Biochemical inuence of amphidromous shrimps on stream ecosystems in Japan

Not only through species interactions but also through excretion of nutrients, consumers can have profound effects on the ecosystem structure. While many studies in lentic ecosystems address both effects in combination, little is known about their linkages in lotic ecosystems. By a combination of eld manipulative experiment, excretion measurements of consumes, and eld surveys, we evaluated biochemical effects of amphidromous shrimps on stream ecosystems. The eld manipulative experiment showed that the presence of shrimp suppressed the total aquatic insect biomass by 9% but increased the total benthic macroinvertebrate biomass including the shrimps by 196%. The biomass-specic NH 4+ excretion rate by shrimp was similar to aquatic insects, and the calculated mean NH 4+ excretion by benthic macroinvertebrate assemblage was 144% higher in the presence of shrimps. On the other hand, shrimps excreted much less PO 43− than aquatic insects, and the PO 43− excretion by the benthic macroinvertebrate assemblage did not change by the presence of shrimps. The eld survey showed a positive correlation between NO 3 − concentration in stream water and shrimp density, suggesting that the excess NH 4+ was nitried and raised NO 3− concentration. In contrast, PO 43− concentration in stream water was negatively associated with the shrimp density, indicating that the dominance of shrimps that excrete little PO 43− decreased the PO 43− concentration. While nutrient concentration of stream water is often attributed to the condition of the watershed area, the results of this study indicate downstream connectivity to the ocean can also inuence the nutrient dynamics of the stream through the density of amphidromous shrimps.


Introduction
The number of animals can greatly vary by habitats or by years for various reasons such as habitat connectivity (Hanski & Kuussaari 1995), diseases (Scheibling 1986), and natural catastrophes (Guha-Sapir et. al. 2004). Many studies have shown that the density of animals can directly in uence other organisms through species interactions such as direct consumption (Paine 1980;Carpenter 1985) and resource competition (Tilman 1977). Furthermore, studies in lentic ecosystems have shown that animals also in uence nutrient cycling through their excretion of waste products (Kitchell et al. 1979;Vanni 2002).
On the other hand, in lotic ecosystems, the chemical aspects of consumer effects have been understudied because the water ow makes the evaluation of the effects of animal excretion in lotic ecosystems challenging. Yet, some careful studies have shown that patchy distribution of shes can create spatial heterogeneity in nutrient concentration in stream water (McIntyre et al. 2008), and the presence of a speci c consumer species can alter nutrient cycling in streams . In reality, consumers in uence the ecosystem structures both through species interactions and through nutrient addition by excretion, and the combined evaluations of the biochemical in uence of consumers are needed to truly understand the in uence of consumers on the stream ecosystems (Vanni & Layne 1997).
In intact coastal streams, a large proportion of aquatic organisms exhibit migration between streams and the ocean or lakes (Myers 1949). For obligate migratory organisms, disruptions of the migration routes limit their distribution in streams. Therefore, large variation in consumer densities exists among streams depending on the hydrological connectivity of river networks. Many studies have shown that salmon-run from the ocean can have signi cant direct and indirect effects on stream ecosystems (Gende et al. 2002). Childress et al. (2015) have shown that migration of iteroparous shes can also provide nutrients to stream ecosystems through nutrient inputs by excretion, decomposed unfertilized eggs etc, and increase the productivity in streams. While such anadromous shes that come to streams for spawning transport nutrients from the ocean or lakes to streams (Material subsidy), many other migratory organisms that exhibit amphidromous or catadromous migration rather stay in streams for a prolonged period (McDowall, 1988) and should in uence the local stream ecosystem as a newly added consumer species (Process subsidy) (Flecker et. al., 2010). Yet, studies evaluating the effect of process subsidies are limited compared to the material subsidies (Flecker et. al., 2010).
The presence of a migratory consumer as the process subsidy would not only in uence the resident consumer species through species interactions but also in uence the nutrient cycling through their excretion Flecker et. al., 2010). The addition of a migratory consumer would likely reduce the biomass of other resident consumers through species interactions. Yet, if the added species utilize resources more effectively than others do, the total biomass of consumers would be increased by the addition of the new species (Tilman 1977). Furthermore, mass-speci c excretion rates vary by consumer species (McManamay et al. 2011), and change in dominant species would in uence the nutrient cycling (Evans-White et al. 2005;McIntyre et al. 2007). Therefore, to understand the chemical in uences of a migratory consumer as the process subsidy, we need to know the biological impacts of the consumer species on the whole assemblage as well as the excretion rate by the composing species.
