Environmental DNA Is Effective at Detecting a Salmonid Ectoparasite: A Statewide Survey of the Distribution and Genetic Diversity of the Gill Louse Salmincola californiensis in Colorado, USA

Abstract

ABSTRACT

Invasive species present considerable threats to native biodiversity by disrupting ecosystem processes. The gill louse Salmincola californiensis is a copepod that parasitizes Oncorhynchus species, which includes ecologically, commercially, and recreationally important fishes. As S. californiensis expands its geographic range, there are concerns that conservation efforts focused on Rocky Mountain cutthroat trout Oncorhynchus virginalis may be thwarted. To address these concerns, we (1) assessed upstream range expansions of S. californiensis from known infected waters; (2) compared the efficacy of environmental DNA (eDNA) sampling to traditional sampling methods in detecting S. californiensis; (3) evaluated S. californiensis population genetic structure using DNA barcoding; and (4) assessed gill lice occupancy and detection probabilities using occupancy modeling. We compared the success of detecting S. californiensis using electrofishing versus eDNA sampling methods at 48 sites throughout the state of Colorado. We detected gill lice at 17 sites via electrofishing, and at 10 sites using eDNA. For DNA barcoding, we collected 58 lice at 11 sampling localities and sequenced the mitochondrial cytochrome c oxidase subunit 1 for species identification, to assess genetic diversity across Colorado, and to estimate divergence times. Salmincola californiensis was the only species of gill lice we detected. Divergence time estimates show that it is possible that the highly divergent gill lice lineages originally co‐invaded with cutthroat trout. However, the presence of the most widespread parasite haplotypes across multiple drainages presents a phylogeographic pattern consistent with fish stocking facilitating its range expansion. Occupancy modeling suggests that fluvial processes, temperature, and UV impact gill lice detection using eDNA. We conclude that eDNA is effective at detecting gill lice presence in a system, perhaps best used in conjunction with electrofishing methods for early detection, that gill lice continue to expand their range in Colorado, and that continued monitoring will be an important component of future management efforts.

Introduction

The introduction of invasive species into new environments negatively affects native species in numerous ways, such as competing with native species for limited resources, feeding upon or parasitizing native species, or modifying the physical structure and/or chemical properties of natural habitats (Wilcove et al. 1998; Simberloff 2010; Russell 2012; Gallardo et al. 2016; Chinchio et al. 2020). Moreover, as invasive species become established, some species are capable of hybridizing with native species, contributing to population declines and biodiversity loss, or otherwise impacting ecological processes (Mooney and Cleland 2001; Simberloff 2010; Bellard et al. 2021). Consequently, invasive species are considered to be one of the greatest threats to Earth's biodiversity (Diamond 1984; Gurevitch and Padilla 2004; Bellard et al. 2016).

In addition to facilitating ecosystem imbalances, biological invasions that impact economically important ecosystems can cause substantial financial losses because of the acceleration of population declines in species targeted for harvest, damaged infrastructure, disruptions to tourism, agricultural impacts, and human health effects (Andersen et al. 2004; Pyšek and Richardson 2010; Marbuah et al. 2014; Gallardo et al. 2016). Sometimes, invasive species influence ecosystems via a phenomenon described by the invasional meltdown hypothesis, which states that interactions between invasive species help facilitate subsequent invasions (Green et al. 2011), which can include hitch-hiking parasites. Co-invading parasites have three probable origins: (1) the native range of the invasive species, (2) intermediate locations between the native range and the introduced range of the invading species, and (3) the introduced range of the invasive species (Lymbery et al. 2014; Chalkowski et al. 2018). High proportions of co-invading parasites switch to native hosts (Lymbery et al. 2014; Poulin 2017; Chalkowski et al. 2018), and such horizontally transmitted parasites are often more virulent due to the ecological naïveté of the new native host (Hatcher et al. 2012; Lymbery et al. 2014; Cable et al. 2017).

Newly investigated parasites can be challenging to categorize as native or non-native to an ecosystem. Consequently, they are often referred to as cryptogenic species (Carlton 1996). Cryptogenic parasites are relatively common, partially because human translocations of species often began before taxonomic surveys and species monitoring were regularly conducted, and partially because the taxonomy for many parasites is ambiguous or uncertain (Lymbery et al. 2014). Indicators that a cryptogenic parasite is non-native include low genetic diversity, phylogenetic similarities with foreign lineages, and a lack of overlap between the native host distribution and the parasite distribution (Lymbery et al. 2014). Conversely, native parasites tend to exhibit substantial population genetic structure and greater phylogenetic distances from related parasites distributed in other geographic areas.

The gill louse Salmincola californiensis (Crustacea: Lernaeopodidae; Dana, 1853) is a cryptogenic, ectoparasitic copepod that infects salmonid fishes belonging to the genus Oncorhynchus (Actinopterygii: Salmonidae). The life cycle of S. californiensis is described in detail elsewhere (Kabata and Cousens 1973; Ruiz et al. 2017; Boone and Quinlan 2019; Murphy, Gerth, and Arismendi 2020), but typically adult gill lice attach to the gills, opercula, and fin bases of host fish, feeding on blood, mucus, and tissues (Johnson et al. 2004). Larger fish are more susceptible to infection because they circulate more water over the gills than smaller fish do, increasing potential contact with copepodids (Black 1982; Neal et al. 2021). Infected fish suffer inflamed gill filaments, impaired swimming ability, increased fatigue, and potentially reduced immune responses in older fish, suggesting that S. californiensis impairs gas exchange, osmotic regulation, and other metabolic activities (Herron et al. 2018). This reduced metabolic function may then result in anemia, excessive blood clotting, stunted growth, reduced fecundity, weakened immune system function, reduced resistance to environmental stressors, or even host death (Pawaputanon 1980; Barndt and Stone 2003; Vigil et al. 2016). Previously infected fish are susceptible to re-infection, and may be even more susceptible than their naïve counterparts (Neal et al. 2021).

It is well-documented that several ecologically and economically important species of the salmonid genus Oncorhynchus (Appendix S1) are susceptible to S. californiensis infection (Kabata 1969; Nagasawa and Urawa 2002; Barndt and Stone 2003; Monzyk et al. 2015; Murphy, Gerth, and Arismendi 2020). Originally limited to the North Pacific Rim and its tributaries (Nagasawa 2023), S. californiensis has since expanded its geographic range eastward into the United States, reaching as far inland as New Jersey and West Virginia, presumably by way of infected hatchery fish transported from the Pacific Coast (Piasecki et al. 2004; Gunn 2012; Mullin and Reyda 2020).

In the state of Colorado, S. californiensis was first detected on a non-native salmonid in the early 1900s (Wilson 1909), and there have been more recent detections at several fish hatcheries, private aquaculture facilities, reservoirs, and rivers/streams from the 1990s to the present day (Gunn 2012; Vigil et al. 2016). In the last decade, S. californiensis has been detected in several drainages across Colorado and appears to be rapidly expanding across the state (Hargis et al. 2014; Vigil et al. 2016; Lepak et al. 2022). Of the aforementioned susceptible Oncorhynchus species, only the Rocky Mountain cutthroat trout Oncorhynchus virginalis is native to the state of Colorado, but rainbow trout Oncorhynchus mykiss and kokanee Oncorhynchus nerka have been widely introduced, providing a way for gill lice to co-invade new areas and potentially host-switch to native cutthroat trout. Thus, there are concerns that the expansion of S. californiensis into native cutthroat trout waters will hinder the success of decades-long cutthroat trout conservation efforts, particularly in areas upstream from reservoirs with known infections as fish (especially kokanee) migrate upstream to spawn. Improved survey tools may better facilitate early detections of the parasite, informing fisheries managers about its presence in a timely manner, thus allowing for better protections of native cutthroat trout populations.

Environmental DNA (eDNA) is genetic material that is shed by organisms into their environments (Ogram et al. 1987; Taberlet et al. 2018; Pawlowski et al. 2020), and has become a very useful tool in detecting invasive species, particularly in aquatic ecosystems (Rees et al. 2014; Goldberg et al. 2016; Ruppert et al. 2019; Sepulveda, Nelson, et al. 2020; Sahu et al. 2023; Picq et al. 2024). It is plausible that an eDNA approach can be an effective way of detecting small-bodied gill lice in waterbodies without the need to directly interact with them or their fish hosts. In fact, recent work (Katz et al. 2023) has shown that eDNA methods are effective at detecting the gill louse Salmincola edwardsii (Olsson, 1869), a congeneric relative of S. californiensis that infects brook trout Salvelinus fontinalis (Mitchill, 1814). However, eDNA methods are not failsafe and must be interpreted through the lens of potential pitfalls (Roussel et al. 2015; Larson et al. 2020; Farrell et al. 2021). For example, environmental conditions can impact the longevity of eDNA in aquatic systems (Barnes et al. 2014; Rees et al. 2014; Brandão-Dias et al. 2023). Elevated temperatures have been shown to speed up the process of DNA denaturation and hydrolysis (Pilliod et al. 2014). However, if eDNA is collected before it decays, species can be detected without any need for direct interaction with, or observation of, the target species (Adams et al. 2019; Larson et al. 2020). Moreover, it remains unclear whether eDNA methods can reliably detect S. californiensis individuals considering their small size, as well as the fact that crustaceans tend to shed less eDNA than other taxonomic groups (Allan et al. 2021; Crane et al. 2021; Korbel et al. 2024). Unfortunately, information on factors affecting eDNA shedding by S. californiensis is lacking, although it stands to reason that temperature would affect metabolic rates, and therefore shedding rates, as well as certain life cycle events, such as the release of eggs, would contribute to differences in eDNA shedding rates by gill lice. Nevertheless, implementing eDNA techniques may provide researchers with another tool to assist with conducting biological surveys when effective management depends on reliable detection methods (Peters et al. 2018; Ruppert et al. 2019; Bass et al. 2023). Additionally, imperfect detection of gill lice using eDNA can be accounted for by occupancy modeling, which accounts for false negatives in the field (Schmidt et al. 2013; Burian et al. 2021). Thus, eDNA shows promise for detecting S. californiensis given that eDNA methods have reliably detected S. edwardsii in Wisconsin (Katz et al. 2023; Johnson et al. 2024; Tetzlaff et al. 2024), and the salmon louse Lepeophtherirus salmonis in marine ecosystems and aquaculture facilities (Peters et al. 2018; Krolicka et al. 2022).

Direct comparisons between traditional sampling methods (e.g., electrofishing) and eDNA methods in gill lice surveys may provide useful information for fisheries managers to consider when planning and conducting surveys. Katz et al. (2023) concluded that eDNA is as effective as electrofishing in detecting the presence of S. edwardsii in brook trout populations. It remains to be seen if the same holds for detecting S. californiensis in western North American rivers. Traditional electrofishing methods, while effective for capturing fish, can be labor-intensive, stressful to individual fish, and limited by habitat accessibility or water depth. In contrast, eDNA sampling provides a broader, less invasive means of detecting species presence and may offer increased detection probabilities in difficult-to-access areas or low-density populations. Both methodological approaches can be susceptible to imperfect (or false negative) species detections, wherein a species is deemed to be absent when it is actually present at a site, but this can be accounted for using an occupancy modeling approach. Site occupancy modeling takes imperfect detections into consideration to assess the “true” presence of a species. In recent years, occupancy models in freshwater systems have been useful in understanding distribution patterns of common and imperiled native species (e.g., Falke et al. 2012; Groce et al. 2012), invasive species (e.g., Sepulveda 2018; Rogosch and Olden 2021), and aquatic pathogens (e.g., Schmidt et al. 2013; Tetzlaff et al. 2024), and thus should inform a better understanding of gill lice distributions in our study area.

In this study, we set out to (1) test the hypothesis of upstream range expansion of S. californiensis from known infected waters in Colorado using a tandem sampling approach that compared traditional physical sampling methods (e.g., electrofishing and subsequent visual inspections of captured fish to see if they carry gill lice) with non-invasive eDNA testing of water samples; (2) compare the efficacy of eDNA sampling to traditional electrofishing methods to detect S. californiensis in an area, specifically assessing whether eDNA offers advantages over electrofishing in terms of detection sensitivity, spatial coverage, and non-invasive collection; (3) evaluate population genetic structure among S. californiensis populations in Colorado and elsewhere using a DNA barcoding approach to assess whether gill lice exhibit a population structure consistent with human-mediated dispersal; and (4) employ an occupancy modeling approach to determine the effects of environmental factors, including water temperature and ultraviolet radiation index, on gill lice detection probability using eDNA.

