1 2 3 The interaction between the spatial distribution of resource patches and 4 population density: consequences for intraspecific growth and 5 morphology 6 7 Running headline: Spatial resource distribution and population density 8 9 B. Jacobson1*, J.W.A. Grant2 and P.R. Peres-Neto1† 1 Département des Sciences Biologiques, Université du Québec à Montréal, C.P. 8888, Succ. Centre-Ville, Montréal, Québec H3C3P8 Canada 2 Department of Biology, Concordia University, 1455 de Maisonneuve Blvd. West, Montréal, Québec, H3G1M8, Canada 10 * Corresponding author: jacobson.bailey@courrier.uqam.ca 11 12 13 14 † Canada Research Chair in Spatial Modelling and Biodiversity 15 16 17 1 18 Summary 19 1. How individuals within a population distribute themselves across resource patches 20 of varying quality has been an important focus of ecological theory. The ideal free 21 distribution predicts equal fitness amongst individuals in a 1:1 ratio with 22 resources, whereas resource defence theory predicts different degrees of 23 monopolization (fitness variance) as a function of temporal and spatial resource 24 clumping and population density. 25 2. One overlooked landscape characteristic is the spatial distribution of resource 26 patches, altering the equitability of resource accessibility and thereby the effective 27 number of competitors. While much work has investigated the influence of 28 morphology on competitive ability for different resource types, less is known 29 regarding the phenotypic characteristics conferring relative ability for a single 30 resource type, particularly when exploitative competition predominates. 31 3. Here we used young-of-the-year rainbow trout (Oncorhynchus mykiss) to test 32 whether and how the spatial distribution of resource patches and population 33 density interact to influence the level and variance of individual growth, as well as 34 if functional morphology relates to competitive ability. Feeding trials were 35 conducted within stream channels under three spatial distributions of nine 36 resource patches (distributed, semi-clumped and clumped) at two density levels (9 37 and 27 individuals). 38 4. Average trial growth was greater in high-density treatments with no effect of 39 resource distribution. Within-trial growth variance had opposite patterns across 40 resource distributions. Here, variance decreased at low-population, but increased 2 41 at high-population densities as patches became increasingly clumped as the result 42 of changes in the levels of interference vs. exploitative competition. Within-trial 43 growth was related to both pre- and post-trial morphology where competitive 44 individuals were those with traits associated with swimming capacity and 45 efficiency: larger heads/bodies/caudal fins and less angled pectoral fins. 46 5. The different degrees of within-population growth variance at the same density 47 level found here, as a function of spatial resource distribution, provide an 48 explanation for the inconsistencies in within-site growth variance and population 49 regulation often noted with regard to density-dependence in natural landscapes. 50 51 Key-words: intraspecific competition, landscape structure, morphometrics, negative 52 density dependence, phenotype-dependent growth, stream fish 53 54 55 56 57 58 59 60 61 62 63 3 64 65 Introduction How individuals within a population distribute themselves across resource patches 66 of varying quality has been a central focus of ecological theory. The ideal free 67 distribution (IFD) proposes that competitively equivalent individuals with perfect 68 knowledge of and the cost-free ability to move throughout the landscape distribute 69 themselves in a 1:1 ratio with resources such that fitness across patches is equal (Fretwell 70 & Lucas 1970; Lomnicki 1988; Kennedy & Gray 1993). In reality, individual fitness 71 within populations often varies as a function of unequal competitive abilities (Abrahams 72 1986; Kennedy & Gray 1993); however, the presence of unequal competitors within a 73 landscape does not necessarily result in variation in resource gain (Grand & Grant 1994). 74 Tregenza, Hack & Thompson (1996), for instance, found that good competitors only had 75 a fitness advantage at low densities because at high densities individuals switched from 76 interference to exploitative (alternatively, contest to scramble) competition. 77 The influence of environmental properties (sensu Grant 1993) on the ability for an 78 individual to monopolize resources has been well investigated within the realm of 79 behavioural ecology. At the scale of an individual resource patch, resource defence 80 theory predicts a dome-shaped relationship between the economic defensibility of a 81 resource patch and either the degree of temporal and spatial resource clumping or the 82 number of potential competitors within the population (Grant 1993; Robb & Grant 1997; 83 Grant, Gaboury & Levitt 2000; Ward, Webster & Hart 2006). Taken together, within- 84 population fitness (growth) variation will not only depend on the amount of aggression 85 (i.e. interference competition) and resource monopolization (Grant & Guha 1993; Weir & 4 86 Grant 2004; Noël, Grant & Carrigan 2005), but also on the equality of resource 87 accessibility (Boujard, Labbé & Aupérin 2002). 