In many coastal streams in the mid-low latitude region, amphidromous shrimps often dominate the headwater streams (Covich 1988;Mantel & Dudgeon 2004;Cross et al. 2008). Because of their migratory life cycles, where their larvae exhibit planktonic form in estuaries (McDowall, 1988), their densities can signi cantly vary among streams depending on the physical accessibility from the ocean (Greathouse et al. 2006). Benstead et al. (2010) have estimated the total nutrient excretion by amphidromous shrimps in tropical streams and concluded that shrimp excretion should play a signi cantly large role in nitrogen cycling in tropical streams. On the other hand, a number of studies have shown that the presence of amphidromous shrimps can suppress the aquatic insect biomass through direct consumption or resource competition (Pringle et al. 1993;Greathouse et al. 2006). While a signi cant amount of nutrient excretion by shrimps is added by their presence, the nutrient excretion by aquatic insects may be reduced by the reduction in their biomass. The effect of shrimps may also differ by nutrient elements because previous excretion studies have shown that crustaceans including shrimps excrete very little phosphorous compared to aquatic insects while their excretion rate of nitrogen is equivalent to aquatic insects (McManamay et al. 2011).
Here we aim to evaluate the biochemical in uence of amphidromous shrimps on stream ecosystems. First, we conducted a eld manipulative experiment to examine how the presence of shrimps in uence benthic communities. Then to examine the change in nutrient cycling by the shrimps, we calculated nutrient output from the whole benthic macroinvertebrate assemblages in the presence/absence of shrimps by excretion measurement of major aquatic insect taxa and shrimp. Finally, we conducted a broad eld survey of streams with various shrimp densities to examine how the difference in nutrient export from benthic macroinvertebrates induced by the shrimp densities in uences the material cycling in streams. As a whole, we discern the biochemical in uences of the amphidromous shrimps on stream ecosystems, by a combination of a eld manipulative experiment, excretion measurement of consumers, and eld surveys.

Study system
This study was conducted in the southern part of Wakayama Prefecture, Honshu, Japan (Fig. 1). Wakayama Prefecture is located in Kii Peninsula that faces the North Paci c Ocean. In this area, the climate on aquatic ecosystems is largely in uenced by Kuroshio that is the warm current, and provides the habitats of tropical-subtropical diadromous organisms such as a giant mottled eel Anguilla marmorata and some species of atyid shrimps (Saito et al. 2012;Nagasawa et al. 2020). In these streams, various amphidromous shes and invertebrates including shrimps inhabit where the migration route from the ocean is sustained (Tanaka et al. 2020), and benthic macroinvertebrate assemblages consist of diverse aquatic insects and amphidromous shrimps.

Field manipulative experiment
To investigate the biological effect of shrimps on the benthic communities directly, we carried out a eld manipulative experiment in the Takase-river. By excluding shrimps from certain areas of the streams, we examined the change in the aquatic insect assemblages as well as their potential food resources to examine the processes of the species interactions between the shrimps and the aquatic insect assemblages. Within a 500 m segment of the river, we set four blocks in distinct pools, and within each block, we set four experimental cages. Each cage consists of ~ 10 L of pebbles lled in a 30 cm by 50 cm 8 cm deep tray with 2 cm mesh, and an electric fence (Pringle & Blake 1994;Moerke et al. 2017; Fig. S1), which selectively remove large animals including shrimps was placed on top. Over the experimental period, the electric fences of two cages were electri ed, and the other two were not electri ed. Three 3 cm by 3 cm tiles were tied on each cage to examine the chlorophyll-a and benthic ne particulate organic matter (BFPOM). Two binds of major riparian tree leaves, Mallotus japonicus and Quercus serrata, were tied to each cage to examine the decomposition rate of leaves during the experiment. The eld manipulative experiment was initiated on March 17, 2020, and retrieved on April 21, 2020.
The number of shrimps on each cage was counted in daytime and nighttime in the rst week of the experiment. At retrieval of the experiment, all aquatic insects in each cage were collected to estimate their densities. Aquatic insects were sorted at the family level, then the total dry weight of all aquatic insects in each chamber was measured nally. The tiles were collected, and loose organic materials on the tiles were ushed with water onto 70 µm mesh as BFPOM, and their dry weight was measured at the lab. Periphyton on each tile was brush washed with < 100ml stream water and collected into a bottle, then ltered onto a GF/F lter (Whatman, Maidstone, UK). Chlorophyll-a was extracted with DMF (Dimethylformamide) and spectro-uorometry (RF-5300PC, Shimadzu, Japan) was used to estimate the mean chlorophyll-a per area on each sampled area following the methods described by Suzuki and Ishimaru (1990). The bound leaves were dried and weighed after the experiment, and their decomposition rate was calculated.