Methods

Field Sampling

We sampled 48 sites within 19 river systems representing four major drainage basins of Colorado: the Arkansas River, Colorado River, Rio Grande, and South Platte River drainages (Table 1). The sampling sites we selected were based partially on results of previous survey information. Eleven rivers had prior records of the presence of S. californiensis (Table 1). The majority of prior surveys were conducted from 2013 to 2015 (Vigil et al. 2016), after a 2006 gill lice outbreak in Elevenmile Reservoir alerted Colorado Parks and Wildlife that gill lice were becoming more of a problem in the state. We selected the other sites based on proximity to known infested waters with native cutthroat trout populations upstream to better understand S. californiensis range expansion among suitable cutthroat trout habitats in Colorado. For this study, each site was defined as a 100-m stretch of river, and we thoroughly sampled all available habitat within that stretch, working our way upstream via electroshocking, after taking 10 water samples for eDNA testing so as not to collect disturbed sediments (see Sections 2.2 and 2.3 for more on our electrofishing and eDNA methods). Within each river, we sampled three sites (designated as lower, middle, and upper) located 2-km apart whenever possible (but occasionally accessibility and other logistics prevented this). If the accessible portion of a river was too short for the 2-km distance between sites, the accessible river section was divided into three equally sized sites. We sampled between September 2022 and October 2023.

TABLE 1 Sampling locations included in our statewide survey of gill lice in Colorado.

Location	Site	Lat/long	Drainage basin	Date	Gill lice detected	Mean volume H2O filtered (L)	Fish species sampled	Water temp. (°C)	UV Index
Prior	Electrofishing	eDNA
Arkansas River	Uppera	38.274 N, 104.650 W	Arkansas River	11/17/2022	Unknown	0	0/10	0.5 ± 0.2	Rainbow Trout, Brown Trout	10.3	3
Middlea	38.271 N, 104.648 W	Arkansas River	11/17/2022	Unknown	0	0/10	0.8 ± 0.1	Rainbow Trout, Brown Trout	10.7	3
Lowera	38.271 N, 104.642 W	Arkansas River	11/17/2022	Unknown	0	0/10	0.7 ± 0.3	Rainbow Trout, Brown Trout	10.7	3
Big Thompson River	Upperb	40.377 N, 105.485 W	South Platte River	11/22/2022	Present	1	0/10	0.8 ± 0.2	Rainbow Trout	0.0	3
Middleb	40.390 N, 105.463 W	South Platte River	11/22/2022	Present	0	2/10	0.7 ± 0.1	Rainbow Trout	0.3	2
Lowerb	40.397 N, 105.421 W	South Platte River	11/22/2022	Present	0	0/10	0.7 ± 0.2	Rainbow Trout	0.3	2
Blue River	Uppera	40.025 N, 106.385 W	Colorado River	9/10/2022	Present	5	0/10	0.5 ± 0.2	Rainbow Trout, Cutbow Trout, Brown Trout	5.9	6
Middlea	40.027 N, 106.385 W	Colorado River	9/10/2022	Present	1	10/10	0.3 ± 0.1	Rainbow Trout, Brown Trout	6.0	5
Lowera	40.032 N, 106.385 W	Colorado River	9/10/2022	Present	0	8/10	0.6 ± 0.3	Rainbow Trout, Brown Trout	6.0	5
Conejos River	Uppera	37.354 N, 106.526 W	Rio Grande	9/25/2022	Present	0	0/10	0.5 ± 0.1	Rainbow Trout, Brown Trout	21.3	6
Middlea	37.335 N, 106.474 W	Rio Grande	9/25/2022	Present	0	0/10	0.5 ± 0.1	Rainbow Trout, Brown Trout	21.2	5
Dolores River	Upperb	37.620 N, 108.067 W	Colorado River	10/13/2023	Unknown	0	0/10	1.8 ± 0.2	Brown Trout, Brook Trout, Rainbow Trout	—	—
Gunnison River	Upper	38.709 N, 106.851 W	Colorado River	12/03/2022	Present	n/a	4/10	0.8 ± 0.2	Rainbow Trout, Brown Trout	1.4	0
Middle	38.581 N, 106.923 W	Colorado River	12/03/2022	Present	n/a	0/10	0.7 ± 0.1	Rainbow Trout, Brown Trout	4.4	2
Lower	38.546 N, 106.951 W	Colorado River	12/03/2022	Present	n/a	5/10	1.0 ± 0.3	Rainbow Trout, Brown Trout	4.3	2
Henson Creek	Upperb	38.018 N, 107.335 W	Colorado River	9/10/2022	Absent	0	0/10	0.9 ± 0.3	Rainbow Trout, Cutthroat Trout	20.0	5
Middleb	38.021 N, 107.319 W	Colorado River	9/10/2022	Absent	0	0/10	0.8 ± 0.1	Rainbow Trout, Brook Trout	20.2	6
Lowerb	38.026 N, 107.315 W	Colorado River	7/21/2022	Absent	1	0/10	0.7 ± 0.2	Rainbow Trout, Brown Trout	20.5	4
Huerfano River	Upperb	37.702 N, 105.381 W	Arkansas River	12/20/2022	Unknown	0	0/10	1.2 ± 0.4	Rainbow Trout, Brown Trout	5.2	0
Middleb	37.705 N, 105.377 W	Arkansas River	12/20/2022	Unknown	0	1/10	1.0 ± 0.2	Rainbow Trout, Brown Trout	5.1	2
Lowerb	37.708 N, 105.373 W	Arkansas River	12/20/2022	Unknown	0	1/10	0.9 ± 0.2	Rainbow Trout, Brown Trout	5.1	2
Roan Creek	Upperb	39.560 N, 108.624 W	Colorado River	8/03/2022	Unknown	0	0/10	0.7 ± 0.1	Cutthroat Trout	23.2	6
Middleb	39.559 N, 108.623 W	Colorado River	8/03/2022	Unknown	0	0/10	1.3 ± 1.3	Cutthroat Trout	23.2	8
Lowerb	39.545 N, 108.598 W	Colorado River	8/03/2022	Unknown	0	0/10	1.0 ± 0.3	N/A	24.2	8
Roaring Fork River	Upperb	39.368 N, 107.047 W	Colorado River	10/31/2022	Present	0	0/10	0.8 ± 0.2	Rainbow Trout, Brown Trout	2.1	3
Middleb	39.378 N, 107.085 W	Colorado River	10/31/2022	Present	0	0/10	1.0 ± 0.1	Rainbow Trout, Brown Trout	2.5	2
Lowerb	39.414 N, 107.222 W	Colorado River	10/31/2022	Present	0	0/10	1.1 ± 0.2	Rainbow Trout, Brown Trout	2.3	3
Rio Blanco	Upperb	37.212 N, 106.796 W	Colorado River	10/14/2023	Present	8	0/10	1.3 ± 0.2	Rainbow Trout, Cutbow Trout	—	—
Lowerb	37.202 N, 106.807 W	Colorado River	10/14/2023	Present	0	0/10	1.1 ± 0.2	Rainbow Trout	—	—
Rio Grande	Uppera	37.759 N, 107.343 W	Rio Grande	9/14/2022	Present	0	0/10	0.8 ± 0.2	Rainbow Trout, Brown Trout	22.3	4
Middlea	37.794 N, 106.981 W	Rio Grande	9/14/2022	Present	0	0/10	0.9 ± 0.1	Rainbow Trout, Brown Trout	22.2	5
Lowera	37.679 N, 106.591 W	Rio Grande	9/14/2022	Present	0	0/10	0.7 ± 0.3	Rainbow Trout, Brown Trout	22.2	5
San Miguel River	Upperb	37.993 N, 108.021 W	Colorado River	12/08/2022	Unknown	0	0/10	0.8 ± 0.2	Rainbow Trout	8.1	0
Middleb	37.998 N, 108.036 W	Colorado River	12/08/2022	Unknown	0	1/10	0.7 ± 0.1	Rainbow Trout	8.1	1
Lowerb	38.004 N, 108.040 W	Colorado River	12/08/2022	Unknown	0	0/10	0.7 ± 0.1	Rainbow Trout	8.0	1
South Apache Creek	Upperb	37.853 N, 104.979 W	Arkansas River	12/10/2022	Unknown	0	0/10	0.7 ± 0.1	Rainbow Trout	0.6	3
Middleb	37.854 N, 104.977	Arkansas River	12/10/2022	Unknown	0	0/10	0.8 ± 0.1	Rainbow Trout	0.5	3
Lowerb	37.852 N, 104.966 W	Arkansas River	12/10/2022	Unknown	0	0/10	0.7 ± 0.2	Rainbow Trout	0.6	3
South Platte River	Upperb	38.969 N, 105.609 W	South Platte River	11/07/2022	Present	35	0/10	1.4 ± 0.4	Rainbow Trout, Cutbow Trout, Kokanee Salmon, Brown Trout	8.8	3
Middleb	38.978 N, 105.577 W	South Platte River	11/07/2022	Present	11	1/10	1.2 ± 0.3	Rainbow Trout, Cutbow Trout, Kokanee Salmon, Brown Trout	8.9	3
Taylor River	Uppera	38.775 N, 106.636 W	Colorado River	10/18/2022	Present	48	0/10	2.0 ± 0.2	Rainbow Trout, Cutbow Trout, Brown Trout	5.8	0
Middleb	38.762 N, 106.662 W	Colorado River	11/05/2022	Present	2	0/10	1.2 ± 0.4	Rainbow Trout, Brown Trout	5.8	1
Whale Creek	Uppera	37.925 N, 106.698 W	Rio Grande	8/30/2022	Present	0	0/10	1.0 ± 0.1	Brook Trout	21.2	7
White River 3	Uppera	40.013 N, 108.089 W	Colorado River	10/04/2022	Present	1	0/10	0.5 ± 0.2	Rainbow Trout, Brown Trout	9.9	5
Middlea	40.014 N, 108.102 W	Colorado River	10/04/2022	Present	1	1/10	0.4 ± 0.1	Rainbow Trout, Brown Trout	10.1	5
White River 4	Uppera	39.950 N, 107.694 W	Colorado River	10/03/2022	Present	1	0/10	0.3 ± 0.0	Rainbow Trout, Brown Trout	10.1	4
Middlea	39.949 N, 107.705 W	Colorado River	10/03/2022	Present	1	0/10	0.3 ± 0.1	Rainbow Trout, Brown Trout	10.2	5
Yampa River	Upperb	40.499 N, 106.919 W	Colorado River	10/11/2022	Present	4	0/10	0.7 ± 0.1	Rainbow Trout, Brown Trout, Cutbow Trout	8.9	3
Middleb	40.487 N, 106.839 W	Colorado River	10/11/2022	Present	0	0/10	0.6 ± 0.2	Rainbow Trout, Brown Trout	9.0	3

Electrofishing

We used various electrofishing methods (i.e., backpack electrofishing, barge electrofishing) to determine the presence or absence of gill lice at each site via physical sampling methods. Our electrofishing approach depended on stream size and the availability of personnel on a given date. Following all Colorado Parks and Wildlife field safety protocols, we electroshocked at all sites and collected stunned fish using dip nets. We held fish in live pens for subsequent species identification and examination for gill lice, which included visually inspecting the gills, opercula, fin attachments, and the mouth. We carefully inspected susceptible fish species of genus Oncorhynchus (rainbow trout, cutthroat trout, and the occasional kokanee) but immediately released non-susceptible trout (e.g., brook trout, brown trout) near their site of capture with a cursory inspection because S. californiensis does not infect those species. For each captured fish, we recorded taxonomic identification, weight (g), total length (mm), and the presence or absence of gill lice. When we did observe gill lice, we recorded the location of each louse on the fish and removed lice using sterile forceps and surgical scissors. We then placed lice into individually labeled 2-mL o-ring tubes filled with 70% ethanol, then placed them on ice for transport and subsequent DNA barcoding (see Section 2.4 below). To minimize handling stress on each infected fish, a maximum of two lice were removed from an infected individual, and then fish were released near the site of capture.

We did not complete electrofishing surveys in the Gunnison River, CO, because temperatures were too cold for fish to recover well from being shocked on that sampling date (and subsequent dates). Therefore, we opted to just take eDNA samples (see below) to prevent harming fish. However, previous data show relatively high gill lice abundance in the Gunnison River (Vigil et al. 2016; Colorado Parks and Wildlife, unpublished data), so it is already established that gill lice are prevalent in that system.

Environmental DNA Sampling

When sampling each site within a river, eDNA samples (10 replicate samples and a field blank) were taken at the downstream end of each site (lower, middle, and upper) prior to electrofishing in an effort to prevent disturbed sediments from affecting our results. Environmental DNA cross-contamination was prevented across sites by sampling at the downstream end first (lower), and then moving upstream for subsequent sites (middle, then upper). After sampling each site, we followed a decontamination protocol to prevent accidental spread of gill lice among localities. We cleaned all sampling equipment with a 10% bleach solution, rinsed with distilled water, then allowed it to dry for 24–48 h. As a precaution to prevent the accidental spread of pathogens where we had multiple rivers within a drainage system to sample, we sampled sites for the upstream tributaries first (in the same sequence as described above for lower, middle, then upper sites within that river). Then, after decontaminating the equipment and allowing sufficient time (> 24 h) for disturbed sediments to re-settle, we sampled the river sites located lower in the drainage. For example, in the Gunnison River drainage, where gill lice have been previously detected, we sampled the Taylor River (a major tributary to the Gunnison River) first, then sampled downstream Gunnison River sites 3 weeks later.

