Growth Rates of Non-Native Bighead and Silver Carp in the Upper Mississippi River

: Fish age and growth analyses can be used to infer spawning success, recruitment, and population age structure. Understanding these dynamics is important when assessing the impacts of invasive species, such as bighead carp, Hypophthalmichthys nobilis , and silver carp, H. molitrix , (bigheaded carp). These species have established throughout much of the Mississippi River Basin; however, Lock and Dam 19 (LD19) has slowed their establishment in the Upper Mississippi River, resulting in low-density populations with limited reproduction upstream. The age and growth of Bigheaded Carp in recently established populations, especially in this region, are poorly characterized. Therefore, we compared the age and growth of bigheaded carp for four low-density navigation pools of the Mississippi River near the invasion front with one long-established high-density population, separated by LD19. To assess growth rates and age at maturity, we estimated consensus age from pectoral spines, postcleithra, and vertebrae from 1229 bigheaded carp, 30 ﬁsh per 50 mm size class. Length at age of capture and back-calculated length-at-age were used to create growth curves using VBGC analysis. The analysis indicated higher growth rates of bighead carp and silver carp upstream of Lock and Dam 19 than downstream. Comparison of growth models with a likelihood ratio test indicated differences in length at time zero between the pools for bighead carp but not for silver carp. However, maximum length and growth rates were signiﬁcantly different for silver carp upstream and downstream of the dam but not for bighead carp. These results will be used to better understand the population dynamics of this emerging population to inform control and containment actions.


Introduction
Fish age and growth are important tools for the management and conservation of fish species [1]. Many factors can influence growth rates (i.e., population density, water chemistry, water temperature, and prey abundance; [2][3][4]) and are often dependent on available food resources, competition, and habitat [5]. With the ability to grow quickly, reproduce copiously, and exploit available resources, invasive species can cause reduced growth rates in native species. Individuals at invasion fronts are often 'bigger and better' with high phenotypic plasticity [6]. These fish often display boldness and increased dispersal behavior [7], leading to competition with native species. Although more attention is given to the invasive species' impact on growth of native species, intraspecific growth effects of invasive species do occur [8].
Bighead carp, Hypophthalmichthys nobilis, and silver carp, H. molitrix, (bigheaded carps) are invasive and abundant planktivorous species introduced into the Mississippi River system in the USA in the early 1970s. Due to their mobility, high fecundity, and ability to quickly colonize new areas, increases in their density have been associated with the decline of native planktivore body condition [9] and their own body condition [10]. Body condition and growth are often closely correlated and behave similarly in response to

Fish Collection
Gill and trammel nets (7.0-12.7 cm bar mesh, 2.4-9.1 m deep × 91.4-182.9 m long), and hoop nets (5, 1.5-m diameter rings with 5 cm bar mesh in front and 3.8 cm bar mesh at cod end) were used to collect fish on the Upper Mississippi River in pools 16-20 from July 2015 to June 2017 ( Figure 1). Multiple sampling gears were used to obtain a robust sample over a wide range of sizes, which a single gear could not achieve. Pool 20 (downstream of Lock and Dam 19, river kilometers 552-586) has a high-density bigheaded carp population, relative to pools [16][17][18][19] (river kilometers 586-778; [17]). To represent the range of ages in the populations, we attempted to collect at least 30 fish from each 50 mm size bin (bighead carp = 600-1450 mm, silver carp = 350-1200 mm), but this was often not possible and resulted in unequal numbers across length bins. Due to low densities of bigheaded carp above LD19, samples were collected over multiple seasons to obtain robust sample sizes across length categories. Backwaters, side channels, and areas upstream of spillways were primary targets for sampling as informed by residency and movement of acoustically transmittered fish.