88 One overlooked landscape characteristic (but see Silver et al. 2000) that has the 89 potential to influence resource accessibility, and thus, relative fitness amongst individuals 90 is the spatial distribution of resource patches. When patches are spread across the 91 landscape, individuals may be able to settle into patches with few interactions between 92 competitors, thus leading to more even resource access, partitioning and fitness (Noël, 93 Grant & Carrigan 2005). Conversely, as patches become increasingly clumped in space, 94 the effective number of competitors within the population will increase due to a decrease 95 in foraging area (Fausch 1984; Noël, Grant & Carrigan 2005), in turn decreasing the 96 equality of resource accessibility and potentially increasing fitness variation. In stream 97 fishes, where much of the work on density-dependent growth and regulation (e.g. 98 mortality, emigration) has been conducted, within-site growth variance has largely been 99 attributed to variation in foraging site (patch) quality (Site Quality Hypothesis; SQH) 100 (Newman 1993; Ward et al. 2007; Lobón-Cerviá 2010). Thus, the potential influence of 101 the spatial distribution of resources on growth variation may provide an alternative 102 explanation for the differential patterns of individual growth (Imre, Grant & Cunjak 103 2010) and distribution found across natural populations. 104 Just which individuals in the population are able to monopolize resources and 105 have higher relative fitness is a function of their relative competitive ability. In stream 106 fishes, competitive ability is often linked to body size, with larger individuals able to gain 107 access to the best foraging patches (Fausch 1984; Ward, Webster & Hart 2006), though 108 the causal order of this relationship has been questioned (Ward, Webster & Hart 2006). 5 109 While the resource polymorphism literature has investigated the influence of morphology 110 on competitive ability for different resources (Smith & Skúlason 1996; Bolnick 2004; 111 Araújo et al. 2008; Edelaar, Siepielski & Clobert 2008), less is known about the 112 phenotypic characteristics dictating relative competitive ability for a single resource type 113 when size variation is inconsequential. This is a particularly relevant question within a 114 single young age class where size may initially be somewhat invariant, but growth ability 115 has large impacts for subsequent survival (Ward, Nislow & Folt 2009; Lobón-Cerviá 116 2010). In such cases, competitive ability may be based on an individual’s ability to use its 117 environment more efficiently. Within riffle habitat sections of streams, for example, 118 where greater water velocity is related to increased resource delivery (Gotceitas & Godin 119 1992), an individual’s ability to gain resources may be linked to its swimming capacity 120 and ability to hold station against the current (Ward, Webster & Hart 2006; Leavy & 121 Bonner 2009). 122 In this study, we set out to test whether and how the spatial distribution of 123 resource patches with equal quality (contrary to the SQH) and population density interact 124 to influence the mean and variance of within-population growth. In addition, we tested 125 whether an individual’s morphology relates to its ability to obtain resources (compete). 126 To do so, we conducted feeding trials within flow-through stream channels using young- 127 of-the-year rainbow trout (O. mykiss) under three different spatial distributions of nine 128 resource patches: evenly distributed, semi-clumped and clumped (Fig. 1). Such trials 129 were conducted at two density levels: low, one individual per patch and high, three 130 individuals per patch. We expected (i) mean growth to be greater within high-density 131 trials due to a decrease in energy expended on aggression (i.e. less interference 6 132 competition), as per resource defence theory (Kim & Grant 2007); (ii) within-trial growth 133 variance to increase with increased resource patch clumping due to changes in effective 134 competitor number (Fausch 1984; Noël, Grant & Carrigan 2005) and resource 135 accessibility; and (iii) individual growth to be most similar in low-density distributed 136 resource trials due to a 1:1 competitor to resource patch ratio, as per the IFD (Noël, Grant 137 & Carrigan 2005). Lastly, we expected (iv) competitive ability, irrespective of density 138 and resource distribution, to be based on functional morphology and its relation to 139 swimming ability (Ward, Webster & Hart 2006). 140 141 Materials and Methods 142 Species and experimental set-up 143 We acquired 500 young-of-the-year rainbow trout (Oncorhynchus mykiss) (Fig. 144 1a) 5cm in length from Pisciculture des Arpents Verts, Québec, Canada for use in our 145 experiments. When not in use fish were housed within two circular 133L constant-flow 146 tanks (flow at 50% depth=0.25m/s; water temperature=18-19.5°C) on a 12h light:12h 147 dark cycle (Brown & Brown 1993) (lights on at 7am) and fed a maintenance ration of dry 148 food pellets (Skretting extruded salmonid feed). Housing conditions were monitored, and 149 daily feedings provided by animal care staff at Concordia University, Québec, Canada. 