1

Diet composition affects the rate and N:P ratio of fish excretion

2

Eric K. Moody1*, Jessica R. Corman1, James J. Elser1 and John L. Sabo1,2

3

1

4

2

5

*Corresponding author:

6

Eric Moody

7

Arizona State University, School of Life Sciences, P.O. Box 4601, Tempe, AZ, USA, 85287

8

ekmoody@asu.edu

Arizona State University, School of Life Sciences, Tempe, AZ, USA
Arizona State University, Global Institute of Sustainability, Tempe, AZ, USA

9
10

Running title: Diet affects fish excretion

11

Keywords: nutrient recycling, ecological stoichiometry, food quality, diet manipulation,

12

assimilation efficiency

13

14
15

SUMMARY
1. Nutrient recycling by fish can be an important part of nutrient cycles in both freshwater

16

and marine ecosystems. As a result, understanding the mechanisms that influence

17

excretion elemental ratios of fish is of great importance to a complete understanding of

18

aquatic nutrient cycles. As fish consume a wide range of diets that differ in elemental

19

composition, stoichiometric theory can inform predictions about dietary effects on

20

excretion ratios.

21

2. We conducted a meta-analysis to test the effects of diet elemental composition on

22

consumption and nutrient excretion by fish. We examined the relationship between

23

consumption rate and diet N:P across all laboratory studies and calculated effect sizes for

24

each excretion metric to test for significant effects.

25

3. Consumption rate of N, but not P, was significantly negatively affected by diet N:P.

26

Effect sizes of diet elemental composition on consumption-specific excretion N, P and

27

N:P in laboratory studies were all significantly different from 0, but effect size for raw

28

excretion N:P was not significantly different from zero in laboratory or field surveys.

29

4. Our results highlight the importance of having a mechanistic understanding of the drivers

30

of consumer excretion rates and ratios. We suggest that more research is needed on how

31

consumption and assimilation efficiency vary with N:P and in natural ecosystems in order

32

to further understand mechanistic processes in consumer-driven nutrient recycling.

33

34
35

Introduction
Consumers can play an essential role in nutrient cycles in marine and freshwater

36

ecosystems by controlling the storage and fluxes of key nutrients such as nitrogen (N) and

37

phosphorus (P) (Kitchell et al., 1979; Elser et al., 1988; Vanni, 2002; McIntyre et al., 2007;

38

Allgeier, Yeager & Layman, 2013). Through the excretion of dissolved inorganic nutrients,

39

consumers can supply significant amounts of limiting nutrients to primary producers and

40

decomposers (McIntyre et al., 2008; Small et al., 2011). While a considerable body of literature

41

has developed around investigations of the importance of consumers to nutrient cycles in aquatic

42

ecosystems, a mechanistic understanding of what influences rates and elemental ratios of

43

nutrients excreted by consumers has lagged behind. Consumers can create biogeochemical

44

hotspots simply by achieving locally high biomass (McIntyre et al., 2008; Atkinson et al., 2013;

45

Capps & Flecker, 2013a), but the digestion, metabolism, storage and retention of consumed

46

nutrients in consumer bodies, in combination with overall biomass, control the role individual

47

species play in altering ecosystem function (Vanni et al., 2002; Small et al., 2011; Capps &

48

Flecker, 2013b; Vanni, Boros & McIntyre, 2013). As a result, both the elemental composition of

49

an organism and its diet should impact the rates and ratio at which it excretes nutrients (Sterner,

50

1990; Elser & Urabe, 1999; Sterner & Elser, 2002). While the effect of organismal elemental

51

composition on nutrient recycling by aquatic vertebrates has been investigated (e.g., Vanni et al.,

52

2002; Hood, Vanni & Flecker, 2005), empirical studies of the impacts of diet elemental

53

composition on excretion ratios have provided mixed results. The positive relationship between

54

diet N:P and excreted N:P predicted by Sterner (1990) has been found in Daphnia, crayfish and

55

mottled sculpin (Cottus bairdi) (He & Wang, 2008; McManamay et al., 2011), but no significant

56

relationship has been found for a number of other species of fish and invertebrates (Schindler &

57

Eby, 1997; Verant et al., 2007; McManamay et al., 2011; Taylor et al., 2012). We investigate the

58

impacts of diet on consumer excretion ratios in fish, a group of consumers that is both abundant

59

in aquatic ecosystems and exhibits a great diversity of dietary strategies over which to examine

60

excretion responses.

61

Fish are both abundant and diverse in many aquatic ecosystems, and as a result they have

62

been frequently identified as the most important nutrient recyclers or retainers in a diverse range

63

of aquatic systems (e.g., McIntyre et al., 2007; Small et al., 2011; Allgeier et al., 2013; Capps &

64

Flecker, 2013b). Fish are diverse taxonomically as well as functionally, with known diets

65

ranging widely in elemental composition from plant and algal detritus to invertebrates and other

66

vertebrates (González-Bergonzoni et al., 2012). While some fish species are highly specialized

67

to feed on specific foods, many fish are omnivorous to some degree and thus may consume diets

68

that vary widely in quality over time or ontogeny (e.g., Grimm, 1988; Pilati & Vanni, 2007;

69

González-Bergonzoni et al., 2012). Diets that are animal-based are generally relatively higher in

70

P content than plant- or algae-based diets (e.g., Green, Hardy & Brannon, 2002), thus the

71

impacts of animal- vs. plant- or algae-based diets on organismal physiology are informed by the

72

mass balance of multiple chemical elements and energy in ecological systems employed by

73

ecological stoichiometry (Sterner & Elser, 2002). Following a mass balance model of fish

74

growth assuming no difference in growth rate between diets, the difference between the amount

75

of a given nutrient in the diet and that used by the consumer will equal the total released, which

76

includes both nutrients excreted as dissolved inorganic and organic molecules and those egested

77

as particulate waste (Kitchell et al., 1974; Sterner, 1990; Schindler & Eby, 1997; Fig. 1).

