BASELINE HEALTH AND DISEASE ASSESSMENT OF FOUNDER EASTERN QUOLLS (DASYURUS VIVERRINUS) DURING A CONSERVATION TRANSLOCATION TO MAINLAND AUSTRALIA

The eastern quoll (Dasyurus viverrinus) is a medium-sized (700–2,000 g), predominantly terrestrial, carnivorous marsupial within the family Dasyuridae (Godsell 1983; Jones and Rose 2001). Eastern quolls are found in a wide range of habitats but occur at highest densities in open grasslands and woodland mosaics (Godsell 1983). Invertebrates constitute the bulk of the diet, but quolls will also consume small vertebrates, carrion, and vegetable matter (Jones and Rose 2001). Eastern quolls are sexually dimorphic, with synchronized patterns of breeding, resulting in the birth of young in early winter (Godsell 1983). Eastern quolls were formerly abundant and widespread throughout southeastern Australia. The species is presumed extirpated from the Australian mainland with the last confirmed sighting in 1963 (Frankham et al. 2017). The species remains relatively widespread and common on the island of Tasmania, but recent evidence suggests this population may be under threat and in decline (Fancourt et al. 2013). As part of an ongoing multispecies, multitrophic, grassy box-woodland restoration project (Manning et al. 2011; Shorthouse et al. 2012), eastern quolls were translocated to Mulligans Flat Woodland Sanctuary, in the Australian Capital Territory (ACT), in two separate conservation translocations in March 2016 and June 2017.

Wildlife health surveillance is frequently advocated and is increasingly undertaken, during Australian conservation translocations. However, the baseline health and disease status of some Australian marsupials remain largely unknown. Performing comprehensive health evaluations during conservation translocations serves several functions (Mathews et al. 2006; Portas 2019). When this information is unknown, health evaluations allow for the establishment of baseline health and disease data that can be applied to conservation management of both the source and translocated populations. These data can also be used in disease risk analysis and mitigation strategies for subsequent conservation translocations and to form the basis for longitudinal health and disease monitoring of the translocated populations. Health evaluations also allow for the selection of candidates of appropriate health status, enhancing both conservation and welfare outcomes in translocations.

We undertook health evaluations of eastern quolls, during a conservation translocation, using techniques previously established in another medium-sized marsupial, the eastern bettong (Bettongia gaimardi; Portas et al. 2014, 2016). We assessed quolls for the presence of potential pathogens (Toxoplasma gondii, herpesviruses, Salmonella serovars, protozoan hemoparasites, and ectoparasites) and whether sex, provenance (wild caught versus captive bred), and two potential pathogens (T. gondii and herpesviruses) influenced hematologic and biochemical variables or body weight. Additionally, we assessed the relationship between provenance and infection with T. gondii and herpesviruses.

Eastern quoll founders were either wild caught or captive bred. Captive-bred founders were sourced from Mount Rothwell Biodiversity Interpretation Centre, which is 60 km southwest of Melbourne, Australia. Wild-caught founders were sourced from free-ranging populations across six locations in Tasmania (Fig. 1). Wild-caught quolls were removed from traps and placed into plastic carrier cages for transport. Quolls were transported by road within Tasmania and were subsequently freighted by air to the ACT (total transport times ranged from 7 to 18 h). Quolls were not sedated for transport.

On arrival in the ACT, quolls were anesthetized with isoflurane (ISO Inhalation Anaesthetic, Veterinary Companies of Australia Pty. Ltd., Kings Park, Australia) in oxygen, delivered via mask. All quolls received a complete physical examination under anesthesia; weight was determined to the nearest gram with an electronic scale, and any physical abnormalities were noted and investigated as required.

Up to 6 mL of blood was collected from the jugular or femoral vein and placed into ethylenediaminetetraacetic acid (EDTA) and serum separator tubes (BD Vacutainer, Becton, Dickinson and Company, Plymouth, UK). Serum tubes were allowed to clot for a minimum of 4 h and centrifuged, and the serum was harvested into nylon storage vials (Nalgene^® Cryovials, Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA). Sera and EDTA blood were stored at 4 C until processing for hematologic and biochemical analyses. Sera for T. gondii serology were stored at –20 C until processing within 1 mo of collection. We stored 0.5 mL of EDTA blood at –20 C for up to 6 mo before analysis for hemoparasite DNA.