150 Experimental trials were conducted within four flow-through experimental stream 151 channels (Fig. 1b) under the same light regime. Each channel was lined to a depth of 152 2.5cm with small natural-coloured aquarium gravel wherein nine terracotta “saucers” 153 were embedded, flush with the bed of gravel, to act as resource patches. Flow in each 154 channel was held constant across all trials (average flow at 50% depth=0.15-0.18m/s 7 155 across channels). Water temperature within the system (c. 18°C; Li & Brocksen 1977) 156 was controlled by the amount of de-chlorinated city water entering each channel and was 157 continuously recorded every 15 min by a HOBO temperature/light data logger (Onset 158 Computer Corporation, Bourne, MA, USA) placed in each channel. Loggers were 159 checked in the morning before feeding and again at the end of the light cycle (average 160 across channels ± SD: week 1 19.1± 0.6°C; week 2 18.7 ± 0.2°C; week 3 17.7 ± 0.8°C; 161 week 4 18.2 ± 0.2°C; and week 5 17.7 ± 0.5°C) to ensure that temperatures remained 162 within the preferred range for rainbow trout (Wood, Grant & Belanger 2012). To record 163 feeding dynamics and inter-individual interactions throughout the light cycle, digital 164 colour CCD bullet cameras were connected to a surveillance system (GeoVision, Inc., 165 Irvine, CA, USA) and mounted directly above each channel; with the exception of 166 recording periods, channels were covered with large-weave netting to prevent fish from 167 jumping. 168 169 170 Experimental procedure To determine the influence of spatial resource distribution and competitor density 171 on growth, feeding trials were conducted under a 3x2 factorial design testing three 172 different resource distributions [distributed (D), semi-clumped (S) and clumped (C), Fig. 173 1b] at each of two densities (low n=9 and high n=27) with three replicates per treatment. 174 Trials were conducted four at a time from 14 October to 22 November 2011, where each 175 week the individuals and treatments tested were both randomly selected and assigned to 176 each stream channel. 8 177 Each trial lasted a total of 8 days. On day 1, individuals were selected from 178 housing tanks, anesthetized using a 1 clove oil:10 ethanol mixture (active agent eugenol) 179 (Anderson, McKinley & Colavecchia 1997; Keene et al. 1998) and each given a unique 180 identifier by way of two subcutaneous VIE tags (visible implant elastomer; Northwest 181 Marine Technologies, Shaw Island, WA, USA) in one or two of four fluorescent colours 182 in one or two of several body locations. Individuals were then photographed on one 183 lateral side for pre-trial morphology, weighed for initial mass and measured for initial 184 body and caudal widths (Fig. 1a). After recovering from anesthesia, individuals were 185 introduced into a stream channel and remained unfed for the day (Olsson, Svanbäck & 186 Eklöv 2007). On days 2-7, resources were added once daily to stream channels, evenly 187 divided across the nine patches, given at a level of 10% (Brown & Brown 1993) of the 188 total initial mass of individuals within the trial per day in order to encourage growth. 189 Feeding was recorded on days 2 and 7 (i.e. initial and final feedings) from the time food 190 entered the system until the end of the light cycle. Note that while no mortality occurred 191 as the result of handling or experimental treatment, there were some incidences of 192 individuals jumping out of the channels during initial and final recording and thus final 193 densities did not always match initially set levels (average density per treatment, low 194 density: D=8.67, S=8, C=7.67, high density: D=25.67, S=24.67, C=26; see also outliers 195 below). When this occurred during day 1 or 2, individuals were identified and the food 196 level adjusted to remain at 10% of the remaining trial initial mass. Lastly, on day 8 197 individuals were not fed (Noël, Grant & Carrigan 2005) to ensure the same level of 198 gastric evacuation (Fausch 1984; Currens et al. 1989) and near the end of the light cycle 199 were anaesthetized, identified, weighed for final mass and photographed on the same 9 200 lateral side for post-trial morphology. Individuals were afterwards returned to a separate 201 holding tank to ensure that no individual was used more than once. 202 203 204 Data analyses Three growth metrics were calculated for each individual and used as response 205 variables within analyses: (i) individual growth Gind was calculated as the difference 206 between log10 transformed final and initial mass (Noël, Grant & Carrigan 2005); (ii) 207 frowth variance Gvar was calculated as the absolute value of the difference between Gind 208 and the average Gind of the trial in which the individual participated. This metric, when 209 used in an analysis of variance (ANOVA), results in a test for variance differences rather 210 than means across treatments (Zar 2010); and lastly, (iii) relative growth Grel was 211 calculated as the difference between Gind and the average Gind of the trial in which the 212 individual participated and classified individuals based on their growth relative to their 213 trial average (i.e. either less or greater growth than the average). 