At each sampling site, we used reusable, modified 1-L Nalgene bottles, each with a Luer-Lok (Becton Dickinson) fitting attached to the bottom for filter attachment, and a Schrader valve on the cap for attaching a bicycle pump with a pressure gauge (see ) to collect water samples prior to electrofishing efforts so as to avoid disturbing sediments and potentially altering eDNA results (as mentioned above). Between uses, we sterilized each bottle using a 10% bleach solution followed by a thorough rinsing with deionized water. Subsequently, we thoroughly rinsed each sterilized bottle with river water prior to taking a new sample for eDNA testing to wash away any bleach residue. We collected water samples from random points in the river channel by submerging the bottle to approximately halfway between the bottom substrate and the water surface and letting it fill, but took care not to sample downstream from a previous eDNA sampling point where sediments may have been disturbed while taking a prior sample. We collected 10 water samples per site for eDNA analyses and included one field blank (deionized water) that was randomly assigned a sample number for each set of 10 samples, as recommended to detect contamination (Thomsen and Willerslev 2015; Sepulveda, Hutchins, et al. 2020). For each eDNA sample taken, we tightly attached a sterile Swinnex-25 filter unit pre-loaded with a nylon mesh filter (pore size = 5 μm, 25 mm diameter) to the Luer-Lok fitting at the bottom of the bottle to collect eDNA, shed cells, and cellular debris. This pore size has been demonstrated to be effective in field experiments because even though some eDNA passes through the pore, it generally allows for the filtration of higher volumes of water without clogging, thus resulting in more eDNA captured (Thomas et al. 2018; Dass et al. 2024). Once attached to the bottle, we used a bicycle pump to pressurize the bottle at 35–40 PSI, maintaining the pressure until all water passed through the filter. If the entire water sample passed through the filter, we collected another water sample, reattached the same filter, pumped more air into the bottle, and stopped when the filter clogged with sediment and debris and we could no longer force water through the filter. Once each eDNA sample was collected, we removed the Swinnex unit from the bottle, injected the filter with 1 mL of 100× TE buffer (Fisher Scientific) using a disposable syringe, tightly sealed it with a Luer-Lok cap, and placed it on ice in an individually labeled plastic bag for transport. We transferred eDNA samples from the cooler to a −20°C freezer until they could be shipped on ice to Pisces Molecular LLC (Boulder, CO) for subsequent analyses.

At Pisces Molecular LLC, eDNA samples were processed as follows: Each filter was removed from the Swinnex housing, folded with sterile forceps, and placed into a 1.7 mL microcentrifuge tube containing tissue lysis buffer (Qiagen Buffer ATL and Proteinase K) and 20 μL of carrier DNA (Salmon Sperm DNA, CAS #438545–06-03, Sigma-Aldrich, St. Louis, MO, as a precaution against losing target DNA in water samples with unknown DNA concentrations during the DNA extraction process). Tubes containing individual eDNA samples and lysis buffer were incubated at 55°C for 1 h with vortexing every 15 min to release DNA from the filter. DNA was then extracted using Qiagen DNeasy Blood and Tissue kits, following the manufacturer's recommended protocols. All DNA samples were assayed for the presence of S. californiensis DNA with a quantitative polymerase chain reaction (qPCR) assay targeting an 80 base amplicon of the 28S rRNA gene (Appendix S2). The qPCR reactions used primers PM829 (5′-CCC GTC TCC TTA GCT TTA TCG A-3′) and PM830 (5′-GCA AGC CGT GTC AGA CAA G-3′), and probe PMP61 (5′-FAM-CGG CAG GTA TTT AAG CGG GCT CGT TT-BHQ1-3′). The primers and probe were developed at Pisces Molecular LLC and were previously demonstrated to amplify 28S for S. californiensis and S. edwardsii with internal testing (Appendix S3). The reaction master mix used a PCR inhibitor resistant Taq polymerase (PerfeCTa Multiplex ToughMix, Quantabio) and a VIC-labeled internal positive control (Life Technologies) to detect PCR inhibition. Reaction cocktails were 20 μL in volume and included the following: 4.0 μL of template DNA, 9.4 μL of nuclease free water, 4.0 μL of ToughMix, 0.5 μL of 10× exogenous internal positive control mix, 0.1 μL of 50× exogenous internal positive control DNA, and 2.0 μL of primer and probe working stock (1 μL of each primer and probe in 97 μL of nuclease free water). The qPCR used a 10-fold dilution series of known positive S. californiensis genomic DNA or a gBlocks synthetic DNA fragment (see Appendix S2) ranging from 4.32 × 10⁶ target sequence molecules per reaction to 4.32 relative target sequence molecules per reaction used as controls. All qPCR reactions were performed on an Agilent Technologies AriaMx real-time instrument using the following thermal profile: An initial 3 min denaturation at 95°C, followed by 45 cycles of 95°C for 15 s and 60°C for 60 s. Positive fluorescent signals from sample DNA were quantified by comparison of the cycle at which a sample fluorescent signal crossed the threshold relative to the cycles at which serial dilutions of the genomic or synthetic control DNA crossed the threshold (see Appendix S2 for additional information).

DNA Barcoding

Infected fish yielded 58 S. californiensis specimens from 11 localities during our physical sampling efforts. We collected these individual lice from their host fish as described above (see Section 2.2 Electrofishing Methods).

We implemented a DNA barcoding approach to assess the phylogenetic and population genetic structure of these samples. Because of their small body size, we made the decision to destructively process individual specimens to ensure high enough DNA concentrations to proceed, but unfortunately that means no vouchers of these samples were retained. For whole genomic DNA extraction, Pisces Molecular LLC lab technicians used Qiagen DNeasy Blood and Tissue kits following the manufacturer's recommended protocol to extract genomic DNA from each louse. Those genomic DNA samples were then used as template DNA in the subsequent polymerase chain reactions (PCRs).

The mtDNA cytochrome c oxidase subunit 1 (COI) gene was targeted for amplification via PCR using universal barcoding primers LCO1490 (5′-GGT CAA CAA ATC ATA AAG ATA TTG G-3′) and HCO2198 (5′-TAA ACT TCA GGG TGA CCA AAA AAT CA-3′) published by Folmer et al. (1994). Each PCR reaction cocktail was 20 μL in volume and included the following reagents: 2 μL of template DNA, 1.6 μL (16 pmols) of each oligonucleotide primer, 0.2 μL of Amplitaq Gold (Thermofisher), 1.6 μL of dNTPs (New England Biolabs, Beverly, MA), 1.2 μL of 25 mM MgCl₂, 2.0 μL of 10× AmpliTaq Gold Core buffer (Thermofisher), and 9.8 μL of nuclease-free water. The thermal profile consisted of an initial denaturation at 94°C for 9 min, followed by 35 cycles of denaturation at 95°C for 30 s, annealing at 50°C for 60 s, extension at 72°C for 60 s, and finishing with a final extension at 72°C for 60 s. Positive and negative controls were included in each batch of PCR amplifications (with positive controls including previously amplified S. californiensis template DNA and negative controls including nuclease-free water instead of DNA). Successful PCR amplifications were qualitatively assessed by visualizing ethidium bromide-stained 1.0% agarose gels under ultraviolet radiation following gel electrophoresis. Each amplified PCR sample was treated with an ExoSAP PCR clean-up reaction to remove unincorporated primers and dNTP molecules. The purified PCR products were then used as templates for dideoxy-cycle sequencing reactions for forward- and reverse-strand Sanger sequencing (ABI BigDye Terminator v.3.1). Light and heavy DNA strands were sequenced on a capillary electrophoresis DNA sequencer (ABI 3130 Genetic Analyzer, 36 cm array, POP7) and base-called with Sequence Analysis v.6 software. For each individual COI sequence, light and heavy strand chromatograms were assembled into a contig sequence using Geneious Prime v.2019.2.3 (), and a consensus sequence was generated for each contig. The protocol for handling any discrepancies between forward and reverse nucleotide sequences was to choose the sequence with the highest Q value, resulting in all base positions exhibiting a Q value greater than 30, thus indicating no ambiguities or base-calling errors in this dataset. The COI sequences from these lice were combined with sequences from an additional 14 S. californiensis individuals that were collected from Mt. Massive Lakes, Colorado (Arkansas River drainage), Iron Lake #2, Idaho (Salmon River drainage), Lookout Point Lake, Oregon (Middle Fork Willamette River drainage), and Rabbit Run Cooperative Nursery (hereafter, Rabbit Run), Pennsylvania, as well as two S. edwardsii individuals from Leadville Hollow, Pennsylvania, that were previously shipped to Colorado Parks and Wildlife for DNA barcoding at Pisces Molecular LLC.

We included reference COI sequences from GenBank for phylogenetic analyses as well (see Table 2). These reference sequences included five S. californiensis individuals from four locations in Russia (B. Anaur River, Raduga River, Utkholok River, Maximoyka River in Primorye and Kamchatka); one Salmincola carpionis individual from the B. Anaur River, B. Shantar Island, Russia; one Salmincola edwardsii sample from Squaw Creek, Wisconsin, USA; one Salmincola markewitschi individual from the Amgu River, Primorye, Russia; one Salmincola siscowet individual from Lake Michigan, Michigan, USA; and one Salmincola stellata individual from the Amgu River, Primorye, Russia. All COI sequences were then aligned using MAFFT v.7 (Katoh and Standley 2013), the alignment was viewed in Geneious and edited by eye, and the final alignment was exported for phylogenetic analyses.

TABLE 2 List of Salmincola COI sequences obtained from GenBank, as well as the GenBank accession number, the sampling locality, and the publication reference for each.

Species	GenBank Accession#	Geographic Distribution	Source
S. californiensis	OQ844022	B. Anaur River, B. Shantar Island, Russia	Shedko et al. (2023)
	OQ844027	Raduga River, Kamchatka, Russia	Shedko et al. (2023)
	OQ844033	Utkholok River, Kamchatka, Russia	Shedko et al. (2023)
	OQ844039 & OQ844041	Maximovka River, Primorye, Russia	Shedko et al. (2023)
S. carpionis	OQ844005	B. Anaur River, B. Shantar Island, Russia	Shedko et al. (2023)
S. edwardsii	OQ355023	Squaw Creek, Wisconsin, USA	Katz et al. (2023)
S. markewitschi	OQ843994	Amgu River, Primorye, Russia	Shedko et al. (2023)
S. siscowet	OP830005	Lake Michigan, Michigan, USA	Marshall et al. unpublished data
S. stellata	OQ844054	Amgu River, Primorye, Russia	Shedko et al. (2023)

Phylogenetic Analyses

To select an appropriate model of DNA sequence evolution for the Salmincola COI alignment, we used the program jModelTest2 (Darriba et al. 2012) as implemented in PHyML v.3.0 (Guindon and Gascuel 2003) under the Akaike Information Criterion. We then reconstructed a maximum likelihood (ML) phylogeny using RAxML-HPC2 v.8.2.12 (Stamatakis 2014) on XSEDE using the CIPRES portal (Miller et al. 2010) and assessed nodal support using RAxML rapid bootstrapping with 100 rapid bootstrap inferences and subsequent ML search (Stamatakis et al. 2008).

To better visualize intraspecific COI haplotype diversity in S. californiensis, we created a median-joining haplotype network using TCS v.1.21 (Clement et al. 2000). For this analysis, we excluded the outgroup taxa (i.e., other species of Salmincola) and only included COI sequences from S. californiensis individuals (including the Russian sequences obtained from GenBank). We selected the 90% connection probability limit for this analysis. We color-coded each haplotype in the network according to geographic location (i.e., states and river drainages) from which gill lice were sampled.

To qualitatively compare divergence times of Salmincola clades with previously published divergence times for native cutthroat trout divergence, we estimated divergence times of the variation observed within S. californiensis and among outgroups using BEAST2 (Bouckaert et al. 2019). To ensure proper mixing of chains, we removed all redundant haplotypes, randomly selecting which individual carrying a given shared COI haplotype to retain in the dataset. We used jModelTest2 (Darriba et al. 2012) as implemented in PHYML (Guindon and Gascuel 2003) to select an appropriate model of sequence evolution for the reduced dataset under the Bayesian Information Criterion, and based on those results, proceeded using the GTR + G substitution model. We implemented the optimized relaxed clock model (Drummond et al. 2006; Douglas et al. 2021) with a lognormal prior distribution and a mean clock rate of 0.023 nucleotide base substitutions per site, per lineage, per million years with lower and upper bounds set at 0.013 and 0.033, respectively (which spans the 95% credible interval for estimated COI mutation rates in Salmincola; see Shedko et al. 2023). Commonly, researchers have used a rate of 1.4% per million years for arthropod COI sequence evolution (White 1994; Cornils and Held 2014). However, it has been demonstrated that copepodids have accelerated COI mutation rates compared to other groups of arthropods (Shao and Barker 2007). The 2.6%–6.6% per million year rates that we used span a range of accelerated rates, and this range does match recently published arthropod COI mutation rates that are faster, ranging from 3.2% to 5.5% per million years, respectively (e.g., Copilaş-Ciocianu and Petrusek 2017; Loeza-Quintana et al. 2019). Therefore, we proceeded with these rates in our analyses. We used the Yule model to set the tree prior (Bouckaert 2022). We ran the Markov Chain Monte Carlo sampling (MCMC) chain for 30,000,000 steps, sampling every 1000 steps, and discarding the first 3,000,000 steps (10%) as burn-in. We used Tracer v.1.7.2 (Rambaut et al. 2018) to ensure adequate effective sample sizes (ESS > 200) and proper mixing of chains. We then used TreeAnnotator v.2.7.6 (Heled and Bouckaert 2013) to summarize the sample of trees produced by BEAST and recover the tree in the posterior sample of trees that exhibited the highest sum of posterior probabilities (i.e., the “best” tree), then visualized that tree in FigTree v.1.4.4 ().

