Female reproductive biology of an exotic suckermouth armored catfish ( Loricariidae ) in the San Marcos River , Hays Co . , Texas , with observations on environmental triggers

Invasive populations of suckermouth catfishes (Loricariidae) are native to Central and South America, but have become established in US and Mexican waters since the 1950’s and have been reported to have negative impacts on North American freshwater ecosystems. Two genera of loricariids have been reported from Texas waters (Hypostomus spp. and Pterygoplichthys spp.), both of which have become established in aquatic ecosystems where there are warm-water refugia, or suitable and seasonally stable temperatures. In an effort to better understand the invasive dynamics of these loricariids in novel ecosystems, aspects of their reproductive biology such as fecundity, seasonality of spawning activity, and spawning frequency of the individual fish were studied for Hypostomus cf. niceforoi living in the spring-fed San Marcos River, Texas. Fecundity was similar to Hypostomus spp. in the native range. There did not appear to be any synchronicity of spawning between individual fish within the invasive population, and there was a hint in the oocyte size-frequency data that some of the fish may be spawning multiple times per year. The timing of spawning activity in the novel range was seasonally inverted compared to the pattern typical of loricariids in the native range. The season of peak reproductive activity was also less distinct in the novel range than in the native range, and the period of reproductive quiescence was more seasonally compressed than in the native range. We provide data and supporting arguments suggesting that photoperiod is the primary proximate factor triggering the onset of reproductive quiescence, as well as a return to reproductive activity for these fish in both their native and novel ranges.


The Loricariids
The Loricariidae is the most diverse family of siluriform fishes, with more than 840 species having been described among 90+ genera and 6 subfamilies (Nico et al. 2012).All are native to fresh and brackish waters of Costa Rica, Panama, and South America (Nelson 2006).Loricariids typically consume algae or detritus (Flecker 1992;Power 1984b) and inhabit lotic environments (Sakurai et al. 1993;Sterba and Mills 1983).Part of their invasiveness is due to them being facultative air-breathers that can absorb oxygen through the stomach, allowing them to survive periods of low dissolved oxygen (Graham and Baird 1982;Mattias et al. 1998) to which native fish may succumb.

Statement of problem
Because of their hardy nature and their reputation as cleaners, several genera of loricariids are routinely imported into North America for the aquarium trade (Walker 1968).Interest in loricariids by aquarists has grown substantially in the last several decades.Hoover et al. (2014) found only three loricariid genera addressed in the popular aquarium literature between 1933 and 1935, but found 35 genera addressed between 1987 and 2009.Loricariids are also sometimes intentionally released into freshwaters outside their native range by natural resource managers hoping to control algae and aquatic plants (TISI 2014).These findings suggest that loricariid introductions are likely to continue expanding in frequency, taxonomic diversity, and geographic location, and that the consequences of these introductions will likely become more complex.
Release of loricariids has led to established populations in several states in the USA (Hoover et al. 2004), including Nevada, Hawaii, Texas, Arizona, Colorado, Connecticut, Louisiana, Florida, andPennsylvania (Courtenay andDeacon 1982).The earliest report of loricariids in Texas waters was in the San Antonio River in 1956 (Barron 1964), and loricariids have since been reported from several other Texas drainages (Hubbs et al. 1978;López-Fernández et al. 2005;Nico and Martin 2001).The first record of loricariids in the San Marcos River, a sensitive habitat that is home to several state and federally listed endemic taxa, was in the early 1990's (Perkin 2009;Perkin and Bonner 2011).The feeding preferenda of the San Marcos River Hypostomus was elucidated by Pound et al. (2011).
Although the San Marcos River Hypostomus was identified as Hypostomus plecostomus (Linnaeus, 1758) in some earlier literature, the local loricariid that we studied from the San Marcos River has been declared to not be H. plecostomus, and has been tentatively identified (personal communication from Jonathan Armbruster at Auburn University) as Hypostomus cf.niceforoi (Fowler, 1943).The native range of this species is the Japurá River system, State of Amazonas, Brazil (Froese and Pauly 2015).We have also recently noted the presence of an unidentified species of Pterygoplichthys (Loricariidae) now thriving in the San Marcos River, but this species was not included in our study.