Aluminum-shafted swabs (Sterile Swab Applicators, Copan Italia, Brescia, Italy) were used for combined sampling of the mucosa of the conjunctiva, oropharynx, and urogenital tract of individual animals for herpesvirus DNA, then stored at room temperature until processing. Sterile bacteriologic swabs (Sterile Transport Swabs, Interpath Service Pty. Ltd., Melbourne, Victoria, Australia) were used to collect rectal fecal samples and were submitted to a commercial laboratory for Salmonella serovar culture. Representative ectoparasites were collected with forceps (ticks) or by combing the fur with a flea comb (fleas and mites), and burdens were scored as none (zero ectoparasites detected), low (<10 ectoparasites detected), moderate (10–20 ectoparasites detected), or high (>20 ectoparasites detected). Ectoparasites were stored in 70% ethanol until identification using published literature (Roberts 1970; von Kéler 1971; Dunnet and Mardon 1974).

Hematologic and biochemical analyses and Salmonella species culturing were performed within 24 h of collection (Vetnostics, North Ryde, New South Wales, Australia). Hematologic analysis was performed on EDTA whole blood, with initial processing on a Sysmex XT-2000i Automated Hematology Analyzer (Sysmex America Inc., Mundelein, Illinois, USA), followed by a manual differential cell count. Hemoglobin, red blood cell count, mean corpuscular volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin content, platelet count, neutrophils, lymphocytes, monocytes, eosinophils, and basophils were measured. Serum biochemistry was performed on a Roche Modular EVO Analyzer (Roche Products Pty. Ltd., Castle Hill, New South Wales, Australia). Sodium, potassium, chloride, bicarbonate, anion gap, urea, creatinine, glucose, bilirubin, aspartate amino transferase (AST), alanine transaminase (ALT), gamma glutamyl transferase, alkaline phosphatase (ALP), total protein, albumin, globulin, albumin:globulin ratio, calcium, phosphate, creatine kinase, cholesterol, and triglyceride were also measured. For Salmonella species culture, swabs were inoculated onto xylose lysine deoxycholate (XLD) agar medium and mannitol selenite broth and incubated at 35 C for 48 h in an aerobic atmosphere. The mannitol selenite broth was then subcultured onto an additional XLD agar plate at 24 h after inoculation. Both XLD plates were then read at 48 h after inoculation.

Toxoplasma gondii serology was performed at the Department of Primary Industries, Water and Environment (Mount Pleasant Laboratories, Launceston, Australia) using the modified agglutination test for antibodies to T. gondii (Fancourt et al. 2014). A titer of >64 was considered positive for T. gondii infection (Dubey and Desmonts 1987).

For the detection of herpesviruses DNA, DNA was extracted from pooled swab samples from each individual animal, as previously described (Vaz et al. 2011). Extracted DNA was then used as a template in a generic panherpesvirus PCR, using primers targeting a conserved region of the herpesvirus DNA polymerase gene, approximately 210–230 base pairs in length (Chmielewicz et al. 2003), and amplicon melt profiles were compared as per Vaz et al. (2013). The PCR products were purified, and their DNA sequence was determined. The predicted amino acid sequence was compared with publicly available sequences using BLAST in GenBank online algorithm (National Center for Biotechnology Information 2019), and subsequently aligned using ClustalW2 (Larkin et al. 2007) and Geneious software (Biomatters Ltd., Auckland, New Zealand), with representative members from the three Herpesviridae subfamilies from a range of host species. These sequence alignments were used to generate phylogenetic trees through maximum-likelihood distance analysis using the Jones-Taylor-Thornton model of amino acid replacement, with 100 replicates.