Gill Lice Occupancy Modeling

We estimated site occupancy using an occupancy modeling framework in the “unmarked” package (Fiske and Chandler 2011; Kellner et al. 2023) as implemented in R v4.3.1 (R. Core Team 2023) to assess gill lice occupancy at our sample sites. Occupancy models address the issue of imperfect detection of animals such as gill lice by estimating a probability of detection. This probability is calculated using data from repeated visits to a site during a period of consistent occupancy. Here, we considered each of the 10 water samples a repeat visit to the same site and considered a positive gill lice eDNA detection as “present” and a negative gill lice eDNA detection as “absent.” We did not include electrofishing results in occupancy modeling efforts because we worked with crews of varying sizes (without recording catch per unit effort), used different electroshockers at various sites (e.g., backpack vs. barge electrofishing), and could not conduct repeated visits at each site due to budgetary and time constraints.

We created a set of 20 candidate models where gill lice occupancy was either constant or affected by site, and detection probability was either constant or affected by different combinations of reach, water temperature, and UV radiation, as well as the precise amount of water filtered to obtain each sample (Table 3). We included water temperature as an environmental covariate because of the indirect effects it has on microbial enzymatic DNA hydrolysis (Strickler et al. 2015; Jo et al. 2019; Jo 2023). We included ultraviolet radiation index as an environmental covariate because of the damaging impact of UV light on DNA (Strickler et al. 2015). Both water temperature and UV radiation index at the time of sampling were obtained from the United States Geological Survey () at the nearest gaging station to the collection site (usually within 40 km). We included site as a covariate because of the potential effect fluvial processes could have on eDNA distribution and longevity. We used Akaike's Information Criterion (AIC; Akaike 1973) to rank occupancy models. We considered models to be significantly different if their AIC differed by more than two (Akaike 1973).

TABLE 3 Model selection table showing the effect of covariates on occupancy probability (Ψ) and detection probability (p) compared to constant models (.). The model name, number of parameters (k), difference in AIC from the highest-ranked model (ΔAIC), and model weight for each model in the candidate set are provided.

Model	k	ΔAIC	Weight
Ψ (Site) p (Temperature + UV + Site)	8.00	0.00	0.65
Ψ (Site) p (Temperature + UV + Site+LitersFiltered)	9.00	1.96	0.24
Ψ (.) p (Temperature + UV + Site)	6.00	5.08	0.05
Ψ (.) p (Temperature + UV + Site+LitersFiltered)	7.00	7.08	0.02
Ψ (Reach) p (Temperature + UV)	6.00	7.67	0.01
Ψ (.) p (Temperature + UV)	4.00	8.03	0.01
Ψ (Reach) p (Temperature + UV + LitersFiltered)	7.00	9.23	0.01
Ψ (.) p (Temperature + UV + LitersFiltered)	5.00	9.57	0.01
Ψ (Reach) p (UV + Site+LitersFiltered)	8.00	24.83	0.00
Ψ (Reach) p (UV + Site)	7.00	25.78	0.00
Ψ (.) p (UV + Site+LitersFiltered)	6.00	27.40	0.00
Ψ (.) p (UV + Site)	5.00	28.54	0.00
Ψ (Reach) p (Temperature + Site+LitersFiltered)	8.00	31.30	0.00
Ψ (.) p (Temperature + Site+LitersFiltered)	6.00	31.74	0.00
Ψ (Reach) p (LitersFiltered)	5.00	32.65	0.00
Ψ (.) p (LitersFiltered)	3.00	33.72	0.00
Ψ (Site) p (.)	4.00	40.37	0.00
Ψ (Site) p (Temperature + Site)	7.00	41.14	0.00
Ψ (.) p (Temperature + Site)	5.00	41.47	0.00
Ψ (.) p (.)	2.00	41.73	0.00

Power Analysis

The number of repeated surveys that should be conducted is a key aspect of designing occupancy studies (Mackenzie and Royle 2005). Therefore, we performed a power analysis to inform future studies using eDNA for gill lice detection. Following Stauffer et al. (2002), we determined the number of surveys required to detect gill lice at specific probability levels given their presence. We extracted the back-transformed detection probability estimate and standard errors from the highest-ranked occupancy model based on the average value of covariates in that model. From those, we estimated the cumulative probability of detecting gill lice given their presence at a site based on increasing numbers of repeated surveys, up to a total of 15 repeated surveys.

Results

Sampling and Gill Lice Occurrences

In total, we sampled 48 sites within 19 river systems (Figure 1; Table 1). At these sites, we captured 1575 total fish representing Rocky Mountain cutthroat trout (n = 19; 1.2% of the total fish captured), rainbow trout (n = 1501; 95.3%), cutbow trout (O. mykiss x O. virginalis hybrids; n = 27; 1.7%), and kokanee (n = 28; 1.8%). Of those fish, 148 (9.4%) were infected with gill lice, and 147 (9.3%) of the infected fish were captured in waters with a previously recorded history of gill lice, but one infected fish (< 0.01%) was captured in Henson Creek, where gill lice presence was previously undocumented. Across Colorado, gill lice infection load ranged from 1 to 27 lice per infected fish when gill lice were present. The average infection load of rainbow trout was three lice per fish (range of 0–20 lice per fish), whereas the average infection load of cutthroat trout was zero lice per fish (range of 0 lice per fish). However, the zero abundance of infected cutthroat trout is likely an underestimate because of the relatively low number of cutthroat trout individuals examined. The average infection load of infected kokanee was approximately 14 lice per fish (range of 5–27 lice per fish), whereas the average infection load of infected cutbow trout was approximately four lice per fish (range of 0–8 lice per fish).

[IMAGE OMITTED. SEE PDF]

We detected gill lice at 10 of the 48 sites (21%) using eDNA techniques, at 17 of the 45 sites (38%) using electrofishing methods (because electrofishing efforts were not completed at three sites in the Gunnison River due to temperature concerns), and at five of the sites (11%) using both detection methods. On average, we filtered 0.83 (±0.54) liters of water for each eDNA sample (see Table 1 for site-specific averages). Within rivers that had positive eDNA tests, 10 of 20 sampled sites (50%) tested positive for the presence of gill lice. Of the sites that tested positive for the presence of gill lice using eDNA, three were from areas where gill lice were previously undocumented (Huerfano River [two positive sites], San Miguel River [one positive site]). Using electrofishing methods, we detected gill lice in one area where they had not previously been documented: Henson Creek (one positive site). We did not observe DNA amplification in any of the field blanks.

To further assess gill lice distributions in river drainages where they were present, results from physical sampling via electroshocking and eDNA are summarized here. We considered a site positive for gill lice if detecting one or more lice on fish stunned via electrofishing, or if at least one of the ten eDNA samples at a site tested positive for gill lice. Using physical sampling, we detected gill lice in 50% of upstream sites, 33.3% of midstream sites, and 5.6% of downstream sites within a set of sampling sites in a given river. Using eDNA methods, we detected gill lice at 5.6% of upstream sites, 33.3% of midstream sites, and 16.7% of downstream sites within a river.

DNA Barcoding and Phylogenetic Analyses

We sequenced 663 base pairs of the COI gene from 72 individual lice (the 58 we sampled along with the 14 that were previously sent to Pisces Molecular for DNA barcoding) and deposited these newly generated DNA sequences in GenBank (accession numbers PP942459–PP942530). Maximum likelihood phylogenetic analysis produced a best tree with a ML score of −1326.71 (Figure 2). Salmincola californiensis appears to be a monophyletic group that is sister to a clade containing S. edwardsii and S. siscowet. The intrageneric relationships depicted here are consistent with those previously published (Shedko et al. 2023). All samples we collected in Colorado fall within the monophyletic S. californiensis clade. There are two clades containing S. californiensis haplotypes that vary in size and geographic composition. Clade 1 includes all the haplotypes from Whale Creek, Colorado. Clade 2, comprising haplotypes distributed across all four major drainages in Colorado as well as haplotypes from Idaho, Oregon, Pennsylvania, and Russia, exhibits little phylogenetic resolution and contains multiple unresolved polytomies.

[IMAGE OMITTED. SEE PDF]

There were 19 unique COI haplotypes in our S. californiensis dataset (Haplotypes A–S), four of which (Haplotypes A, B, F, and I) were shared among multiple individuals and geographically widespread (Figure 3). There is little genetic distance between eleven of these (Haplotypes A–K), with only one mutational step separating unique haplotypes in most cases. Greater genetic distances appear between the others (Figure 3): One haplotype from the White River (Haplotype P), one haplotype from Oregon (Haplotype Q), four haplotypes from Russia (Haplotypes L, M, N, and O), and two haplotypes from Whale Creek (Haplotypes R and S). The two haplotypes in Whale Creek did not connect to the others under the 90% confidence threshold.

[IMAGE OMITTED. SEE PDF]

Divergence time estimates reveal a relatively recent diversification (Pliocene—Pleistocene) within S. californiensis (Figure 4). The initial divergence among our samples and those obtained from GenBank occurred approximately 10.8 (95% credible interval = 5.1–18.6) million years before present (mya) for the common ancestor of all Salmincola species included herein. Diversification among S. californiensis haplotypes is estimated to have begun approximately 4.3 (1.6–7.9) mya, which is when the Whale Creek population diverged from the rest of the ingroup. Diversification among haplotypes in (the not well-resolved) Clade 2 began approximately 3.2 (1.1–6.2) mya. Our 95% credible intervals only included the present day in three instances: Between the two Whale Creek haplotypes (WHC_157771 and WHC_160955), which are estimated to have diverged in the Pleistocene, 0.6 (0.0–2.1) mya; between haplotypes from the Gunnison River (GUN_166958), the South Platte River (STP_166264), and Idaho (IDA_171845), estimated to have last shared a common ancestor 0.6 (0.0–1.4) mya; and between two haplotypes from the Taylor (TAY 166275) and South Platte (STP_166256) rivers, with diversification estimated at 0.4 (0.0–2.0) mya.

[IMAGE OMITTED. SEE PDF]

Gill Lice Occupancy Modeling and Power Analysis

The highest ranked occupancy model (weight = 0.65) from our candidate model set indicated that site occupancy and detection probability for gill lice were affected by site, and detection probability was further affected by the water temperature and UV index (Table 3). One model was within two AIC (ΔAIC = 1.96, weight = 0.24) from the highest-ranked model, indicating that there is no evidence that the two models are significantly different. The second-ranked (but not statistically distinct) model was similar to the highest-ranked model but had the amount of water filtered as an additional effect on detection probability (Tables 3 and 4). However, the confidence interval for the slope of this covariate overlapped with zero (slope = 0.16, SE = 0.75), and was therefore uninformative (Table 4; Appendix S4). All other coefficients from this second-ranked model were within 3% of those from the highest-ranked model; therefore, we only demonstrate the highest-ranked model here (see Appendix S4 for the second ranked model). Generally, upstream sites had lower occupancy probabilities and higher detection probabilities (Figure 5). Detection probabilities also increased with UV radiation index but decreased with increasing temperature (Table 4, Figure 5).

TABLE 4 Coefficients for the highest-ranked and second-ranked occupancy models.

Parameter	Covariate	Highest-ranked model	Second-ranked model
Occupancy probability	Site—Lower (Intercept)	−0.06	(SE = 0.82)	−0.06	(SE = 0.82)
Site—Middle	0.98	(SE = 1.22)	1.01	(SE = 1.24)
Site—Upper	−2.41	(SE = 1.33)	−2.41	(SE = 1.33)
Detection probability	Site—Lower (Intercept)	−2.93	(SE = 0.93)	−3.11	(SE = 1.25)
Site—Middle	−0.65	(SE = 0.70)	−0.64	(SE = 1.70)
Site—Upper	3.30	(SE = 1.17)	3.37	(SE = 1.23)
Temperature	−0.59	(SE = 0.12)	−0.60	(SE = 0.13)
UV	1.61	(SE = 0.31)	1.64	(SE = 0.35)
Liters Filtered	—	—	0.16	(SE = 0.75)

[IMAGE OMITTED. SEE PDF]

Our overall probability of detection of gill lice from the highest ranked model was 0.29 (SE = 0.05). Cumulative probability of detection increased with increasing numbers of repeated surveys, and first reached a value greater than 0.95 at 9 surveys with a detection probability of 0.954 (SE = 0.14). A power curve depicting probability of detection with increasing numbers of eDNA sample sizes is shown in Figure 6. With ten eDNA samples per site, averaging 0.83 L of water filtered per eDNA sample, we had a 96.8% (±0.14 SE) probability of detecting gill lice when they were present at a site.