Reproductive biology of loricariids
The reproductive biology of loricariids has been studied in their native habitats using such analytical tools as the gonadosomatic index (GSI) and the maturation patterns of ovaries and oocytes (Duarte and Araújo 2002;Duarte et al. 2007;Mazzoni and Caramaschi 1997a,b).Loricariids living in the native habitats of South America have developed reproductive habits coincident with the seasonal availability of food and cover for young fish caused by seasonally predictable flooding of the riparian forest floor, sometimes for weeks at a time (Duarte and Araújo 2002;Duarte et al. 2007;Mazzoni and Caramaschi 1997a,b).Power (1984a) speculated that loricariids in lotic South American habitats had synchronized their reproductive activity to take advantage of the increased food availability and cover (the ultimate factors) during these periods, but were probably using increased water temperatures and rainfall rates in spring and summer months as triggers (proximate factors) to prepare for breeding (Power 1984a).However, the exotic population of H. cf.niceforoi living in the thermally stable spring run of the San Marcos River would not be influenced by the same weather-related proximate factors that are experienced annually by fish in the surface-fed waters of their native range.Moreover, seasonal variations in the ultimate factors (that originally synchronized the reproductive activities of Hypostomus to proximate factors in the native habitat) do not occur in the San Marcos River.
During the period of increased reproductive activity, some Hypostomus species in South America are known to spawn asynchronously, with individual females spawning multiple times within the year (Duarte and Araújo 2002;Duarte et al. 2007;Mazzoni and Caramaschi 1997a,b).
The purpose of our study is to provide information regarding the fecundity and the seasonal patterns of reproductive activity of the San Marcos River population of Hypostomus cf.niceforoi.Our findings are compared to reports from similar studies in the Northern and Southern Hemispheres, and help shed light on whether or not the presumptive proximate factors that have been proposed as triggers for the seasonal reproductive patterns observed for loricariids in their native habitats (Duarte and Araújo 2002;Duarte et al. 2007;Mazzoni and Caramaschi 1997a,b) are transferrable to exotic populations in the northern hemisphere.We hypothesize that both native and exotic populations are responding not to temperature, rainfall, or stream discharge, as suggested by other workers, but to another proximate factor (photoperiod) that is a covariate with the ultimate factors in the native range, but not in the novel range in North America.

Study site
The San Marcos River originates at headsprings issuing from the Edwards Aquifer.The headsprings were first impounded in 1848 (Slattery and Fahlquist 1997) and this resulted in what is now known as Aquarena Lake, or Spring Lake.The current study was restricted to a 300-meter reach of the San Marcos River spring run immediately downstream from the dam impounding the headsprings (29º53′24.04″N,97º56′4.12″W).This reach is in the USGS 12-digit Hydrologic Unit HUC 12-121002030302, and includes the segment of the river channel shown in Figure 1.
The exceptional water quality, persistent flow, and stable temperatures of the San Marcos River (Groeger et al. 1997) provide a stable habitat for a variety of endemic aquatic life; some species of which    Mean daily water temperature (2005) are listed by the state of Texas and the U.S. federal government.Year-round temperature fluctuation of these waters is minimal compared to local surface waters, with the mean daily water temperature, measured every four hours just downstream from the study reach during the study year (2005), varying between 20.7 and 23.5 with an annual mean of 22.7 °C (Figure 2).The mean annual discharge rate of the San Marcos Springs between 1994 and 2014 was 5.2 m 3 /s and, although mean monthly discharge from the springs fluctuated substantially during that period (2.4-20.6 m 3 /s, USGS 2014), there are no annually consistent flood seasons during which the stream inundates the flood plain for extended periods.

Study fish
Eighty seven Hypostomus cf.niceforoi were collected from the study reach by gigging from January through December of 2005 (≤10 fish/month).Length and weight summary statistics for males and females collected are provided in Table 1.Immediately following collection, the fish were euthanized by pithing and then weighed and measured for total length.Each fish was then incised ventrally from the pharynx to the vent for the study of ovaries.Only 50 females had discernable ovaries and were considered to be suitable females for reproductive study.