78

Therefore, fish excretion ratios should be proportional to diet elemental composition across a

79

gradient of food elemental ratios unless fish differentially assimilate N and P (Sterner, 1990;

80

Schindler & Eby, 1997; Sterner & Elser, 2002). However, if fish differentially excrete and egest

81

waste products, these ratios may not be directly proportional. Such a scenario arises when

82

assimilation efficiency changes with diets of varying composition.

83

To assess how diet composition affects fish excretion ratios, direct manipulations of

84

organismal diets in a controlled setting are required. Here we review the literature for studies in

85

which multiple diets were fed to fish in a controlled setting and consumption rates and excretion

86

rates and/or ratios were measured. Specifically, we draw on the field of experimental aquaculture

87

research which represents a rich source of data on physiological responses of consumers to

88

differing diets, the value of which is only beginning to be recognized by ecologists (Boersma &

89

Elser, 2006; Benstead et al., in press). We employ a meta-analysis using standardized effect sizes

90

to quantify how both consumption and composition of diet may affect excretion ratio in fish.

91

Finally, we discuss the implications of the results from controlled settings to nutrient recycling in

92

natural ecosystems.

93

Methods

94

We used a meta-analytic approach to determine if fish consumption rates and nutrient

95

excretion ratios are influenced by the N and P composition of their diet. We used the ISI Web of

96

Science database to search the peer-reviewed literature for studies of fish where diet was directly

97

manipulated and a dissolved excretion response was measured. While faecal egestion is

98

undoubtedly important in the N and P budgets of organisms (Fig. 1; Halvorson et al., 2015), we

99

focus on dissolved excretion because it is in this form that excreted nutrients can have significant

100

immediate impacts at the ecosystem scale (e.g., Kitchell et al., 1979; McIntyre et al., 2008;

101

Small et al., 2011). We included studies that measured mass-specific excretion as a rate and

102

those that reported it only as a loading per unit of fish biomass. We performed this search using

103

the terms fish, diet and excretion. Our search included articles published between 1970 and 2013.

104

This search initially returned >600 articles, which were cursorily examined by title to determine

105

whether they were likely relevant to the meta-analysis; for example, articles discussing only

106

modeled excretion and growth or the use of fishmeal as a feed for other animals were

107

disregarded. We identified 74 articles that appeared to be relevant by suggesting some type of

108

study of fish N and P excretion among different diets which were then searched in greater detail

109

to determine whether they met our criteria of inclusion. Studies included in the meta-analysis

110

were those that were conducted on fish from a single population, included multiple diets that

111

were directly manipulated or measured over natural gradients, measured N and P composition

112

and fish consumption rates of those diets and measured N and/or P excretion in some form. In

113

the few instances where our search returned multiple studies of a single species by the same

114

research group, we selected only one of them with a random number generator to avoid violating

115

test assumptions of independence. We categorized studies as those with direct diet manipulations

116

in laboratory settings and those that measured natural variation diets in field settings and also

117

noted whether dietary P was manipulated by varying the level of organic or inorganic P. We

118

found no studies that conducted diet manipulation experiments in a natural setting.

119

As raw excretion rates may be influenced by differences in diet elemental composition

120

and changes in consumption rate caused by diet differences, we used linear models of mass-

121

specific consumption rate (g * g fish-1 * day-1) of N, P and total food consumption predicted by

122

diet N:P to calculate and test for significance of effect sizes. From these models, we calculated

123

effect size as the Pearson correlation coefficient r, which we transformed to Z-scores using the

124

Fisher transformation (Rosenthal & DiMatteo, 2001). Then, we tested whether mean effect sizes

125

differed from 0 using t-tests with Bonferroni corrections to adjust α when performing multiple

126

comparisons with the same dependent variable (Rice, 1989; Rosenthal & DiMatteo, 2001). We

127

then calculated consumption-specific excretion measurements for each study by dividing N, P

128

and N:P excreted by the mass-specific consumption rate when feeding on a given diet and used

129

the above methods to calculate effect sizes for both consumption-specific and raw N, P and N:P

130

excreted as a response to diet N:P in diet manipulation studies. Field surveys did not measure

131

consumption rates and some did not report N and P excretion data individually, thus we could

132

not calculate consumption-specific and single nutrient excretion effect sizes for those studies.

133

To assess whether effect sizes may have been influenced by other factors aside from diet

134

composition, we tested study heterogeneity in the effect size measures. First, we used Cochran’s

135

Q to test for significance of study heterogeneity for each effect size measure. Cochran’s Q

136

follows a χ2 distribution and is a widely used and relatively conservative test of study

137

heterogeneity in meta-analyses (Takkouche, Cadarso-Suarez & Spiegelman, 1999). For those

138

effect sizes with significant heterogeneity, we fit linear regression models for each effect size

139

measurement as a response to difference in N:P between the diet end-members, average water

140

temperature during the experimental period, initial fish mass and experimental duration

141

(Rosenthal & DiMatteo, 2001). Our sample size was not sufficient to estimate the interaction

142

terms between all of these variables thus we examined only main effects. We assessed

143

homoscedasticity and normality of residuals visually for each model with a plot of model

144

residuals vs. fitted values and a normal probability plot, respectively. We could not construct

145

linear regression models for field studies due to a lack of data presented in those manuscripts and

146

small sample size. All analyses were conducted in the software R v2.15 (R Core Team, 2013).