For hemoprotozoans, genomic DNA was extracted from 200 µL of whole blood using a MasterPure^™ DNA Purification Kit (Epicentre^® Biotechnologies, Madison, Wisconsin, USA), according to the manufacturer's instructions. Mock extractions were performed from an equal volume of sterile, molecular-grade water, to exclude DNA contamination in reagents and consumables. A nested set of universal piroplasm primers was used to amplify an approximately 1.2-kb fragment of the 18S rRNA gene, as previously described (Paparini et al. 2012). Partial Trypanosoma 18S rRNA sequences were amplified by PCR from quoll DNA samples using a heminested PCR assay and the primers S825F and S662R (Maslov et al. 1996), as previously described (Barbosa et al. 2017a). The PCRs were performed in 25-µL reaction volumes and then run on a 2% agarose gel containing SYBR Safe Gel Stain (Invitrogen, Camarillo, California, USA). Gel bands were visualized with a dark reader transilluminator (Clare Chemical Research, Dolores, Colorado, USA), and those corresponding to the expected length were excised and purified using an in-house filter tip method (Yang et al. 2013). The DNA extraction blanks and positive and negative controls from each PCR batch produced appropriate results. All purified PCR products were sequenced using corresponding internal reverse primers at a concentration of 3.2 pmol with an ABI Prism^™ Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, California, USA) on an Applied BioSystems 3730 DNA Analyzer (Applied Biosystems, Beverly, Massachusetts, USA). Chromatograms containing the DNA sequences were imported into Geneious R7 (Kearse et al. 2012), for curating and BLAST-searching against the GenBank database.

Hematologic and biochemical variables were analyzed to assess whether there were significant differences between male and female quolls and between wild-caught and captive-bred quolls. Before analysis, each variable was tested for outliers using a Dixon Q test for the rejection of outliers (Dixon 1950, 1951). This was repeated until all outliers were removed. To establish whether the data were normally distributed a Shapiro-Wilk test of normality was performed on each variable (Royston 1995). If the data were not normally distributed then the variable was natural log transformed (see Supplementary Material Table 1). Linear models were used to test whether there were significant differences between values for males and females and between captive and wild-caught quolls. Additionally, linear models were used to test whether there were significant differences in the body weights of quolls infected and not infected with herpesvirus and T. gondii and whether that applied to males and females differently. A post hoc Tukey honestly significant test was used to test the significance of the differences among all groups in significant models containing multiple terms (Tukey 1949). Finally, the significance of any relationship between males and females and wild-caught and captive-bred quolls infected and not infected with herpesviruses and T. gondii were tested using a chi-square test of independence. Analyses and plotting were performed using the outliers (Komsta 2011), ggplot2 (Wickham 2016), plyr (Wickham 2011), ggpubr' (Kassambara 2018), and emmeans (Lenth 2018) packages in R (R Development Core Team 2019).

A total of 31 (eight male and 23 female) adult eastern quolls underwent health evaluation during translocation. Sixteen (five male and 11 female) eastern quolls were wild caught and 15 (three males and 12 females) were captive bred. Mean±SD body weights for male and female quolls were 1.17±0.19 kg and 0.84±0.09 kg, respectively.

All quolls, both wild caught and captive bred, were considered to be in good general health, with the exception of one female that was retained in captivity for treatment before release. Specific abnormalities noted included mild ulcerative glossitis and cheilitis in one captive-bred male; marked ulcerative glossitis, cheilitis, and ulceration of the hard palate in one wild-caught female; focal granulomatous and necrotizing dermatitis and panniculitis just caudal to the pouch, from which Hafnia alvei and an Enterococcus spp. were cultured, in one captive-bred female; and a fractured upper canine in one wild-caught male. The female with the necrotizing dermatitis and panniculitis was retained in captivity for 22 d and treated with antibiotics before release.

Eighteen of 18 (100%) of the rectal swab samples were culture negative for Salmonella serovars. Sera from six of 21 quolls (29%) tested were positive for antibodies to T. gondii, with antibody titers ranging from 1/64–1/16,000. Eighteen of 31 (58%) combined conjunctival-pharyngeal-urogenital swabs from quolls were positive for herpesvirus DNA via polymerase PCR. The PCR product melt profiles were identical for all animals, indicating the same virus was present in all individuals. Sequencing of DNA confirmed all quolls were infected with the same gammaherpesvirus, closely related to dasyurid gammaherpesvirus 2 (Fig. 2). The newly detected virus was tentatively named dasyurid gammaherpesvirus 3.

Ectoparasites were present on 19 of the 31 quolls (61%) examined. Three species of tick, two species of mites, and four species of fleas were collected from wild-caught quolls, whereas a fifth species of flea, Echidnophaga myremecobii, was found exclusively on captive-bred quolls (Table 1). Of the 19 quolls with ectoparasites present, 16 (84%) had low ectoparasite burdens, two (11%) had moderate ectoparasite burdens, and one (5%) had a high ectoparasite burden.