[IMAGE OMITTED. SEE PDF]

Discussion

General Findings

This study provides important insights into the advantages of using eDNA sampling methods in conjunction with conventional sampling methods in detecting gill lice in freshwater rivers and streams. Our results demonstrate that: (1) eDNA is a useful, but not infallible tool in monitoring gill lice populations in the field and is perhaps most useful when used in combination with physical sampling methods; (2) increasing water temperatures negatively influence gill lice detection probabilities using eDNA methods, but increasing UV radiation index increases detection probability; (3) S. californiensis was the only species of gill louse detected in our samples across the state of Colorado; (4) S. californiensis has expanded its range in Colorado, now occurring in every major drainage basin; and (5) S. californiensis populations distributed across Colorado exhibit relatively low genetic diversity and do not display population genetic structure consistent with in situ evolution on the landscape, with the exception of the Whale Creek haplotypes and one highly divergent haplotype from the White River. Rather, S. californiensis exhibits a population genetic structure that is consistent with a pattern of recent, rapid, and possibly human-facilitated expansion. Moreover, this study provides additional support that eDNA methods may also be effectively used to detect gill lice in hatchery settings, where chemical treatments may be feasible to control the parasite.

Comparing eDNA and Electrofishing Methods

Small-bodied parasites, such as gill lice, are difficult to detect and can spread into naïve ecosystems undetected until they are well established, thus presenting significant problems for natural resource managers. Once established, invasive species are notoriously difficult to eradicate or control; thus, early detection is imperative in invasive species management (Victorian Government 2010; Simberloff 2010; Vander Zanden et al. 2010). Gill lice are difficult to manage via biological, chemical, or physical controls in invaded environments, so eDNA may be particularly useful in early detections in terms of their management. For example, a better understanding of gill lice distributions can inform fish stocking or translocation efforts. Moreover, if gill lice are detected early in downstream reaches of a system but have not yet infiltrated sensitive cutthroat trout populations upstream, barriers to fish passage could be installed to prevent upstream migration. Fisheries managers could also stop stocking susceptible hosts in affected areas in attempts to break the gill lice life cycle. Additionally, managers could allocate resources to conduct follow-up surveys to evaluate gill lice distributions on a finer scale within drainages where they occur, investigate population demographics of lice at those sites to assess population growth patterns to better understand the extent of infection, and/or use eDNA metabarcoding to explore biological communities in search of biological controls among co-occurring species (Ahmed et al. 2022; Katz et al. 2023; Johnson et al. 2024; Tetzlaff et al. 2024).

In the field, eDNA sampling may be slightly less effective than electrofishing when the two approaches are considered independently, given that we detected gill lice at more sites via electrofishing than we did via eDNA methods. However, we did detect gill lice presence at two sites in the Arkansas River drainage, and one site in the San Miguel River using eDNA, even though we did not directly observe any infected fish at those sites, so eDNA is informative for improving understanding of current Salmincola distributions. While false positives are possible with qPCR, we do not think it is likely to be the case in these instances given that the field blanks at these sites did not amplify, and there was no significant cross-reactivity in silico (Appendix S2). While there is no strong way to rule out potential false positives other than DNA sequencing of amplicons, and these are too short for high-quality Sanger sequencing, there are no heterospecific gill lice in Colorado's watersheds based on our surveys and those of previous investigators. Hence, intrageneric detections of gill lice are not really as much of a concern in this study area as it would be in others with S. californiensis and S. edwardsii (and potentially other congenerics) occupying the same system. In both drainages, positive eDNA results came from middle and lower sites, which experience warmer temperatures, and brown trout abundance increases while rainbow trout abundance decreases, which may have contributed to us not detecting gill lice in those areas via electroshocking, especially if gill lice population sizes are low in those areas.

Environmental DNA could be particularly effective in detecting gill lice in hatchery settings, where chemical treatments may be feasible prior to stocking. This is important if fish stocking practices have contributed to their spread, which is plausible. Thus, the effectiveness of eDNA as a tool for detection of aquatic parasites has far-reaching implications for management efforts. Honing the ability to detect parasites will become more important as globalization accelerates the rates at which nonnative species move into novel ecosystems (Early et al. 2016), and eDNA appears to be a very useful tool in that regard.

It seems likely that our methodological approach contributed to discrepancies between gill lice detections using eDNA vs. electrofishing methods, particularly given successes reported by Katz et al. (2023) in using eDNA to detect S. edwardsii. For example, our sampling approach spanned multiple seasons, with sampling beginning in September 2022 and ending in October 2023. As such, it is likely that metabolic rates, and therefore eDNA shedding, would have varied across seasons, potentially confounding our results. Given that kokanee migrate upstream to spawn in the Fall, and rainbow and cutthroat trout spawn in the Spring, gill lice could be passively dispersing accordingly during these seasons and perhaps be more readily detected (although spring runoff could also dilute eDNA concentrations). Additionally, electrofishing efforts generally took more effort than collecting eDNA samples did, and electrofishing effort varied among sites according to crew size on any given sampling date (although we did not record this information). Catch–per–unit–effort would likely have been informative in interpreting these results.

It is also worth noting that the relatively large pore size (5 μm) of the eDNA filters we used allowed more water to be filtered and collect intact cells from which DNA could be extracted, but may have also let some extracellular eDNA molecules through, potentially influencing gill lice detection probabilities via eDNA. While this pore size has been shown to be optimal in some freshwater systems (Thomas et al. 2018; Dass et al. 2024), a smaller pore size (0.1–0.2 μm) may have facilitated increased gill lice detections by not allowing eDNA through the filter pores (Hunter et al. 2019; Dass et al. 2024), although likely with much lower volumes of water filtered before clogging. It is also possible that some positive eDNA samples failed to amplify in the laboratory, but we observed a 95% PCR efficiency in our qPCR validation (see Appendix S3). Nevertheless, the power analysis showed that even the larger eDNA filter was effective and that we had a > 95% probability of detecting gill lice in a system when they were present, with 10 eDNA samples taken per site. There is seemingly much more to be understood about gill lice detectability via eDNA in future research efforts (Lamb et al. 2022). To better understand the reliability and consistency of eDNA detection of gill lice, additional environmental covariates that may accelerate eDNA decay in the environment should be examined. Incorporating environmental covariates such as on-site measurements of temperature and ultraviolet radiation, salinity, pollution levels, water turbidity, stream width, stream depth, community species composition, etc. into occupancy models may provide additional insights into the reliability of gill lice detection using eDNA and may allow for the collection of more reliable, consistent presence/absence data.

Our combined eDNA and physical sampling results suggest that most infected fish in our study were found in upstream and midstream reaches, which is a pattern consistent with the spread of gill lice via migrating salmonid fishes, particularly spawning kokanee, which trended toward higher parasite loads per individual. Conversely, eDNA results showing higher gill lice detection in midstream and downstream reaches may be due to eDNA being carried downstream via hydrologic processes, particularly in cold-water streams where eDNA is expected to degrade more slowly. Cool, highly oxygenated waters benefit cool-adapted salmonid fishes, such as those in the genus Oncorhynchus, and cool temperatures likely slow the degradation of eDNA based on our occupancy modeling results. The top occupancy model suggests that increasing temperatures negatively impact gill lice detection probabilities using eDNA, and this temperature effect on eDNA longevity is reflected in other studies as well (Pilliod et al. 2014; Strickler et al. 2015; Andruszkiewicz et al. 2017; Allan et al. 2021; Rounds et al. 2024). Thus, there may be a seasonal aspect to gill lice eDNA detections (Rounds et al. 2024), which would be worth further exploration. This may help explain negative eDNA results in the Taylor River, which is known to have high gill lice concentrations based on our electrofishing efforts as well as prior surveys conducted by fisheries managers (Colorado Parks & Wildlife, unpublished data). We sampled the Taylor River late in the year, when water temperatures were cold, and eDNA should have been better preserved, but metabolic rates and eDNA shedding may also have slowed. It is also possible that shed eDNA at the uppermost Taylor River site was carried downstream by hydrologic processes into privately owned areas that were inaccessible for sampling, and thus remained undetected in our samples.

It remains unclear why increased UV radiation index would increase the eDNA detection probability, as shown in our top occupancy model. In fact, we expected to see the opposite effect, given the well-known effects of UV light facilitating the degradation of DNA molecules. However, it is possible that UV radiation does not have the same effect on DNA in aquatic environments. Microcosm experiments have demonstrated that eDNA degradation rates are relatively low in cold, alkaline water (Strickler et al. 2015), when there are high amounts of total eDNA from all organisms in the system (Barnes et al. 2014), and when waters are less turbid, have low amounts of chlorophyll, and have fewer suspended particles (Barnes et al. 2021). Several of these conditions are met in Colorado's rivers that support salmonid fishes. Ultraviolet radiation does not seem to impact eDNA detection in field experiments either (Mächler et al. 2018). So, it may be that eDNA is not closely tied to UV radiation, even though there is a temporal component that needs to be considered, as eDNA does break down over time (Andruszkiewicz et al. 2017). Different taxa also show different responses to UV radiation (Allan et al. 2021; Holman et al. 2022). It is also possible that the relationship between UV radiation and eDNA detection rates is an entirely spurious effect, caused by other unmeasured parameters, or that our use of UV data recorded at USGS gaging stations included too much distance to adequately represent the true UV index at each site. Given the wide range of habitat types, elevational changes, stream orders, community compositions, and other environmental factors, and the distance between USGS gaging stations and our sampling sites, it is possible that the UV intensity measured at these locations is correlated with other confounding variables that were not incorporated into our models, leading to the unexpected positive relationship. One possibility may be the occurrence of greater UV radiation at higher elevations. Smaller, high-elevation (i.e., lower Strahler stream order) streams generally have lower biodiversity and water volume than larger, lower-elevation streams. This phenomenon may lead to greater dilution of target eDNA in larger streams at lower elevations and relatively higher concentrations of eDNA in smaller streams at high elevations. This would result in corresponding detection probabilities that are greater in locations with higher UV radiation indices. It is also possible that a high-elevation, low-order stream that is heavily shaded by terrestrial vegetation does not actually experience enough direct sunlight for UV radiation to have an impact on eDNA longevity at the time of sampling, even when the UV index is high in the general area, thus potentially allowing for easier gill lice detection despite a high UV index. Clearly, more field studies investigating the effect of UV light on eDNA degradation in freshwater systems are warranted to better understand its effects on eDNA longevity and detectability.

Gill Lice Distribution and Population Structure

Gill lice were detected in every major Colorado drainage basin using eDNA and electrofishing methods. In the Arkansas River, Colorado River, South Platte River, and Rio Grande drainages, we detected gill lice in 46% of our sampling sites using eDNA and electrofishing methods combined (Figure 1).

Gill lice do appear to have expanded their range upstream from known infected waters in Colorado. For example, the detection of a single infected rainbow trout specimen in Henson Creek, where gill lice were previously undetected, is perhaps unsurprising given that Henson Creek is a tributary to the Lake Fork of the Gunnison River, which in turn is a tributary to Blue Mesa Reservoir where gill lice have proliferated in recent years (Lepak et al. 2022) and migrating fish can move upstream unimpeded.

It can be difficult to approximate the exact time of introduction of invasive species that are unintentionally introduced into a region (Maebara et al. 2020), or whether cryptogenic parasites are native or introduced (Lymbery et al. 2014). However, molecular data can provide some answers as to an approximation of when an invasion occurred, the origin of invasion, and the range expansion process (Maebara et al. 2020). As mentioned above, low genetic diversity across geographically isolated samples, lack of overlap between native hosts and non-native parasite distributions, and phylogenetic similarities with foreign lineages can all indicate that a cryptogenic parasite is not native to an area (Lymbery et al. 2014). The gill lice we sampled across Colorado have a shallow phylogenetic structure with many unresolved relationships (Figure 2), share multiple haplotypes across drainage basins (Figure 3), exhibit mismatches between native cutthroat trout populations and areas where gill lice are currently detected (Colorado Parks and Wildlife 2018), and exhibit shared haplotypes or haplotypes with close genetic distances to haplotypes from other states (Idaho, Oregon, and Pennsylvania; Figures 3 and 4). Therefore, it is likely that the current S. californiensis range expansion is at least partially invasive.

In Colorado, it is plausible that gill lice originally co-invaded with Rocky Mountain cutthroat trout as that lineage diversified from other cutthroat trout lineages (i.e., coastal cutthroat trout, Lahontan cutthroat trout, and westslope cutthroat trout), but more recently have been transported across the state by human activities. Rocky Mountain cutthroat trout is postulated to have diverged from those other cutthroat trout lineages during the Pleistocene as it dispersed further inland from other cutthroat populations into Colorado (Shiozawa et al. 2018; Kokkonen et al. 2024). Subspecies of the Rocky Mountain cutthroat trout include Rio Grande cutthroat trout (O. v. virginalis), greenback cutthroat trout (O. v. stomias), and Colorado River cutthroat trout (O. v. pleuriticus), which are all naturally distributed in Colorado. These lineages are hypothesized to have diverged from each other 1.1–1.4 million years before present (Shiozawa et al. 2018). Divergence times within the Rio Grande cutthroat trout subspecies (the subspecies that naturally occupies Whale Creek), based on fossil calibrations and mtDNA mutation rates, are estimated to be approximately 610,000 and 390,000 years before present, respectively (Shiozawa et al. 2018). While divergence time estimates should be interpreted with caution, particularly when based on a single mtDNA locus resulting in broad uncertainty around means, as ours are, these estimated dates for cutthroat trout diversification coincide with the 95% credible intervals for our estimates of diversification within S. californiensis (Figure 4) and could explain the highly divergent Whale Creek population of gill lice. However, the remaining phylogeographic patterns are not consistent with long-term persistence on the landscape in areas where gill lice are spreading. Aside from the Whale Creek and Russian samples, and one White River haplotype, the current geographic structure across the state appears to be one of admixture rather than structure associated with drainage divides or other geographic boundaries (Figures 3 and 4).

