Introduction

Some scientists create with their work a watershed divide, a tipping point in time when much of the research and progress that comes afterwards in that field is strongly influenced by their unique contributions. In ecology, Giorgii Gause’s integration of mathematical models of population growth with laboratory experiments to test competition between two species of unicellular organisms is one such example (Gause 1934). Raymond Lindeman’s quantitative synthesis of physiological and community ecology based on his data on trophic levels in a lake food web is another such intellectual watershed divide (Lindeman 1942), and provided the genesis for ecosystem ecology (McIntosh 1985). Indeed, many previously unexplained and sometimes unrelated observations suddenly may be explained by a new theory or model, leading to a “paradigm shift” (Kuhn 1962) that sets a new baseline for future work.

In many ways, Shigeru Nakano was this sort of scientist, whose work helped catalyze a shift by ecologists, especially stream ecologists, toward new and more productive research and synthesis. His early studies on the behavior and ecology of fish in Japanese streams later expanded to include entire food webs linked across streams and their riparian forests. To develop this synthesis, Nakano combined what he learned from studying individual components of the stream food web, such as fish and invertebrates, with observations and inferences about the ecology of riparian animals like spiders and birds and bats. Working across the boundaries of taxa and subdisciplines in ecology, he first imagined and then organized and carried out pioneering large-scale experiments that severed the linkages between streams and their riparian forests and measured the drastic responses in both habitats. Despite his tragic death with four other accomplished ecologists in March 2000 (Fausch 2000), the importance of Nakano’s collaborations and contributions has grown steadily since. His publications continue to be widely read and cited, and inspire the research of others. Moreover, Nakano’s early work on the individual behavior of stream salmonids, and his later collaborations to understand their habitats and interrelationships across local-to-riverscape spatial scales, were also groundbreaking research that had strong effects on these subdisciplines of ecology.

In this paper, I address Nakano’s enduring legacy for ecology, nearly two decades after we lost him as a scientist, colleague, and friend. I first describe his background, both personal and academic, which provided a template for both his interests and his drive to pursue them as a scientist. Next I select key papers from each of three main groups in which his work can be categorized, summarize these findings, and show examples of how these ideas crossed boundaries and set a new standard for others who followed. Finally, I offer thoughts on why Shigeru Nakano’s legacy has endured, and is likely to continue to inspire ecologists to cross new boundaries well into the future.

Personal and academic background

Shigeru Nakano grew up in the small town of Kamioka, in the Hida Mountains of Gifu Prefecture in central Japan, a region of beautiful streams and rivers tumbling through steep mountain valleys. Headwater streams there are inhabited by iwana, the form of whitespotted charr (Salvelinus leucomaenis) recognized from this region, and yamame, the river resident form of masu salmon (Oncorhynchus masou masou), which are found from the middle reaches downstream (see Kawanabe 1989 for common names of salmonids in Japan). His family often went camping along local rivers and streams during summer, and Nakano was usually in the water either fishing or snorkeling (Fig. 1). His mother and aunt ran a banquet restaurant in their home, and the chef told Shigeru stories about traveling deep into the mountains to fish for charr, which captivated his interest (see Fausch 2015, Chapter 2). He pleaded with his parents to let him attend a private high school in Tokyo so he could have a better chance to enter a good university, and they let him go. During his first year of high school, Shigeru read an article in a nature magazine written by Tetsuo Furukawa, then a graduate student at Kyoto University, about studying the microhabitat use and interactive segregation of whitespotted charr and masu salmon by making observations underwater while snorkeling (Furukawa 1978). This sparked Shigeru’s imagination about becoming a scientist and using this method to study stream salmonids himself one day.

Fig. 1
figure 1

(image by Y. Tokuda)

Shigeru Nakano with a rainbow trout he caught in the Takahara River near his home in Kamioka, Gifu Prefecture, Japan, in August 1987

Full size image

After high school Shigeru Nakano entered the Department of Fisheries at Mie University, located on the Kii Peninsula in south central Japan. There he met classmate Yukinori Tokuda, who had read the same article by Furukawa (1978) while in high school, and they became best friends. During their second year they began traveling to small streams in the mountains nearby to fish, and bought wetsuits so they could snorkel in the cold water to watch fish for fun. By their third year they were helping Dr. Makoto Nagoshi, a professor at Mie University, with his research on the population ecology of amago, the fluvial redspotted masu salmon (O. masou ishikawai) found in southern Japan, which inhabits Hirakura Stream on the Mie University Forest. As fourth-year students they also sampled fish by angling to conduct stomach analysis and snorkeled to measure the depths and velocities at their stream positions (focal points), and these studies resulted in Nakano’s first publications (Nakano and Nagoshi 1985; Nagoshi et al. 1988).

When Nakano graduated with his undergraduate degree in Fisheries in March 1985, Nagoshi invited him to pursue a Master’s degree, which he started immediately. Shigeru remembered the article that he and Tokuda had read in high school, and approached Dr. Furukawa-Tanaka (his married name) to learn more about his methods of measuring focal points and individual behavior of stream salmonids by snorkeling (Furukawa-Tanaka 1985, 1988, 1989). Nakano used these methods in his research on amago in Hirakura Stream, which began in spring 1985 (Nakano 1994). In addition, he was invited to study cichlid fishes in Lake Tanganyika in August and September 1985, as part of a project led by Dr. Hiroya Kawanabe of Kyoto University and Nagoshi. There he was exposed to detailed methods of underwater observation and recording fish behavior used by other scientists (e.g., Hori 1983; Kuwamura 1986). Unfortunately, Shigeru contracted malaria during the expedition, which severely limited the time available for research. He managed to complete underwater observations on brood defense by a species of cichlid and published a paper from it (Nakano and Nagoshi 1990), but the experience confirmed that his real interest lay in studying salmonids in mountain streams.