147

Results

148

Of the 74 candidate papers identified as possibly relevant, we found 19 independent

149

studies that met our criteria for inclusion in the meta-analysis (Table 1). Of these, two studies

150

featured only two experimental diets; these studies were excluded from the meta-analysis

151

because effect sizes could not be calculated from two data points. Of the remaining 17 studies,

152

15 were diet manipulation experiments conducted in controlled laboratory facilities and two were

153

field surveys conducted over natural gradients of diet elemental composition. Of the diet

154

manipulations, 12 studies manipulated the levels of animal vs. plant-based protein while three

155

studies directly manipulated dietary P content by adding phosphate salts to the same base diet;

156

however these three studies did not measure N excretion. The majority of laboratory studies fed

157

fish to apparent satiation, although several fed fish at specific levels based on fish body mass

158

(Ballestrazzi et al., 1994; Green et al., 2002; Sumagaysay-Chavoso, 2003; Yang et al., 2011).

159

The laboratory studies included involved 10 fish species in seven families while the field studies

160

included involved seven fish species in seven families (Table 1). Resource N:P ratios (by mass)

161

ranged from 2.5 to 56 in laboratory studies (mean=8.2, SD=8.3) and from 2.4 to 174 in field

162

studies (mean=44.7, SD=42.4). All field studies measured excretion N:P, but only 12 of 15

163

laboratory studies presented N excretion data that allowed us to calculate N:P ratios of excretion.

164

Additionally, all laboratory studies measured average initial fish mass, the average water

165

temperature and the length of the experimental period between when the diet switch began and

166

when excretion was measured.

167

We first examined whether consumption rates differed with diet composition. Total mass-

168

specific consumption was not significantly affected by diet N:P (two-tailed t-test, t=-1.796, ν=11,

169

P=0.10). Mass-specific consumption rate of N was also unaffected by diet N:P (two-tailed t-test,

170

t=-0.270, ν=11, P=0.480) across studies but mass-specific P consumption rate significantly

171

decreased with increasing diet N:P (two-tailed t-test, t=-3.650, ν=11, P=0.004) (Fig. 2).

172

Diet effects on excretion ratios were similar for laboratory and field studies; however we

173

had fewer results for field studies due to the lack of consumption and separated N and P

174

excretion data. For diet manipulation studies, effect size of diet N:P was significantly below 0 for

175

P excretion (two-tailed t-test, t=-2.606, ν=14, P=0.021), and positive, but not significantly

176

different from 0 for N excretion (two-tailed t-test, t=1.381, ν=11, P=0.195) (Fig. 3). However,

177

effect sizes for consumption-specific excretion of both P (two-tailed t-test, t=-2.244, ν=14,

178

P=0.042) and N (two-tailed t-test, t=2.915, ν=11, P=0.014) were significantly different from 0

179

(Fig. 3). Mean effect size of diet elemental composition on excretion N:P was not significantly

180

different from 0 in diet manipulation studies (two-tailed t-test, t=2.00, ν=11, P=0.071) nor field

181

surveys (two-tailed t-test, t=-0.002, ν=6, P=0.999), but was significantly different from 0 when

182

corrected for consumption in diet manipulations (two-tailed t-test, t=2.42, ν=11, P=0.034) (Fig.

183

4). Of all excretion response effect sizes in diet manipulation studies, only raw P excretion

184

exhibited significant heterogeneity (Q=23.82, ν=11, P=0.014). However, this heterogeneity was

185

not significantly related to temperature, body mass, experimental duration or the difference in

186

diet elemental composition (P>0.35 for all slopes). Additionally, there was significant

187

heterogeneity in the response of N:P excretion in field studies (Q=12.83, ν=6, P=0.046), but we

188

could not further explore any potential sources of this heterogeneity with the data available.

189

Discussion

190
191

In this study we synthesize a variety of empirical studies to show that diet can influence
the ratio of dissolved nutrients excreted by aquatic consumers and suggest mechanisms by which

192

it may do so. We found that dietary composition can have significant impacts on fish excretion

193

ratios in controlled aquaculture settings. In particular, fish feeding on low N:P diets with higher

194

amounts of animal protein excreted at a lower N:P ratio when accounting for the amount

195

consumed (Fig. 4). While these effects were strong in laboratory studies, other sources of

196

variation must be examined to improve our mechanistic understanding of consumer-driven

197

nutrient recycling in the field.

198

The mass-balance used in ecological stoichiometry (Sterner & Elser, 2002) provides a

199

simple framework for making predictions about organismal growth and nutrient recycling (Elser

200

et al., 1988; Sterner, 1990; Elser & Urabe, 1999; Elser, Hayakawa & Urabe, 2001). In a mass-

201

balance model of organismal growth, an animal should excrete and/or egest the excess nutrients

202

consumed beyond what is needed for somatic growth and reproduction (Kitchell et al., 1974;

203

Sterner & Elser 2002; Fig. 1). As animals often exhibit strong stoichiometric homeostasis, their

204

body elemental composition should not change substantially with diet; therefore excess

205

consumed nutrients should be excreted or egested (Sterner & Elser, 2002). Some recent studies

206

have suggested fish can be stoichiometrically flexible in some cases (McManamay et al., 2011;

207

El-Sabaawi et al., 2012a,b; Benstead et al., in press), thus offering a potential explanation for the

208

lack of strong correspondence of diet to excretion ratios in prior field studies (Schindler & Eby,

209

1997; McManamay et al., 2011). However, in finding that consumption-specific excretion of N

210

and N:P increases and P decreases with increasing diet N:P, our results support the predictions of

211

stoichiometric theory. By accounting for consumption rates, we have gained new insights into

212

how diet affects excretion ratios, insights that we could not from field studies for which

213

consumption is extremely challenging to measure.