214 To determine the influence of treatment type on trial growth, Gind (differences in 215 average growth across treatments) and Gvar (differences in growth variance across 216 treatments) were used as response variables in separate mixed-model ANOVAs with the 217 fixed effects of density, spatial resource distribution and the density x spatial distribution 218 interaction and the random effects of channel, week in which the trial was conducted and 219 initial mass or initial mass variance (calculated as per Gvar). Given the differences in 220 variances across treatments (a measure of interest of this study, i.e. Gvar), weights 221 proportional to group variances were used (Pinheiro & Bates 2000). 10 222 In order to determine the influence of treatment type on the spatial dynamics of 223 resource patch use as well as the amount of inter-individual aggression (i.e. interference 224 competition) all recordings of the final feeding (day 7) were analyzed, as it was assumed 225 that recorded behaviour had at this point become established (Noël, Grant & Carrigan 226 2005). Spatial feeding dynamics were characterized and recorded as the number of 227 seconds within a minute that at least one individual was at least one-half of a body length 228 within each patch, recorded every 5 min for the first 15 min of feeding during which time 229 feeding activity was at its highest. As all replicates for any given treatment type had 230 similar dynamics, information garnered from the final replicate only is presented. 231 Aggressive interactions were characterized as chases, defined as a unidirectional 232 swimming burst of one fish towards another lasting for at least one body length (Noël, 233 Grant & Carrigan 2005; Kim & Grant 2007; Wood, Grant & Belanger 2012), and scored 234 for the same 3 min as spatial feeding dynamics for each video. Due to issues of low video 235 quality and the tightly packed fast movements that occurred during feeding at high 236 densities, each video was scored twice and the total number of observed chases averaged 237 for each replicate. This average was then divided by the number of fish within the trial to 238 control for differences in density (Robb & Grant 1998; Kim & Grant 2007) and the 239 number of chases per fish used as the response variable within a mixed-effect ANOVA 240 with density, spatial distribution and the density x spatial distribution interaction entered 241 as fixed effects and the channel and week in which the trial was conducted entered as 242 random effects. 243 244 To determine if there was an influence of an individual’s morphology on their ability to gain resources (grow) within trials, geometric morphometrics were used. 11 245 Twenty-two landmarks were placed on both pre- and post-trial photographs (Fig. 1a) 246 using the program tpsDIG2 (Rohlf 2005a) and converted, separately, into partial warps 247 (shape variables) using the programs Coordgen6 and PCAgen6 (Sheets 2004a, 2004b). 248 Pre- and post-trial warps were then entered as separate fixed effects within two mixed- 249 model ANOVAs using Gind and Grel as the response variables with channel and week 250 entered as random effects. Note that three individuals, randomly distributed across 251 treatments, were identified as outliers with respect to post-trial morphology and were 252 removed from all analyses. To visualize which morphological features differed between 253 individuals with greater or less relative growth, deformation grids were produced for pre- 254 and post-trial morphologies of Grel classifications using the program tpsRegr (Rohlf 255 2005b). Lastly, as competitive ability within fish is often linked with length/size (Ward, 256 Webster & Hart 2006) and this effect is removed through standardization in geometric 257 morphometrics (Zelditch et al. 2004), regressions were performed for each treatment to 258 test for the presence of a relationship between Gind (growth) and each of log10 initial mass 259 (Massi), initial standard length (Lengthi), initial body width (BodyWi) and initial caudal 260 width (CaudalWi) (size). The significance of these correlations was determined with 261 mixed-effect regression using Gind as the response and each variable, separately per 262 treatment, as the fixed effect with random factors as above. All mixed-model ANOVAs 263 were run using the nlme package (Pinheiro et al. 2012) in R version 2.13.0 (R 264 Development Core Team, 2011 Vienna, Austria) and regressions were performed using 265 Statistica 6.0 (StatSoft, Inc., Tulsa, OK, USA). 