Measurement of consumer excretion
To calculate the whole nutrient excretion by benthic macroinvertebrates in the presence/absence of shrimp, we measured the excretion rate of major aquatic insect taxa and shrimp following the methods used by Benstead et al. (2010) and McManamay et al. (2011). The excretion measurement was carried out at night time (20:00-25:00) on July 23, 2020, and daytime (8:00-12:00) on July 24, 2020, in Takaseriver. Five individuals of major shrimp and aquatic insect taxa were collected with a dip net respectively. Each individual was held less than 15 minutes before the start of the excretion experiment. Incubations were carried out by introducing each shrimp in 100ml and each insect in 30ml of stream water ltered with GF/F lter. As chironomid midges were small, ve individuals were incubated together in 30ml of water. Each individual was incubated for 50 to 70 minutes, and we con rmed all individuals were behaving normally through the incubations. Then at the end of the experiment specimens were frozen in individual bottles, and each incubated water was ltered with a 0.45µm membrane lter (25CS045AN, Toyo Roshi Kaisha, Japan) and frozen until analysis. The dry weight of each specimen was measured. All the water samples were frozen in the eld and transported to the lab facility at the Center for Ecological Research, Kyoto University. Inorganic nitrogen (NO 3 − , NO 2 − , and NH 4 + ) and phosphorous (PO 4 3− ) concentrations in the water samples were measured colorimetrically using an autoanalyzer (QuAAtro 2-HR, BLTEC, Japan). Because the NO 2 − concentration was immeasurably low, we show the sum of NO 3 − and NO 2 − concentration as NO 3 − concentration in the present study.

Field survey
To examine if biochemical impacts of shrimps found in the eld manipulative experiment and the measurement of consumer excretion can explain (extrapolate) the natural variation of water nutrient concentration among streams, we conducted a biochemical eld survey at sites with a wide range of shrimp density. The eld survey was conducted at 13 sites in four watersheds, the Tonda River, the Takase River, the Hiki River, and the Koza River (Fig. 1). The Tonda River and the Takase River meet at their river mouth and share the estuary. All sites were located in distinct tributaries of the rivers, and the watershed area of the study sites ranged from 0.8km 2 to 10.4km 2 (Table S1). Most of the watershed area was covered by natural broadleaf forest or arti cial coniferous forest. The geology of the study watersheds is dominated by the Cenozoic accretionary complexes composed of sandstone, mudstone, and shale (Takao et al. 1981). No study sites had dams upstream, but ve sites had large dams downstream (Tonoyama-dam 64-meter height, and Shichikawa-dam 59-meter height). All samplings were conducted in November 2017 within 5 days, and there was no precipitation during the sampling period. At each site, samplings were repeated twice in day time and night time on the same day because the shrimps were strictly nocturnal in the system. Daytime sampling was conducted between one hour after the sunrise and one hour before the sunset, and nighttime sampling was conducted between one hour after the sunset and one hour before the sunrise.
Shrimp density was estimated with 50cm ×50cm quadrat for ve times at random locations in each site at night. Aquatic insect assemblages were sampled with 30cm×30cm Surber-net sampler three times at randomly selected locations in each site and time (day time and night time) to estimate the density.
Sampled aquatic insects were preserved in 99% ethanol on-site, and > 0.5mm individuals were sorted to family level under compound microscope at lab. The total dry weight of all aquatic insects collected at each site was measured nally. Algae were sampled from three cobbles at each site: 36cm 2 of the cobble surface was scrubbed with a toothbrush to collect algae, then chlorophyll-a density was estimated as described above. 100ml water sample was collected at each site, ltered with GF/F within 24 hours. Then the ltered water samples were frozen until analysis. Inorganic nitrogen (NO 3 − , NO 2 − , and NH 4 + ) and phosphorous (PO 4 3− ) concentrations in the water samples were analyzed as described above.

Data analysis
Field manipulative experiment-The eld manipulative experiment enabled us to compare the effect of shrimps on major aquatic insect taxa and their potential food sources between shrimp and no shrimp treatments. We assessed the differences of their abundance between shrimp and no shrimp treatment using a one-way ANOVA including block as random factor by Linear Mixed Model using the lmer() function in lme4 package and the anova() function in the lmerTest package in Program R. We logtransformed aquatic insect density to satisfy assumptions of normality prior to the analysis.