214

Our results highlight the importance of consumption to excretion ratios. Most

215

importantly, we found that while fish excretion rate of N did not significantly differ with diet

216

composition, the excretion rate of N per gram of food consumed did (Fig. 3). In contrast,

217

excretion of P significantly decreased with increasing dietary N:P both independent of

218

consumption and per gram consumed (Fig. 3). This result could stem from fish eating less total

219

food when feeding on high N:P diets and/or the fact that those diets had less P. The fact that

220

mass-specific P consumption declined with increasing diet N:P is likely a consequence of most

221

studies manipulating diet N:P primarily by manipulating P rather than N contents. As dietary P

222

contents of fish can vary substantially through space and time (e.g., Mehner et al., 1998;

223

Zandonà et al., 2011), this mechanism certainly impacts fish excretion ratios in natural settings.

224

Further, mass-specific consumption rates tended to decline with increasing dietary N:P (P=0.10),

225

thus this mechanism may be important in some, but not all situations. If fish consume less

226

material when feeding on high N:P foods, and they also excrete more N and less P per gram of

227

diet consumed, then the ratio of N:P excreted will be altered through both direct and

228

consumptive effects of diet stoichiometry. However, the underlying fact that both N and P

229

excretion per gram consumed differed with diet N:P ratio is itself an interesting result that merits

230

further examination.

231

In many of these studies, and often in natural systems, shifts in diet elemental

232

composition co-occur with differences in the abundance of animal, plants or algae in the diet. In

233

systems where consumers are largely consuming entirely one group of diet items, such as

234

zooplankton feeding on phytoplankton, dietary N:P alone should largely determine how diet

235

impacts excretion ratios (e.g., Sterner, 1990). However, when animals consume diets with co-

236

varying elemental composition and protein sources, these confounding sources of variation can

237

produce differing effects on excretion ratios. Differences in the biochemical form of nutrients

238

present could alter assimilation efficiency, which could in turn lead to differential egestion and

239

excretion of individual nutrients. Although previous researchers have assumed constant

240

assimilation efficiencies across diets in fish, this assumption is unrealistic for fish that consume

241

diets consisting of multiple food types (Lall, 1991). Since excess undigested nutrients should be

242

egested as particulate waste products (Wotton & Malmqvist, 2001; Halvorson et al., 2015),

243

concurrent changes in digestibility with diet N:P could confound effects of diet on dissolved

244

excretion rates. For example, variation in protein digestibility among plant- or algae-based and

245

animal-based diet items could lead to differences in the amount of N egested as opposed to

246

excreted without substantially affecting the amount of P egested or excreted (Robbins et al.,

247

2005). However, P digestibility often differs between plants, algae and animals because plants

248

often contain large amounts of P in phytate or phytic acid, which is difficult for many fish to

249

digest (Lall, 1991). In our study a large number of plant-based diets were treated with phytase to

250

increase P digestibility, thus we expected effects of P digestibility to be lower in magnitude than

251

those of N digestibility. However, this digestibility difference is likely important to consumers in

252

natural settings where fish cannot easily digest phytic acid. Our results support this prediction, as

253

consumption-specific excretion rates of both N and P differed with diet N:P (Fig. 3), suggesting

254

that N and P assimilation efficiency differed when feeding on high N:P plant-based diets vs. low

255

N:P fishmeal-based diets. If the proportion and elemental ratios of material egested and excreted

256

differ as a function of diet elemental composition and/or protein source, no strong relationship

257

between diet elemental composition and excretion ratios may be observed (McManamay et al.,

258

2011). As a result, our results support the idea that factors other than diet N:P such as protein

259

digestibility, phytate contents and consumption rates must be taken into account when assessing

260

the impacts of diet on consumer excretion ratios.

261

In spite of the considerable interest in excretion ratios such as N:P due to the importance

262

of stoichiometric ratios of nutrients supplied to primary producers (e.g., Elser et al., 1988;

263

Sterner, Elser & Hessen, 1992), studies of excretion ratios are complicated by the fact that

264

physiological regulation of N and P is largely controlled separately in fish. The majority of P

265

consumed by fish and other vertebrates is used for bone mineralization (Lall, 1991; Hendrixson

266

et al., 2007; Huitema et al., 2012), yet a large amount of N consumed is used for the synthesis of

267

protein (Sterner & Elser, 2002). However, stoichiometric theory offers a link between these

268

disparate physiological pathways. Since fish are generally stoichiometrically homeostatic over an

269

individual life stage (Sterner & Elser, 2002), those excess nutrients not assimilated must be

270

excreted and/or egested. Therefore, the ratio of what is consumed to what is needed by a fish can

271

still be used to predict excretion ratios even if the individual pathways of those elements within

272

the organism are not tightly connected. Another potential factor that may confound dietary

273

effects on excretion is that excretion rates of N and P scale differently with body mass (Torres &

274

Vanni, 2007). If consumers grow at different rates when feeding on diets of differing elemental

275

composition, differences in body mass alone could account for differences in excretion ratios

276

(Villéger et al., 2012a,b). We were unable to correct for the different allometries of N and P

277

excretion because the units in which excretion was reported varied between studies, but all

278

studies reported excretion as some function of fish mass. We believe that our conclusions are

279

robust to the lack of an allometric correction in our analyses since specific growth rate was not

280

significantly affected by diet N:P in the studies analyzed. However, P-limitation of growth in fish

281

is possible at ecologically relevant dietary P levels (Hood et al., 2005; Benstead et al., in press),

282

thus we do believe that organismal growth and size differences caused by feeding on different

283

diets could lead to differences in excretion ratios in natural settings.