Aging
Identification of species, sex, total length (mm), and weight (g) were recorded for each fish in the laboratory. Three aging structures were obtained for consensus age estimates (postcleithra, principal pectoral fin ray, and vertebrae) to avoid biases associated with spatial-specific lumen formation in structures between the species. Specifically, pectoral fin ray lumen formation and annulus obstruction is more prevalent in bigheaded carp in high-density areas, but the opposite is true for postcleithra in bigheaded carp from the lowdensity locations (J. Lamer, Illinois Natural History Survey, unpublished data). Postcleithra (left) and first pectoral fin ray (left) were dissected with a knife, placed into a labeled coin envelope, and dried for 48 h, at approximately 32 degrees C, in a custom-made dryer. Both structures were taken from the left side of the fish to ensure consistency. All structures were sectioned using a low-speed Isomet saw (Buehler, Lake Bluff, Illinois). Postcleithra were sectioned at the middle of the curve on the transverse plane [18], while pectoral spines were sectioned directly above the condyle and the basal recess (2 sections from each structure, 500-µm each). Five vertebrae were removed using a hand or reciprocating saw, just posterior to the first two fused vertebrae of the spinal column, anterior to the dorsal fin. The vertebrae were then boiled for 20 s to remove residual flesh and allowed to air dry for 48 h, and one undamaged vertebra was sectioned twice (500-µm sections) through the nucleus. Each section was photographed at 5× magnification using a Leica S8 APO stereomicroscope and a Leica MC170 HD microscope camera. Two readers recorded age estimates, and if disagreement between two readers occurred, structures were looked at in concert and an agreement reached. If disagreement between readers persisted, a third reader was used to resolve the age. One year was added to assigned age if the fish was caught during winter months, and no annuli was seen near the outer edge.

Aging
Identification of species, sex, total length (mm), and weight (g) were recorded each fish in the laboratory. Three aging structures were obtained for consensus estimates (postcleithra, principal pectoral fin ray, and vertebrae) to avoid biases associa with spatial-specific lumen formation in structures between the species. Specifica pectoral fin ray lumen formation and annulus obstruction is more prevalent in bighea carp in high-density areas, but the opposite is true for postcleithra in bigheaded carp fr the low-density locations (J. Lamer, Illinois Natural History Survey, unpublished da Postcleithra (left) and first pectoral fin ray (left) were dissected with a knife, placed in labeled coin envelope, and dried for 48 h, at approximately 32 degrees C, in a cust made dryer. Both structures were taken from the left side of the fish to ensure consisten All structures were sectioned using a low-speed Isomet saw (Buehler, Lake Bluff, Illino Despite our attempts to catch smaller individuals in gill nets (7 cm, 7.6 cm, and 8.9 cm bar mesh) and electrofishing, the effort to sample bigheaded carp was biased towards catching larger older fish because the primary effort was with large mesh nets (10.2 cm and 12.7 cm bar mesh gill nets) for removal by contract fishers for control purposes. To avoid sampling bias when creating growth curves, a stratified random sample of 30 fish per each 50 mm size class was used. Due to the high annual recruitment variability observed in these species [19,20], it is likely that many of these smaller size classes were not present or not very abundant in the population during our sampling. Therefore, to anchor our growth curves and to bolster our sample sizes for length-at-age for smaller age classes of silver carp (less than 600 mm) and bighead carp (less than 800 mm), we used vertebral cross sections to back-calculate length-at-age using the Dahl-Lea method. The Dahl-Lea method assumes that fish length is directly proportional to scale (vertebrae) radius [21]. We used vertebrae cross sections from 130 individuals based on the clarity of annuli, lack of central lumen, and radial shape. We modified the Dahl-Lea formula for use with vertebrae, and it is as follows: where L i is the estimated length at ith age, V i is the radius to the ith annulus, V C is the total vertebrae radius at time of capture, and L C is the total length of the fish at time of capture. Vertebral annulus radii measurements, in µm, were recorded in Leica Application Suite software (Leica Microsystems, Wetzlar, Germany). Only back-calculated lengths up to age four were used for analysis to avoid the potential effects of Lee's Phenomenon [22], and no fish were used to produce multiple length-age estimates (e.g., if a fish was used to estimate length at age 4 with the Dahl-Lea method, it could not be used to estimate length at age 3, 2, or 1) to avoid bias from the dependent nature of estimates from the same fish [23].

