3.1 Predation Risk Layers

Catch was variable between stocking events and years for several predator taxa (Tables 2 and 3). For example, we handled 37 < preferred length Hybrid Striped Bass during Walleye stocking in 2017 but only 9 in 2018. Similarly, we handled 0 < preferred length Hybrid Striped Bass during White Bass stocking in 2017 and 23 in 2018. The temporal variability in catch may be related to species-specific movement patterns. Various species have been noted to move near or away from shore to spawning locations (Beck and Willis 2000; Jackson and Hightower 2001), seek out suitable water quality (Beck and Willis 2000; Farquhar and Gutreuter 1989; Henesy, Goetz, and Mullican 2022; Prchalová et al. 2009; Zale et al. 1990) and appropriate prey (Brandt 1980; Olson, Guy, and Koupal 2007). In addition, Walleye and Moronids in this study could be vertically separated due to depth preferences (Lincoln, Aday, and Rice 2016; Olson, Guy, and Koupal 2007; Van Den Avyle et al. 1983; Willis, Paukert, and Blackwell 2002). Previous sampling on Lake McConaughy also noted differences in the spatial distribution of different predators, though these differences were noted between seasons (Schall, Schoenebeck, and Koupal 2019) rather than within the shorter time frame of our collections within each year. The observed spatial and temporal variability in our study amplifies the need to develop time-specific predator maps in this system; similar differences in predator distributions may be found elsewhere (Specziár, György, and Erős 2013).

We collected food habits from 472 and 186 predators with food in their stomachs during Walleye and White Bass stocking, respectively (Tables 2 and 3). Every length category of predators was observed consuming either Walleye or White Bass fingerlings, except ≥ the preferred length of Channel Catfish (Tables 2 and 3). For most predator length categories, O_i was inconsistent between years and between the two stocking events (Tables 2 and 3). The O_i was higher for stocking events in 2018 compared to 2017. The O_i values were consistently higher during the Walleye stocking period than during the White Bass stocking period. Hybrid Striped Bass, Smallmouth Bass and White Bass that were < preferred length had the highest O_i for both stockings. The only occurrences of a larger length category having a higher O_i than the smaller one were Smallmouth Bass during the 2017 Walleye stocking and Walleye during the 2018 White Bass stocking. In both instances, the number of larger predators handled was low, which could have influenced the overall calculation of O_i. Taxa with a higher O_i could be selecting for stocked fingerlings when they are introduced, while taxa with a lower O_i could have a preferred food source (i.e., Alewife) that limits fingerling consumption or could be exhibiting opportunistic feeding strategies (Bowen 1996). However, Bowen (1996) warns against over-inferring the importance of O_i results as encounter probability could differ among predator taxa. The variety of species found consuming stocked fingerlings confirms the need to evaluate all possible predators within a system to determine which ones are potentially impacting hatchery products.

The predators that consumed both species of stocked fish at the highest rates per individual were Hybrid Striped Bass, Smallmouth Bass and White Bass (Tables 2 and 3). Which predators are riskier to stocked fishes may relate to a number of factors, including characteristics of the predators themselves (Graeb et al. 2005; Mihalitsis and Bellwood 2017) and their ability to feed on a variety of prey (Gilliland 1981; Schake, Dembkowski, and Wuellner 2014; Schultz et al. 2002; Weidel, Josephson, and Krueger 2000; Willis, Paukert, and Blackwell 2002) or specialise on a specific prey (i.e., stocked fish) (Scheibel et al. 2016). Interestingly, all three top predators are ecomorphologically similar (Edwards, Gebhart, and Maughan 1983; Kilpatrick 2004; Lueckenhoff 2011); body morphology can influence predation rates as it relates to swimming speed and capture efficiency (Mihalitsis and Bellwood 2017; Videler 1993; Videler and Wardle 1991; Wolter and Arlinghaus 2003). However, various ecological aspects of the three species differ. Hybrid Striped Bass and White Bass most often swim in schools within the epipelagic zone (Hamilton and Nelson 1984; Kilpatrick 2004; McNaught and Hasler 1961), whereas Smallmouth Bass often concentrate around habitat structure but do not shoal (Edwards, Gebhart, and Maughan 1983; Miranda, Lakin, and Faucheux 2021). Schooling piscivores, such as Hybrid Striped Bass and White Bass, could have higher capture rates than individual predators when encountering a large group of prey fish (Major 1978). In contrast, solitary predators, like Walleye or Smallmouth Bass, are better suited for capturing individual prey that have broken away from the larger group (Major 1978; Winemiller and Taylor 1987). In addition to feeding habits, the habitat of the predators may also influence predation rates. For example, littoral species such as Smallmouth Bass (Edwards, Gebhart, and Maughan 1983) may already be present around sites where stocking occurs. Other species, such as White Bass (Devine and Shiozawa 1984) and Walleye (Kelso 1978), may move into the stocking locations during the crepuscular period. Waterbody-specific consideration of which predators are present and what habitats they occupy could help determine where to stock hatchery products.