Five (four wild caught and one captive bred) of 15 blood samples (33%) were positive for hemoprotozoan DNA via PCR. A novel Hepatozoon species genetically most similar (97.5%) to Hepatozoon species DG1 from a colocolo opossum (Dromiciops gliroides) in Chile (Merino et al. 2009), and Hepatozoon banethi (97.3%) isolated from I. tasmani ticks removed from domestic dogs (Canis familiaris) in Tasmania (Greay et al. 2018), were detected in the blood from one quoll. In addition, a novel Theileria species and Theileria paparinii were detected in the blood of one and three quolls, respectively. Both the novel Theileria sp. and the T. paparinii isolated from quolls have been further described using multiple genetic markers and microscopy (Barbosa et al. 2019b). One of the quolls positive for T. paparinii was concurrently infected with Trypanosoma copemani.

There were evident differences between captive-bred and wild-caught quolls (Figs. 3, 4 and Supplementary Table 2) for hemoglobin (P=0.008), red blood cell count (P=0.021), neutrophils (P=0.003), sodium (P=0.002), urea (P=0.008), AST (P=0.000), ALP (P=0.000), protein (P=0.001), globulin (P=0.002), and albumin (P=0.021). There were evident differences between male and female quolls (Figs. 3, 4 and Supplementary Table 3) for hemoglobin (P=0.024), red blood cell count (P=0.022), sodium (P=0.016), and phosphate (P=0.049). Quolls seropositive for T. gondii had significantly lower potassium (P=0.024), anion gaps (P=0.047), urea (P=0.037), and triglycerides (P=0.043) than those that were seronegative (Supplementary Table 4). There was no significant difference between hematologic or biochemical variables of quolls that were positive by PCR for dasyurid gammaherpesvirus 3 and those that were negative by PCR. Quolls that were seropositive for T. gondii weighed significantly more (P=0.035, F_2,19=18.08, slope=159.40, SE=70.43) than those that were seronegative, although that result was unrelated to sex (Fig. 5 and Supplementary Table 6). Wild-caught quolls were more likely than captive-bred quolls to be seropositive for T. gondii (χ²=5.73, P=0.0167). There was no significant difference (P=0.151, F_{2, 28}=26.77, slope=–73.13, SE=49.59) in body weights of quolls that were positive by PCR for dasyurid gammaherpesvirus 3 and those that were negative by PCR. More wild-caught quolls were positive by PCR than were captive-bred quolls, although that difference was not significant (χ²=2.590, P=0.108).

Significant differences were detected between male and female quolls for a small number of hematologic (hemoglobin and red blood cell count) and biochemical (sodium and phosphate) variables. Fancourt and Nicol (2019) found significant sex-related differences between mean corpuscular hemoglobin content, ALP, AST, ALT, lipase, total protein, creatine kinase, and bilirubin in wild-caught eastern quolls. Potential reasons for the limited sex-related differences observed in our study include the small sample size for males and that the blood samples were collected from males in the non-breeding season.

A previous study found no differences between wild-caught and captive-bred eastern quolls for hematologic variables (Melrose et al. 1987). In our study, captive-bred quolls had higher hemoglobin and red blood cell counts than did wild-caught quolls, which might be accounted for by a higher nutritional plane in captive-bred animals. The higher neutrophil counts observed in wild-caught quolls might potentially be explained by greater pathogen exposure or by a greater stress response to capture and handling, when compared with habituated captive-bred quolls (Clark 2004; Stockham and Scott 2008; Portas et al. 2016; Batson et al. 2017). Wild-caught quolls had higher sodium, AST, ALT, total protein, albumin, globulin, and phosphate, but lower urea values than did captive-bred quolls. The observed differences in biochemical variables are not readily explained but could relate to differences in nutrition, pathogen exposure, or physiological responses to capture and handling during translocation.