After finishing his Master’s degree in 1987 (Nakano 1987), Shigeru found a temporary position with the Takahara River Cooperative Fishery Union in Kamioka, where Yukinori Tokuda had secured a permanent position, and they often worked together. For example, they conducted some of the first surveys of fish distributions throughout headwater streams of the region. After a year, Nakano landed a permanent job as a biologist in a small museum near Kamioka. These positions afforded him a familiar environment in which to continue pursuing his passion for underwater research. Overall, the studies Nakano undertook during the summers of 1985 through 1987 while conducting his Master’s degree and soon after while working near Kamioka were the first to attract the attention of other ecologists. And, the chance to attend a symposium on charrs and masu salmon in Hokkaido in 1988 proved to be a turning point in both his life and mine.

Contributions in three subdisciplines of ecology

Even though Shigeru Nakano’s scientific career was limited to an approximately 15-year period from his first paper in 1985 until his death in 2000, he and his collaborators were amazingly productive scientists, publishing 111 journal articles and book chapters from the research they conducted, 94 in refereed journals, and 86 in English (see Electronic Supplementary Material). Of these, 41 have been cited in other published papers at least 41 times each (the h-index; based on Google Scholar accessed 23 August 2017), and 76 have been cited at least 10 times (the i-10 index). Overall, papers published by Nakano either alone or with collaborators have been cited about 400 times per year or more every year since 2007, and more than 6300 times in total, an astounding number based on a 15-year career. This is especially impressive for any scientist whose first language was not English.

Research by Shigeru Nakano alone and with his collaborators can be broadly categorized into three groups: (a) his early work on individual behavior of stream salmonids in dominance hierarches, (b) research on the ecology of salmonids at population to landscape scales, and (c) research on stream and forest food webs, and linkages across the terrestrial-aquatic boundary. He also conducted additional studies on various other organisms and topics in aquatic and terrestrial ecology (e.g., Hino et al. 2002; Taniguchi et al. 2003). Each of these categories includes papers that have been cited at least 100 times, which I arbitrarily consider to be the key papers from this body of research. Although most of Nakano’s work is considered to fall within the realm of stream ecology, what made it unique was that it crossed many boundaries to employ new methods, span across different habitats and taxa, and bridge across ecological subdisciplines. Therefore, my goal here is to consider the key papers in each of the three categories above, and discuss why they are unique in crossing these boundaries and thereby setting new standards. In each case I also review a few subsequent papers that used Nakano’s work as a foundation on which to build additional new contributions in those fields, forming an enduring legacy for his work.

Individual behavior in stream salmonid dominance hierarchies

During Shigeru Nakano’s research for his Master’s degree and the summer afterwards (1985–1987), he conducted three highly detailed studies of the individual behavior of salmon and charr and the mechanisms creating their dominance hierarchies in mountain streams of central and southern Japan (Fig. 2). Two were carried out in Hirakura Stream (Nakano 1994, 1995b) where Nakano had worked with Dr. Makoto Nagoshi on amago, the fluvial redspotted masu salmon. The third considered interspecific dominance hierarchies of yamame (resident masu salmon) and iwana (whitespotted charr) in a stream near Kamioka (Nakano 1995a). These studies were my first introduction to Nakano’s work, and ultimately led to our work together.

Fig. 2
figure 2

(image by Y. Tokuda)

Shigeru Nakano snorkeling in Shimosaya, Ibaraki Prefecture, Japan, in summer 1988

Full size image

Studies of individual behavior of salmonids that live in streams have a long history (Newman 1956; Kalleberg 1958), but most had been conducted in artificial stream channels which have the potential to induce artificial behaviors. A few early studies in the western United States were conducted by making observations from stream banks or platforms (Jenkins 1969) or by snorkeling (Edmundson et al. 1968; Everest and Chapman 1972; Griffith 1972). In turn, these led to theories and tests of the mechanisms by which salmonids select foraging position in streams (Chapman 1966; Fausch 1984; see Fausch 2014 and Piccolo et al. 2014 for reviews) and early field and laboratory experiments testing how different species compete for these resources (e.g., Fausch and White 1981, 1986).

However, underwater observations of stream fish have a longer history in Japan than North America, owing to the pioneering research of Dr. Hiroya Kawanabe on ayu (Plecoglossus altivelis) starting in 1955 (e.g., Kawanabe et al. 1956; Kawanabe 1970; see Fausch and Nakano 1998 for a review). Moreover, this research began in the Laboratory of Animal Ecology at Kyoto University where investigators were already studying the individual behavior and social systems of primates (Japanese macaques; Macaca fuscata) in the mountains nearby. As a result, research on fish populations in Japan, whether salmonids in streams or cichlids in African Rift lakes, began by considering fish as individuals with unique properties and behaviors, rather than as populations of similar individuals as was often assumed in western science (Fausch and Nakano 1998). In turn, these individual properties can have effects that influence how populations of individuals and communities of interacting species function. For example, Kawanabe (1969) found that the production of ayu depended on a shift in territorial behavior of individuals that was influenced by density. Nakano and Nagoshi (1990) and Kohda (1991) reported that territorial defense among cichlid fishes in Lake Tanganyika was different depending on the function. Feeding territories were defended against species with similar foraging habits, whereas breeding territories were defended against conspecifics, or potential predators of broods of small fish.