Statistical Analysis
Mean length-at-age was calculated for each species collected from each pool and compared at each age using an ANOVA in R [24]. Residual plots were used to test for normality. A Tukey's Honestly Significant Difference (HSD) test was used to determine which pools differed at each age. Growth was modeled using the von Bertalanffy model implemented by the FSA [25] and AICcmodavg [26] packages in R [24] to create growth curves and statistically compare estimated parameters among the pools: where l is the length at time t, L ∞ is the average maximum length, K is the growth coefficient, and t 0 is the theoretical time at which l t = 0 [27]. Using the methods outlined by Ogle [28], starting values were found for the model parameters L ∞ , K, and t 0 for each pool separately, using length-at-age data via polynomial regression. These were then fit to the general von Bertalanffy model as well as two and single parameter models to find the best fit. A likelihood ratio test from the FSA package [25] was used to compare models to determine which parameters differed by pool.

Fish Collection and Aging
Postcleithra and pectoral spines were obtained from 1203 fish, including 501 bighead carp and 702 silver carp from Pools 16 to 20 (Table 1). Of the bighead carp sampled, 55% were female and 45% were male, and for the silver carp sampled, 65% were female and 35% male. Pool 16 is at the leading edge of the invasion front, and fish were difficult to ob from this pool, resulting in lower sample sizes of both species compared to pools 17 (Figures 2 and 3).  Pool 20 sample sizes were also lower, because the majority of fish sampled were of similar size and fell into a narrow length range (650-750 mm). Complete age agreement between readers for all structures was 70.3%, and 98.3% agreement ±1 year. Agreement ±1 year for pectoral ray and postcleithra, pectoral ray and vertebrae, and postcleithra and vertebrae were 99.4%, 100%, and 92.5% respectively. Moreover, to estimate growth for fish younger than age 4, mean length-at-age was determined from back-calculated length-at-age from vertebrate increments (n = 128 fish: n = 67 bighead carp, n = 61 silver carp; Table 1). A combination of estimated (Dahl-Lea) and observed length-at-age were used for ages 3 and 4. Only estimated length-at-age was used for ages 1 and 2.  Pool 20 sample sizes were also lower, because the majority of fish sampled wer similar size and fell into a narrow length range (650-750 mm). Complete age agreem between readers for all structures was 70.3%, and 98.3% agreement ±1 year. Agreem ±1 year for pectoral ray and postcleithra, pectoral ray and vertebrae, and postcleithra a vertebrae were 99.4%, 100%, and 92.5% respectively. Moreover, to estimate growth fish younger than age 4, mean length-at-age was determined from back-calculated leng at-age from vertebrate increments (n = 128 fish: n = 67 bighead carp, n = 61 silver ca Table 1). A combination of estimated (Dahl-Lea) and observed length-at-age were u for ages 3 and 4. Only estimated length-at-age was used for ages 1 and 2.

Mean Length-at-Age
Mean length-at-age after age five for both species was significantly lower (α = 0 for Pool 20 populations (high-density) compared to pools 16-19 (low-density; Table  The maximum age for bighead carp collected in Pool 20 was 9 years, and 11, 11, 12, a 12 years for pools 16-19, respectively. The maximum age for silver carp collected in P 20 was 7 years, and 8, 8, 9, and 10 years for pools 16-19, respectively. In pools 16-18, most prominent age class of bigheaded carps present was 6-8 years. In Pool 19, this shif to 5-7 years, and in pool 20, the most common age was 5 years. At age five, bighead c upstream of LD19 were an average estimated length of 1056 (±35) mm versus 782 (±

Von Bertalanffy Growth Curves
Comparison of potential growth models by the likelihood ratio test indicated a best fit model of L ∞ and t 0 (4 df; α = 0.01) for bighead carp, and L ∞ (4 df; α = 0.005) for silver carp. Residual plots were examined to verify fit of growth models.  (Figure 4). Bighead carp also exhibited different lengths at t 0 between pools 16-18, Pool 19, and Pool 20 (Table 3).

