SEARCH

SEARCH BY CITATION

Keywords:

  • headwater streams;
  • stream network;
  • forest clear-cutting;
  • debris flows;
  • population extinction;
  • amphipod;
  • dispersal

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

To understand the impacts of debris flows on the distribution of an amphipod with limited dispersal ability in the context of stream networks, we surveyed the presence of Gammarus nipponensis in 87 headwater streams with different legacies of debris flow occurrence within an 8.5-km2 mountain catchment. The amphipod was present in only 7% of the streams impacted by debris flows after 1977; in contrast, it was present in 69% of the streams that had older or no debris flow occurrence. The absence of the amphipod in certain headwater streams did not appear to be related to water chemistry because pH and calcium concentrations differed little among streams within the catchment. In addition, survival rates of individuals incubated in streams with the amphipod present and absent did not differ significantly. Debris flows appeared to displace amphipod populations, and the absence of amphipods in streams for more than 30 years after debris flow occurrence suggests that considerable time is required for the recovery of populations. Because of geographic isolation from the source of colonists, headwater streams in the uppermost sections of the catchment and those indirectly connecting to the main stream via tributaries appear to be at greater disadvantage for receiving colonists from other areas and thus population recovery. Copyright © 2012 John Wiley & Sons, Ltd.


INTRODUCTION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Headwater streams, the uppermost channels in stream networks, have attracted attention for their roles in transporting sediments, nutrients and organic matter to downstream ecosystems (Gomi et al., 2002; Benda et al., 2003; Wipfli et al., 2007), and contributing species diversity to the entire stream network (Meyer et al., 2007; Clarke et al., 2008). Headwater streams are often isolated from each other separated by virtue of ridges and larger streams within the larger catchment network. For invertebrate species that are adapted to headwater streams, physical and biological barriers can limit the migration of individuals from one headwater stream to another (Richardson and Danehy, 2007; Clarke et al., 2008). In general, upstream reaches are the key source areas of colonists for stream invertebrates (Mackay, 1992). However, because headwater streams are the uppermost reaches, populations in headwater streams typically lack sources of colonists in the upstream direction. Thus, the period for recovery may be long once local populations disappear, and dispersal of individuals among headwater streams would play a key role in population recovery in headwater streams (Lake et al., 2007; Richardson and Danehy, 2007; Clarke et al., 2008).

Debris flows are discrete and severe disturbances that shape and modify headwater mountain streams (Lamberti et al., 1991; Swanson et al., 1998; Bilby et al., 2003). Landslides that occur on hillslopes often travel to headwater channels and transform into debris flows through the accumulation of water, sediment and wood (Sidle and Ochiai, 2006). Human activities on hillslopes such as logging and forest road construction can induce slope failures and subsequent debris flows in channels (Sidle et al., 2006; Imaizumi et al., 2008). Scour, transport and deposition of sediment because of debris flows substantially alter in-channel physical characteristics (Benda et al., 2003; Gomi et al., 2003) and also displace organisms during the transport of sediment. Substantial decreases in invertebrate density immediately after an occurrence of debris flows have been reported in mountain streams (Lamberti et al., 1991; Kiffney et al., 2004). Therefore, debris flows, especially those passing through an entire reach of a headwater channel, may act as a local extinction event for invertebrate populations.

Impacts of debris flows on invertebrate populations cannot be fully understood by observations from a limited number of headwater streams. Because of the episodic and unpredictable occurrence of debris flows, few studies have reported the ecosystem status both before and after the occurrence of a debris flow in a given stream (Lamberti et al., 1991). Previously, we examined the responses of the macroinvertebrate community to the occurrence of debris flows by looking at a unique set of headwater streams with different legacies of debris flow occurrence in a steep mountain catchment (Kobayashi et al., 2010). On the basis of monthly macroinvertebrate surveys in ten headwater streams, we clarified the absence of amphipods (Gammarus nipponensis Ueno) in streams with recent debris flows (<20–30 years), despite their dominance in streams with older or no debris flows. Because no evidence of lethal conditions associated with water chemistry or flow regime was apparent, the amphipod populations were assumed to be obliterated by debris flows, with several decades required for populations to recover. Such long periods prior to recovery would be unexpected unless difficulties in recolonization of individuals existed from other headwater streams. Among-stream dispersals seem to be opportunistic and rare for amphipods of mountain streams because of a lack of specialized dispersal stage as well as flight ability (Marchant, 1981; Siegismund and Müller, 1991), which results in genetic differentiation among different catchments (Gooch, 1989; Gooch, 1990; Siegismund and Müller, 1991). Furthermore, recolonization of individuals to disturbed headwater streams can be greatly affected by the presence of nearby sources of colonists (i.e. headwater streams with intact populations) within a stream network (Lowe and Bolger, 2002; Grant et al., 2007). Thus, despite the correspondence of the absence of G. nipponensis with debris flow occurrences in our previous study, ten streams are too few to understand the effects of debris flows on populations of G. nipponensis within an entire stream network.