Reproductive biology
Ovaries from the 50 study females were excised, weighed to the nearest milligram, and preserved in 10% buffered formalin for later assessment.The preserved ovaries were later assigned to one of four stages of ovarian maturity categories based on Mazzoni and Caramaschi (1997a,b).This system employs an estimate of the percentage of the body cavity occupied by the ovary, in combination with a visual-based classification scale.Subsequent to our collections and dissections, a new standard for terminology pertaining to reproductive stages of oocytes in fish ovaries was published by Brown-Peterson et al. (2011).The latter system requires histological examination for reliable assignment to ovarian development categories, and our fish were not examined histologically.However, the categories in the system we adapted from Mazzoni and Caramaschi (1997a,b) are roughly equivalent to those of Brown-Peterson et al. (2011), and we provide the equivalent terms here for future reference.The four categories we used from Mazzoni and Caramaschi (1997a,b) are: (1) "Mature-1" (ovary <15% of body cavity and pale cream in color with subtle granulation, which is roughly equivalent to early "Developing" of Brown-Peterson et al. 2011); (2) "Mature-2" (ovary >15% but <50% of body cavity and yellow in color with light vascularization and large light-yellow oocytes discernable, which is roughly equivalent to late "Developing"); (3) "Ripe" (ovary >50% of body cavity and orange in color with strong vascularization, thin ovarian wall, and containing large yellow oocytes 5.5 mm in diameter, which is apparently identical to "Spawning Capable"); and (4) "Recovering" (ovary < 30% of body cavity and flaccid with slight vascularization, which is apparently equivalent to "Regressing").While other workers have included a category "Immature," this category was omitted from our study because only one of the 51 females collected was considered to be sexually immature.However, we included an additional category ("Resting") for six of the females which were large enough to be sexually mature (by being larger than some other females that were obviously sexually mature), but which had ovaries with no obvious sign of past or present oocyte development (perhaps roughly equivalent to the "Regenerating" category of Brown-Peterson et al. (2011)).Seasonal variation in reproductive activity of the females was studied by calculating mean monthly gonadosomatic index (GSI; typically defined as ovary weight divided by total body weight multiplied by 100%) and by examining the relative monthly frequencies of the ovarian maturity categories described above.However, the Mature-2 category was so sparsely represented that, for purposes of exploring seasonal patterns, we pooled Mature-1 and Mature-2 into "Mature," which corresponds rather well to the Developing category of Brown-Peterson et al. (2011).
Fecundity was estimated following the McGregor (1922) sub-sampling-by-weight technique recommended in the International Biological Programme Handbook No. 3 (Ricker 1968).The majority of oocytes from the 17 Ripe ovaries appeared to be in a mid-to late-vitellogenic growth phase, as per Patiño and Sullivan (2002), which is roughly equivalent to the vitellogenic stages 2 and 3 (Vtg2 and Vtg3) of Brown-Peterson et al. (2011).The oocytes from these Ripe ovaries were then set aside and later included in the oocyte size-frequency study.The spawning pattern of individual Hypostomus in the San Marcos River was investigated by generating oocyte size-frequency charts from the 21 fish having ovaries classified as either Mature-2 (n=4) or Ripe (n=17).Approximately 100 oocytes (as per Hunter and Goldberg 1985) were extracted from each of these 21 ovaries and measured under a dissecting microscope to the nearest 100 m with an ocular micrometer.These measurements were then cast into separate size-frequency charts for each of the 21 fish, and arranged by month of capture.
All calculations were performed in Excel.Charts were produced in Excel and some were enhanced in PowerPoint.All geographic references are from Google Earth Pro, using the WGS84 datum.

Seasonal variation in reproductive activity
The patterns of seasonal variation in the five individual categories of ovarian maturity (including the Resting category) are represented by percent monthly totals in Figure 3. Fish with Ripe ovaries only occurred March through September, and four of the six fish with Resting ovaries were observed September through November.Fish with Mature ovaries were observed throughout the 11 months we collected fish, except for August.
Although no definitive peak in reproductive activity was noted for the San Marcos River Hypostomus, the fluctuations in mean monthly GSI's appeared to correspond seasonally with photoperiod.All elevated mean monthly GSI's (greater than 2%) occurred from March through August, and five of the six lowest mean monthly GSI's (less than 2%) occurred September through January (Figure 4).As with the Pterygoplichthys population in Volusia Springs, Florida (Gibbs et al. 2008), there appeared to be a period of reproductive quiescence, with five of the six GSI values below 2% occurring during the five months between when day length had declined more than half way to minimum value, until just after the day length had begun to increase again.