Discussion
Differences in growth rates upstream and downstream of Lock and Dam 19 for bighead carp and silver carp indicate a potential density-dependent growth effect downstream. The growth curves suggest that bighead carp have a similar age at reproduction both upstream and downstream of Lock and Dam 19 but exhibit greater growth upstream of LD19. Age-at-reproduction is indicated by a decrease in the growth

Discussion
Differences in growth rates upstream and downstream of Lock and Dam 19 for bighead carp and silver carp indicate a potential density-dependent growth effect downstream. The growth curves suggest that bighead carp have a similar age at reproduction both upstream and downstream of Lock and Dam 19 but exhibit greater growth upstream of LD19. Age-at-reproduction is indicated by a decrease in the growth rate as the growth curve begins to reach an asymptote [29]. Silver carp growth rates began to decrease and reach the maximum asymptote 1-2 years earlier downstream of Lock and Dam 19 than upstream. Resource limitation or early onset of reproduction below LD19 may explain this difference. The lack of resource limitation upstream of Lock and Dam 19 may mask the shift to reproduction in the growth curves [29]. Supporting our findings that suggest densitydependent intraspecific competition below LD19, a recent study by Coulter et al. [10] found that the bigheaded carp's condition decreased as density increased in the Illinois River.
In newly established populations, where resources are not limiting, high growth rates and early maturation are common [13,30]. A similar trend can be seen with African jewelfish in the Everglades. Populations from the invasion front had higher fitness, better body condition, and higher reproductive investment than the already established populations [31]. With reduced competition, individuals on the leading edge of an expanding population can trigger an increase in reproductive investment in order to establish the population quickly [32]. If the population of bigheaded carp reaches numbers in which interspecific competition occurs frequently, native populations of filter feeders and juveniles of other species may be adversely affected by diminished food resources.
The population of bigheaded carp upstream of Lock and Dam 19 has significantly greater mean length-at-age, growth rate, and larger sized individuals than the population in Pool 20 downstream of the dam. Consistent with our findings, increased growth rates of low-density bigheaded carp populations were also observed in newly established silver carp in the middle Mississippi River in 2003 [13]. They found silver carp to average 317.7 mm of growth in the first year and reach sexual maturation at 2-3 years. Lampo (Illinois Department of Natural Resources, unpublished data) found that silver carp in the high-density population of the LaGrange reach of the Illinois River were only 80-100 mm at the end of the first year of growth. Age 1 silver carp in our study averaged 216 mm in the pools upstream of the dam, intermediate of the two studies. However, these estimates are similar to length-at-capture of young-of-year bigheaded carp collected in 2016 from Pool 18 and 19, Kentucky and Barkley Lakes, and Gobindsagar Reservoir, India at the end of the first year of growth [13,[33][34][35].
The frequency of hybrids in the population may also play a role in differing growth rates. Morphologically, later generation bighead carp and silver carp hybrids are very difficult to distinguish from each parental species. We were able to filter out early generation hybrids from our sample morphologically, indicated by twisted gill rakers [36]. Lamer et al. [36] found that early generation hybrids of bigheaded carp have a much lower body condition than that of parental species or later generation hybrids, which could also relate to lower growth rates. However, Lamer et al. (2015) found that early generation hybrids were more abundant at the leading edge of populations; thus, the growth effects we found (increased growth at the leading edge relative to established populations) would be contradictory to the effect suggested by Lamer et al. [17]. Had we not filtered out early generation hybrids, our results may not have been contradictory to Lamer et al. 2015.
Low densities of bigheaded carp above Lock and Dam 19 led to difficulties obtaining sufficient sample sizes in a single season. Samples were collected in an even effort throughout seasons each year to obtain robust samples. This may have resulted in increased variability in length-at-age estimates among individuals. Low sample sizes from Pool 16 and the lack of smaller individuals and underrepresented age classes may slightly bias growth curves. Our estimates also may be biased by the incorporation of back-calculated lengths.
Growth rates among pools upstream of Lock and Dam 19 were similar. This similarity might be explained by similar low densities and high longitudinal connectivity among pools. Lock and dams 16, 17, and 18 reach open river conditions frequently to control water levels in these pools [37]. During open river conditions, bigheaded carps can travel upstream between pools through dam spillways and open main channel gates rather than through lock chambers. Although many fish have a strong fidelity to a particular reach, tracking of telemetered bigheaded carp has determined that interpool movement in the reach between LD 15 and LD19 is common during high water conditions [38].