The occurrence of hydrogeomorphic disturbances (e.g. landslides, debris flows, hyperconcentrated flows) and their response to land cover change is often patchy and manifested in space and time within a stream network (Benda et al., 2003; Gomi and Sidle, 2003; Wilford et al., 2004), though few studies have demonstrated how aquatic organisms and ecosystems respond to such disturbances in a landscape context. Our primary objective in this study is to understand the impacts of debris flows as a local extinction event on G. nipponensis in the context of a stream network. We investigated the presence and absence of G. nipponensis in 87 headwater streams with different legacies of debris flows (i.e. years since the last debris flow occurrence) in a steep mountain catchment. Amphipods are often dominant in headwaters and springs with relatively stable flow regimes (Glazier and Gooch, 1987; Barquín and Death, 2004; Meyer et al., 2007), play important roles in the breakdown of plant detritus (Dangles and Guerold, 2001; Tiegs et al., 2008) and have strong impacts on other benthic invertebrates through predation or competition (Kelly et al., 2003; Barquín and Death, 2004). We hypothesized that G. nipponensis is absent in headwater streams with recent debris flows and without nearby sources of colonists (i.e. streams where the amphipod was present). Water chemistry such as pH and calcium concentrations has been reported to restrict the distribution of amphipods (Okland and Okland, 1985; Glazier et al., 1992; Zehmer et al., 2002). These as well as other water chemistry parameters were relatively uniform among headwater streams in our study catchment (Fukushima and Tokuchi, 2008), and there was no evidence that water chemistry affected the presence of G. nipponensis (Kobayashi et al., 2010). Nevertheless, solutes that were not measured may inhibit the survival of G. nipponensis in the streams; thus, in this study, we also examined the survival of G. nipponensis in headwater streams with and without native populations of G. nipponensis.

STUDY SITE

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

The study was conducted in a mountainous catchment (8.5 km2, 860–1370 m above sea level) in central Kii peninsula, Japan (34°04′N, 135°35′E, Figure 1a). The area is upstream of Kanno River, within the Kumano River basin. Surficial geology in the area consists of alternating beds of sandstone and shale or mudstone. Mean hillslope gradient is around 40° with thin soils (generally <1 m deep). Annual precipitation in the area ranged from 2122 to 3508 mm from 1998 to 2002. Forests within the entire catchment have been clear-cut and planted, predominantly with Japanese cedar (Cryptomeria japonica D. Don), since 1912. Trees are typically planted in the year following clear-cutting. Stands are thinned by cutting approximately 25% of the trees 30 years after planting. All forest management entries have been conducted either within the context of individual small headwater catchments or within sets of adjoining catchments. In subsequent years, these harvest activities have been rotated within the larger catchment. Consequently, the greater study catchment is composed of a mosaic of headwater catchments with stand ages ranging from 1 to 91 years (in 2005) (Figure 1b). Stream water chemistry of headwater streams in this catchment was studied by Fukushima and Tokuchi (2008). They reported that nitrate concentrations were greater (>5-fold) in streams with recent clear-cutting than in streams with older (>10 years) forests. The probability of new landslide occurrence increased during the period from 1 to 25 years after clear cutting (Imaizumi et al., 2008), which corresponds to a decrease in slope stability because of root strength deterioration (Sidle, 1992).

image

Figure 1. (a) Map of the study catchment and 87 headwater streams that were surveyed. Several locations in the main stream and tributaries that connect headwater streams were also surveyed. (b) The spatial distribution of forest age and debris flow occurrence in the catchment.

Download figure to PowerPoint

Our survey included all headwater streams (i.e. mostly first-order streams) in the catchment (a total of 87 streams, Figure 1a) except for those we could not access or those with intermittent flow. According to a 10-m contour topographic map, the catchment area of these streams ranged from 0.01 to 0.22 km2 with a mean area of 0.06 km2; channel length ranged from 16 to 1300 m with a mean of 350 m. We identified occurrences of debris flows in each headwater stream using aerial photographs taken in 1948, 1964, 1967, 1971, 1976, 1984, 1989, 1994, 1998 and 2003. The photographs provided a range of possible years for the debris flow occurrences rather than the exact year of the occurrence. For example, if a debris flow was first detected on photographs from 2003, the event occurred sometime between 1998 and 2003. Because we focused on debris flows that are large enough to damage populations of amphipods, we defined debris flows as those scouring events that initiated from the upper part of the channel reach.