Studies of individual behavior among fishes continued in the Laboratory of Animal Ecology, which Kawanabe directed (e.g., Katano 1985, 1991), but Nakano was particularly attracted to those by Tetsuo Furukawa-Tanaka (Furukawa-Tanaka 1985, 1988, 1989). As described above, his early magazine article (Furukawa 1978) had inspired Shigeru to pursue his own studies on focal points of stream salmonids in dominance hierarchies. In turn, Furukawa-Tanaka had been inspired by an early paper I wrote developing a theory to explain such positions (Fausch 1984) and subsequently invited me to Japan to present a paper at the International Symposium on Charrs and Masu Salmon in Sapporo in 1988 (Noakes 1989), perhaps with some urging from Nakano (see Nakano 1989).

Shigeru Nakano drew inspiration from this previous work on fish behavior in North America, Europe, and Japan to create his unique research. His three studies showed clearly that stream salmonids like the redspotted masu salmon and whitespotted charr whose behavior he measured in pools of mountain streams in Japan are arranged in linear dominance hierarchies (Nakano 1994, 1995a, 1995b). The ranks of individuals in intraspecific hierarchies of the salmon are apparently determined by their size, and were often perfectly correlated with their weight or length. In the most detailed analysis, Nakano (1995b) reported that the summer intraspecific dominance hierarchies of salmon were quite stable, with dominant fish holding positions near the head of the pool and closer to the water surface, where they fed more frequently and on larger prey, and grew faster. In contrast, subordinate fish had more agonistic encounters, often were forced to flee from attacks, shifted positions often, had larger foraging ranges but lower foraging rates, grew slower, and more often emigrated from pools than dominant fish (Nakano 1994). However, in his study of both species in an interspecific dominance hierarchy, Nakano (1995a) found that salmon dominated charr, so that only charr more than about 40% greater in mass than salmon were clearly dominant. Moreover, salmon grew faster than charr during summer, causing some individual salmon to surpass some charr in rank.

What made these studies unique? After all, earlier investigators had described linear dominance hierarchies based on observations in the field (Jenkins 1969) and laboratory (Fausch 1984). A primary reason is that Shigeru Nakano crossed boundaries to adapt and develop new methods for observing and mapping positions and habitat of stream salmonids underwater in natural streams (see Fausch 2015, Chapter 2). Moreover, he collected so much data, and on so many different details of individual behavior, that no one could argue with his conclusions. For example, during the three studies he recorded 856, 2191, and 2835 aggressive encounters from which he constructed the dominance hierarchies, and for one study he snorkeled to collect data in one pool during an amazing 41 consecutive days (Nakano 1995a). He also collected highly detailed data on fish foraging in another study (Nakano 1995b), as well as measuring the diet and growth of individual fish, which allowed him to relate dominance for stream positions with the prey fish ate and their growth, a measure of fitness. Most importantly, all of this work was done under entirely natural conditions in the field, allowing him to confirm and extend findings previously reported only from the laboratory (e.g., Chapman and Bjornn 1969; Fausch 1984). Finally, the highly detailed habitat maps he created from measurements of depth, velocity, and substrate in grid cells often only 10 cm in each dimension, coupled with simple and elegant analyses and succinct writing, ensured that his papers provided the highest-quality data and synthesis to that date, and will stand the test of time into the future. It is not surprising that along with a subsequent study of charr foraging (Nakano and Furukawa-Tanaka 1994), these papers were cited by one of the leading textbooks of ecology as among the best examples of intraspecific and interspecific dominance hierarchies (Ricklefs and Miller 2000).

As I have described and written about in more detail elsewhere, Shigeru Nakano showed me the results from these studies during our first meeting at the symposium in Sapporo in October 1988 (see the RiverWebs documentary film, http://www.riverwebs.org; Fausch 2015). His careful work and boundless enthusiasm impressed me so much that I later arranged to collaborate with him, and we conducted four field studies in mountain streams in Hokkaido, Japan, and Montana, USA, using the same methods of snorkeling in pools to measure detailed microhabitat use and foraging behavior by native charr and trout (e.g., Nakano et al. 1992, 1998). In Hokkaido, we followed leads from his earlier collaboration with Furukawa-Tanaka (Nakano and Furukawa-Tanaka 1994) and a famous book by Ishigaki (1984) about the mechanisms causing shifts by oshorokoma or Dolly Varden charr (Salvelinus malma) from defending relatively fixed positions and feeding on drifting invertebrates to ranging widely and picking benthic invertebrates from the substrate (Fausch et al. 1997; Nakano et al. 1999a). We used a combination of field experiments during one summer and underwater observations across four summers (Fig. 3) to demonstrate that this flexible behavior allowed oshorokoma and amemasu (the common name for whitespotted charr in Hokkaido) to partition scarce food resources as drifting prey subsided with declining stream flow through the summer in these streams. We proposed this as a key mechanism to explain the coexistence of these two closely-related charr in narrow zones of sympatry in Hokkaido mountain streams (Fausch et al. 1994; Taniguchi and Nakano 2000), and in doing so crossed boundaries to link individual behavior with the community ecology of this native salmonid guild.