284

Physiological responses to differing diets that are not accounted for in field studies of diet

285

effects on excretion ratios may explain the difficulty of translating laboratory results into field

286

settings. While heterogeneity in the only effect size measured in field studies, excretion N:P, was

287

significantly greater than 0, only one of the six effect size measurements, raw P excretion,

288

exhibited significant heterogeneity in laboratory studies. One source of this discrepancy may be

289

the lack of correspondence between measured resources and actual fish diets. There are

290

considerable difficulties associated with measuring the true elemental composition of the diet

291

consumed and assimilated in the field. If the resources sampled by the researchers do not

292

specifically match what the fish are consuming and assimilating, conclusions about the effect of

293

diet on excretion ratios may be invalid (Hood et al., 2005). This may be particularly true of

294

omnivorous fish, which may consume different proportions of animals, plants and algae at

295

different sites or times of the year (e.g., Grimm, 1988). Further, local selection pressures such as

296

the degree of predation can lead to differences in fish dietary habits and life history traits

297

between sites (Zandonà et al., 2011; El-Sabaawi et al., 2012a). While differences between fish in

298

each treatment were controlled for in aquaculture studies by selecting all fish from one

299

population, such as a single hatchery source and keeping all fish under the same conditions aside

300

from the diet they were fed, field studies often compare individuals from separate populations.

301

Evolutionary differences between populations in the field studies may also represent a

302

covariate that cannot be separated from diet differences, thus complicating interpretation. That is,

303

comparisons of diet differences of a given species between sites, e.g., different streams or lakes,

304

represent populations of that species that likely experience at least some degree of genetic

305

separation. Therefore, differences in genotypes between populations cannot be ruled out as a

306

confounding variable in these studies. While stoichiometric theory predicts that individuals of a

307

given animal species and life history stage should have a given C:N:P stoichiometric

308

composition (Sterner & Elser, 2002), this does not apply across organisms with differing

309

genotypes. Indeed, P homeostasis is known to be genetically controlled in developing fish

310

(Huitema et al., 2012). Therefore, differential selection pressures between populations may

311

affect a fish’s response to diet quality. Differences in selection pressures such as temperature,

312

salinity, resource quality and predation pressure also drive evolution of organismal traits and life

313

histories that can affect body elemental composition (e.g., Zandonà et al., 2011; El-Sabaawi et

314

al., 2012a,b; Liess et al., 2013). Since interpopulation differences may be a source of

315

unmeasured variance in studies across natural gradients, linking evolutionary divergence to

316

consumer-driven nutrient recycling represents a promising area of future research.

317

Since Vanni (2002) reviewed the importance of nutrient recycling by consumers in

318

freshwater ecosystems, we have gained a greater appreciation for the role animals play in the

319

way nutrients cycle through ecosystems. Indeed, many studies have investigated how important

320

the transportation and transformation of nutrients by consumers can be to ecosystem function

321

(McIntyre et al., 2007; Layman et al., 2011; Small et al., 2011; Atkinson et al., 2013). However,

322

more work is needed to improve our understanding of the mechanisms that influence consumer

323

excretion rates and ratios. Our results suggest that diet is one of these mechanisms, but relatively

324

few studies have examined the effects of diet composition on consumer-driven nutrient recycling

325

in the field (McManamay et al., 2011). We show that dietary N:P can affect excretion ratios

326

across several fish species when correcting for consumption (Fig. 4). As raw N excretion was not

327

significantly affected by dietary N:P (Fig. 3), we hypothesize that differences in protein

328

digestibility can weaken the relationship between dietary N:P and excreted N:P for consumers

329

that feed on both animal and plant or algal material. While the application of stoichiometric

330

theory provides a promising framework through which to investigate consumer impacts on

331

ecosystem function, effective testing of stoichiometric theory may require that future work

332

examining dietary effects on excretion rates and ratios should consider not only dietary N:P but

333

specifically the forms in which these nutrients are present in the diet, how much is consumed and

334

how efficiently consumers assimilate dietary elements. Additionally, it is worth investigating

335

whether evolutionary differences between populations impact intraspecific consumer nutrient

336

recycling rates. While our study suggests that dietary composition can play a significant role in

337

altering excretion rates and ratios, more careful tests of this effect in the field across a range of

338

diets are needed before the impact of resource quality changes on consumer-driven nutrient

339

recycling and its importance to ecosystem function can be fully understood and integrated into

340

conceptual and theoretical frameworks.

341

Acknowledgments

342

We thank Albert Ruhí and two anonymous reviewers for comments on prior drafts of this

343

manuscript that greatly improved its quality. We also thank the editors of this special issue for

344

providing a forum to discuss these ideas and their own feedback on the manuscript. Kate Chanba

345

provided the fish illustration in Figure 1. EKM was supported by a research fellowship from

346

Arizona State University and the Smithsonian Tropical Research Institute. JJE acknowledges

347

support from the National Science Foundation (DEB-0950175).

348

References

349
350

Allgeier J.E., Yeager L.A. & Layman C.A. (2013) Consumers regulate nutrient limitation
regimes and primary production in seagrass ecosystems. Ecology, 94, 521-529.