METHODS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

A field examination of the existence of G. nipponensis was conducted for all the headwater streams from June to September 2005 except for nine streams for which we already knew their presence or absence by the previous study (Kobayashi et al., 2010). It is usually difficult to conclude that an invertebrate species is absent without intensive sampling. However, G. nipponensis was both abundant and dominant if they were present, irrespective of season (Kobayashi et al., 2010). Thus, we recorded the presence or absence of G. nipponensis based on a 10-min sampling using a kicknet (mesh size: 1 mm) in each stream. As a first step, we sampled for litter accumulations in slow-current areas where G. nipponensis tended to be abundant. If no individual was found there, then we searched for other microhabitats such as cobbles in fast-current areas. We argue that the 10-min sampling is sufficient to judge the presence of G. nipponensis and appropriate for surveying many streams within a short period, although the abundance of individuals cannot be quantified by this method. The same sampling procedure was conducted at 17 sites in second-order and third-order tributaries that connect headwater streams and the main stream (third-order or higher order streams) (Figure 1a). We also recorded the locations of sediment control dams and waterfalls (in the main stream and the major tributaries) that may act as physical barriers to the migration of G. nipponensis. A logistic regression model was used to test whether the probability P of the presence of G. nipponensis in a headwater stream was associated with forest age within the catchments (1–91 years) or the year since the last debris flow (2–57 years).

  • display math

where x is forest age or the year since the last debris flow and β0 and β1 are regression parameters.

Field incubation of G. nipponensis was conducted in two streams in August 2005: one with and one without G. nipponensis. The former stream was located in an undisturbed forest (2.7 km west from the catchment) and the latter stream was no. 73 in Figure 2. In each stream, flow was partially diverted by a PVC pipe and routed into 12 separate plastic chambers (area: 225 cm2, depth: 6 cm). The inflow volume of each chamber was adjusted to 0.06 l s−1 using a stopcock at the outlet of the pipe. Ten individuals (>5 mm in body length) of G. nipponensis and five to ten pieces of alder leaves (about 4 g dry weight) for food were enclosed in a 1-mm mesh net and placed into each chamber. Individuals of two different populations were used for the incubation: one from the stream in the undisturbed forest and the other from a stream (i.e. stream no. 76 in Figure 2) that was located close to the incubated stream without G. nipponensis (no. 73). In each of the incubated streams, six randomly selected chambers were used for each population. The number of individuals that survived was recorded after 7 and 21 days for each chamber. Repeated measures analysis of variance (ANOVA) was conducted to examine the difference in survival between the two streams and between the two populations.

image

Figure 2. Presence or absence of G. nipponensis in 87 headwater streams (gray lines with numbers) and at several locations in the main stream and tributaries (blue lines). Only small numbers of individuals (less than ten) were collected in a few streams. Several sediment control dams and waterfalls existed along the main stream and in a few tributaries.

Download figure to PowerPoint

RESULTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Distribution of amphipods in the catchment

At least one debris flow occurred in 67 of 87 headwater catchments between 1964 and 2005 according to aerial photos. All debris flows occurred within 30 years after clear-cutting (Figure 3). The time lag between clear-cutting and debris flow occurrence was 0–4 years for 23% of the debris flows, 5–9 years for 39%, 10–14 years for 27%, 15–19 years for 8% and 20–30 years for 4%. Almost 90% of the debris flows occurred <15 years after clear-cutting (Figure 3, most occurrences fall between y = x and y = x + 15 dashed lines).

image

Figure 3. Relationship between forest age and the year of debris flow occurrence detected by aerial photos. x-Axis shows both forest age and planted year (i.e. a year after clear-cutting, in parenthesis). Dashed lines indicate debris flow occurred soon after clear-cutting (y = x) or occurred 15 years after clear-cutting (y = x + 15).

Download figure to PowerPoint

Gammarus nipponensis was present in 26 of the 87 headwater streams (Figure 2). More than ten individuals were found during the first minute of the initial step of net collection except for two streams where less than ten individuals were found throughout the 10-min sampling (shown by open triangles in Figure 2). Individuals were found in only two of nine tributaries (second-order or third-order streams) that connect headwater streams, and in one of these, only a few individuals were found throughout the 10-min sampling. No individuals were found throughout the main stream (i.e. >third-order streams).