Fecundity estimates
The fecundity estimates for the 17 Ripe ovaries varied from 871 to 3,367 oocytes per ovary, with a mean estimate of 2,124 per ovary ( s X  217).

Spawning frequency within a breeding season
Oocyte size-frequency distributions for each of the ovaries classified as Mature (n=4) and Ripe (n=17) are plotted in separate charts for each ovary in Figure 5.In at least three of the charts for Ripe ovaries (charts P, R, and T), two distinct and separate cohorts of oocytes are discernible.

Spawning frequency
Native Hypostomus populations in South America have been reported to spawn multiple times per year (Duarte and Araújo 2002;Mazzoni and Caramaschi 1997a,b).Multiple-batch spawning has also been reported for an introduced population of Pterygoplichthys disjunctivus in Volusia Blue Spring, Florida (Gibbs et al. 2008).We investigated the pattern of spawning frequency of individual fish in our study by examining the size-frequency charts portraying oocyte development within individual ovaries.We found some evidence of bimodality, indicating two different oocyte-size cohorts (Figure 5; charts P, R, & T).The ovaries of these three females suggest that the Hypostomus in the San Marcos River system is a determinate batch spawning species (as per Brown-Peterson et al. 2011) and will likely spawn at least twice during the reproductive season.However, we cannot be certain that these individual fish are actually spawning more than once per season because we did not conduct any real-time spawning observations of individual fish.
All of the oocyte size-frequency charts in Figure 5 show most of the oocytes within an ovary clustered rather tightly around one or two length values.However, while the central lengths around which the oocytes clustered varied from ovary to ovary, there was no consistent seasonal increase in oocyte size among the fish as the seasons progressed, as one would expect if these fish were determinate synchronous spawners.These observations are consistent with patterns of oocyte development reported for loricariids in South America that are asynchronous spawners (Macchi and Acha 2000;Wallace and Selman 1981).Therefore, the observed lack of seasonal progression in the oocyte size frequencies for the San Marcos Hypostomus could be an indication of a similar asynchronous spawning pattern for our fish in the invaded Northern Hemisphere habitat.

Seasonal patterns
The highest individual GSI values (Figure 4), and also the highest monthly percentages of fish with Ripe ovaries (Figure 3), were recorded from early spring (March) through late summer (August), though as females appeared to be reproductively active through most months of the spring, summer, and early fall, we could not reliably estimate the exact seasonal peak in breeding activity.However, we were able to discern the seasonal period of minimal reproductive activity (reproductive quiescence).The low and progressively declining GSI values recorded in the last quarter of 2005 suggest decreasing levels of spawning activity for the population at that time.Even though 30% of the females that were rated for ovarian maturity were collected during the months of Oct-Jan, not one of these 15 females had ovaries classified as Ripe (Figure 3).In addition, the lowest four mean monthly GSI values in the study, involving 40% of the 50 fish rated for GSI, occurred in progressively declining order from September through December (Figure 4).
These findings are consistent with those for an exotic population of Pterygoplichthys disjunctivus established in Volusia Blue Spring Florida (28º56′51.00″N,81º20′22.50″W;Gibbs et al. 2008), and also for a different population of P. disjunctivus at El Infiernillo Reservoir in Mexico (18º16′21.27″N,101º53′34.13″W;Rueda-Jasso et al. 2013).Findings by Gibbs et al. (2008) suggest that the period of reproductive quiescence for P. disjunctivus in Volusia Blue Springs starts in the last quarter of the calendar year and continues through January (lowest numbers of large oocytes Oct, 2005-Jan, 2006and Nov, 2006-Jan, 2007; lowest mean GSI and lowest variation in GSI during same periods).In addition, Rueda-Jasso et al. ( 2013) reported the lowest raw fecundity and lowest egg number/wt.of females in January, and also reported lowest GSI values from November through January (Figure 6).