Relative predation risk for both stocked species was three times higher among the smaller (< preferred) categories when compared to the larger (≥ preferred) categories (Tables 2 and 3). Larger predator food habits were largely composed of Alewife (between 150 and 170 mm TL), while smaller categories, when not consuming stocked fingerlings, contained more macroinvertebrates than other fish (see appendix I in Farrier 2023). The pattern of increased consumption of stocked fingerlings by smaller predators may be related to the optimum foraging theory, where energy gained by consuming a prey item is expected to be higher than energy expended (MacArthur and Pianka 1966). While larger prey will provide more energy than smaller prey, larger prey are often more difficult to capture (Boisclair and Leggett 1989). Mouth gape (Schmitt and Holbrook 1984; Slaughter and Jacobson 2008) and swimming speed of predators both increase as fish TL increases, which contributes to higher prey capture efficiency (Domenici 2001; Hammer 1995; Webb 1976; Wolter and Arlinghaus 2003). Smaller predators might not be large enough to capture and consume the larger Alewife that are available during fingerling stockings (Schall 2019). Compared to larger predators, smaller predators do not have as high energy demands, but their metabolism rates are higher on a per gram basis (Arim, Borthagaray, and Giacomini 2016). Thus, to meet high metabolic energy demand, the increase in capture efficiency by smaller predators on naïve stocked fingerlings could make these fish a better food source in relation to the optimum foraging theory by decreasing capture time and allowing for higher energy utilisation (MacArthur and Pianka 1966).

The ecology of each prey (stocked) species, such as movement rates, may influence their relative predation. For instance, if hatchery fingerlings remain near the location they are stocked, it could influence the number of potential encounters with predators. Parsons and Pereira (1997) found that fingerling Walleye stocked in the fall often stayed near (< 1–4 km over multiple lakes) stocking locations for several years. By remaining in the same location, the stocked fish could be exhibiting shoaling (Walleye; Pratt and Fox 2001) or schooling (White Bass; Hamilton and Nelson 1984) behaviour, which would leave the fingerlings in larger groups that would favour predators that employed capture techniques more effective at exploiting grouped prey. Similarly, in Iowa, large Walleye fingerings (> 200 mm TL) dispersed more than 400 m within 1 day post-stocking and moved nearly 1400 m within 13 days (Weber et al. 2020). If movement is more individualised, this behaviour could make these fish more vulnerable to some predators (Winemiller and Taylor 1987). Unfortunately, dispersal of smaller Walleye and White Bass fingerlings (< 40 mm TL) is not well understood and needs further investigation. Body size, species preference towards pelagic or demersal habitats (Kilpatrick 2004; Pratt and Fox 2001) and wind (direction and speed) could further influence dispersal from stocking locations (Ellison and Franzin 1992). Thus, consideration into how stocked species disperse once released could be important to the identification of potential stocking locations and development of stocking maps (Radinger and Wolter 2014; Weberg et al. 2020).

Spatial distribution of predicted predation risk varied across Lake McConaughy between Walleye and White Bass stockings (Figures 3 and 4). During Walleye stocking, higher predation risk was concentrated in the middle and lower part of the reservoir, while higher predation risk for stocked White Bass was more concentrated in the upper portion. Changes in predicted predation highlights the need for individualised plans for different stocked species within the same reservoir. Often, stocking plans only consider the specific species being stocked, and adjustments focus on the rate (number per hectare) and developmental stage (size) at the time of stocking (Fielder 1992). These blanket strategies are deployed without considering how stocked species and predator needs (e.g., habitat, water quality, food, etc.) vary and could limit stocking success (Agostinho et al. 2010; Molony et al. 2003). For example, adult White Bass can be found near and away from shore, depending on temperature, cloud cover, precipitation and moon phase (Beck and Willis 2000; Willis, Paukert, and Blackwell 2002). If the timing of stockings could be manipulated to avoid times when more predators are near shore, predation on stocked fishes may be reduced (Buckmeier, Betsill, and Schlechte 2005).