266 267 Results 12 268 The degree of individual growth (Gind) within trials was found to differ as a 269 function of density, with average growth greater within high-density compared to low- 270 density treatments irrespective of the spatial distribution of resource patches (Table 1 and 271 Fig. 2a). Interestingly, the degree to which growth varied amongst individuals within 272 trials (Gvar) differed significantly across all factors, spatial distribution, density and their 273 interaction (Table 1); the variance decreased in low-density trials but increased in high- 274 density trials as resources became more spatially clumped (Fig. 2b). Such greater growth 275 variance in high-density spatially clumped treatments occurred due to the weight loss 276 (negative growth) of 14 individuals throughout the trials; only two individuals lost weight 277 in the distributed and one each in the semi-clumped and clumped treatments at low 278 densities. 279 Observation of videos recorded during the final feeding revealed different spatial 280 dynamics (Fig. 3a) and amounts of aggressive interactions between individuals as a 281 function of density and resource distribution (Table 1 and Fig. 3b). Overall, more 282 aggressive interactions (interference competition) occurred between individuals within 283 low-density compared to high-density trials, with the number of chases per fish 284 decreasing as spatial resource clumping increased. As reflected by the spatial dynamics of 285 patch use (location and shading within Fig. 3a), within low-density distributed trials, 286 individuals tended to chase and exclude others from the resource arena wherein foraging 287 attempts were made mainly at previously sampled patches. Conversely, within clumped 288 trials, patches were relatively more evenly utilized. Likewise, the spatial dynamics of 289 patch use for high-density trials also reflected more even resource partitioning with 290 individuals divided amongst and continually circulating over all patches. However, 13 291 within clumped trials, several individuals never appeared to attempt foraging. Across all 292 treatments, when not foraging, individuals remained within the downstream portion of the 293 stream channel. 294 The degree to which individuals grew within a trial, irrespective of treatment, was 295 found to be a function of their morphology; both pre- and post-trial morphology 296 correlated with individual growth (Gind) and differentiated between individuals of 297 opposite relative growth (Grel) (Table 1). Such effects could also be seen in deformation 298 grids of pre- and post-trial morphology as a function of Grel (Fig. 4), where the primary 299 differences between groups in pre-trial morphology were differences in pectoral fin 300 angle, head/body and caudal fin size and head shape and body height with regards to 301 post-trial morphology. Growth was, with few exceptions, unrelated to initial fish 302 length/size, with regressions overall largely non-significant (Table 2). 303 304 Discussion 305 The main goal of our study was to test whether the spatial distribution of 306 resources and population density interact in determining the growth and growth variance 307 (i.e. fitness) amongst individuals within a landscape. We found that this interaction can 308 generate different levels of fitness variation within a population through differences in the 309 amount of interference competition, thus causing individuals with specific morphologies 310 (phenotype) to be conferred more or less relative fitness dependent on treatment. Such 311 results have important consequences for the demographics of populations and degree of 312 regulation expected in natural systems of varying structure, potentially explaining several 14 313 inconsistencies found in the literature regarding density-dependent growth patterns (see 314 Implications). 315 316 317 Inter-treatment growth patterns: variance and competition Results from our study clearly show that the spatial distribution of resource 318 patches themselves, irrespective of quality, can influence the degree and distribution of 319 relative fitness (i.e. growth) amongst individuals within a population through changing 320 the effective number of competitors for patches within the system. We expected the 321 distribution to follow an IFD, with growth variance lowest in low-density distributed 322 resource trials, but increasing with increased resource clumping (Noël, Grant & Carrigan 323 2005). Instead, opposite patterns of growth variance were found across resource 324 distributions between density treatments; variance was greatest when resources were 325 clumped at high densities but when resources were distributed at low densities. The high 326 variance in growth found in low-density distributed trials, as evidenced from feeding 327 recordings, was a function of increased interference competition; individuals were 328 observed attempting to defend patches and others attempting to forage at only those 329 already sampled. As patches became increasingly clumped at low densities, the effective 330 number of competitors increased. Consequently, interactions switched from interference 331 to exploitative competition, which accounted for the decrease in growth variance found 332 here, likely because defense was too costly with increasing competitor pressure (Grant 333 1993; Grant & Guha 1993; Tregenza, Hack & Thompson 1996; Syarifuddin & Kramer 334 1996; Kaspersson, Höjesjö & Pedersen 2010). 