Abiotic factors covarying with ovarian dynamics
The period of reproductive quiescence for Hypostomus populations in the river Paraiba do Sul in South America near the Tropic of Capricorn (at approximately 21º43′7.10″S,42º16′29.25″W)typically occurs in the second quarter of the year (Duarte and Araújo 2002;Mazzoni and Caramaschi 1997a,b).This is almost exactly one half year out of phase with what we observed for the three North American populations, which straddle the Tropic of Cancer (from about 18ºN to 30ºN).In contrast to the South Figure 6.Comparison of periods of reproductive quiescence (based on declining GSI) reported for loricariids in southern North America (Florida, Texas, and Mexico) vs.South America (Rio de Janeiro), with the annual photoperiod cycles for the latitudes of interest.A -Based on Rueda-Jasso et al. ( 2013), GSI values less than 0.8%, lowest GSI in January (0.36%); B -Based on Mazzoni and Caramaschi (1997b), GSIs were reported by these authors with resolution of 2 months, so beginning and ending of indicated quiescent period is represented herein as middle of 2-month periods reported by the authors (GSIs<2%); C -Based on our study (San Marcos River, TX, 2005); GSIs during indicated quiescent period were all less than 2% and declining until increase in January; D -Based on Gibbs et al. (2008); quiescent period was compressed by two months during the second year of study, so quiescent period represented herein is bounded by median of two beginning months and two ending months reported by authors.Day lengths determined from USNO (2015).
American populations, there were no apparent relationships between patterns of reproductive quiescence for Hypostomus in the San Marcos River and local rainfall rates, spring discharge, or ambient air temperature.Gibbs et al. (2008) noted that P. disjunctivus spread out into the connecting St. John's River in warmer months, but returned to the warmer spring run in large numbers during the winter.Temperature may therefore have played a role in influencing the fourth quarter period of reproductive quiescence in the Volusia Blue Springs population, even though that period of quiescence coincided well with the San Marcos Springs population, which was not exposed to temperature swings.Rueda-Jasso et al. (2013) noted that the rainy season happened to be coincident with the peak of reproductive activity in a population of P. disjunctivus they studied in Mexico.However, the single common factor that ties all these periods of reproductive quiescence together in both hemispheres is that they all become reproductively quiescent during the second half of the period of waning day length at their respective localities.Moreover, the end of the period of reproductive quiescence (increase in reproductive activity) occurs shortly after the photoperiod begins to lengthen again (arrows between dotted lines in Figure 6).Note that the Mexico site is at a lower latitude than the other sites under comparison and experiences less seasonal amplitude in photoperiod (Figure 6), perhaps explaining why the quiescent period in Mexico is not as seasonally distinct as for the Volusia Blue Springs and San Marcos Springs populations.
Our suggestion that photoperiod is the main proximate stimulus triggering the onset of reproductive quiescence for these loricariids does not eliminate the possibility that other proximate factors, such as rainfall and water temperature, originally played a role in the evolution of patterns of reproductive activity for loricariids in their native habitat.
Proximate factors have often been implicated in determining the evolution of behavioral patterns (Brown and Lawson 2010;Hoar 1976;Kramer 1987;Sorensen and Stacey 1999).While water temperature is typically thought to influence hormonal cycles in fish more so than photoperiod (Vlaming 1972), mean daily water temperatures during periods of reproductive quiescence for the Hypostomus population in the San Marcos River were not more than 3 °C lower than during the extended period of reproductive activity during the summer (Figure 2).Therefore, while proximate factors other than photoperiod might covary with reproductive behavior in the native and novel habitats, the proximate factor that actually stimulates the onset of increasing reproductive activity of these fishes in the Northern Hemisphere appears to be lengthening photoperiod following the annual minimum.Gibbs et al. (2008) observed that the period of reproductive quiescence in the Volusia Blue Springs population was shorter than the period of reproductive quiescence reported for native populations, i.e. the period of reproductive activity had apparently lengthened at the expense of the quiescent period.Moreover, even during their two-year study, the quiescent period was compressed in the second year compared to the first (Gibbs et al. 2008).They conjectured that, while the day-length cycle may have served as an adaptive trigger for the South American populations signaling the onset of favorable breeding conditions, that trigger has no adaptive value for the exotic populations in the seasonally constant spring habitat.Photoperiod seems to be the proximate factor that is most likely to be triggering the observed seasonal spawning cycle in Northern Hemisphere populations at this time, but is probably also doing so for populations in their native ranges as well.For populations of loricariids established in environmentally stable spring runs in the Northern Hemisphere there is no obvious selection pressure that would discourage spawning during the season of waning day length.Therefore, while there may be residual (but seasonally inverted) patterns of reproductive activity in newly introduced populations in the Northern Hemisphere, with time, reproductive activity in these exotic populations might become more seasonally uniform as the period of reproductive quiescence is gradually compressed and ultimately disappears.