A major limitation of the predation layer was the lack of available adult predator distributions that were collected during each stocking event. Stockings of both Walleye and White Bass typically occur annually in early and late June, respectively. We used the best historical spatial distribution data available, but it was collected in May and July of 2015 and 2016, respectively. As a result, we introduced variation into the stocking maps (different months and years from food habit collections) that could limit their overall accuracy. Future studies should focus on collecting predator spatial distributions from specific stocking timeframes.

3.2 Zooplankton Availability Layers

The density of Calanoida and Cyclopoida spp. ranged from 9 to 241 individuals/L (mean ± standard error = 80 ± 10 individuals/L) during Walleye stocking and 7–154 individuals/L (65 ± 4.8 individuals/L) during White Bass stocking (see appendix 2 in Farrier 2023). Zooplankton density was consistently lower throughout the reservoir during White Bass stocking, just 20 days after Walleye stocking. However, similar spatial patterns were observed between both stocking events, with the highest densities observed around southcentral Lake McConaughy (Figures 5 and 6). Studies at other waterbodies have found similar relative spatial and temporal zooplankton density patterns (Bernot et al. 2004; Maline et al. 2011). The spatial differences in zooplankton abundances are often greater than temporal variation (Threlkeld 1983) and correlate to phytoplankton spatial distribution (Hart 1988; Striebel et al. 2012), which, in turn, is influenced by several abiotic (e.g., turbidity, wind direction and speed, water temperature) factors (George and Edwards 1976; Grobbelaar 1989; Gyllström et al. 2005; Shuman 1990; Winder and Sommer 2012). The moderate temporal decreases in zooplankton density observed in our study could have been influenced by Alewife (i.e., grazing) (O'Gorman, Mills, and DeGisi 1991), seasonal drought (Olds et al. 2014), temperature (Kelly et al. 2016) or seasonal succession patterns (Urabe 1989).

Predicted spatial distributions of zooplankton in Lake McConaughy suggest stocking in locations based on convenience could be detrimental to stocking success (Agostinho et al. 2010). Traditionally, stockings in this reservoir have occurred from the upwind side on high wind (> 16 km/h) days to avoid displacing fingerlings out of water as a result of heavy wave action (Darrol Eichner, personal communication). However, strong wind and wave action can also influence zooplankton distribution (George and Edwards 1976), which might result in stocked fish being released in areas with lower zooplankton densities on those windier days. If stockings could occur near higher abundances of zooplankton, even with high winds, survival and growth of stocked fish may be improved (Houde 1975). This premise is further supported with the match-mismatch hypothesis, which suggests that high growth and survival are expected when young fish overlap spatially and temporally with favourable biotic and abiotic environmental conditions (Cushing 1975). Roseman et al. (2005) found larval Walleye, in mid-June, were more likely to be found in areas where densities of Calanoida, Cyclopoida and Cladocera were at least 40 individuals/L and had strong positive associations to areas with less transparent water and higher relative water temperatures. The observed positive associations to less transparent water and higher relative water temperatures (Roseman et al. 2005) could indicate the appropriateness of including a habitat metric in this conceptual model.

Within our study, the appropriateness of zooplankton as the preferred food for Walleye fingerlings could be scrutinised due to their length being close to the threshold when ontogenetic shifts to different prey items have been reported (Galarowicz, Adams, and Wahl 2006). While literature supports both species still consuming zooplankton at this developmental stage (Miller, Schoenebeck, and Koupal 2019; Perrion 2016; Uphoff et al. 2019; Woiak 2014), food habits are often specific to waterbodies and hatchery products. Further investigation into fingerling food habits post-stocking could provide clarity on the appropriateness of zooplankton as a food source in Lake McConaughy.

3.3 Habitat Layers

Secchi disc depth readings during Walleye stocking ranged from 70 to 203 cm (131 ± 3.9 cm). The reservoirs was generally less transparent on the west end and became more transparent closer to the dam; the south shore was generally more turbid than the north shore. Overall, the observed transparency pattern followed similar limnological patterns compared to other irrigation reservoir (Olds et al. 2011; Soares et al. 2008). Most Secchi depth readings fell within the optimal range for Walleye (100–300 cm; McMahon, Terrell, and Nelson 1984) except in portions of the west end (< 100 cm). The predicted transparency layer provides some insight into whether historic stocking locations (Figure 1) are the most suitable to support stocking success (Figure 7). Stockings that have occurred from shore in the upper portion of the reservoir were in areas where transparency values were higher than the optimal range (McMahon, Terrell, and Nelson 1984). Adequate turbidity is necessary for juvenile Walleye retinal tapetum development (Braekevelt, Mclntyre, and Ward 1989; Bristow, Summerfelt, and Clayton 1996) and could serve as a form of cover from sight predators such as Smallmouth Bass (Carter et al. 2010). If turbidity is excessive, the stocked fingerlings could have negative feeding performance (Bulkowski and Meade 1983). Feeding efficiency has been linked to both the short- and long-term survival of stocked fish (Brown, Davidson, and Laland 2003; Ersbak and Haase 1983; Hansen et al. 2022; Vandenbyllaardt et al. 1991).

