance the growth of fish from an early breeding, whereas the recruitment occurring during the dry season, with less favourable trophic conditions, would not allow for optimal growth (Gelineau and Saget, 2016). Studies of the fry growth rate in captivity provide additional data. Under controlled conditions, young individuals fed daily increased their weight from 0.57 g to 3.5 g and their length from 30 - 35 to 60 - 70 mm in less than 3 months (Hoarau, 2009; Gelineau and Saget, 2016). During the first month, the maximal weight gain recorded was 166 % versus 125 % in average (Hoarau, 2009). Growth and longevity of subadults and adults In the Pacific Ocean, Lewis and Hogan (1987) showed a growth of 2 cm. y – 1 and a longevity reaching 14 and 20 years for males and females, respectively. In Okinawa, Japan, maximal longevity is 8 years (Oka and Tachihara, 2017, pers. comm.). In Queensland, a mark-recapture experiment associated with scale analysis on 145 individuals showed an adult growth rate of 2 - 4 cm. y – 1 with 1 year old averaging 80 mm and the oldest specimen being 327 mm long and 13 years old. In captivity, fin ray increments suggested that a female weighing 3.54 kg was 20 years old (ER, Hutchinson, 2015). Growth can be much faster in captivity compared to the wild and some females were able to gain 1 kg. y – 1. Additionally, sexual dimorphism is well recognized for K. rupestris. Females have a higher longevity and faster growth compared to males (Lewis and Hogan, 1987; Hutchinson et al., 2009 a) although the fastest growth ever recorded was 184 mm in 200 days in a male (Henderson, 2010). This study also showed that males generally do not exceed 7 years, whereas all individuals 10 years old and above are exclusively females. Population size and trend It is difficult to estimate the number of individuals present in a specific river or at a basin scale and K. rupestris is no exception with very limited data available. In Queensland, some investigations were conducted in order to assess the abundance and distribution of K. rupestris. Indeed, based on a broad data collection, collected from natural history museums, universities, fishing associations, magazines and scientific journals, Hutchison et al. (2002) were able to demonstrate that the species had declined in recent times. The increase in construction of barriers to migration was identified as the main cause for this decline. In Réunion Island, an electrofishing survey is ongoing since 2010 on all 13 perennial rivers. Kuhlia rupestris is captured intermittently, with low abundances (up to 1 - 2 specimens. 100 m – 2) and variability between the rivers (Gelineau et al., 2015). Experts have identified no particular trend there since 2000 (ER, Lagarde, Valade, 2015). Behaviour, locomotion and moving capacity Kuhlia species are pelagic mostly feeding from the surface (Resh et al., 1999). Several authors identified juveniles swimming in schools (Hutchinson, 2002; Boseto et al., 2007; Feutry, 2011). Schools of juveniles can sometimes be accompanied by larger individuals assumed to be adults (ER, Henderson, 2015). Juveniles are generally more mobile and less wary than adults, which have preferentially been observed in static position near river bank overhang cavities and crevices (ER, Anamparéla, 2015). Fish movement depends on the inherent physiological capacity of the species, the abiotic conditions of their environment (temperature, hydrology) and the presence / absence of physical barriers, which can reduce access to suitable habitats. K. rupestris is considered a strong swimmer (ER, Marquet, Boseto, 2015), with good sprinting capacity (ER, Henderson, 2015). The optimum swimming speed for individuals of 10, 15 and 20 cm is 1 m. s – 1, 1.5 m. s – 1 and 2 m. s – 1, respectively (Antemi et al., 2011 - 2013). However, the species is only able to maintain such speed across few meters. For larger individuals, the swimming speed is closer to 1.5 m. s – 1 and is generally maintained for 5 to 20 seconds (Kapitzke, 2010). In Australia, Veitch and Burrows (2006) observed this species overcome partial or complete submerged barrier during flood events, therefore demonstrating good upstream swimming capacities. Although its swimming characteristics are quite clearly understood and unanimously supported by international researchers, the jump capacity is not well known. Jumping could be a particular behaviour in response to stress conditions (Valade, 2015, pers. comm.). Observations in Australia attest that they have observed some juvenile overcoming barriers of 10 to 30 cm high (Henderson, Hutchinson, 2015 pers. comm.). Individuals of about 20 cm are theoretically able to overcome barriers of 40 to 50 cm high but only in specific conditions such as: flow velocity ranged between 0.5 to 1 m. s – 1 with an adequate draft (> 15 cm), presence of taking off ramp and sufficient temperature (Antemi et al., 2011 - 2013). Other factors may affect the jump ability such as number of barriers, which have been previously overcame.