Comparing haplotypes from populations in a region of probable origin (Oregon), a region where introduction was highly likely (Pennsylvania), and major river drainages across Colorado conveys a plausible invasion route of S. californiensis across the United States. Based on the limited number of haplotypes observed across Colorado and other states, we can reasonably infer that S. californiensis in Colorado, Idaho, and Pennsylvania originated in the same geographic area given that Haplotype B, the most widespread haplotype, is shared across those areas (Figure 3). Though there was no direct overlap in Colorado haplotypes and the two from Oregon, Haplotypes G and P, the former was only 2–3 mutational steps away from the geographically widespread Haplotype B and other haplotypes in Colorado. Given that there are only 11 haplotypes detected in our Colorado samples (two from Whale Creek, and nine from all other sampling sites), it may be that S. californiensis was introduced to the state some time ago from the North American west coast, in the late 1800s or early 1900s. This scenario is consistent with the first time gill lice were documented in the state (Wilson 1909). External conditions permitting, gill lice produce an average of four generations per year (Gunn 2012), meaning that if they were introduced to the state 100–150 years ago, hundreds of generations have persisted in Colorado since their presumed introduction. While this is not likely enough time for this many haplotypes to have evolved, it is likely that, if introduced, the source population would have had multiple closely related haplotypes. Additional sampling from the Pacific Northwest region of western North America would be informative in that regard. Moreover, our divergence time estimates for Salmincola are slightly more recent than those estimated by Shedko et al. (2023), despite using the same mutation rates, likely due to the use of different priors on the trees. This does highlight the need for better taxonomic sampling, inclusion of multi-locus data (rapidly evolving markers in particular), and fossil calibrations, if possible, to evaluate phylogenetic structure and better understand the timing of diversification within Salmincola (and among copepodids in general).

Management Implications

Management of S. californiensis in Colorado is dependent on several external factors. As there is no current effective method of gill lice control in wild settings, continued monitoring of the prevalence and expanding range of gill lice in Colorado is important as environmental conditions continue to change. The current 24-year megadrought plaguing the southwestern United States (which began in the year 2000; Williams et al. 2022) presents significant problems for native fishes via habitat loss/fragmentation, thermal stress, decreased dissolved oxygen concentrations, and increased disease susceptibility (Gido et al. 2023). Extended periods of warmer water do stress cold-adapted salmonids (Bui et al. 2022) and give S. californiensis the opportunity to develop faster and produce larger broods more often than usual (Vigil et al. 2016). Crowded conditions and warmer temperatures produced by low water megadrought conditions also provide copepodids with concentrated host populations, increasing infection success rates and contributing to the rapid proliferation of parasites (Sutherland and Wittrock 1985; Vigil et al. 2016; Murphy, Gerth, Pauk, et al. 2020; Ugelvik et al. 2022). For example, Blue Mesa Reservoir reached historically low lake levels in Summer 2018 and again in Summer 2022, concentrating kokanee and facilitating increased gill lice infections (Colorado Parks & Wildlife, unpublished data). Other Colorado reservoirs have experienced similar proliferation of gill lice in recent years (e.g., Elevenmile Reservoir and Green Reservoir). Studies on gill lice in other areas have shown similar increases in populations of Salmincola species (Mitro and Griffin 2018; Boone and Quinlan 2019; Hasegawa and Koizumi 2024), so this is not likely to be a regional phenomenon but appears to be occurring on a much broader scale. As the southwestern megadrought continues, it is paramount that monitoring gill lice prevalence continues simultaneously with native cutthroat trout population monitoring.

It seems prudent to screen for gill lice in hatchery operations prior to fish stocking in areas where salmonid fisheries are supplemented with stocking practices. While this would require additional resources for the implementation of such screenings, it is possible to treat for gill lice in a hatchery setting, whereas it is virtually impossible to treat for gill lice in a wild setting. Chemical treatments (e.g., hydrogen peroxide, pyrethroid, azamethiphos and/or imidacloprid bath treatments, and emamectin benzoate feed treatments [Gunn et al. 2012, Aldrin et al. 2023]) may effectively remove gill lice in hatchery settings, although these treatments may lose their effectiveness over time. Biological control agents have been effective at removing sea lice from marine salmonids (e.g., wrasses and other cleaner fish that consume sea lice in Atlantic salmon fish farms); however, biological controls for freshwater gill lice have not yet been identified.

We acknowledge that the target region for the 28S eDNA assay we used is identical for S. californiensis and S. edwardsii (see Appendix S2). [An anonymous reviewer pointed out that this sequence is also identical for S. markewitschi based on GenBank sequences they evaluated.] Therefore, this assay may be more useful in detecting Salmincola at the genus level than it is for species-specific detections in areas where multiple Salmincola species co-occur, or where gill lice species composition of an area of interest is not well understood. While this was unlikely to impact our study, given that S. californiensis is the only species of louse we encountered in Colorado (confirmed by our COI barcode sequencing and phylogenetic analyses), it is an important consideration for researchers considering using this assay outside of our study area. Future specificity testing may help determine the utility of this assay for other species/genera in Family Lernaeopodidae. It seems as though cross-amplification of S. californiensis, S. edwardsii, and S. markewitschi would be likely. Nevertheless, this assay could be useful in eDNA detections of those species in areas where they are the only Salmincola species to occur.

It warrants mentioning that eDNA extracts are preserved at −80°C at Pisces Molecular LLC. While we focused on S. californiensis eDNA in our study, these samples are a valuable resource for future studies focusing on metabarcoding the biological community at each site, retrospectively focusing on host eDNA at each site, or screening for potential biocontrol species present at sites where gill lice were not detected.

Conclusions

Our study confirms that the gill louse S. californiensis continues to expand its range across the state of Colorado with detections in three rivers where gill lice were not previously documented (Henson Creek, Huerfano River [two sites], and the San Miguel River; see Figure 1). The phylogeographic pattern exhibited by our DNA barcoding data is consistent with human-mediated dispersal of S. californiensis, likely from decades-long fish stocking efforts, given the shared COI haplotypes across drainage divides and substantial geographic distances. Environmental DNA is effective at detecting the presence of gill lice in a system, but perhaps a combined approach using both eDNA and traditional electroshocking methods gives fisheries managers the best chance of detection, particularly in the early stages of gill lice invading a new area. Continued monitoring will be an important component of future cutthroat trout conservation and management efforts. Future research efforts focused on additional sampling of gill lice across western North America, better understanding environmental factors that influence the probability of gill lice detection using eDNA, the efficacy of eDNA in screening hatchery populations (perhaps before and after chemical treatment), and attempts at developing biological controls would enhance our understanding of this system and offer management solutions.

Author Contributions

G.J.S., S.J.P., and D.D.H. designed the project. S.J.P. collected all the samples and data with assistance from D.D.H. and E.M.V. (and many others listed in the acknowledgements). J.S.W. developed the qPCR assay and oversaw the molecular laboratory work. S.J.P., D.D.H., M.K., and J.S.W. conducted the data analyses. S.J.P. and D.D.H. led the writing, but all authors contributed to writing, editing, and revising the manuscript. All authors contributed to and approved the final version of the manuscript.

Acknowledgments

We thank the many Colorado Parks & Wildlife biologists who made significant contributions in planning and implementing the field work conducted in this study, specifically Dan Brauch, Jim White, Tory Eyre, Bill Atkinson, Jon Ewert, Carrie Tucker, Benjamin Felt, Matt Kondratieff, Tyler Swarr, and Paul Winkle. We also thank several people from Western Colorado University who assisted in the field: Ashley Lee, Amanda Aulenbach, Liam Duggan, Andrew Martinez, and Jett Moore. Emily Underwood (Idaho Fish and Game) provided S. californiensis samples from Idaho, Ethan Gardner (Oregon State University) provided samples from Oregon, and Jason Detar (Pennsylvania Fish and Boat Commission) and Meredith Batron (U.S. Fish & Wildlife Service) provided samples from Pennsylvania. Sarah Spotten, Patrick Power, and Jackson Strobel (Pisces Molecular LLC) helped conduct the lab work for eDNA qPCR and DNA barcoding. Alex Peasley (Colorado Parks & Wildlife) was the first to discover gill lice in Whale Creek, CO. We thank Dr. Brian Dalton (Western Colorado University) for early insights regarding project design. This research was conducted under Colorado Parks & Wildlife YT-1 Aquatic Scientific Collection license.

Ethics Statement

All animals included in this study were ethically and humanely sampled and processed under valid state collecting permits and following proper field safety protocols.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

All DNA sequences newly generated in this study have been deposited in GenBank (accession numbers PP942459–PP942530).

References

Adams, C. I. M., M. Knapp, N. J. Gemmell, et al. 2019. “Beyond Biodiversity: Can Environmental DNA (eDNA) Cut It as a Population Genetics Tool?” Genes 10: 1–20.

Ahmed, D. A., E. J. Hudgins, R. N. Cuthbert, et al. 2022. “Managing Biological Invasions: The Cost of Inaction.” Biological Invasions 24: 1927–1946.

Akaike, H. 1973. “Information Theory and an Extension of the Maximum Likelihood Principle.” In Second International Symposium on Information Theory, edited by B. N. Petrov and B. F. Csaki, 267–281. Akademiai Kiado.

Aldrin, M., R. B. Huseby, L. C. Stige, and K. O. Helgesen. 2023. “Estimated Effectiveness of Treatments Against Salmon Lice in Marine Salmonid Farming.” Aquaculture 575: 739749.

Allan, E. A., W. G. Zhang, A. C. Lavery, and A. F. Govindarajan. 2021. “Environmental DNA Shedding and Decay Rates From Diverse Animal Forms and Thermal Regimes.” Environmental DNA 3: 492–514.

Andersen, M. C., H. Adams, B. Hope, and M. Powell. 2004. “Risk Assessment for Invasive Species.” Risk Analysis 24: 787–793.

Andruszkiewicz, E. A., L. M. Sassoubre, and A. B. Boehm. 2017. “Persistence of Marine Fish Environmental DNA and the Influence of Sunlight.” PLoS One 12: e0185043.

Barndt, S., and J. Stone. 2003. “Infestation of Salmincola californiensis (Copepoda: Lernaeopodidae) in Wild Coho Salmon, Steelhead, and Coastal Cutthroat Trout Juveniles in a Small Columbia River Tributary.” Transactions of the American Fisheries Society 132: 1027–1032.

Barnes, M. A., W. L. Chadderton, C. L. Jerde, A. R. Mahon, C. R. Turner, and D. M. Lodge. 2021. “Environmental Conditions Influence eDNA Particle Size Distribution in Aquatic Systems.” Environmental DNA 3: 643–653.

Barnes, M. A., C. R. Turner, C. L. Jerde, M. A. Renshaw, W. L. Chadderton, and D. M. Lodge. 2014. “Environmental Conditions Influence eDNA Persistence in Aquatic Systems.” Environmental Science and Technology 48: 1819–1827.

Bass, D., K. W. Christison, G. D. Stentiford, L. S. J. Cook, and H. Hartikainen. 2023. “Environmental DNA/RNA for Pathogen and Parasite Detection, Surveillance, and Ecology.” Trends in Parasitology 39: 285–304.

Bellard, C., C. Bernery, and C. Leclerc. 2021. “Looming Extinctions due to Invasive Species: Irreversible Loss of Ecological Strategy and Evolutionary History.” Global Change Biology 27: 4967–4979.

Bellard, C., P. Genovesi, and J. M. Jeschke. 2016. “Global Patterns in Threats to Vertebrates by Biological Invasions.” Proceedings of the Royal Society B: Biological Sciences 283: 2454.

Black, G. A. 1982. “Gills as an Attachment Site for Salmincola edwardsii (Copepoda: Lernaeopodidae).” Journal of Parasitology 68: 1172–1173.

Boone, E. C., and H. R. Quinlan. 2019. “Occurrence of Salmincola sp. on Brook Charr Salvelinus fontinalis in Lake Superior.” Hydrobiologia 840: 301–308.

Bouckaert, R., T. G. Vaughan, J. Barido‐Sottani, et al. 2019. “BEAST 2.5: An Advanced Software Platform for Bayesian Evolutionary Analysis.” PLoS Computational Biology 15: 1–28.

Bouckaert, R. R. 2022. “An Efficient Coalescent Epoch Model for Bayesian Phylogenetic Inference.” Systematic Biology 71: 1549–1560.