351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396

Atkinson C.L., Vaughn C.C., Forshay K.J. & Cooper J.T. (2013) Aggregated filter-feeding
consumers alter nutrient limitation: consequences for ecosystem and community
dynamics. Ecology, 94, 1359-1369.
Ballestrazzi R., Lanari D., D’Agaro E. & Mion A. (1994) The effect of dietary protein level and
source on growth, body composition, total ammonia and reactive phosphate excretion of
growing sea bass (Dicentrarchus labrax). Aquaculture, 127, 197-206.
Benstead J.P., Hood J.M., Whelan N.V., Kendrick M.R., Nelson D., Hanninen A.F. et al. (In
Press) Coupling of dietary phosphorus and growth across diverse fish taxa: a metaanalysis of experimental aquaculture studies. Ecology, doi: 10.1890/13-1859.1.
Boersma M. & Elser J.J. (2006) Too much of a good thing: On stoichiometrically balanced diets
and maximal growth. Ecology, 87, 1325-1330.
Bureau D.P. & Cho C.Y. (1999) Phosphorus utilization by rainbow trout (Oncorhynchus mykiss):
estimation of dissolved phosphorus waste output. Aquaculture, 179, 127-140.
Capps K.A. & Flecker A.S. (2013a) Invasive fishes generate biogeochemical hotspots in a
nutrient-limited system. PLoS ONE, 8, e54093.
Capps K.A. & Flecker A.S. (2013b) Invasive aquarium fish transform ecosystem nutrient
dynamics. Proceedings of the Royal Society B – Biological Sciences, 280, 20131520.
Dias J., Conceição L.E.C., Ramalho Ribeiro A., Borges P., Valente L.M.P. & Teresa Dinis M.
(2009) Practical diet with low fish-derived protein is able to sustain growth performance
in gilthead seabream (Sparus aurata) during the grow-out phase. Aquaculture, 293, 255262.
El-Sabaawi R.W., Kohler T.J., Zandonà E., Travis J., Marshall M.C., Thomas S.A. et al. (2012a)
Environmental and organismal predictors of intraspecific variation in the stoichiometry of
a neotropical freshwater fish. PLoS ONE, 7, e32713.
El-Sabaawi R.W., Zandonà, E., Kohler T.J., Marshall M.C., Moslemi J.M., Travis J. et al.
(2012b) Widespread intraspecific organismal stoichiometry among populations of the
Trinidadian guppy. Functional Ecology, 26, 666-766.
Elser J.J., Elser M.M., MacKay N.A. & Carpenter S.R. (1988) Zooplankton-mediated transitions
between N and P limited algal growth. Limnology and Oceanography, 33, 1-14.
Elser J.J. & Urabe J. (1999) The stoichiometry of consumer-driven nutrient recycling: theory,
observations, and consequences. Ecology, 80, 745-751.
Elser J.J., Hayakawa H. & Urabe J. (2001) Nutrient limitation reduces food quality for
zooplankton: Daphnia response to seston phosphorus enrichment. Ecology, 82, 898-903.

397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440

González-Bergonzoni I., Meerhoff M., Davidson T.A., Teixeira-de Mello F., Baatrup-Pedersen
A. & Jeppesen E. (2012) Meta-analysis shows a consistent and strong latitudinal pattern
in fish omnivory across ecosystems. Ecosystems, 15, 492-503.
Green J.A., Hardy R.W. & Brannon E.L. (2002) Effects of dietary phosphorus and lipid levels on
utilization and excretion of phosphorus and nitrogen by rainbow trout (Oncorhynchus
mykiss). 1. Laboratory-scale study. Aquaculture Nutrition, 8, 279-290.
Grimm N.B. (1988) Feeding dynamics, nitrogen budgets, and ecosystem role of a desert stream
omnivore, Agosia chrysogaster (Pisces: Cyprinidae). Environmental Biology of Fishes,
21, 143-152.
Halvorson H.M., Fuller C., Entrekin S.A. & Evans-White M.A. (2015) Dietary influences on
production, stoichiometry and decomposition of particulate wastes from shredders.
Freshwater Biology, xx, xxx-xxx.
He X.J. & Wang W.X. (2008) Stoichiometric regulation of carbon and phosphorus in P-deficient
Daphnia magna. Limnology and Oceanography, 53, 244-254.
Hendrixson H.A., Sterner R.W. & Kay A.D. (2007) Elemental stoichiometry of freshwater fishes
in relation to phylogeny, allometry, and ecology. Journal of Fish Biology, 70, 121-140.
Hood J.M., Vanni M.J. & Flecker A.S. (2005) Nutrient recycling by two phosphorus-rich grazing
catfish: the potential for phosphorus-limitation of fish growth. Oecologia, 146, 247-257.
Hossain M.A., Pandey A. & Satoh S. (2007) Effects of organic acids on growth and phosphorus
utilization in red sea bream Pagrus major. Fisheries Science, 73, 1309-1317.
Huitema L.F.A., Apschner A., Logister I., Spoorendonk K.M., Bussmann J., Hammond C.L. et
al. (2012) Entpd5 is essential for skeletal mineralization and regulates phosphate
homeostasis in zebrafish. Proceedings of the National Academy of Sciences of the USA,
109, 21372-21377.
Jahan P., Watanabe T., Satoh S. & Kiron V. (2002) Different combinations of protein ingredients
in carp diets for reducing phosphorus loading. Fisheries Science, 68, 595-602.
Kaushik S.J., Covès D., Dutto G. & Blanc D. (2004) Almost total replacement of fish meal by
plant protein sources in the diet of a marine teleost, the European seabass, Dicentrarchus
labrax. Aquaculture, 230, 391-404.
Kitchell J.F., Koonce J.F., O’Neill R.V., Shugart H.H., Magnuson J.J. & Booth R.S. (1974)
Model of fish biomass dynamics. Transactions of the American Fisheries Society, 103,
786-798.

