Some published articles contributed by early-career scientists serve to confirm or debunk a prevailing assumption in science. As such, they can either make or break a doctoral thesis currently in progress, depending upon whether that article is discovered by the doctoral candidate during their literature searches or, in the most unfortunate of circumstances, during their dissertation defence. One such paper, which was a crucial contribution to freshwater research in the 1980s, was Pot et al. (1984). That article was published in the first year of my doctoral studies at the Université Claude Bernard – Lyon I on a Canada-France exchange fellowship. I was fortunate in that first year to gain guidance from a fellow PhD candidate (Georges Carrel) and from three key established ichthyologists whom he recommended I contact. One of those eminent persons, Prof. Eugene K. Balon (University of Guelph, Canada), was based at the same university where the Pot et al. (1984) study had been carried out. The other two ichthyologists were DSc Milan Peňáz (Institute of Vertebrate Biology, Brno, Czechia) and Dr Antonín Lelek (Forschungsinstitut Senckenberg, Germany), whom I visited in the initial months of my PhD. In the case of Prof Balon, I was able to visit his laboratory at the University of Guelph during a trip home to Canada in 1985—a particularly enlightening experience for a debutant PhD candidate.

To the best of my recollection, I do not believe I had the pleasure of meeting Prof. Noakes during that visit, except perhaps in passing. During my discussions with Prof. Balon, it soon became clear that Prof. Balon maintained firm opinions on certain aspects of fish biology, most of which influenced my subsequent research. These firm opinions included two concepts he elaborated from the originals by S.G. Kryzhanovsky, i.e. saltatory ontogeny of fishes (Kryzhanovsky 1949) and the reproductive guilds of fishes (Kryzhanovsky 1974). The former of these concepts provided the rationale for the re-examination of my entire collection of young-of-the-year (age 0 +) fishes to ensure that specimens were correctly allocated to their period of ontogenetic development (i.e. Copp and Peňáz 1988). Another of Prof. Balon’s firm opinions was the need for density estimates of freshwater fishes to be validated using an ‘absolute’ (or total) sampling approach, as was used in the Pot et al. (1984) study on the University of Guelph campus. This total sampling approach, preceded or not by a conventional capture method, involved the application of rotenone (e.g. Penczak 1981; Halyk and Balon 1983). This plant-derived neural toxin has been widely used for ‘quantification of fish populations’ (Ling 2003, p 24), based on the assumption that absolutely all of the fish in the sampled area would be collected in the delineated sampling area (e.g. Mahon 1980; Penczak 1981; Halyk and Balon 1983).

Why ‘absolute’ or ‘total’ sampling? In the decades leading up to, and during the 1980s, a major theme in freshwater biology was the estimation of ‘production’ (Ivlev 1966), in particular that of fishes (Gerking 1978). Production estimates require data on the number of specimens of a species and their weight, which was frequently estimated from body length (Le Cren 1951; Froese et al. 2014). An early leader in the study of riverine fish production in relation to the duration and extent of inundation was Grigore Antipa (1921, 1928), who postulated in his publications on floodplain water bodies of the River Danube’s ‘Delta’ area (in Romania) that ‘la production (des pêcheries) est directement proportionnelle à la superficie inondée et à la durée des inondations’ (Botnariuc 1968, p. 61). Focus on fish production, including consideration of the role of the hydrological regime, continued through the decades (e.g. Stanković 1938; Léger 1948; Le Cren 1959; Ivlev 1966; Backiel 1971; Gerking 1978; Bruton and Jackson 1983). Most of those studies involved the use of a conventional, quantitative sampling method, which in water courses is usually referred to as the ‘depletion’ or ‘successive-removal’ method (Beaumont 2016), followed by the application of rotenone. However, interest in fish production began to wane during the 1980s, though there have been some studies on, or involving, fish production estimates in subsequent decades (e.g. Holčík 1996; Boisclair 2001).