Fig. 3
figure 3

Shigeru Nakano, Satoshi Kitano, and Kurt Fausch (left to right) on the last day of their first summer of research in Poroshiri Stream, Hidaka Mountains, Hokkaido, July 1991

Full size image

Nakano’s most detailed study (Nakano 1995b) is among his four most highly cited papers, and catalyzed subsequent research by others worldwide into the mechanisms of salmonid foraging and behavior. Many investigators cite his three papers (Nakano 1994, 1995a, 1995b) as standard references to describe the first principles of stream salmonid microhabitat use (e.g., Gowan 2007; Harvey and Railsback 2014), and have used this foundation to expand our understanding of how and why these fishes move to select positions both within and among pools across reaches. For example, Gowan and Fausch (2002) reported that energetically-favorable foraging positions declined throughout summer in a Colorado, USA mountain stream as flows and drifting prey declined, and that dominant trout moved within and among pools, or focused their foraging on terrestrial prey inputs at specific locations within pools, to meet their energetic needs. Akbaripasand et al. (2014) found that subordinate galaxiids in New Zealand streams, and those fish that grew more slowly, were more likely to move among pools than dominant fish. These and other recent studies (e.g., Urabe et al. 2010; Wall et al. 2016) have also crossed boundaries to link individual behavior in dominance hierarchies and the energetic tradeoffs at focal points to population dynamics across entire stream reaches.

Other investigators report that behavioral experience and genetic factors have strong effects on how dominance hierarchies in stream salmonids are structured. For example, White and Gowan (2013) found that juvenile brook trout (Salvelinus fontinalis) in laboratory streams were able to discern their relative rank in dominance hierarchies in part by observing the outcome of interactions among pairs of other fish, a phenomenon called transitive inference. This proved most successful when the subjects had interacted directly with one of the pair (see also Hotta et al. 2015 for a test with a cichlid fish). Such social learning allows fish to avoid contests and injury, thereby increasing fitness, and additional research showed that brook trout in natural streams could learn to forage on novel prey from conspecifics (White and Gowan 2014). Hughes et al. (2016) reported that two life history forms of brown trout (Salmo trutta) from a Scottish loch showed clear differences in dominance, even when matched for size and reared in a common environment, indicating that some combination of genetic makeup or maternal effects accounted for the differences.

Although it has been more than 20 years since Shigeru Nakano published his seminal papers on dominance hierarchies in stream salmonids, and more than 30 years since he conducted the research, to date I have yet to see more comprehensive field studies on this aspect of the behavioral ecology of these fishes than those he conducted. New imaging systems and software are providing more accurate measurements of fish positions (Vivancos and Closs 2015; Neuswanger et al. 2016), but Nakano’s papers are still the standard for defining the principles of how salmonid behavior creates the dominance hierarchies that are ubiquitous in natural streams.

Ecology of salmonids across local to landscape scales

By the early 1990s, it was becoming evident to animal ecologists in general, and those studying salmonids in streams in particular, that predicting the distribution and abundance of fish and other animals depended on the spatial and temporal scale at which their populations, habitat, and other interacting biota were measured (Wiens 1986, 1989; Fausch et al. 1988; Bozek and Rahel 1991). Stream habitat is inherently hierarchical, with climate, geology, and geomorphology of catchments setting the broad-scale template of physical features in which the habitat at reach, channel-unit, and microhabitat scales develops (Frissell et al. 1986; Fausch et al. 2002b). In turn, fish evolve in response to this physical template, and develop life histories, species interactions, population dynamics, and individual behaviors that interact across all these scales (Baxter 2002). As a result, relationships between, for example, salmonid density and habitat characteristics like pool volume are unlikely to be constant across species or biomes or spatial scales, owing to other interacting factors (Fausch et al. 1988; Inoue et al. 1997; Inoue and Nakano 1998).

Five of the most highly cited papers reporting research on which Shigeru Nakano collaborated address the ecology of salmonids across spatial scales from channel units to entire drainage basins, and across levels of ecological organization from individuals to communities (Fausch et al. 1994; Inoue et al. 1997; Inoue and Nakano 1998; Taniguchi and Nakano 2000; Fausch et al. 2001). In 1989, Shigeru earned a position as an Assistant Professor at the Nakagawa Experimental Forest of Hokkaido University in northern Hokkaido (see Fausch 2000 for a chronology of his career). After gaining funding for our first collaborative field studies in 1991 and 1992, Nakano and I conducted research with Ph.D. student Satoshi Kitano in a small mountain stream in Hokkaido, part of which was focused on measuring what factors accounted for the distribution of the native salmonid guild in that catchment and others across Hokkaido. We found that temperature was a strong predictor of the largely non-overlapping boundary between Dolly Varden charr distributed upstream and whitespotted charr found downstream in catchments across Hokkaido Island, and that the altitude of this boundary changed predictably across the island owing to regional climate driven by ocean currents (Fausch et al. 1994; see Ishigaki 1984 for a previous description of the pattern with altitude). However, within the single catchment we studied in detail, charr distributions differed between the mainstem and tributary, apparently owing to stream discharge and flood disturbance that affected habitat and populations at the reach scale.