The distribution of G. nipponensis among headwater streams was associated with forest age and the year since the last occurrence of a debris flow (Figure 4 and Table 1). G. nipponensis was present in >75% of headwater streams in forests of both the youngest (0–9 years) and the oldest (91 years) age classes (Figure 4a, the number of surveyed streams in each class is also shown). In contrast, G. nipponensis was present in <40% of the streams in forests 10–47 years old, and no individuals were present in streams in the middle-aged class forests (20–29 years). Based on a logistic regression model, the probability of the presence of G. nipponensis significantly decreased with forest age for streams with forests <30 years and increased with forest age for streams with forests >30 years (Table 1). G. nipponensis was present in only 7% of the streams impacted by debris flows after 1976 (Figure 4b); however, they were present in >60% of the streams with older debris flows (in the 1960s) and in >80% of the streams with no debris flows. The probability of the presence of G. nipponensis significantly increased with time since the last debris flow (Table 1).

image

Figure 4. Number of streams surveyed and the proportion of streams with G. nipponensis present in (a) different forest age classes and (b) different debris flow history classes.

Download figure to PowerPoint

Table 1. Summary of logistic regression model [P(x) = 1/(1 + exp(−(β0 + β1x)))] for predicting the probability of the presence of G. nipponensis using forest age of the catchments or the year since the last debris flow
  CoefficientWald zP
x = forest age (0–29), n = 47 
β04.192.6390.008
β1−0.36−3.2610.001
x = forest age (30–91), n = 40 
β0−3.25−3.0360.002
β10.062.6190.009
x = the year since the last debris flow, n = 87
β0−3.74−5.068<0.001
β10.084.746<0.001

The distribution of G. nipponensis was spatially biased at the scale of the entire catchment (Figure 2). G. nipponensis was absent in all headwater streams in the two major uppermost tributaries of the main stream (i.e. streams 1–12, 39–45, 46–56 in Figure 2); most of these streams drained middle-aged forests (20–39 years) and had recent (after 1976) debris flows (Figure 1b). G. nipponensis was also absent in most headwater streams that connect indirectly to the main stream through tributaries (i.e. streams 19–20, 24–25, 28–31, 34–37, 63–74, 82–85 in Figure 2). Some headwater streams that connect indirectly to the main stream through tributaries (i.e. streams 28–31, 82–85) drained older forests (>40 years) and had older debris flows (1960s and 1970s; Figure 1b). In other words, the presence of G. nipponensis was limited to headwater streams that directly connect to the main stream (specifically along the lower half of the main stream) and streams 14–15 and 17–18 that connected indirectly to the main stream. The upstream limits of distribution of G. nipponensis did not correspond to the locations of either sediment control dams or waterfalls (Figure 2).

Survival of individuals in stream water

On average, more than seven of ten individuals survived during the 21-day experiment that was conducted in streams with and without G. nipponensis (Figure 5). Because only a few dead individuals were found in chambers, missing individuals were possibly consumed by the other individuals or escaped from the chambers. Effects of stream and population on the survival were not detected by repeated-measures ANOVA (Table 2). The mean survival during the first 7 days was lower in the stream with G. nipponensis than in the stream without G. nipponensis (Figure 5). In the stream with G. nipponensis, the survival was relatively low (five to seven individuals) in three chambers, all of which had deposits of fine particulate organics on leaves. If results from these three chambers are excluded, then the mean survival was similar between these two streams. Thus, effects of differences in stream water on the survival of G. nipponensis between streams with and without G. nipponensis were not detected.

image

Figure 5. The proportion of individuals that survived in experiment chambers in streams where G. nipponensis was absent and present. Populations of two headwater streams were used in this experiment. Error bars denote standard deviation (±1) among six replicates.

Download figure to PowerPoint

Table 2. Summary of repeated-measures analysis of variance for the incubation experiment of G. nipponensis
FactorDegrees of freedomMean squaresFP
Stream110.0833.3330.083
Population11.3330.4410.514
Stream × population14.0831.350.259
Error203.025

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Factors affecting the distribution of amphipods among headwater streams

Our surveys of 87 headwater streams in an 8.5-km2 catchment revealed that the spatial distribution of G. nipponensis corresponded with the last occurrence of a debris flow as well as forest age at the catchment scale. The absence of the G. nipponensis in streams with recent debris flows supports our assumption that debris flows act as local extinction events of the populations in headwater streams. Debris flows can damage the population of G. nipponensis directly through killing or removing individuals from a headwater stream (Lamberti et al., 1991). Alternatively, debris flows may damage the population indirectly by changing hydraulic conditions and food availability for G. nipponensis. Streams with recent debris flows tend to have less slow-flow habitat, where G. nipponensis is usually abundant, and less accumulations of coarse particulate organic matter, an important food resource of G. nipponensis (Kobayashi et al., 2010). Reduced habitat and food availability may increase the probability of local extinction or delay the recovery of the population.