In retrospect, it was quite unfortunate that I did not have the opportunity to make Prof. Noakes’ acquaintance, because the article published with his student (Pot et al. 1984) was instrumental in debunking the pervasive assumption that the total number of fish in freshwater ecosystems could be obtained using rotenone (and indeed any other sampling method). The aim of their study was ‘to test two of the most basic and commonly used techniques, capture-mark-recapture and total sampling after rotenone poisoning’ (p. 249, Pot et al. 1984) in order to address the lack of systematic testing of these two approaches. In their case, minnow traps and lift nets were used to capture, mark and recapture fishes in a freshwater pond (‘a small and rather uniform system’, also p. 249), which was then treated with rotenone. Following that treatment, only 62.5% of the marked fish (four species) were recovered, with no significant differences among species in their recovery rates. Although Pot et al. (1984) reported a higher recovery rate than those of previous studies (30–40% recovery), the poor recovery of all fish was attributed (amongst other possible explanation) to fish becoming trapped in the pond’s substratum during the rotenone treatment. The same explanation for unrecovered fishes following rotenone treatment was reported by Fraser (1981) in their study of four Precambrian Shield lakes (Ontario, Canada). In the summary of the Pot et al. (1984) article, the authors concluded that (1) even in a small system, the conventional capture-mark-recapture method was less than adequate as a means to estimate fish numbers, (2) post-rotenone collection of fishes ‘probably produces a much less-than-complete sample of a population’ and (3) ‘adequate density estimates were obtained for only a minority of the species present’, with most species either not captured at all or only selectively during the pre-rotenone sampling.

The consequence of the Pot et al. (1984) article for me personally was the evidence it provided, which helped me justify, during the public defence of my dissertation, the sampling approach taken in my doctoral studies. The aim of my PhD was to assess the role of floodplain channels and backwaters as fish nursery habitats (Copp and Peňáz 1988; Copp 1989b). This required a systematic means of sampling young-of-the-year fishes (i.e. free-embryos, larvae and juveniles) in virtually all of the various floodplain ecosystems present in the River Rhône during the mid-1980s (Roux and Copp 1993). In large-river floodplains, partially abandoned and abandoned channels (of braided and meander origin) represent a mosaic of fish nursery areas. These floodplain water bodies are characterised by submerged vegetation, fallen tree branches and bushes, dense aquatic vegetation and the presence of increasingly muddy and silted substrata as they advance through ecological succession (e.g. Amoros et al. 1987; Copp 1989b). These various encumberments, combined with the range of water velocity conditions, render ineffective the conventional approaches for sampling 0 + fishes (e.g. minnow traps, seine nets, drop and pop-up nets) in all habitats of floodplain ecosystems (Copp and Peňáz 1988).

To address this complexity of field conditions, and to enhance the robustness of the data, I employed a slightly modified version of point abundance sampling by electrofishing (PASE), which had been adapted by Henri Persat for larger and older fishes from a sampling approach developed to estimate nesting bird numbers in forests (Nelva et al. 1979; Persat et al. 1981). To apply this systematic sampling approach for young-of-the-year (0 +) fishes, the size of the anode was reduced from 30 cm diameter to 10 cm (Copp and Peňáz 1988; Copp 1989a). This adaptation of the anode created a current field that effectively increased the resistivity of the water (Cuinat 1967); this renders the 0 + fish more conductive than the encircling water, thereby inducing galvanotaxis/narcosis to permit collection (Copp 1989a). The PASE approach is effectively a form of catch-per-unit-effort, CPUE, i.e. estimates of relative fish density for each species (Persat and Copp 1989; Copp and Garner 1995; Copp 2010). Like all other such CPUE approaches widely used in fisheries, PASE is based on the statistical premise that a sufficiently large number of samples collected in a systematic manner will result in a more statistically-robust dataset than one comprised of a few small samples (Gerard and Berthet 1971; Chessel 1978). For the purposes of studies on species richness, relative abundance and species-habitat relationships (Copp and Garner 1995), there is little need to know the absolute number of fishes in a study site, but rather to obtain a representative sample (Copp 2010). In such a PASE application by electrofishing, all 0 + fishes immobilised by the electricity are captured with a dip net (Copp and Garner 1995; Copp 2010), whereas for relative density estimates using PASE, a point sample consists of the immobilised 0 + fishes captured by one upwards lifting of the net through the water column at the sampling point (Copp and Garner 1995; Garner 1997).