We also conducted detailed research at the scale of individual pools along the narrow zone of sympatry, and found that distribution of the two species apparently depended on interspecific competition, which also was ultimately modified by temperature. Long-term laboratory experiments by Ph.D. student Yoshinori Taniguchi and Shigeru Nakano testing competition between the species at cold and warm temperatures subsequently showed that the downstream limit of Dolly Varden charr was apparently set by the temperature at which they lost contests for stream positions to whitespotted charr, but the upstream limit of whitespotted charr was most likely set by the physiological intolerance of that species to cold temperature, rather than by strong effects of competition from Dolly Varden (Taniguchi and Nakano 2000). The two species are likely to coexist in the narrow zones of sympatry where they both have positive survival rates (Fausch 2015), and Dolly Varden probably persist in these places in part because of their ability to shift foraging modes during late summer when prey resources are low (Fausch et al. 1997; Nakano et al. 1999a). Overall, this set of field and laboratory studies crossed new boundaries to link the zoogeography of this salmonid guild with their ecology, physiology, and behavior, and provides data invaluable for conserving their populations in an era marked by warming temperatures and a more variable disturbance regime.

The role of large wood that falls into streams in creating habitat for stream salmonids also emerged as a major focus of research during this period (e.g., Nakamura and Swanson 1993; Richmond and Fausch 1995), but here too the effects are likely to vary across spatial scale. In the early 1990s, graduate student Mikio Inoue and Shigeru Nakano embarked on research across an extensive set of study reaches (n = 36–48 reaches) to test how large wood and its effects in streams differed in northern Hokkaido compared to other regions, and how masu salmon responded. They reported that the relatively small diameter and short length of the willow (Salix spp.) and alder (Alnus hirsuta) pieces that fell into the streams they studied had little effect on channel morphology and pool spacing at the reach scale, but did create overhead cover, visual isolation, and velocity refuges within individual channel units (pools, runs, and riffles; Inoue and Nakano 1998). Salmon density was low in reaches adjacent to grasslands, owing to unsuitably high water temperatures (Inoue et al. 1997), but in forest reaches it was positively correlated with the physical structure created by the large wood at both reach and channel-unit scales (although with various contingencies). Overall, they concluded that the role of large wood in these streams, and the response of masu salmon to it, depended strongly on the geomorphic template, past land uses, and the spatial scale at which it was measured. Additional research revealed many important relationships between fish and habitat at even finer scales within pools (Inoue and Nakano 1999; Inoue and Nunokawa 2002), and at broader scales between forest conditions or forestry practices and salmon and charr populations in streams throughout Japan (Inoue and Nakamura 2004; Inoue et al. 2013).

Nonnative salmonids have strong effects on native species worldwide, and these also vary across scales. Rainbow trout (O. mykiss), originally native only to the Pacific coast drainages north of Hokkaido and in western North America (Behnke 2002), were introduced to Hokkaido Island and began actively invading in the 1980s, but were not invading in Honshu Island, Japan despite large numbers having been stocked there for more than 100 years (Fausch et al. 2001). I became intrigued by what caused this difference. Working from a hypothesis about differences in flooding regimes across the two islands and among other regions around the world where rainbow trout were native or had been introduced, I collaborated with Taniguchi and Nakano and several other investigators to assemble and analyze data on these environmental variables. We found that rainbow trout invasion success was high in regions where the seasonal flooding regime matched that in their native range, or where the probability of flooding was low during the period of trout fry emergence, as in Hokkaido (Fausch et al. 2001). In contrast, invasion success was low in regions like Honshu and Colorado where floods during fry emergence wash fry away and reduce their survival and growth. Additional research by Inoue and Taniguchi and their students in streams of southwest Hokkaido showed that rainbow trout are more likely to invade and establish reproducing populations in catchments where discharge is stable, such as in spring-fed streams draining volcanic topography, than in streams with higher flow fluctuations (Inoue et al. 2009).

Determining the spatial scales at which salmonid populations and communities are related to physical and biotic features is an active area of research. These publications coauthored by Shigeru Nakano crossed boundaries to assess relationships of salmonids to habitat and native and nonnative salmonids across a wide range of scales, and thereby played an important role in fostering new ideas. Although originally counterintuitive, salmonids are often found to be more strongly related to physical features that emerge at riverscape scales than to factors measured at local scales (e.g., Fausch et al. 2002b; Isaak et al. 2007; Flitcroft et al. 2012; Falke et al. 2013). Ongoing research in this arena, including advances in methods for measuring and modeling habitats across the entire spatial hierarchy (Peterson et al. 2013; Fullerton et al. 2015), will continue to yield insights and improve management of salmonids in riverscapes.