Forest clear-cutting may indirectly affect G. nipponensis by increasing the likelihood of debris flow occurrence. There was a time lag between forest clear-cutting and debris flow occurrence in our catchment (from 0 to 30 years with most debris flows occurring 5–9 years after clear-cutting). All of these debris flows appeared to occur at the same time as the associated hillslope landslides; i.e. 1–25 years after clear-cutting (Imaizumi et al., 2008). Thus, the time lag can be explained by a gradual decrease in slope stability because of root strength deterioration after clear-cutting and the increased probability of slope failure during this ‘window’ of susceptibility prior to substantial forest regeneration (Sidle, 1992). Although some landslides in our catchment (91 of 362 distinct landslides that occurred from 1976 to 2003) were associated with the ridgeline road (F. Imaizumi, unpublished data), given the distance of the road from the headwater channels, many of these landslides did not initiate debris flows in the channel. The frequent occurrence of G. nipponensis in streams with forests of the youngest age class (Figure 4a) suggests that environmental changes associated with immediate effects of clear-cutting, such as increases of solar radiation and stream temperature (Kobayashi et al., 2010), did not eliminate or affect G. nipponensis.

The absence of G. nipponensis in streams with recent debris flows is likely not associated with unsuitable water chemistry. As mentioned previously, water chemistry including pH and calcium concentration, which can restrict the distribution of species in the same genus (Okland and Okland, 1985; Glazier et al., 1992; Zehmer et al., 2002), is relatively uniform in our study catchment (Fukushima and Tokuchi, 2008; Kobayashi et al., 2010). The incubation experiment also showed no evidence that water chemistry inhibited the survival of G. nipponensis in streams with recent debris flows. This finding can be applied to other headwater streams in the greater catchment with similar water chemistry; water in the streams without G. nipponensis was neither lethal nor different compared with the streams with G. nipponensis.

Although correspondence existed between the last occurrence of debris flows and the absence of G. nipponensis, it is unknown whether G. nipponensis existed in the streams prior to the last occurrence of debris flows. Because all headwater streams are proximate to one another (less than 200–300 m) within the catchment and no physical barriers exist at the fronts of their distribution, we assume that G. nipponensis was present at least once in each headwater stream. In a neighbouring catchment across the ridgeline to the west, which is dominated by undisturbed forest, G. nipponensis was present in almost all headwater streams. Thus, it is plausible that G. nipponensis was absent in some headwater streams in this study because of an extinction event such as debris flow occurrence. Because forest clear-cutting has been conducted since 1912, the populations of G. nipponensis may have been obliterated by debris flows that occurred earlier than those detected by aerial photos.

The main stream and the numerous tributaries that connect headwater streams are probably not very suitable habitats for G. nipponensis. Because of higher hydraulic competence in the main stream, little plant litter was observed and refugia for G. nipponensis during high flows are probably limited throughout the main stream. In addition, the main stream harbors a predatory fish (a salmonid, Oncorhynchus masou), which can potentially feed on and reduce the abundance of G. nipponensis. In general, steep headwater streams are usually free of large predators, and thus considered as refugia for invertebrates (Meyer et al., 2007; Richardson and Danehy, 2007). Because individuals of various life stages (<1 to >10 mm in size) of G. nipponensis have been observed in headwater streams, they probably complete their life cycles within individual streams. Thus, individuals in different headwater streams appear to be isolated by the main stream and tributaries, and thus belong to different populations.

Effects of position of headwater streams on population recovery

As we expected, G. nipponensis was absent in streams without a nearby source of colonists as evidenced by a biased distribution of absent streams to certain areas in the catchment. First, G. nipponensis was absent in headwater streams in the uppermost portion of the catchment (Figure 2). Because of a lack of strong swimming abilities of G. nipponensis, their dispersal through the main stream is biased in the downstream direction, similar to other Gammarus species (see Marchant, 1981; Humphries and Ruxton, 2003). Thus, the opportunities for colonization in headwater streams are probably lower if potential sources of colonists are located downstream along the main stem. Headwater streams in the upper portion of the larger catchment are probably at a disadvantage for population recovery once nearby populations disappear compared with the headwater streams in the lower portion of the larger catchment.