Bank slopes within Lake McConaughy ranged from 0.2%–5.7% (2% ± 0.1%). Shallow flats with low percent change were distributed throughout the reservoir (Figure 8), and some historic stocking locations (Figure 1) fell outside of the optimal range. If fish are stocked outside of suitable nursery habitat (shallow littoral areas), they may be more vulnerable to predation and have reduced growth (Beck et al. 2001; Brosse, Grossman, and Lek 2007). Research in four Czech Republic reservoirs found that combining substrate and slope together was the most accurate way to identify gentle sloping, sandy areas (Šmejkal et al. 2014), providing evidence of the appropriateness of using bank slope to describe habitat. At Lake McConaughy, White Bass stock contribution has been lower than Walleye stock contribution (Rowles 2019), which could be related to the higher percent slope found at past stocking locations (boat ramps). Stocking White Bass in areas with lower percent slope could positively impact their survival but needs validation in future studies. While Secchi depth and bank slope were selected for this study, other variables such as water temperature (Peake 1999) or vegetation (Hanson and Margenau 1992) could be more appropriate for other species. Consideration of habitat variables specific to the species being stocked is crucial when developing the habitat layer within the conceptual model.

3.4 Stocking Maps

The Walleye stocking map (Figure 9) identified the top 12.8% of the total area of Lake McConaughy that would be expected to support higher post-stocking survival, and the White Bass stocking map (Figure 10) identified the top 13.9%. The largest continuous area of optimal stocking locations for both models was concentrated in the middle of the reservoir near the south shoreline. However, there was some variation between the two layers, with the White Bass optimal stocking area on the south shoreline extending further west than that of Walleye. In addition, the Walleye layer identified more locations on the north shoreline than that of White Bass. While some similarities could be related to the overlapping parameters (predation and zooplankton) used, it is surprising the two predicted layers were largely similar, considering the differences observed in each map layer of the conceptual model.

Time and personnel to collect the encompassing amounts of data to model stocking layers using the conceptual model will always be a concern. Fortunately, parameters like zooplankton, habitat and water temperature are often easily investigated by fishery managers. However, stocked fish interaction, both direct and indirect, with other species should be considered when selecting conceptual model parameters. For example, Fayram, Hansen, and Ehlinger (2005) found Largemouth Bass (M. salmoides) predation not only limited the survival of stocked Walleyes, but by stocking Walleyes, Largemouth Bass populations increased. By selecting multiple parameters to model stocking layers with the conceptual model, it leaves each parameter open to interpretation. Managers can have flexibility based on characteristics of specific waterbodies and use only map layers thought to be influential at that location or they could elect to use only specific layers instead of the derived stocking layer. For example, managers could decide predation is the only parameter they are worried about; therefore, only use the predation layer when making stocking decisions.

Determining the long-term accuracy of each stocking model could help clarify if the extra time and effort to collect and analyse data for each individual layer comprising the composite model was justified. Without validation of each parameter within Lake McConaughy, it will be difficult to determine if using the conceptual model increases the survival of stocked Walleye and White Bass. Ideally, validation of the stocking models could be implemented at Lake McConaughy to determine stocking success related to optimal and suboptimal stocking locations. However, the economic pressures to be efficient with hatchery products could supersede the desire to validate these layers, as the logical development of the final stocking layer could provide enough evidence to fisheries administration. An evaluation might be needed at a different reservoir where the risk of a missing year class does not have the same ramifications as Lake McConaughy.

Despite the lack of current validation, use of the conceptual model to identify important abiotic and biotic factors that could contribute to the mortality of stocked fish can be an improved paradigm for approaching stocking efforts. We recognise there are limitations in this study that will need consideration before developing stocking models for other waterbodies. However, the concept that applies a multitude of factors to the stocking decision-making process using logical interpretations of preferred abiotic and biotic conditions has the potential to improve survival and help managers reach relative abundance goals in the waterbodies they oversee.