However, at the state of freshwater science in the 1980s, most freshwater fisheries scientists believed that electrofishing was ineffective for sampling 0 + fish (Copp 1989a), and total sampling to determine fish production predominated (e.g. Mahon 1980; Penczak 1981; Halyk and Balon 1983). So, for my public thesis defence, I needed a study that supported my ‘relative density’ sampling approach in order to respond to critical comments from Prof. Balon, a principal jury member (in France, known as a Rapporteur) and a firm proponent of ‘absolute sampling’. As anticipated, during the public defence of my dissertation, Prof. Balon raised this sampling issue as a potential weakness of my study. Thankfully, the Pot et al. (1984) study, which corroborated Fraser (1981), provided the essential evidence to demonstrate that absolute/total sampling was impossible in water bodies with muddy bottoms, even very small ponds, because some fish will escape capture by immersing themselves in the muddy bottom to escape the rotenone—a phenomenon well known with electrofishing (e.g. Balayave 1981; see Beaumont 2016 for a review). Indeed, during the application of rotenone in the sampling of small tributaries of the river Sinnamary, French Guyana (Ponton and Copp 1997), field operatives with dip nets had to be positioned at each end of the netted-off study site (in the small boats that traversed the stream to hold up the block net) in order to capture the (flying) fishes that attempted to escape the rotenone by jumping over the stop nets to escape the rotenone-treated area.

Despite the importance of a field sampling approach (strategy, method, technique) to establish the validity of the resulting dataset, the Pot et al. (1984) article has received relatively few citations since its publication (16 recorded by both Web of Science and Google Scholar, with the later listing four non-peer-reviewed citations). Nonetheless, the Pot et al. (1984) study was very likely the ‘make or break’ publication in the defence of my doctoral dissertation. And for that, I am very grateful to Prof. Noakes’ obvious insight into a key issue associated with field studies of freshwater fishes and their fisheries and in retrospect his apparent mentorship and promotion of early-career scientists as lead and supporting co-authors of that publication.

The moral of this story for PhD candidates and other early-career scientists is that the advancement of science requires emerging scholars to question and test existing assumptions. Resistance to change is a human trait, so endeavours to test entrenched assumptions are very likely to encounter disbelief and even rejection from journal reviewers and editors. Therefore, it is important to consult a range of scientists, not only your immediate supervisor(s) or advisors but also experts external to your lab. And if these mentors remain unconvinced, then you may have to ‘go it alone’. That was indeed the case with the above-mentioned PASE—none of the people consulted at the start of my doctoral research believed that electricity could induce galvanotaxis/narcosis in 0 + fishes. Indeed, one reviewer of Copp (1989a) absolutely refused, despite the evidence and supporting literature provided in original French and the published English translation thereof (i.e. Cuinat 1967), to accept that use of a smaller (10 cm) anode would create a current field capable of immobilising 0 + fishes. So, persistence is an important attribute for emerging scientists, because despite resistance elsewhere there will be established scientists who are willing to entertain and encourage the use of new approaches. A case in point is the PASE approach for 0 + fishes, which following its development in France has been used in several European countries and at least once in Canada (reviewed in Copp 2010). As such, it is important that early-career scholars be persistent and employ, as aptly stated by one of the reviewers of the present manuscript, a ‘healthy dose of conceptual/methodological scepticism (not cynicism) that we all should seek to challenge our often-entrenched values and assumptions’.