Calculation of total excretion by benthic macroinvertebrates assemblages-Total nutrient excretion by benthic macroinvertebrates in the presence/absence of shrimps were calculated based on the results of the electroshock experiment and the excretion measurement of major aquatic insect taxa and shrimp. For each nutrient element, we multiplied the taxon-speci c excretion rate by the density of each taxon in the presence/absence of the shrimps in the experiment and summed up the excretion rates by all taxa. For minor taxa (< 5% in total) whose taxon-speci c excretion rates were not measured, we applied the excretion rate of Diptera as they were mostly as small as Diptera. Similarly, we calculated the total dry weight of benthic macroinvertebrates in the presence/absence of the shrimps.
Field survey-We compared the effect of shrimps and several watershed characters on water chemistry and major aquatic insect density across surveyed sites. As watershed characters, we included elevation of the sampling site, drainage area, forest rate, natural forest rate, shrimp presence, and shrimp density measured at each site (Table S1). Watershed characters were estimated using the ArcGIS system, and forest rate and natural forest rate of each watershed were estimated based on the 1:25,000 vegetation map provided by the Biodiversity Center of Japan, Ministry of the environment, Japan (http://gis.biodic.go.jp/webgis/index.html). For the analysis, we de ned the natural forest as the sum of primary and secondary broadleaf forest, and de ned the forest as the sum of natural forest and the arti cial plantation conifer forests. We rst checked the collinearity among all variables included in each analysis using correlation coe cients and variance in ation factors (VIF). A VIF score greater than 4 and a correlation coe cient greater than 0.7 were used to eliminate habitat variables considered to have a high degree of collinearity (Zuur et al. 2010). There was a high degree of collinearity and VIF scores between the shrimp presence and the elevation and between the shrimp presence and shrimp density (Table S2). Therefore, we excluded the shrimp presence from the analyses. We log-transformed shrimp and aquatic insect density to satisfy assumptions of normality prior to all the analysis.
We generated a list of linear mixed-effects models with the stream as a random effect and all combinations of two or fewer predictor variables as xed effects without interactions. We standardized the independent data to a mean of 0 and standard deviation of 2 so that effect sizes of independent variables could be compared. Models with ΔAICc < 4 were retained to form candidate model sets and were averaged using the MuMIn package in R (Table S2). To evaluate the effect of shrimps and watershed character variables on water chemistry and aquatic insect density, we considered the magnitude and direction of the averaged coe cient, whether the 95% con dence intervals spanned zero, and the relative variable importance (RVI) of each variable. The latter is calculated as the sum of the model weights of all the models in the nal con dence set in which the variable appears (Burnham and Anderson, 2002). All analyses were performed in the software R.
The calculated dry weight of shrimp was 1867mg m − 2 . On the other hand, aquatic insect biomass was 909 mg m − 2 in the absence of shrimps, and 823 mg m − 2 in the presence of shrimps. Therefore, the total macroinvertebrate biomass was elevated were elevated from 909 mg m − 2 to 2690 mg m − 2 and about three times as high in the presence of shrimps (Fig. 3A). The calculated total NH 4 + excretion rate by shrimps was 14.4 µmol hr − 1 m − 2 . While total NH 4 + excretion by benthic macroinvertebrates except shrimps was slightly decreased when shrimps were present (9.7 µmol hr − 1 m − 2 ) compared to when shrimps were absent (9.9 µmol hr − 1 m − 2 ), the total NH 4 + excretion rate by benthic macroinvertebrates including shrimps were elevated from 9.9 µmol hr 1 m − 2 to 24.2 µmol hr − 1 m − 2 and about 2.4 times as high in the presence of shrimps (Fig. 3B). The calculated total PO 4 3− excretion rate by shrimps were 0.18 µmol hr − 1 m − 2 . While total PO 4 3− excretion by benthic macroinvertebrates except shrimps was slightly decreased when shrimps were present (1.28 µmol hr − 1 m − 2 ) compared to when shrimps were absent (1.40 µmol hr − 1 m − 2 ), the total NH 4 + excretion rate by benthic macroinvertebrates in absence of shrimp 1.40 µmol hr − 1 m − 2 was almost the same as the total NH 4 + excretion in presence of shrimps 1.45 µmol hr − 1 m − 2 (Fig. 3C).