Second, G. nipponensis was absent in most headwater streams that indirectly connect to the main stream through a tributary (and was also absent in the tributary) in the lower portion of the catchment, whereas they were present in streams directly connected to the main stream (Figure 2). As such, our results differ with studies of a headwater salamander in North America (Lowe and Bolger, 2002; Grant et al., 2009) in which headwater streams connecting to the same tributary system were efficient sources of colonists for each other. In our catchment, headwater streams in the same tributary system have similar legacies of forest clear-cutting and subsequent debris flow occurrences in most cases (Figure 1b). Thus, these headwater populations were probably damaged simultaneously, and thus cannot be the source of the colonists for adjoining streams. Once all populations in headwater streams in the same tributary system disappear, geographical isolation from a source of colonists is greater compared with the headwater streams that directly connect to the main stream. Exceptions to this pattern were streams 14–15 and 17–18, which also indirectly connect to the main stream but where G. nipponensis was present. Streams 17–18 have not experienced debris flows according to the aerial photos, and thus, the presence of G. nipponensis would be expected.

The previous arguments are possible reasons that would support a delay in the recovery of populations after debris flow occurrence in systems without any dispersal barriers. Some headwater streams without G. nipponensis that indirectly connect to the main stream had not experienced debris flows for more than 30–40 years. On the contrary, one of the headwater streams with G. nipponensis that directly connect to the main stream experienced a debris flow between 1984 and 1989 (stream 26) indicating that G. nipponensis might recover within 20 years. Although our data do not allow a quantitative comparison between the distance from the source and the time for recovery, the differences in the presence/absence of G. nipponensis between neighbouring headwater streams with similar legacies of forest clear-cutting and debris flow occurrence (e.g. streams 24–25 and stream 26, or stream 80 and streams 82–85) implies the effect of connectivity to the main stream on the time for recovery of G. nipponensis.

CONCLUSIONS AND IMPLICATIONS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

This study demonstrated that the distribution of G. nipponensis among headwater streams within a mountain catchment was largely associated with legacies of debris flows. Although debris flows probably induced local population extinctions through displacement of individuals, associated changes in hydraulic conditions and food availability in the channel may also delay population recovery. Our investigation suggests that more than several decades are potentially required for populations to recover after local extinction. Forest clear-cutting indirectly affects G. nipponensis by increasing the probability of debris flows (e.g. Sidle and Ochiai, 2006).

In these mountain sites, the presence/absence of G. nipponensis in headwater streams is a key ecosystem attribute. G. nipponensis is the only invertebrate in these headwater systems that achieves a high biomass (except for large-bodied invertebrates such as crabs and dobsonflies) via growth and abundance; total shredder and non-predator biomass from monthly collected drift samples differed substantially between streams with and without G. nipponensis (roughly 300-fold for shredders and sevenfold for total non-predators; Kobayashi et al., 2010). Additionally, streams with and without G. nipponensis differed in the rate of leaf breakdown and fine particulate organic matter concentrations in stream water (S. Kobayashi, unpublished data). Such differences in invertebrate biomass may also affect the main stem of the stream ecosystem if invertebrates in headwater streams are transported to the main stream and contribute significantly to the diet of predatory fish (Wipfli et al., 2007). Thus, if G. nipponensis populations are removed from the entire catchment, then the effects would include substantial reductions in animal biomass and production and related organic matter in the entire stream system, not only the loss of a headwater species.

The geographic location of headwater streams within a given catchment is a key factor for population recovery. First, the amphipod populations are potentially more difficult to recover after extinction in headwater streams at the uppermost reaches of stream networks compared with those in downstream reaches. Second, geographic isolation from the source of colonists is greater for headwater streams that are indirectly connected to the main stream through tributaries compared with headwater streams that are directly connected to the main stream. The latter case occurs in situations where all headwater populations in the same tributary system are damaged simultaneously (e.g. streams 24–25, 28–31 and 82–85, Figure 1b). Because debris flows are important agents that provide sediment and increase habitat heterogeneity over 100–1000 years time scales (Benda et al., 2003; Bilby et al., 2003), it is important to consider the spatial patterns of disturbance for the conservation of both headwater and larger stream ecosystems. From the standpoint of the amphipod, headwater streams in upper reaches need to be protected because they are an important source of colonists for downstream reaches. Controlling the occurrence of disturbances so that they are not spatially concentrated may be one option for securing neighbouring headwater streams as the potential source of colonists.