Density differences might not be the only factor contributing to growth differences upstream and downstream of LD19. The types of aquatic areas available for bigheaded carps upstream and downstream of the dam also differ. The pools upstream of Lock and Dam 19 have a greater proportion of off-channel habitats (i.e., backwaters, impounded habitats, and tributary mouths) and a lower proportion of main and side channel habitats as compared to Pool 20. For example, Pool 20 aquatic areas consist of 80.4% main channel, 19% side channel, and only 0.6% backwater. That is compared to a much smaller percent of main channel in Pool 19 (29.3%), less side channel habitat (17.4%), and more backwater (5.6%) and impounded habitat (47.7%) [39]. These off-channel habitats have lower flows and potentially greater primary and secondary productivity than channel habitats and, thus, might be more energetically favorable for growth of bigheaded carp. Lock and Dam 19 was designed to generate hydroelectric power with a constant water level elevation in the downstream 32 km of the pool to maintain minimum hydraulic head across the dam [40]. This constant water level and low flow creates a lacustrine environment upstream of the dam, which in turn, supports higher zooplankton and phytoplankton densities [41], the primary food source of bigheaded carp.
Length  [14,42]. These data suggest that increased growth rates upstream of Lock and Dam 19 are not driven by genetic factors, but rather by lower densities and potentially differences in habitat suitability. Better food and habitat conditions above LD19 as compared to below suggests that if bigheaded carp continue to immigrate and reproduce upstream of Lock and Dam 19, Pool 19 may be able to support a larger population than in Pool 20 before individual body condition begins to decline from density-dependent effects.
Bigheaded carp in pools 16-19 are reaching early maturation [43], and individuals are living well over eight years, similar to what is observed in their native range [44]. Abundant food and quality (low flow) habitat might be allowing these fish to use resources for gonadal growth and somatic growth simultaneously, continuing to grow at a high rate, while still being able to reproduce [45]. Lenaerts [43] found a single bighead carp could produce up to 3.2 million eggs per batch spawn. With early maturation, egg volume increasing exponentially with fish length [46], and large numbers of eggs spawned, the invasion front of bigheaded carp can produce large numbers of offspring quickly. Without control methods, the population above LD19 could increase quickly, increasing the risk of upstream expansion in the Mississippi River.
The growth differences between the low-density populations upstream of Lock and Dam 19 and the high-density populations downstream suggest that growth might be a good indicator of the effectiveness of management efforts to reduce populations in recently established, low-density populations. Current management efforts include contracted harvest; however, in the absence of reliable indicators of abundance, it is difficult to gauge the effectiveness of current harvest efforts and how they need to be scaled to achieve zero to negative population growth. Growth could act as a surrogate for general abundance. Observing growth rates over time could indicate changes in abundance of fish in this leading edge. For instance, if growth would negatively deviate from its current level over time, this may indicate that abundance is increasing and more intensive efforts to decrease populations are needed. Without adequate control, bigheaded carps may affect native filter feeding fish in the Upper Mississippi River, leading to decreased body condition and abundance, as observed in the Illinois River [9,47]. However, one might anticipate that intraspecific density-dependent growth effects, as indicated by reduced growth, would occur before interspecific effects. Thus, monitoring growth would also provide a tool to assess the likelihood of interspecific effects.
Early population dynamics data are essential to assess mortality, maturity, year class strength, and recruitment rates. These vital rate estimates can also be used to track changes over time and help guide ongoing contracted harvest in the Upper Mississippi River. Specifically, this information is necessary to parameterize population models such as the Spatially Explicit Asian Carp Population model (SEICarP), which can be used in scenario planning to evaluate alternative management actions such as contract removal and deterrent placement [48]. These population models also inform managers as to where and at what intensity management actions are necessary to achieve the desired goals. This is the first age and growth information on bigheaded carp from the invasion front in the Upper Mississippi River and will be useful in developing a SEICarP-like model for this important region at the leading edge of the invasion.