441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486

Kitchell J.F., O’Neill R.V., Webb D., Gallepp G.W., Bartell S.M., Koonce J.F. et al. (1979)
Consumer regulation of nutrient cycling. BioScience, 29, 28-34.
Lall S.P. (1991) Digestibility, metabolism and excretion of phosphorus in fish. In Cowey C.B. &
Cho C.Y. (eds.) Nutritional Strategies and Aquaculture Waste. Proceedings of the First
International Symposium on Nutritional Strategies in Management of Aquaculture Waste.
University of Guelph, Guelph, Ontario, Canada, pp. 21-36.
Layman C.A., Allgeier J.E., Rosemond A.D., Dahlgren C.P. & Yeager L.E. (2011) Marine
fisheries declines viewed upside down: human impacts on consumer-driven nutrient
recycling. Ecological Applications, 21, 343-349.
Liess A., Rowe O., Guo J.W., Thomsson G. & Lind M.I. (2013) Hot tadpoles from cold
environments need more nutrients – life history and stoichiometry reflects latitudinal
adaptation. Journal of Animal Ecology, 82, 1316-1325.
McIntyre P.B., Jones L.E., Flecker A.S. & Vanni M.J. (2007) Fish extinctions alter nutrient
recycling in tropical freshwaters. Proceedings of the National Academy of Sciences of the
USA, 104, 4461-4466.
McIntyre P.B., Flecker A.S., Vanni M.J., Hood J.M., Taylor B.W. & Thomas S.A. (2008) Fish
distributions and nutrient cycling in streams: can fish create biogeochemical hotspots?
Ecology, 89, 2335-2346.
McManamay R.A., Webster J.R., Valett H.M. & Dolloff C.A. (2011) Does diet influence
consumer nutrient cycling? Macroinvertebrate and fish excretion in streams. Journal of
the North American Benthological Society, 30, 84-102.
Mehner T., Mattukat F., Bauer D., Voigt H. & Benndorf J. (1998) Influence of diet shifts in
underyearling fish on phosphorus recycling in a hypertrophic biomanipulated reservoir.
Freshwater Biology, 40, 759-769.
Pilati A. & Vanni M.J. (2007) Ontogeny, diet shifts, and nutrient stoichiometry in fish. Oikos,
116, 1663-1674.
Plath K. & Boersma M. (2001) Mineral limitation of zooplankton: Stoichiometric constraints and
optimal foraging. Ecology, 82, 1260-1269.
R Core Team (2013) R: A Language and Environment for Statistical Computing. R Foundation
for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
Rice W.R. (1989) Analyzing tables of statistical tests. Evolution, 43, 223-225.
Robbins C.T., Felicetti L.A. & Sponheimer M. (2005) The effect of dietary protein quality on
nitrogen isotope discrimination in mammals and birds. Oecologia, 144, 534-540.

487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530

Rodehutscord M., Gregus Z. & Pfeffer E. (2000) Effect of phosphorus intake on faecal and nonfaecal phosphorus excretion in rainbow trout (Oncorhynchus mykiss) and the
consequences for comparative phosphorus availability studies. Aquaculture, 188, 383398.
Rosenthal R. & DiMatteo M.R. (2001) Meta-analysis: recent developments in quantitative
methods for literature reviews. Annual Review of Psychology, 52, 59-82.
Sarker M.S.A., Satoh S. & Kiron V. (2007) Inclusion of citric acid and/or amino acid-chelated
trace elements in alternate plant protein source diets affects growth and excretion of
nitrogen and phosphorus in red sea bream Pagrus major. Aquaculture, 262, 436-443.
Sarker P.K., Satoh S., Fukada H. & Masumoto T. (2011) Effects of dietary phosphorus level on
non-faecal phosphorus excretion from yellowtail (Seriola quinqueradiata Temminck &
Schlegel) fed purified and practical diets. Aquaculture Research, 40, 225-232.
Schindler D.E. & Eby L.A. (1997) Stoichiometry of fishes and their prey: implications for
nutrient recycling. Ecology, 78, 1816-1831.
Small G.E., Pringle C.M., Pyron M. & Duff J.H. (2011) Role of the fish Astyanax aeneus
(Characidae) as a keystone nutrient recycler in low-nutrient tropical streams. Ecology, 92,
386-397.
Sterner R.W. (1990). The ratio of nitrogen to phosphorus resupplied by herbivores – zooplankton
and the algal competitive arena. American Naturalist, 136, 209-229.
Sterner R.W., Elser J.J. & Hessen D.O. (1992) Stoichiometric relationships among producers,
consumers, and nutrient cycling in pelagic ecosystems. Biogeochemistry, 17, 49-67.
Sterner R.W. & Elser J.J. (2002) Ecological Stoichiometry. Princeton University Press,
Princeton, NJ.
Storebakken T., Shearer K.D. & Roem A.J. (1998) Availability of protein, phosphorus and other
elements in fish meal, soy-protein concentrate and phytase-treated soy-proteinconcentrate-based diets to Atlantic salmon, Salmo salar. Aquaculture, 161, 365-379.
Sukumaran K., Pal A.K., Sahu N.P., Debnath D. & Patro B. (2009) Phosphorus requirement of
Catla (Catla catla Hamilton) fingerlings based on growth, whole-body phosphorus
concentration, and non-faecal phosphorus excretion. Aquaculture Research, 40, 139-147.
Sumagaysay-Chavoso N.S. (2003) Nitrogen and phosphorus digestibility and excretion of
different-sized groups of milkfish (Chanos chanos Forsskål) fed formulated and natural
food-based diets. Aquaculture Research, 34, 407-418.