Brandão‐Dias, P. F. P., D. M. C. Hallack, E. D. Snyder, et al. 2023. “Particle Size Influences Decay Rates of Environmental DNA in Aquatic Systems.” Molecular Ecology Resources 23: 756–770.

Bui, S., A. Madaro, J. Nilsson, et al. 2022. “Warm Water Treatment Increased Mortality Risk in Salmon.” Veterinary and Animal Science 17: 100265.

Burian, A., Q. Mauvisseau, M. Bulling, S. Domisch, S. Qian, and M. Sweet. 2021. “Improving the Reliability of eDNA Data Interpretation.” Molecular Ecology Resources 21: 1422–1433.

Cable, J., I. Barber, B. Boag, et al. 2017. “Global Change, Parasite Transmission and Disease Control: Lessons From Ecology.” Philosophical Transactions of the Royal Society, B: Biological Sciences 372: 20160088.

Carlton, J. T. 1996. “Biological Invasions and Cryptogenic Species.” Ecology 77: 1653–1655.

Chalkowski, K., C. A. Lepczyk, and S. Zohdy. 2018. “Parasite Ecology of Invasive Species: Conceptual Framework and New Hypotheses.” Trends in Parasitology 34: 655–663.

Chinchio, E., M. Crotta, C. Romeo, J. A. Drewe, J. Guitian, and N. Ferrari. 2020. “Invasive Alien Species and Disease Risk: An Open Challenge in Public and Animal Health.” PLoS Pathogens 16: 1–7.

Clement, M., D. Posada, and K. a. Crandall. 2000. “TCS: A Computer Program to Estimate Gene Genealogies.” Molecular Ecology 9: 1657–1659.

Copilaş‐Ciocianu, D., and A. Petrusek. 2017. “Phylogeography of a Freshwater Crustacean Species Complex Reflects a Long‐Gone Archipelago.” Journal of Biogeography 44: 421–432.

Cornils, A., and C. Held. 2014. “Evidence of Cryptic and Pseudocryptic Speciation in the Paracalanus parvus Species Complex (Crustacea, Copepoda, Calanoida).” Frontiers in Zoology 11: 1–17.

Crane, L. C., J. S. Goldstein, D. W. Thomas, K. S. Rexroth, and A. W. Watts. 2021. “Effects of Life Stage on eDNA Detection of the Invasive European Green Crab (Carcinus maenas) in Estuarine Systems.” Ecological Indicators 124: 107412.

Darriba, D., G. L. Taboada, R. Doallo, and D. Posada. 2012. “jModelTest 2: More Models, New Heuristics and Parallel Computing.” Nature Methods 9: 772.

Dass, M. A., C. D. H. Sherman, R. A. H. van Oorschot, et al. 2024. “Assessing eDNA Capture Method From Aquatic Environment to Optimise Recovery of Human Mt‐eDNA.” Forensic Science International 361: 112085.

Diamond, J. 1984. ““Normal” Extinction of Isolated Populations.” In Extinctions, edited by M. H. Nitecki, 191–246. Chicago University Press.

Douglas, J., R. Zhang, and R. Bouckaert. 2021. “Adaptive Dating and Fast Proposals: Revisiting the Phylogenetic Relaxed Clock Model.” PLoS Computational Biology 17: 1–30.

Drummond, A. J., S. Y. W. Ho, M. J. Phillips, and A. Rambaut. 2006. “Relaxed Phylogenetics and Dating With Confidence.” PLoS Biology 4: 699–710.

Early, R., B. A. Bradley, J. S. Dukes, et al. 2016. “Global Threats From Invasive Alien Species in the Twenty‐First Century and National Response Capacities.” Nature Communications 7: 12485.

Falke, J. A., L. L. Bailey, K. D. Fausch, and K. R. Bestgen. 2012. “Colonization and Extinction in Dynamic Habitats: An Occupancy Approach for a Great Plains Stream Fish Assemblage.” Ecology 93: 858–867.

Farrell, J. A., L. Whitmore, and D. J. Duffy. 2021. “The Promise and Pitfalls of Environmental DNA and RNA Approaches for the Monitoring of Human and Animal Pathogens From Aquatic Sources.” Bioscience 71: 609–625.

Fiske, I. J., and R. B. Chandler. 2011. “Unmarked: An R Package for Fitting Hierarchical Models of Wildlife Occurrence and Abundance.” Journal of Statistical Software 43, no. 10: 1–23.

Folmer, O., M. Black, W. Hoeh, R. Lutz, and R. Vrijenhoek. 1994. “DNA Primers for Amplification of Mitochondrial Cytochrome c Oxidase Subunit I From Diverse Metazoan Invertebrates.” Molecular Marine Biology and Biotechnology 3: 294–299.

Gallardo, B., M. Clavero, M. I. Sánchez, and M. Vilà. 2016. “Global Ecological Impacts of Invasive Species in Aquatic Ecosystems.” Global Change Biology 22: 151–163.

Gido, K. B., M. J. Osborne, D. L. Propst, T. F. Turner, and J. D. Olden. 2023. “Megadroughts Pose Mega‐Risk to Native Fishes of the American Southwest.” Fisheries 48: 204–214.

Goldberg, C. S., C. R. Turner, K. Deiner, et al. 2016. “Critical Considerations for the Application of Environmental DNA Methods to Detect Aquatic Species.” Methods in Ecology and Evolution 7: 1299–1307.

Green, P. T., D. J. O'Dowd, K. L. Abbott, M. Jeffery, K. Retallick, and R. Mac Nally. 2011. “Invasional Meltdown: Invader‐Invader Mutualism Facilitates a Secondary Invasion.” Ecology 92: 1758–1768.

Groce, M. C., L. L. Bailey, and K. D. Fausch. 2012. “Evaluating the Success of Arkansas Darter Translocations in Colorado: An Occupancy Sampling Approach.” Transactions of the American Fisheries Society 141: 825–840.

Guindon, S., and O. Gascuel. 2003. “A Simple, Fast, and Accurate Algorithm to Estimate Large Phylogenies by Maximum Likelihood.” Systematic Biology 52: 696–704.

Gunn, C. 2012. “Salmincola in Colorado (Salmincola sp.) Biosheet.”

Gunn, C., D. Carty, P. W. Walker, P. A. Colburn, and J. D. Bowker. 2012. “Pilot Field Trial to Evaluate SLICE (0.2% Emamectin Benzoate)‐Medicated Feed to Reduce a Natural Infestation of Salmincola californiensis in Freshwater‐Reared Rainbow Trout.” North American Journal of Aquaculture 74: 424–427.

Gurevitch, J., and D. K. Padilla. 2004. “Are Invasive Species a Major Cause of Extinctions?” Trends in Ecology & Evolution 19: 470–474.

Hargis, L., J. M. Lepak, E. M. Vigil, and C. Gunn. 2014. “Prevalence and Intensity of the Parasitic Copepod (Salmincola californiensis) on Kokanee Salmon (Oncorhynchus nerka) in a Reservoir in Colorado.” Southwestern Naturalist 59: 126–129.

Hasegawa, R., and I. Koizumi. 2024. “Consistent Negative Correlations Between Parasite Infection and Host Body Condition Across Seasons Suggest Potential Harmful Impacts of Salmincola markewitschi on Wild White‐Spotted Charr, Salvelinus leucomaenis.” Zoological Science 41: 192–200.

Hatcher, M. J., J. T. A. Dick, and A. M. Dunn. 2012. “Disease Emergence and Invasions.” Functional Ecology 26: 1275–1287.

Heled, J., and R. R. Bouckaert. 2013. “Looking for Trees in the Forest: Summary Tree From Posterior Samples.” BMC Evolutionary Biology 13: 221.

Herron, C. L., M. L. Kent, and C. B. Schreck. 2018. “Swimming Endurance in Juvenile Chinook Salmon Infected With Salmincola californiensis.” Journal of Aquatic Animal Health 30: 81–89.

Holman, L. E., Y. Chng, and M. Rius. 2022. “How Does eDNA Decay Affect Metabarcoding Experiments?” Environmental DNA 4: 108–116.

Hunter, M. E., J. A. Ferrante, G. Meigs‐Friend, and A. Ulmer. 2019. “Improving eDNA Yield and Inhibitor Reduction Through Increased Water Volumes and Multi‐Filter Isolation Techniques.” Scientific Reports 9: 1–9.

Jo, T., H. Murakami, S. Yamamoto, R. Masuda, and T. Minamoto. 2019. “Effect of Water Temperature and Fish Biomass on Environmental DNA Shedding, Degradation, and Size Distribution.” Ecology and Evolution 9: 1135–1146.

Jo, T. S. 2023. “Utilizing the State of Environmental DNA (eDNA) to Incorporate Time‐Scale Information Into eDNA Analysis.” Proceedings of the Royal Society B: Biological Sciences 290: 20230979.

Johnson, M., S. Tetzlaff, A. Katz, and J. Sperry. 2024. “Comparison of qPCR and Metabarcoding for Environmental DNA Surveillance of a Freshwater Parasite.” Ecology and Evolution 14: 1–10.

Johnson, S. C., J. W. Treasurer, S. Bravo, K. Nagasawa, and Z. Kabata. 2004. “A Review of the Impact of Parasitic Copepods on Marine Aquaculture.” Zoological Studies 43: 229–243.

Kabata, Z. 1969. “Revision of the Genus Salmincola Wilson, 1915 (Copepoda: Lernaeopodidae).” Journal of the Fisheries Research Board of Canada 26: 2987–3041.

Kabata, Z., and B. Cousens. 1973. “Life Cycle of Salmincola californiensis.” Journal of the Fisheries Research Board of Canada 30: 881–903.

Katoh, K., and D. M. Standley. 2013. “MAFFT Multiple Alignment Software Version 7: Improvements in Performance and Usability.” Molecular Biology and Evolution 30: 772–780.

Katz, A. D., S. J. Tetzlaff, M. D. Johnson, et al. 2023. “Molecular Identification and Environmental DNA Detection of Gill Lice Ectoparasites Associated With Brook Trout Declines.” Transactions of the American Fisheries Society 152: 788–808.

Kellner, K. F., A. D. Smith, J. A. Royle, M. Kéry, J. L. Belant, and R. B. Chandler. 2023. “The Unmarked R Package: Twelve Years of Advances in Occurrence and Abundance Modelling in Ecology.” Methods in Ecology and Evolution 14: 1408–1415.

Kokkonen, A. L., P. C. Searle, D. K. Shiozawa, and R. P. Evans. 2024. “Using De Novo Transcriptomes to Decipher the Relationships in Cutthroat Trout Subspecies (Oncorhynchus clarkii).” Evolutionary Applications 17: e13735.

Korbel, K. L., G. C. Hose, C. Karwautz, et al. 2024. “Detection, Movement and Persistence of Invertebrate eDNA in Groundwater.” Scientific Reports 14: 1–14.

Krolicka, A., M. M. Nilsen, B. K. Hansen, M. W. Jacobsen, F. Provan, and T. Baussant. 2022. “Sea Lice (Lepeophtherius salmonis) Detection and Quantification Around Aquaculture Installations Using Environmental DNA.” PLoS One 17: e0274736.

Lamb, P. D., V. G. Fonseca, D. L. Maxwell, and C. C. Nnanatu. 2022. “Systematic Review and Meta‐Analysis: Water Type and Temperature Affect Environmental DNA Decay.” Molecular Ecology Resources 22: 2494–2505.

Larson, E. R., B. M. Graham, R. Achury, et al. 2020. “From eDNA to Citizen Science: Emerging Tools for the Early Detection of Invasive Species.” Frontiers in Ecology and the Environment 18: 194–202.

Lepak, J. M., A. G. Hansen, M. B. Hooten, D. Brauch, and E. M. Vigil. 2022. “Rapid Proliferation of the Parasitic Copepod, Salmincola californiensis (Dana), on Kokanee Salmon, Oncorhynchus nerka (Walbaum), in a Large Colorado Reservoir.” Journal of Fish Diseases 45: 89–98.

Loeza‐Quintana, T., C. M. Carr, T. Khan, et al. 2019. “Recalibrating the Molecular Clock for Arctic Marine Invertebrates Based on DNA Barcodes.” Genome 62: 200–216.

Lymbery, A. J., M. Morine, H. G. Kanani, S. J. Beatty, and D. L. Morgan. 2014. “Co‐Invaders: The Effects of Alien Parasites on Native Hosts.” International Journal for Parasitology: Parasites and Wildlife 3: 171–177.

Mächler, E., M. Osathanunkul, and F. Altermatt. 2018. “Shedding Light on eDNA: Neither Natural Levels of UV Radiation nor the Presence of a Filter Feeder Affect eDNA‐Based Detection of Aquatic Organisms.” PLoS One 13: 1–15.

Mackenzie, D. I., and J. A. Royle. 2005. “Designing Occupancy Studies: General Advice and Allocating Survey Effort.” Journal of Applied Ecology 42: 1105–1114.

Maebara, Y., M. Tamaoki, Y. Iguchi, et al. 2020. “Genetic Diversity of Invasive Spartina alterniflora Loisel. (Poaceae) Introduced Unintentionally Into Japan and Its Invasion Pathway.” Frontiers in Plant Science 11: 1–12.