Linkages between terrestrial and aquatic ecosystems in forested streams

In April 1995, Shigeru Nakano moved to the Tomakomai Experimental Forest of Hokkaido University in southwestern Hokkaido, and began a new program of study on the relationships between streams and forests. Conservation of biodiversity had become an important topic in Japan, as in many countries, and Nakano began to think creatively about why forests are important to streams and their biota, and how the streams he studied also might be important to the surrounding landscape and its biological diversity. This latter question ran counter to most of the theory in stream ecology, which held that streams are primarily recipients of nutrients and materials like leaves and wood from the adjacent forests and grasslands (e.g., Minshall 1967; Fisher and Likens 1973; Wallace et al. 1997). This new research focus was to become Nakano’s most enduring legacy for ecology, and the source of nine of his most cited publications (Nakano et al. 1999b, 1999c; Kawaguchi and Nakano 2001; Nakano and Murakami 2001; Murakami and Nakano 2002; Iwata et al. 2003; Kato et al. 2003; Kawaguchi et al. 2003; Fukui et al. 2006).

Nakano had also become intrigued with manipulative experiments in ecology, after participating in the field experiments that I planned and we conducted during 1991 and 1992 on foraging mode shifts by charr in Hokkaido (Fausch et al. 1997), and those he planned on interspecific interactions among native and nonnative salmonids in Montana (Nakano et al. 1998). I recall when visiting him in northern Hokkaido during summer 1994 that he asked me about experimental designs and the statistics used to analyze them, which he then began studying on his own. Moreover, his former graduate students and collaborators from that era (Mikio Inoue, Yôichi Kawaguchi, Naotoshi Kuhara, Yo Miyake, Hirokazu Urabe) reported that while driving through an agricultural landscape during 1993 or 1994 Nakano realized that the streams he studied could fit inside a standard greenhouse (Fig. 4), and that he could use them to experimentally cut off inputs of terrestrial invertebrates to streams.

Fig. 4
figure 4

(image by M. Murakami)

Shigeru Nakano describing the greenhouse frame used for experiments to cut off flows of invertebrates to and from Horonai Stream, July 1999

Full size image

These ideas led to the first three studies of the importance of inputs of terrestrial invertebrates to stream fish, and the cascading top-down effects they produce in stream food webs (Nakano et al. 1999b, 1999c; Kawaguchi and Nakano 2001). For this trio of studies, Nakano and his colleagues combined observational and experimental approaches, a particularly effective strategy in any ecological research. They crossed both methodological boundaries, by being the first to enclose a stream with a greenhouse to sever its connections with the riparian forest, and disciplinary boundaries by moving beyond a focus on salmonid fishes and their invertebrate prey to food-web interactions that extended to streambed algae. First, in March 1995, Kawaguchi and Nakano began a year-long sampling program of the inputs of terrestrial invertebrates, biomass of benthic invertebrates, biomass of drifting prey of both types, and prey in fish diets. They sampled each component at frequent intervals in stream reaches adjacent to forest versus grassland riparian zones along Horonai Stream, which traverses the Tomakomai Experimental Forest. For a month from late July to late August that year, Nakano and six graduate students and collaborators sampled these same characteristics intensively (e.g., inputs of terrestrial invertebrates were sampled every day, and every 4 h for 6 days), focusing on diet selectivity by rainbow trout. In addition, in June and July 1995, Nakano and several graduate students erected greenhouses covered with plastic sheeting over four 50-m fenced reaches of the stream, and compared the effects of excluding terrestrial invertebrates and adding Dolly Varden charr on stream food webs in a factorial design including controls.

The results of these studies were striking. Salmonid biomass in the forested reach was two to three times that in the grassland reach, and this matched the annual inputs of terrestrial invertebrates, which were nearly twice as high in the forested reach (Kawaguchi and Nakano 2001). Rainbow trout were highly selective for terrestrial invertebrates, which were larger than larvae and adults of aquatic insects, and drifted during periods when the trout fed most (Nakano et al. 1999b). Cutting off inputs of terrestrial prey with the greenhouses forced Dolly Varden to shift to foraging on benthos, which Nakano knew from our earlier work is a shift at which they are adept (Fausch et al. 1997; Nakano et al. 1999a). This intense predation reduced herbivore benthos to about a third the biomass in control reaches (with no greenhouse or charr), which released streambed algae from grazing and caused an increase in algal biomass of about 60% in a trophic cascade (Nakano et al. 1999c). The amazing results of this large-scale manipulation drew immediate attention from stream ecologists, including those who had conducted similar research in northern California (Sabo and Power 2002a, 2002b). On reading the paper, my colleague LeRoy Poff changed the lecture on stream food webs that he was about to present in his course on Stream Ecology to include Nakano’s study (Fausch 2015).

Observations during the first greenhouse experiment, and of Horonai Stream in general, also led Nakano to realize that insects emerging from the stream could be important to riparian predators like birds, bats, and spiders. Each morning Nakano drove along the stream, which ran close to the forest road from his home to his office, and he often observed birds foraging there in all seasons. His former graduate students related that he had also discovered a short paper in a regional journal from the midwestern US which reported the importance of insects emerging from a prairie stream in Kansas to grassland birds, and the concentration of bird foraging near the stream during periods of peak emergence in early summer (Gray 1993; M. Inoue, Y. Kawaguchi, N. Kuhara, personal communication). In addition, during the experiment in 1995, insects emerging from the stream collected rapidly underneath the greenhouse cover, requiring Nakano and his colleagues to first attempt to capture them with sticky traps and then to simply cut a window in each end to allow them to escape (Nakano et al. 1999c; M. Murakami, personal communication). To explore these ideas, he began collaborating with Ph.D. student Masashi Murakami, who had done research on forest birds, their foraging on terrestrial insects that damage riparian trees and shrubs, and the potential for indirect effects among these food-web members (Murakami 1998, 1999; see also Murakami and Nakano 2000).