Field survey
Amphidromous shrimp, Paratya compressa, Caridina multidentate, Palaemon paucidens, and Macrobrachium Formosense were observed in the study. Many shrimps occurred at night time, and less than 5% of shrimps were observed in day time, indicating they are nocturnal. Shrimps were observed at all the sites without dams, but no shrimps were observed upstream of dams. Model selections to explain the nutrient concentrations showed not only the physical characters such as drainage area of the watershed but also shrimp density can in uence the water chemistry (Fig. 4,5, Table S4). Among nutrients analyzed in this study, NH 4 + concentration was not signi cantly in uenced by shrimp density (Fig. 4A,   5A), while NO 3 − concentration signi cantly increased with shrimp density (Fig. 4B, 5B). PO 4 3− concentration was signi cantly decreased with shrimp density, and also decreased with drainage area and elevation (Fig. 4C, 5C). Chlorophyll-a density as well as the total aquatic insect biomass increased with drainage area (Fig. 4D, 5D), while the trend of aquatic insect biomass was not statistically signi cant (Fig. 4E, 5E). Among all aquatic insects, Heptageniidae, Chironomidae, Baetidae, Perlidae, and Leptophlebiidae were the ve most abundant taxa. Their densities were strongly in uenced by the drainage area, and the density of Heptageniidae and Perlidae were signi cantly higher and the density of Leptophlebiidae was signi cantly lower with increased drainage area (Fig. 4F, I, J, 5F, I, J). Perlidae density was signi cantly higher with increased elevation (Fig. 4I, 5I). Generally, the densities of the ve dominant aquatic insects were lower with increased density of shrimps but that was not statistically signi cant (Fig. 4F-J, 5F-J).

Discussion
The present study shows that the density of amphidromous shrimps not only in uence stream biota through species interactions but also in uence stream nutrient cycling through their excretion. While the direct evaluation of the animal excretion on stream nutrient cycling in stream is challenging due to the ow, by a combination of the eld manipulative experiment and eld survey, we showed the mechanisms as well as the natural variation of nutrient concentrations that amphidromous shrimps induced. This study represents one of the limited studies that carefully evaluated both the biological and chemical in uence of process subsidies (Flecker et al. 2010), and the results indicate the potentially large impacts of the underappreciated amphidromous organisms on the stream ecosystem.
The presence of shrimps as strong competitors slightly decreased the biomass of aquatic insects but increased the total biomass of benthic macroinvertebrates in the eld manipulative experiment. This result meets the theory by Tilman et al. (1997), which predicts the increase in total biomass of assemblages when a more e cient species is added. The reduction in BFPOM indicate shrimps were more effective consumer of BFPOM (Pringle & Blake 1994) than other aquatic insects such as Chironomidae, which are also known as detritus feeder (Merritt et al. 2019). It is assumed that shrimp outcompeted aquatic insects such as Chironomidae that share the resource, and more effectively utilized the resource as the whole benthic macroinvertebrate assemblage. Furthermore, increased Chlorophyll-a density in the presence of shrimps in the experiment probably occurred due to the removal of BFPOM by shrimps that cover and shade periphyton in streams (Power 1990;Kupferberg 1997). The increased primary production of periphyton by the removal of BFPOM would explain the biomass compensation of periphyton feeders such as Baetidae (Merritt et al. 2019) in the presence of shrimps. In addition, the longer lifespan of the shrimps (> 2 years; Kawai & Kazuyoshi 2011) compared to the competing aquatic insect species such as Chironomidae (a few months to one year; Kawai & Tanida 2018) contributed to the high biomass of shrimps in contrast to the little reduction of aquatic insect biomass, and increased the total benthic macroinvertebrate biomass.
Excretion measurement of the major aquatic insect taxa and shrimp and the estimation of the total excretion rate by the benthic macroinvertebrate assemblages in the presence/absence of shrimps showed chemical in uence of shrimps. The mass-speci c NH 4 + excretion of shrimp was in the range of variation among aquatic insects as previously documented (Vanni et al. 2017), and increased total biomass of the benthic macroinvertebrates increased total NH 4 + excretion in presence of shrimps. The total NH 4 + excretion by benthos was 24.2 μmol m -2 h -1 in the presence of shrimps, and that was 2.4 times as much as their absence. The contribution of these nutrient excretions by the benthic macroinvertebrates to the ecosystem-level nutrient cycling is not directly evaluated in this study, as it also depends on nutrient uptake rates that were not measured in this study. However, Benstead et al. (2010) have estimated that the same level of nutrient excretion by shrimps (22.3 μmol NH 4 + ) represented 21% of NH 4 + uptake in a tropical stream. Therefore, it is assumed that the elevated NH 4 + excretion by the benthic macroinvertebrates in the presence of shrimps observed in this study also contributed to the whole stream nitrogen cycling.