Marbuah, G., I. Gren, and B. McKie. 2014. “Economics of Harmful Invasive Species: A Review.” Diversity 6: 500–523.

Miller, M. A., W. Pfeiffer, and T. Schwartz. 2010. “Creating the CIPRES Science Gateway for Inference of Large Phylogenetic Trees. Proceedings of the Gateway Computing Environments Workshop (GCE) 1–8.”

Mitro, M. G., and J. D. Griffin. 2018. “Distribution, Prevalence, and Maximum Intensity of the Ectoparasitic Copepod salmincola Cf. Edwardsii in Brook Trout in Wisconsin Streams.” Journal of Parasitology 104: 685–696.

Monzyk, F. R., T. A. Friesen, and J. D. Romer. 2015. “Infection of Juvenile Salmonids by Salmincola californiensis (Copepoda: Lernaeopodidae) in Reservoirs and Streams of the Willamette River Basin, Oregon.” Transactions of the American Fisheries Society 144: 891–902.

Mooney, H. A., and E. E. Cleland. 2001. “The Evolutionary Impact of Invasive Species.” Proceedings of the National Academy of Sciences of the United States of America 98: 5446–5451.

Mullin, B. R., and F. B. Reyda. 2020. “High Prevalence of the Copepod Salmincola Californiensis in Steelhead Trout in Lake Ontario Following Its Recent Invasion.” Journal of Parasitology 106: 198–200.

Murphy, C. A., W. Gerth, and I. Arismendi. 2020. “Hatching and Survival of the Salmon “Gill Maggot” Salmincola californiensis (Copepoda: Lernaeopodidae) Reveals Thermal Dependence and Undocumented Naupliar Stage.” Parasitology 147: 1338–1343.

Murphy, C. A., W. Gerth, K. Pauk, P. Konstantinidis, and I. Arismendi. 2020. “Hiding in Plain Sight: Historical Fish Collections Aid Contemporary Parasite Research.” Fisheries 45: 263–270.

Nagasawa, K. 2023. “Live Freshwater Parasite, Salmincola californiensis (Copepoda: Lernaeopodidae), on the Gills of an Ocean‐Migrating Steelhead Trout (Oncorhynchus mykiss) and Discussion on the Origin and Survival of the Parasite at Sea.” Zoological Science 40: 360–366.

Nagasawa, K., and S. Urawa. 2002. “Infection of Salmincola californiensis (Copepoda: Lernaeopodidae) on Juvenile Masu Salmon (Oncorhynchus masou) From a Stream in Hokkaido.” Bulletin of the National Salmon Resources Center 5: 7–12.

Neal, T., M. L. Kent, J. Sanders, C. B. Schreck, and J. T. Peterson. 2021. “Laboratory Infection Rates and Associated Mortality of Juvenile Chinook Salmon (Oncorhynchus tshawytscha) From Parasitic Copepod (Salmincola californiensis).” Journal of Fish Diseases 44: 1423–1434.

Ogram, A., G. S. Sayler, and T. Barkay. 1987. “The Extraction and Purification of Microbial DNA From Sediments.” Journal of Microbiological Methods 7: 57–66.

Pawaputanon, K. 1980. “Effects of Parasitic Copepod, Salmincola californiensis (Dana, 1852) on Juvenile Sockey Salmon, Oncorhynchus nerka (Walbaum). MS Thesis, University of British Columbia.”

Pawlowski, J., L. Apothéloz‐Perret‐Gentil, and F. Altermatt. 2020. “Environmental DNA: What's Behind the Term? Clarifying the Terminology and Recommendations for Its Future Use in Biomonitoring.” Molecular Ecology 29: 4258–4264.

Peters, L., S. Spatharis, M. A. Dario, et al. 2018. “Environmental DNA: A New Low‐Cost Monitoring Tool for Pathogens in Salmonid Aquaculture.” Frontiers in Microbiology 9: 1–10.

Piasecki, W., A. E. Goodwin, J. C. Eiras, and B. F. Nowak. 2004. “Importance of Copepoda in Freshwater Aquaculture.” Zoological Studies 43: 193–205.

Picq, S., A. N. Curtis, A. George, et al. 2024. “Environmental DNA Metabarcoding for Whole Community Inventories of Vertebrates in Rivers of the Midwestern United States.” Journal of Freshwater Ecology 39: 2382454.

Pilliod, D. S., C. S. Goldberg, R. S. Arkle, and L. P. Waits. 2014. “Factors Influencing Detection of eDNA From a Stream‐Dwelling Amphibian.” Molecular Ecology Resources 14: 109–116.

Poulin, R. 2017. “Invasion Ecology Meets Parasitology: Advances and Challenges.” International Journal for Parasitology: Parasites and Wildlife 6: 361–363.

Pyšek, P., and D. M. Richardson. 2010. “Invasive Species, Environmental Change and Management, and Health.” Annual Review of Environment and Resources 35: 25–55.

Rambaut, A., A. J. Drummond, D. Xie, G. Baele, and M. A. Suchard. 2018. “Posterior Summarization in Bayesian Phylogenetics Using Tracer 1.7.” Systematic Biology 67: 901–904.

Rees, H. C., B. C. Maddison, D. J. Middleditch, J. R. M. Patmore, and K. C. Gough. 2014. “The Detection of Aquatic Animal Species Using Environmental DNA ‐ a Review of eDNA as a Survey Tool in Ecology.” Journal of Applied Ecology 51: 1450–1459.

Rogosch, J. S., and J. D. Olden. 2021. “Comparing Opportunistic and Strategic Removal Efforts to Manage Invasive Fish Species Using a Dynamic Multi‐State Occupancy Model.” Journal of Applied Ecology 58: 2797–2809.

Rounds, C. I., T. W. Arnold, C. Lan, et al. 2024. “Aquatic Invasive Species Exhibit Contrasting Seasonal Detectability Patterns Based on Environmental DNA: Implications for Monitoring.” Freshwater Biology 00: 1–15.

Roussel, J. M., J. M. Paillisson, A. Tréguier, and E. Petit. 2015. “The Downside of eDNA as a Survey Tool in Water Bodies.” Journal of Applied Ecology 52: 823–826.

Ruiz, C. F., J. M. Rash, D. A. Besler, et al. 2017. “Exotic “Gill Lice” Species (Copepoda: Lenaeopodidae: Salmincola spp.) Infect Rainbow Trout (Oncorhynchus mykiss) and Brook Trout (Salvelinus fontinalis) in the Southeastern United States.” Journal of Parasitology 103: 377–389.

Ruppert, K. M., R. J. Kline, and M. S. Rahman. 2019. “Past, Present, and Future Perspectives of Environmental DNA (eDNA) Metabarcoding: A Systematic Review in Methods, Monitoring, and Applications of Global eDNA.” Global Ecology and Conservation 17: e00547.

Russell, J. C. 2012. “Do Invasive Species Cause Damage? Yes.” Bioscience 62: 217.

Sahu, A., N. Kumar, C. Pal Singh, and M. Singh. 2023. “Environmental DNA (eDNA): Powerful Technique for Biodiversity Conservation.” Journal for Nature Conservation 71: 126325.

Schmidt, B. R., M. Kéry, S. Ursenbacher, O. J. Hyman, and J. P. Collins. 2013. “Site Occupancy Models in the Analysis of Environmental DNA Presence/Absence Surveys: A Case Study of an Emerging Amphibian Pathogen.” Methods in Ecology and Evolution 4: 646–653.

Sepulveda, A. J. 2018. “Novel Application of Explicit Dynamics Occupancy Models to Ongoing Aquatic Invasions.” Journal of Applied Ecology 55: 917–925.

Sepulveda, A. J., P. R. Hutchins, M. Forstchen, M. N. Mckeefry, and A. M. Swigris. 2020. “The Elephant in the Lab (And Field): Contamination in Aquatic Environmental DNA Studies.” Frontiers in Ecology and Evolution 8: 609973.

Sepulveda, A. J., N. M. Nelson, C. L. Jerde, and G. Luikart. 2020. “Are Environmental DNA Methods Ready for Aquatic Invasive Species Management?” Trends in Ecology & Evolution 35: 668–678.

Shao, R., and S. C. Barker. 2007. “Mitochondrial Genomes of Parasitic Arthropods: Implications for Studies of Population Genetics and Evolution.” Parasitology 134: 153–167.

Shedko, S. V., M. B. Shedko, I. L. Miroshnichenko, and G. A. Nemkova. 2023. “DNA Identification of Parasitic Copepods salmincola (Copepoda, Siphonostomatoida, Lernaeopodidae): Variability and Rate of Evolution of the Mitochondrial Cytochrome c Oxidase Subunit I Gene.” Russian Journal of Genetics 59: 1022–1031.

Shiozawa, D. K., R. P. Evans, D. D. Houston, and P. J. Unmack. 2018. “Geographic Variation, Isolation, and Evolution of Cutthroat Trout With Comments on Future Directions for Management and Research.” In Cutthroat Trout Evolutionary Biology and Taxonomy, edited by P. Trotter, P. Bisson, L. Schultz, and B. Roper, 129–172. American Fisheries Society.

Simberloff, D. 2010. “Invasive Species.” In Conservation Biology for All, edited by N. S. Sodhi and P. R. Ehrlich, 131–152. Oxford University Press.

Stamatakis, A. 2014. “RAxML Version 8: A Tool for Phylogenetic Analysis and Post‐Analysis of Large Phylogenies.” Bioinformatics (Oxford, England) 30: 1312–1313.

Stamatakis, A., P. Hoover, and J. Rougemont. 2008. “A Rapid Bootstrap Algorithm for the RAxML Web Servers.” Systematic Biology 57: 758–771.

Stauffer, H., C. J. Ralph, and S. L. Miller. 2002. “Incorporating Detection Uncertainty Into Presence‐Absence Surveys for Marbled Murrelet.” In Predicting Species Occurrences: Issues of Accuracy and Scale, edited by J. M. Scott, P. J. Heglund, M. L. Morrison, et al., 357–365. Island Press.

Strickler, K. M., A. K. Fremier, and C. S. Goldberg. 2015. “Quantifying Effects of UV‐B, Temperature, and pH on eDNA Degradation in Aquatic Microcosms.” Biological Conservation 185: 85–92.

Sutherland, D. R., and D. D. Wittrock. 1985. “The Effects of Salmincola californiensis (Copepoda; Lernaeopodidae) on the Gills of Farm Raised Rainbow Trout (Salmo gairdneri).” Canadian Journal of Zoology 63: 2893–2901.

Taberlet, P., A. Bonin, L. Zinger, and E. Coissac. 2018. Environmental DNA: For Biodiversity Research and Monitoring. Oxford University Press.

Tetzlaff, S. J., A. D. Katz, M. D. Johnson, and J. H. Sperry. 2024. “Community Ecology in a Bottle: Leveraging eDNA Metabarcoding Data to Predict Occupancy of Co‐Occurring Species.” Environmental DNA 6: 1–13.

Thomas, A. C., J. Howard, P. L. Nguyen, T. A. Seimon, and C. S. Goldberg. 2018. “eDNA Sampler: A Fully Integrated Environmental DNA Sampling System.” Methods in Ecology and Evolution 9: 1379–1385.

Thomsen, P. F., and E. Willerslev. 2015. “Environmental DNA—An Emerging Tool in Conservation for Monitoring Past and Present Biodiversity.” Biological Conservation 183: 4–18.

Ugelvik, M. S., S. Mæhle, and S. Dalvin. 2022. “Temperature Affects Settlement Success of Ectoparasitic Salmon Lice (Lepeophtheirus salmonis) and Impacts the Immune and Stress Response of Atlantic Salmon (Salmo salar).” Journal of Fish Diseases 45: 975–990.

Vander Zanden, M. J., G. J. A. Hansen, S. N. Higgins, and M. S. Kornis. 2010. “A Pound of Prevention, Plus a Pound of Cure: Early Detection and Eradication of Invasive Species in the Laurentian Great Lakes.” Journal of Great Lakes Research 36: 199–205.

Victorian Government. 2010. “Invasive Plants and Animals Policy Framework.”

Vigil, E. M., K. R. Christianson, J. M. Lepak, and P. J. Williams. 2016. “Temperature Effects on Hatching and Viability of Juvenile Gill Lice, Salmincola californiensis.” Journal of Fish Diseases 39: 899–905.

White, B. N. 1994. “Vicariance Biogeography of the Open‐Ocean Pacific.” Progress in Oceanography 34: 257–284.

Wilcove, D. S., D. Rothstein, J. Dubow, A. Phillips, and E. Losos. 1998. “Quantifying Threats to Imperiled Species in the United States.” Bioscience 48: 607–615.

Williams, A. P., B. I. Cook, and J. E. Smerdon. 2022. “Rapid Intensification of the Emerging Southwestern North American Megadrought in 2020–2021.” Nature Climate Change 12: 232–234.

Wilson, C. B. 1909. “North American Parasitic Copepods: A List of Those Found Upon the Fishes of the Pacific Coast, With Descriptions of New Genera and Species.” National Museum Proceedings 35: 431–481.