Toxoplasma gondii–associated morbidity and mortality have been reported in free-ranging populations of native Australian marsupials, including bare-nosed wombats (Vombatus ursinus; Donahoe et al. 2015), eastern barred bandicoots (Perameles gunnii; Obendorf and Munday 1990; Groenewegen et al. 2018), red-necked wallabies (Notamacropus rufogriseus), and rufous-bellied pademelons (Thylogale billardierii; Obendorf and Munday 1983). However, the effect of T. gondii infection on Australian marsupials at a population level remains largely unknown (Hillman et al. 2016). Eastern quoll survival and reproduction were not affected by T. gondii infection despite a higher seroprevalence in declining populations (Fancourt et al. 2014). Changing climactic conditions were subsequently proposed to have reduced quoll populations to low abundance, leaving them more vulnerable to existing threats (Fancourt 2016). In our study, seropositive quolls had significantly lower potassium, anion gap, urea, and triglycerides, suggesting chronic T. gondii infection has some, albeit subclinical, physiologic effect. We found clinical signs consistent with toxoplasmosis were absent in seropositive animals, and evaluating any effect on fecundity was hindered by the fact that the 15 founders, translocated in 2017, were translocated in the breeding season (seven of which were bearing pouch young), whereas the remaining 16, translocated in 2016, were translocated outside the breeding season. Seropositive quolls were also significantly heavier than those that were seronegative. This significant difference in body weight is not easily explained, although it is possible that heavier quolls might have been more likely to survive initial infection and subsequently to develop latent T. gondii infections. No significant differences in survival or postrelease dispersal were found in captive-bred and wild-caught founders. That wild-caught quolls were more likely to be infected with T. gondii than captive-bred quolls was readily explained by the lack of exposure to oocysts in captive animals maintained behind a predator-proof fence that excluded feral cats.

Herpesvirus infections are common in Australian marsupials, with at least one host-adapted herpesvirus identified in many marsupial species examined to date (Stalder et al. 2015). Gammaherpesviruses have been found in several dasyurids previously, including a novel gammaherpesvirus (dasyurid gammaherpesvirus 1) in yellow-footed (Antechinus flavipes) and agile (Antechinus agilis) antechinuses; (Amery-Gale et al. 2014), and (dasyurid gammaherpesvirus 2) in Tasmanian devils (Sarcophilus harrisii; Stalder et al. 2015). Two eastern quolls examined in a previous study were both PCR negative for herpesvirus DNA (Stalder et al. 2015), although few conclusions can be drawn from such a small sample size. Captivity has been identified as a risk factor for herpesvirus infection in marsupials, including Tasmanian devils (Stalder et al. 2015), but that was not the case in this study with captive-bred quolls no more likely than wild-caught quolls to be PCR positive for dasyurid gammaherpesvirus 3. The reasons for this are unknown. There was no significant difference in the body weight or hematologic and biochemical variables of eastern quolls infected with dasyurid herpesvirus 3 and those that were not, suggesting limited clinically relevant effects on health. However, two eastern quolls, both PCR positive for herpesvirus DNA, exhibited ulcerative lesions on the lips, tongue, and hard palate; clinical signs that have been observed in macropods with herpesvirus-associated morbidity (Finnie et al. 1976). Unfortunately, samples suitable for tissue PCR tests were not collected directly from these lesions and causation remains speculative.

The lack of Salmonella species in cultures from fecal samples is in contrast to reports that Salmonella mississippi is commonly isolated from eastern quolls (Holz 2008). Reasons for this could include sample collection technique, insufficient sample volume, and variations in the culture methodology.

Nine ectoparasite species were found on wild quolls; however, prevalence for individual parasite species, with the exception of the flea Pygiopsylla hoplia, was generally low. Only a single parasite, the flea Echidnophaga myremecobii, was present on captive-bred quolls; however, that flea occurred at a higher prevalence on captive-bred quolls than any parasite found on wild-caught quolls. Echidnophaga myremecobii is principally a parasite of European rabbits (Oryctolagus cuniculus; Kraus 1974). Its presence and the absence of other expected ectoparasites on captive-bred quolls can be explained by the feeding of culled feral rabbits, which likely harbored Echidnophaga myremecobii, and the routine use of ectoparasiticides. Ectoparasite diversity on wild-caught eastern quolls in this study was comparable to that of northern quoll (Dasyurus hallucatus) in Kakadu National Park, Northern Territory (Oakwood and Spratt 2000) and spotted-tailed quolls (Dasyurus maculatus) in Tuggolo State Forest, New South Wales (Vilcins et al. 2008). However, there was no overlap of species with northern quolls, but four species of flea and two species of ixodid ticks we observed on eastern quolls have been observed on spotted-tailed quolls (Vilcins et al. 2008).