Headwater streams are often dominated by invertebrates such as crustaceans (including amphipods), triclads and mollusks, which are flightless and have limited dispersal abilities (Meyer et al., 2007). Even some headwater insect species show evidence of limited dispersal ability (Tojo, 2005; Miyairi and Tojo, 2007). Indeed, headwater-adapted species other than G. nipponensis in our catchment tended to be less abundant in the headwater streams with recent debris flow (Kobayashi et al., 2010). Thus, the processes of local extinction and recovery shown in this study can be also applied to other headwater-adapted fauna.

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

We express our appreciation to Dr N. Tokuchi, Dr N. Ohte, Dr K. Fukushima and the staff of Wakayama Experimental Forest of Kyoto University for supporting our field work and providing insightful comments. Appreciation is extended to Sanko Forestry Ltd for allowing us to conduct research on their property and for providing information related to logging operations. Funding for this study was from the Japan Society for the Promotion of Science (JSPS #16380102 for RCS and #17780123 for SK). This document has been reviewed in accordance with U.S. Environmental Protection Agency policy and approved for publication.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. STUDY SITE
  5. METHODS
  6. RESULTS
  7. DISCUSSION
  8. CONCLUSIONS AND IMPLICATIONS
  9. ACKNOWLEDGEMENTS
  10. REFERENCES
  • Barquín J, Death RG. 2004. Patterns of invertebrate diversity in streams and freshwater springs in Northern Spain. Archiv für Hydrobiologie 161: 329349.
  • Benda L, Veldhuisen C, Black J. 2003. Debris flows as agents of morphological heterogeneity at low-order confluences, Olympic Mountains, Washington. Geological Society of America Bulletin 115: 11101121.
  • Bilby RE, Reeves GH, Dolloff CA. 2003. Sources of variability in aquatic ecosystems: factors controlling biotic production and diversity. In Strategies for Restoring River Ecosystems: Sources of Variability and Uncertainty in Natural and Managed Systems, Wissmar RC, Bisson PA (eds). American Fisheries Society: Bethesda, MD; 129146.
  • Clarke A, Nally RM, Bond N, Lake PS. 2008. Macroinvertebrate diversity in headwater streams: a review. Freshwater Biology 53: 17071721.
  • Dangles O, Guerold F. 2001. Linking shredders and leaf litter processing: insights from and acidic stream study. International Review of Hydrobiology 86: 395406.
  • Fukushima K, Tokuchi N. 2008. Effects of forest clearcut and afforestation on streamwater chemistry in Japanese cedar (Cryptomeria japonica) forests: comparison among watersheds of various stand ages. Journal of Japanese Forest Research 90: 616 (in Japanese).
  • Glazier DS, Gooch JL. 1987. Macroinvertebrate assemblages in Pennsylvania (U.S.A.) spring. Hydrobiologia 150: 3343.
  • Glazier DS, Horne MT, Lehman ME. 1992. Abundance, body composition and reproductive output of Gammarus minus (Crustacea: Amphipoda) in ten cold springs differing in pH and ionic content. Freshwater Biology 28: 149163.
  • Gomi T, Sidle RC. 2003. Bedload transport in managed steep-gradient headwater streams of southeastern Alaska. Water Resources Research 39: 1336. DOI:10.1029/2003WR002440.
  • Gomi T, Sidle RC, Richardson JS. 2002. Understanding processes and downstream linkages of headwater systems. Bioscience 52: 905916.
  • Gomi T, Sidle RC, Woodsmith RD, Bryant MD. 2003. Characteristics of channel steps and reach morphology in headwater streams, southeast Alaska. Geomorphology 51: 225242.
  • Gooch JL. 1989. Genetic differentiation in relation to stream distance in Gammarus minus (Crustacea, Amphipoda) in Appalachian watersheds. Archiv für Hydrobiologie 114: 505519.
  • Gooch JL. 1990. Spatial genetic patterns in relation to regional history and structure Gammarus minus (Amphipoda) in Appalachian watersheds. American Midland Naturalist 124: 93104.
  • Grant EHC, Lowe WH, Fagan WF. 2007. Living in the branches: population dynamics and ecological processes in dendritic networks. Ecology Letter 10: 165175.
  • Grant EHC, Green LE, Lowe WH. 2009. Salamander occupancy in headwater stream network. Freshwater Biology 54: 13701378.
  • Humphries S, Ruxton GD. 2003. Estimation of intergenerational drift dispersal distances and mortality risk for aquatic macroinvertebrates. Limnology and Oceanography 48: 21172124.