Based on these observations, Nakano and Murakami developed the idea that “reciprocal” subsidies from stream to forest could also be important to riparian predators during different seasons, and conducted another trio of comparative and experimental studies to test it. During a 14-month period from May 1997 to June 1998 they measured fluxes of invertebrates in both directions, and use of these prey resources by birds and fish. Then during summer 1999 Nakano and his colleagues constructed a 1.2-km fine-meshed greenhouse over Horonai Stream, and installed it again in 2000. They measured spider abundance next to the greenhouse and in 600-m control reaches upstream and downstream (separated by a 100-m buffer), and bat foraging using ultrasonic detectors adjacent to the treatment reach and in a 1.2-km control reach downstream.

Here again, the results were striking. In the comparative study, Nakano and Murakami (2001) found that, on average, 26% of the annual energy budgets of 10 forest birds came from adult aquatic insects emerging from the stream, based on > 13,400 observations of bird foraging, with the highest proportions occurring during winter and early spring when terrestrial invertebrates were scarce. Conversely, an average of 44% of the annual energy budgets for the five stream fishes was derived from terrestrial invertebrates that fell into the stream, based on > 1400 stomach samples flushed from fish, with the highest proportions during summer and fall after most adult aquatic insects had emerged and drifting aquatic invertebrates were primarily small life stages (instars). Mary Power (2001) reported that this intensive year-round study was the best demonstration in any ecosystem of the seasonal shifts of cross-habitat resources subsidies, and that this complementarity of resource pulses across the habitat boundary maintained higher densities, and possibly diversities, of both birds and fishes than would otherwise be supported.

Results from the two experiments with the long greenhouse were equally remarkable. Spiders that weave horizontal orb webs to catch emerging insects (Family Tetragnathidae) were reduced by about 60–80% during May through July in the riparian zone adjacent to the greenhouse that excluded emerging aquatic insects on which they rely, whereas two other spider guilds that forage on terrestrial arthropods were not reduced (Kato et al. 2003). In a similar fashion, during May when aquatic insect emergence was highest, foraging by bats was 97% lower adjacent to the greenhouse than in the control reach, even though there was no difference in foraging in the two reaches the next year when only the greenhouse frame was left standing (Fukui et al. 2006). These results matched those of Sabo and Power (2002a, 2002b) who found that riparian lizard abundance and growth were much lower where emerging insects were excluded from the riparian zone of the Eel River in northern California, USA compared to controls (see Baxter et al. 2005 for a review).

Nakano and his students and collaborators continued to conduct experiments on these reciprocal subsidies and their effects on biodiversity and food web functions using fine-meshed greenhouses to cover Horonai Stream (Fig. 5), and mesh covers over riparian shrubs, eventually completing a total of five major experiments and eight studies, before and after Nakano’s death (Nakano et al. 1999c; Murakami and Nakano 2002; Kato et al. 2003; Kawaguchi et al. 2003; Baxter et al. 2004, 2007; Fukui et al. 2006; Nakano et al. unpublished). These showed, for example, that emerging aquatic insects had positive indirect effects on riparian shrubs during spring by drawing birds into the riparian zone. These birds not only fed on the emerging insects but also gleaned terrestrial invertebrates from shrubs that normally damage their leaves (Murakami and Nakano 2002). In turn, cutting off the input of terrestrial invertebrates into streams using greenhouses, which provides about half of the annual prey biomass and annual energy budget for stream fish (Kawaguchi and Nakano 2001; Nakano and Murakami 2001), either caused half the fish biomass to emigrate (Kawaguchi et al. 2003) or, if fish were enclosed, it reduced their growth markedly (Baxter et al. 2007) and caused them to crop the benthos drastically (Baxter et al. 2004; see Fausch et al. 2010 for a review). In turn, the reduced benthos resulted in markedly reduced insect emergence, a predictable decline in tetragnathid spider abundance, and a trophic cascade that increased biomass of streambed algae (Baxter et al. 2004). These different forms of indirect effects, depending on whether predators are enclosed or free to move, fostered new ecological theory and modeling to incorporate these important elements of complexity in real food webs (Takimoto et al. 2002, 2009). This feedback whereby empirical research does not simply respond by testing ecological theory, but indeed directs its development, is exactly the boundary that Gary Polis (1991) had proposed needed to be crossed. Likewise, the desire for a more dynamic interaction between theory and empiricism was arguably one of the motivations that drew Nakano, Polis, and Mary Power together during Nakano’s fateful visit to California and Mexico in March 2000 (C. Baxter, personal communication).