531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576

Takkouche B., Cadarso-Suarez C. & Spiegelman D. (1999) Evaluation of old and new tests of
heterogeneity in epidemiologic meta-analysis. American Journal of Epidemiology, 150,
206-215.
Tantikitti C., Sangpong W. & Chiavareesajja S. (2005) Effects of defatted soybean protein levels
on growth performance and nitrogen and phosphorus excretion in Asian seabass (Lates
calcarifer). Aquaculture, 248, 41-50.
Taylor J.M., Back J.A., Valenti T.W. & King R.S. (2012) Fish-mediated nutrient cycling and
benthic microbial processes: can consumers influence stream nutrient cycling at multiple
spatial scales? Freshwater Science, 31, 928-944.
Torres L.E. & Vanni M.J. (2007) Stoichiometry of nutrient excretion by fish: interspecific
variation in a hypereutrophic lake. Oikos, 116, 259-270.
Vanni M.J. (2002) Nutrient cycling by animals in freshwater ecosystems. Annual Review of
Ecology and Systematics, 33, 341-370.
Vanni M.J., Flecker A.S., Hood J.M. & Headworth J.L. (2002) Stoichiometry of nutrient
recycling by vertebrates in a tropical stream: linking species identity and ecosystem
processes. Ecology Letters, 5, 285-293.
Vanni M.J., Boros G. & McIntyre P.B. (2013) When are fish sources vs. sinks of nutrients in
lake ecosystems? Ecology, 94, 2195-2206.
Verant M.L., Konsti M.L., Zimmer K.D. & Deans C.A. (2007) Factors influencing nitrogen and
phosphorus excretion rates of fish in a shallow lake. Freshwater Biology, 52, 1968-1981.
Villéger S., Ferraton F., Mouillot D. & de Wit R. (2012a) Nutrient recycling by coastal
macrofauna: intra- versus interspecific differences. Marine Ecology Progress Series, 452,
297-303.
Villéger S., Grenouillet G., Suc V. & Brosse S. (2012b) Intra- and interspecific differences in
nutrient recycling by European freshwater fish. Freshwater Biology, 57, 2330-2341.
Wotton R.S. & Malmqvist B. (2001) Feces in aquatic ecosystems. BioScience, 51, 537-544.
Yang Y.H., Wang Y.Y., Lu Y. & Li Q.Z. (2011) Effect of replacing fish meal with soybean meal
on growth, feed utilization, and nitrogen and phosphorus excretion on rainbow trout
(Oncorhynchus mykiss). Aquaculture International, 19, 405-419.
Zandonà E., Auer S.K., Kilham S.S., Howard J.L., Lopez-Sepulcre A., O’Connor M.P. et al.
(2011) Diet quality and prey selectivity correlate with life histories and predation regime
in Trinidadian guppies. Functional Ecology, 25, 964-973.

577
578
579
580
581
582
583

Table 1. Species and family identities of fish in studies included in the meta-analysis. Reference
numbers are as follows: (1) Sukumaran et al., 2009; (2) Sumagaysay-Chavoso, 2003; (3) Jahan
et al., 2002; (4) Kaushik et al., 2004; (5) Ballestrazzi et al., 1994; (6) Tantikitti, Sangpong &
Chiavareesajja, 2005; (7) Yang et al., 2009; (8) Green, Hardy & Brannon, 2002; (9) Bureau &
Cho, 1999; (10) Rodehutscord, Gregus, & Pfeffer, 2000; (11) Hossain et al., 2007; (12) Sarker,
Satoh & Kiron, 2007; (13) Storebakken, Shearer & Roem, 1998; (14) Sarker et al., 2011; (15)
Dias et al., 2009; (16) Small et al., 2011; (17) McManamay et al., 2011.
Species

Family

Reference(s)

Diet Manipulations
Catla catla

Cyprinidae

1

Chanos chanos

Chanidae

2

Cyprinus carpio

Cyprinidae

3

Dicentrarchus labrax

Moronidae

4,5

Lates calcarifer

Latidae

Oncorhynchus mykiss

Salmonidae

Pagrus major

Sparidae

Salmo salar

Salmonidae

13

Seriola quinqueradiata

Carangidae

14

Sparus aurata

Sparidae

15

Alfaro cultratus

Poeciliidae

16

Astatheros alfari

Cichlidae

16

Astyanax aeneus

Characidae

16

Atherinella hubbsi

Atherinopsidae

16

Chrosomus erythrogaster

Cyprinidae

17

Cottus bairdi

Cottidae

17

Oncorhynchus mykiss

Salmonidae

17

6
7,8,9,10
11,12

Field Studies

584
585

FISH GROWTH N:P

FOOD N:P
EXCRETION N:P
EGESTION N:P
EXCRETION N = FOOD N – (FISH GROWTH N + EGESTION N)
EXCRETION P = FOOD P – (FISH GROWTH P + EGESTION P)
586
587
588
589

Fig. 1 Mass balance model of N and P budgets for a fish. Our model represents a conceptual
simplification of the major nutrient fluxes in consumers (Kitchell et al., 1974; Sterner, 1990).

1
0
-1
-2
-3

Intake Rate Effect Size

*

590
591
592
593
594
595
596

N Intake

P Intake

Fig. 2 Effect size of diet N:P on intake (g * g fish-1 * day-1) of N and P in diet manipulation
studies. Effect size, η2, was measured as the treatment sum-of-squares divided by total sum-ofsquares from a linear model then transformed into a Z score for ease of analysis. Bars with *
indicates effect size significantly different from zero based on a two-tailed t-test. Column lengths
indicate mean effect sizes and error bars represent 95% confidence intervals.

2

Consumption-Specific

*

0
-1

1

0
-1

Effect Size

1

2

Raw Excretion

*
P Excretion

-2

-2

*
N Excretion

P Excretion

N Excretion

597
598
599
600
601
602
603
604
605
606
607

Fig. 3 Effect size of diet N:P on raw and consumption-specific N and P excretion in diet
manipulation studies. Effect size, η2, was measured as the treatment sum-of-squares divided by
total sum-of-squares from a linear model then transformed into a Z score for ease of analysis.
Consumption-specific excretion was calculated as the excretion measure presented in the study
divided by mass-specific consumption rate. Points with * indicates effect size significantly
different from zero based on a two-tailed t-test. Points indicate mean effect sizes and error bars
represent 95% confidence intervals.

2

2

Raw Excretion

Consumption-Specific

1

1

0
-1
-2

-1

0

N/A

-2

N:P Excretion Effect Size

*

Diet Manipulations Field Studies

608
609
610
611
612
613
614
615

Diet Manipulations Field Studies

Fig. 4 Mean ± 95% confidence interval of effect size of diet N:P on excretion N:P. Effect size,
η2, was measured as the treatment sum-of-squares divided by total sum-of-squares from a linear
model then transformed into a Z score for ease of analysis. Consumption-specific effect sizes are
missing in field studies because those studies did not measure consumption rate. Points with *
indicates effect size significantly different from zero based on a two-tailed t-test. Points indicate
mean effect sizes and error bars represent 95% confidence intervals.