  • Imaizumi F, Sidle RC, Kamei R. 2008. Effects of forest harvesting on the occurrence of landslides and debris flows in steep terrain of central Japan. Earth Surface Processes and Landforms 33: 827840.
  • Kelly DW, Dick JTA, Montgomery WI, MaCneil C. 2003. Differences in composition of macroinvertebrate communities with invasive and native Gammarus spp. (Crustacea: Amphipoda). Freshwater Biology 48: 306315.
  • Kiffney PM, Volk CL, Beechie TJ, Murray GL, Pess GR, Edmonds RL. 2004. A high-sensitivity disturbance event alters community and ecosystem properties in West Twin Creek, Olympic National Park, Washington, USA. American Midland Naturalist 152: 286303.
  • Kobayashi S, Gomi T, Sidle RC, Takemon Y. 2010. Disturbances structuring macroinvertebrate communities in steep headwater streams: relative importance of forest clearcutting and debris flow occurrence. Canadian Journal of Fisheries and Aquatic Sciences 67: 427444.
  • Lake PS, Bond N, Reich P. 2007. Linking ecological theory with stream restoration. Freshwater Biology 52: 597615.
  • Lamberti GA, Gregory ST, Ashkenas LR, Wildman RC, Moore KMS. 1991. Stream ecosystem recovery following a catastrophic debris flow. Canadian Journal of Fisheries and Aquatic Sciences 48: 196208.
  • Lowe WH, Bolger DT. 2002. Local and landscape-scale predictors of salamander abundance in New Hampshire headwater streams. Conservation Biology 16: 183193.
  • Mackay RJ. 1992. Colonization by lotic macroinvertebrates: a review of processes and patterns. Canadian Journal of Fisheries and Aquatic Sciences 49: 617628.
  • Marchant R. 1981. The ecology of Gammarus in running water. In Perspectives in Running Water Ecology, Lock MA, Williams DD (eds). Plenum Press: New York, NY; 225249.
  • Meyer JL, Strayer DL, Wallace JB, Eggert SE, Helfman GS, Leonard NE. 2007. The contribution of headwater streams to biodiversity in river networks. Journal of the American Water Resources Association 43: 86103.
  • Miyairi K, Tojo K. 2007. Biology of the mayfly Bleptus fasciatus Eaton (Insecta: Ephemeroptera, Heptageniidae), with special reference to the distribution, habitat environment, life cycle, and nuptial behavior. Limnology 8: 8593.
  • Okland KA, Okland J. 1985. Factor interaction influencing the distribution of the freshwater “shrimp” Gammarus. Oecologia 66: 364367.
  • Richardson JS, Danehy RJ. 2007. A synthesis of the ecology of headwater streams and their riparian zones in temperate forests. Forest Science 53: 131147.
  • Sidle RC. 1992. A theoretical model of the effects of timber harvesting on slope stability. Water Resources Research 28: 18971910.
  • Sidle RC, Ochiai H. 2006. Landslides: processes, prediction, and land use. Water Resources Monograph 18. American Geophysical Union: Washington, DC.
  • Sidle RC, Ziegler AD, Negishi JN, Rahim Nik A, Siewc R, Turkelboom F. 2006. Erosion processes in steep terrain—truths, myths, and uncertainties related to forest management in Southeast Asia. Forest Ecology and Management 224: 199225.
  • Siegismund HR, Müller J. 1991. Genetic structure of Gammarus fossarum populations. Heredity 66: 419436.
  • Swanson FD, Johnson SL, Gregory SV, Acker SA. 1998. Flood disturbance in a forested mountain landscape. Bioscience 48: 681689.
  • Tiegs SD, Peter FD, Robinson CT, Uehlinger U, Gessner MO. 2008. Leaf decomposition and invertebrate colonization response to manipulated litter quantity in streams. Journal of the North American Benthological Society 27: 321331.
  • Tojo K. 2005. Genetic diversity in a headwater-specific aquatic insect Dipteromimus tipuliformis McLachlan (Ephemeroptera, Dipteromimidae), based on the mitochondrial 16S rRNA gene sequences. Ecology and Civil Engineering 7: 119127 (in Japanese).
  • Wilford DJ, Sakals ME, Innes JL, Sidle RC, Bergerud WA. 2004. Recognition of debris-flow hazard through watershed morphometrics. Landslides 1: 6166.
  • Wipfli MS, Richardson JS, Naiman RJ. 2007. Ecological linkages between headwaters and downstream ecosystems: transport of organic matter, invertebrates, and wood down headwater channels. Journal of the American Water Resources Association 43: 7285.
  • Zehmer JK, Mahon SA, Capelli GM. 2002. Calcium as a limiting factor in the distribution of the amphipod Gammarus pseudolimnaeus. American Midland Naturalist 148: 350362.