Fig. 5
figure 5

(image by C. Baxter)

Nakano’s greenhouse used by Colden Baxter and colleagues to conduct experiments in summer 2002 on the combined effects of nonnative trout and cutting off flows of invertebrates between Horonai Stream and its riparian forest

Full size image

These innovative studies first pioneered by Shigeru Nakano and Mary Power and their colleagues, and the grueling field and laboratory work accomplished by them and their many collaborators, resulted in a paradigm shift (sensu Kuhn 1962) in the study of streams, their riparian habitats, and the diverse plants and animals that occupy them. These investigators crossed many boundaries by employing new methods, spanning across adjacent habitats and different taxa, and most importantly, helping build a strong bridge between the disciplines of food web and landscape ecology (see Polis et al. 2004). This research spawned a host of additional manipulative experiments excluding or adding prey subsidies for either aquatic or riparian predators (e.g., Paetzold et al. 2006; Marczak and Richardson 2007; Eros et al. 2012; Sato et al. 2016). Additional research that Nakano participated in or inspired has addressed the spatial and temporal heterogeneity of invertebrate subsidies and how it influences the distribution of predators like birds and spiders at different spatial scales (Iwata et al. 2003, 2010; Iwata 2007; Uesugi and Murakami 2007), and new research has revealed how spatial subsidies of invertebrates stabilize resource fluxes in spatially and thermally heterogeneous river-tributary networks (Uno and Power 2015; Uno 2016). Recent meta-analyses have shown that fish predation reduces adult aquatic insects that emerge to feed terrestrial predators by about 40% (Wesner 2016), and that even though the flux of terrestrial prey to streams is nearly an order of magnitude higher than aquatic prey to riparian zones, the contribution of aquatic prey to terrestrial predators is apparently much higher than of terrestrial prey to aquatic predators (Bartels et al. 2012).

Another set of studies has crossed disciplinary boundaries by applying these concepts to the conservation biology of managed ecosystems. These investigators have assessed how the effects of human land uses such as cattle grazing (Saunders and Fausch 2007, 2012), logging (Inoue et al. 2013), and mining (Kraus et al. 2016), as well as the introduction and invasion of species like nonnative trout (Baxter et al. 2004, 2007; Benjamin et al. 2011, 2013; Lepori et al. 2012) and global climate change (Larsen et al. 2016), can all have strong effects on the processes that produce or shape the flux of invertebrate prey across the stream-riparian boundary. For example, Benjamin et al. (2013) reported that nonnative brook trout reduced the emergence of adult aquatic insects by 55% compared to native cutthroat trout (O. clarkii), and projected that this was sufficient to eliminate emerging insects from the diets of two-thirds of riparian birds that depend, in part, on this food source. Kraus et al. (2016) reported that trout in streams polluted by heavy metals from mining increased their reliance on terrestrial invertebrate prey as in situ aquatic prey were lost owing to the metal pollution.

Conclusion: an enduring legacy for ecology

It is clear from this brief review of only the most highly cited papers published by Shigeru Nakano and his colleagues that this body of research has created an enduring legacy for ecology. His early work on the behavioral ecology of stream salmonids set a new standard for detail and rigor in natural settings, and has inspired others to conduct additional studies aimed at similarly detailed questions for salmonids and other drift-feeding stream fish in diverse locations worldwide. Likewise, his collaborative research on salmonids across spatial scales has inspired other Japanese and western scientists alike to seek this kind of synthesis in other systems.

And for ecology in general, his innovative manipulations of linkages between stream and riparian food webs set new standards for holism and rigor in food web ecology, expanding our horizons to the landscape scale but with close attention to resolving the details of phenology, diet, and behavior of the species involved (Power 2001; Fausch et al. 2002a). Indeed, I can offer no better tribute than a recent personal communication from Mary Power, whose own work independently sought to answer many similar questions about river-watershed exchanges. She wrote “Shigeru Nakano’s legacy is very strong still today. He inspired a huge number of ecologists, and especially young Japanese ecologists, to do experimental work and to think at landscape scales about terrestrial-river interactions. He was an extraordinary figure in world ecology, and his tragic early loss has not diminished his strong legacy, thanks to devoted friends and colleagues and the many students he inspired.”

One of Shigeru Nakano’s most indelible contributions is the many graduate students and postdoctoral scientists he mentored, his own and those of other academic colleagues. During his 15-year career those with whom he published papers number more than 30, and many of these have become accomplished professors and research scientists in their own right. Each carries unforgettable memories about their years in the “Nakano school” and all that Shigeru taught them about research and life, and each has passed those lessons on to their own students and colleagues. Finally, perhaps an unappreciated aspect of Shigeru Nakano’s legacy is the ongoing interaction and collaboration he inspired among those of us who followed the clues he left and conducted additional research that honors his work (e.g., see RiverWebs documentary film; Fausch 2015). These experiences have forged new friendships and connections that span the globe (Fig. 6) and continue to foster reciprocal collaborations and new research and synthesis among Japanese researchers and western scientists that are advancing ecological research worldwide (e.g., Dunham et al. 2008; Fausch et al. 2010; Richardson and Sato 2015; Sato et al. 2016).

Fig. 6
figure 6

Faculty and graduate students assembled at Tomakomai Experimental Forest on June 15, 2003, many of whom were conducting studies inspired by the research of Shigeru Nakano. From left to right: Back row – Kurt Fausch, Masashi Murakami, Colden Baxter, Masataka Miura, Toshihide Hirao, Koh Hasegawa, Kenta Tanaka, Yoshinori Taniguchi, Scott Laeser, Daisuke Kishi, Tsutomu Hiura. Middle row (right of center): Mikio Inoue, Hiroshi Miyata, Hiromitsu Kamauchi, Dai Fukui, Taro Tosuji. Front row (kneeling): Maiko Hotta, Emiko Kubota, Eri Nabeshima, Michiko Matsuda, Fumie Okabe, Akane Uesugi

Full size image