In the spatial survey, variation in the shrimp density was not re ected in NH 4 + concentration but re ected in NO 3 concentration. NO 3 concentration in streams with high shrimp density was on average twice as high as those without shrimp. While excretion of nitrogen from benthos was in the form of NH 4 + -N, it is assumed that NH 4 + -N from the benthic macroinvertebrate assemblages was nitri ed to NO 3 --N in streams. (Newbold et al. 1983;Day and Hall 2017). This is a contrast to most studies that measured an increase in nutrient level from sh excretion, which documented increased NH 4 + but not NO 3 - McIntyre et al. 2008). We believe that the difference comes from the habitats of the shrimps and shes that have been studied for excretion. Most excretion studies in stream ecosystems to date have been conducted with anadromous salmons or suckers that migrate from the ocean or lakes to downstream reaches of streams for spawning. Therefore, there were not long enough distances downstream of spawning ground for nitri cation. In contrast, amphidromous shrimps we studied migrate up to headwater areas of streams, and the stream water continuously receives shrimp excretion from headwater to the sampling sites over long distances. Furthermore, habitat of the shrimps, the benthic zone of headwater streams, is where nitri cation occurs most actively in streams (Ensign and Doyle 2006;Hall et al. 2013). When NH 4 + concentration is elevated, nitri cation should occur actively in bio lms in oxic sediments in mountainous streams like where this study was conducted (Butturini et al. 1999). Reduction of BFPOM as a result of shrimp activity as observed in the electroshock experiment may have also contributed to activate the nitri cation by oxidizing the sediment (Stief 2013).
In contrast, shrimp excreted much less PO 4 3compared to aquatic insects, and the total PO 4 3excretion by benthic macroinvertebrate assemblages were similar regardless of the presence of shrimps. As crustaceans require a large amount of phosphorous for their carapace (Vrede et al. 1999), crustaceans generally excrete low PO 4 3compared to other aquatic consumers such as aquatic insects and snails (Evans-White et al. 2005;McManamay et al. 2011). Therefore, the dominance of shrimps in benthic macroinvertebrate assemblages would decrease the phosphorous excretion by the benthic macroinvertebrate assemblages. In this study, total PO 4 3excretion by benthic macroinvertebrate assemblages were similar in the presence/absence of shrimps, because the reduction in aquatic insect density was limited. However, the strength of species interaction is condition-dependent (Menge et al.

2004)
, and slightly stronger dominance of shrimps and stronger suppression of aquatic insects compared to this study, which would lead to the reduction of total phosphorous excretion by the benthic macroinvertebrates, could occur under different conditions. The result of the spatial survey showed a negative correlation of shrimp density on PO 4 3concentration in stream water, while PO 4 3concentration had a stronger negative correlation with the drainage area and the elevation of the sampling site. Lower PO 4 3concentration in high elevation indicates that the phosphorous export from the drainage area is lower due to the low weathering rate of the bedrock (Hartmann et. al. 2011;Lintern et al. 2018). Lower PO 4 3concentration with increased drainage area is explained by more active photosynthesis by algae downstream (Finlay et al. 2011). In addition, the dominance of shrimps that excrete little phosphorous could have resulted in the reduction in the PO 4 3concentration in stream water.
Nutrient in stream water is often considered as output from the terrestrial ecosystems (Brookshire et al. 2009), and the variation in the nutrient level is attributed to environmental condition of the watershed including geology (Holloway et al. 1998) and forest type or age (Likens et al. 1970;Wakamatsu et al. 2006). The result of this study shows that not only the upstream watershed conditions but also the downstream connectivity to the ocean can also in uence the nutrient in stream water through the excretion of migratory organisms. It is particularly surprising in such mountainous headwater streams far from the ocean as studied in this research. Greathouse et al. (2006) has shown upstream effects of dams by extirpation of amphidromous organisms. Though they only found the reduction in aquatic insects through species interactions, the result of this study shows the upstream effect of the dams can be extended to nutrient cycling.
Loss of habitat connectivity often leads to the decline in animal diversity (Fahrig et al. 2003;Cardinale et al. 2012). The result of this study shows the loss of a migratory species can lead to change not only in community structures but also in nutrient cycling. While the ecological consequences of the loss in large