The clinical significance of hemoprotozoan infections, in particular mixed infections, in Australian marsupials has not been fully elucidated (Barbosa et al. 2017a, 2019a). The novel Theileria species detected in one quoll in this study was also found in a swamp wallaby (Wallabia bicolor) from the Australian Capital Territory and a quokka (Setonix brachyurus) from Bald Island, Western Australia (Barbosa et al. 2019b). Recent phylogenetic studies revealed that the Theileria genotypes identified in three quolls in this study represent intraspecific variations of the recently described T. paparinii (Greay et al. 2018; Barbosa et al. 2019b). To date, T. paparinii has been identified in an Ixodes tasmani tick removed from a dog and in a range of Australian marsupials: one burrowing bettong (Bettongia lesueur) screened from a wildlife rehabilitation center in Western Australia (Paparini et al. 2012), and in eastern bettongs (Bettongia gaimardi), eastern grey kangaroos (Macropus giganteus), swamp wallabies, a brush-tailed rock-wallaby (Petrogale penicillata), and yellow-footed rock-wallabies (Petrogale xanthopus) from the ACT (Barbosa et al. 2019a). The molecular assay used to detect piroplasms also amplified a novel Hepatozoon species in one quoll. Although one Hepatozoon species (H. peramelis) and several novel Hepatozoon genotypes have been reported in Australian native mammals to date (Bettiol et al. 1996; Wicks et al. 2006; Barbosa et al. 2017b), the potential impact of these parasites requires further investigation.

Trypanosoma copemani has a wide geographical distribution and a diverse range of marsupial hosts in Australia (Barbosa et al. 2019a). This parasite species can suppress innate cell-mediated immunity (Hing et al. 2016), and has been associated with pathological changes and tissue degeneration in the muscles of infected brush-tailed bettongs, which may adversely affect the coordination of the host and, therefore, increase its susceptibility to predation (Botero et al. 2013; Thompson et al. 2014). A recent study on trypanosome-infected quokkas (Setonix brachyurus), identified erythrocyte abnormalities in the blood, typically associated with hemolytic anemias (Austen et al. 2016). However, no erythrocyte abnormalities were detected in the single quoll infected with T. copemani in our study.

Despite the presence of numerous potential pathogens, including T. gondii, a novel gammaherpesvirus, numerous ectoparasites, and three hemoprotozoans, in translocated quolls, there appeared to be limited clinical effects associated with their presence and no impact on reintroduction success was apparent. The current study serves as a reminder that translocations involve not only the movement of the target species but a broad array of microparasites (viruses, bacteria, and fungi) and macroparasites (helminths and arthropods) that the target animal may host. Documenting the presence of those parasites is essential because there is growing evidence that they are an important component of biodiversity, that disease has the potential to contribute to translocation failure, and that information provides data that informs ongoing monitoring of the health of translocated populations.

Translocations were carried out under licenses from the Tasmanian Department of Primary Industries, Parks, Water and Environment (permits TFA 16025 and 17091, export licenses 12818/16 and 13528/17), Victorian Department of Environment, Land, Water and Planning (permit 14505167), and Australian Capital Territory, Territory and Municipal Services (import license L120161261). We thank Andrew Crane, Annika Everaardt, Claire Hawkins, and Robbie Gallney of Tasmanian Department of Primary Industries, Parks, Water and Environment for their assistance and support. The reintroduction procedures were approved by the Australian National University Animal Experimentation Ethics Committee (protocol A2016/02). B.A.W. was supported by a PhD scholarship funded by the Australian National University, the Australian Capital Territory Government, and an Australian Research Council Linkage Grant (LP140100209). This work was conducted as part of the Mulligans Flat-Goorooyarroo Woodland Experiment (www.mfgowoodlandexperiment.org.au). We thank Catherine Ross, Daniel Iglesias, Dave Whitfield, Dean Heinze, Greg Hosking, Helen Crisp, Jelena Vukcevic, Joel Patterson, John Lawler, Kate Grarock, Katherine Jenkins, Katherine Moseby, Kristi Lee, Loren Howell, Lyall Marshall, Margaret Kitchin, Melissa Snape, Michelle White, Corin Pennock, Nick Mooney, Sam Reid, and many more for their assistance during the project. We thank the anonymous reviewers for their valuable feedback.

Supplementary material for this article is online at http://dx.doi.org/10.7589/2019-05-120.