Fecundity
Our fecundity data for the Hypostomus in the lotic habitat of the San Marcos River are similar to those reported for Hypostomus affinis (Steindachner, 1877) populations in lentic and lotic habitats of South America.For example, in the native Brazilian range of H. affinis, Duarte and Araújo (2002) reported mean fecundity to be 2,373 in a lentic habitat, while Mazzoni and Caramaschi (1997b) reported a mean fecundity of 1,784 for P. disjunctivus in a lotic habitat.The mean fecundity estimate for the Hypostomus cf.niceforoi of the San Marcos River was intermediate, with approximately 2,124 oocytes per Ripe ovary.Our fecundity data are also consistent with those reported for other species of loricariids, which typically have less than 2,500 oocytes per ovary (Azevedo 1938;Rueda-Jasso et al. 2013).However, if H. cf.niceforoi in the San Marcos River is, indeed, a multiple-batch spawner, then our fecundity estimates for this species are probably low.

Conclusions and implications for future research
Our findings indicate that the San Marcos River Hypostomus population is probably batch-spawning asynchronously, and that the seasonal onset and cessation of reproductive activity are likely triggered by photoperiod.
Because the variance in such measures as GSI is very high for asynchronous spawners having extended spawning periods, it may be more useful to study the seasonality of reproductive behavior for such fishes by describing the period of reproductive quiescence rather than attempting to estimate the seasonal peak in reproductive activity.The quiescent period in both the Northern and Southern Hemispheres for the loricariids considered herein begins during the period of declining day length at approximately the equinox, and the resumption of reproductive activity occurs at or shortly after the minimum day length, even for invasive populations in spring-dominated physicochemically stable environments.Thus, loricariids in both hemispheres are probably using photoperiod as triggers of reproductive activity, rather than seasonal changes in other covarying environmental factors such as river discharge, water temperature, rainfall, etc.
Since this study was conducted, there have been several anecdotal reports and personal observations of invasive loricariids in Texas River systems not influenced by thermally constant spring runs, such as the Guadalupe River some 12 km downstream from the confluence with the spring-fed Comal River.The fish in these thermally ambient waters have been observed to survive winter thermal minima by crowding into thermally stable refugia formed by hyporheic flows in which winter thermal minima remained above 17 °C (personal obs., DGH).
Loricariids have also been captured along Texas coastal habitats in brackish and salt water bays and estuaries (Chilton et al. 2011).Outside of spring-run habitats, it is unclear what influence abiotic factors may be having on loricariid reproductive behaviors (Nico et al. 2012).The more or less random spawning of individual fish over a protracted spawning period of several months has also been reported for loricariids in the native habitat of South America, but this behavior in the novel habitat of North America, coupled with the hardy nature of these fishes, will undoubtedly lead to continued colonization of new habitats.

Figure 2 .
Figure 2. Water characteristics of the San Marcos River just downstream from the study reach: daily discharge rates for 2005 (USGS 2014), means of monthly mean discharge rates (from 1994-2015, USGS 2014), and water temperatures during 2005 (from local measurements every 4 hours, compliments of BIO-WEST Corporation).

Table 1 .
Length and weight summary statistics for Hypostomus cf.niceforoi collected from the San Marcos River, TX, during 2005.