15 335 Lower levels of interference competition were found within high-density trials, 336 (Brown, Brown & Srivastava 1992; Jobling & Baardvik 1994), as individuals cycled in 337 and out of resource patches (Robb & Grant 1997) and were able to spread themselves 338 more evenly amongst patches. As aggression has important associated fitness costs, such 339 as increasing energy expenditure, social stress and resource loss while defending (Li & 340 Brocksen 1977; Kim & Grant 2007; Kaspersson, Höjesjö & Pedersen 2010), the reduced 341 interference competition costs are likely responsible for the greater average growth, 342 irrespective of resource distribution, found in high-density trials. That said, while the 343 exploitative competition utilized at high densities is proposed to allow for a more even 344 distribution of resources amongst individuals (Grant 1993; Jobling & Baardvik 1994; 345 Boujard, Labbé & Aupérin 2002; Weir & Grant 2004), our results suggest that even 346 partitioning does not occur past a certain threshold. Indeed, the decrease in the equality of 347 resource accessibility at high densities, also found in other studies (Syarifuddin & Kramer 348 1996; Boujard, Labbé & Aupérin 2002), caused individuals to cease feeding, leading to 349 negative growth and increased growth variance, with increased spatial resource clumping 350 (Li & Brocksen 1977; Grant & Kramer 1990). 351 352 353 The influence and shaping of phenotype While competitive ability is often suggested to be related to body size (Post, 354 Parkinson & Johnston 1999; Ward, Webster & Hart 2006; Kaspersson, Höjesjö & 355 Pedersen 2010), we found no significant relationship between an individual’s level of 356 growth and their initial mass, length, body or caudal width. Instead, body shape was the 357 important component in obtaining resources, as levels of individual and relative growth 16 358 correlated with pre-trial morphology. Here individuals with greater than average trial 359 growth had a morphology indicating greater swimming ability and turning stability: 360 relatively deeper heads/bodies and larger caudal fins (Gatz 1979; Webb 1984; Ojanguren 361 & Braña 2003; Peres-Neto & Magnan 2004). These individuals also had pectoral fins that 362 were more flatly angled, potentially indicating more efficient maneuvering (Drucker & 363 Lauder 2003) and reduced energy expenditure through holding position against current on 364 the substrate (Fausch 1984). As swimming ability has been proposed to relate to 365 competitive ability in both exploitative and interference contexts (Ward, Webster & Hart 366 2006), such phenotypic characteristics likely accounted for the greater growth of these 367 individuals. Post-trial morphology was even more highly correlated with trial growth, 368 potentially reflecting both the relative effects of the amount of growth on body shape in 369 and of itself (Currens et al. 1989; Borcherding & Magnhagen 2008), as well as potential 370 investment in plasticity (i.e. morphological modulation) by individuals who had greater 371 levels of growth (Olsson, Svanbäck & Eklöv 2007), acting to further increase their 372 competitive ability. Indeed, individuals with greater relative growth were those with 373 characteristics of benthic habitat use (more dorsally placed eyes and greater body height) 374 (Gatz 1979; Mason et al. 2007) and exploitative ability (increased swimming ability and 375 turning stability) (Webb 1984; Ojanguren & Braña 2003; Peres-Neto & Magnan 2004). In 376 contrast, individuals with less relative growth were characterized by more slender bodies 377 and longer heads, potentially indicative of under-nourishment (Currens et al. 1989; 378 Olsson, Svanbäck & Eklöv 2007; Borcherding & Magnhagen 2008). 379 380 Implications for population regulation in natural landscapes 17 381 Within natural landscapes, as density increases, mean growth is often found to 382 follow a negative power curve (Jenkins et al. 1999; Grant & Imre 2005), with growth 383 variance increasing (as per the SQH) (Newman 1993; Keeley 2001; Lobón-Cerviá 2010). 384 After density-dependent emigration and mortality, however, the resulting patterns are 385 much more variable; exhibiting negative power curves, negative linear relationships of 386 varying strength, to no relationship between average growth and density (Newman 1993; 387 Jenkins et al. 1999; Post, Parkinson & Johnston 1999; Keeley 2001; Imre, Grant & 388 Cunjak 2005), with little consistency in associated variance-density relationships (Imre, 389 Grant & Cunjak 2010). As, here, we found different degrees of growth variance at the 390 same population density as a function of the spatial distribution of resource patches, our 391 results have the potential to shed light on such inconsistencies in natural systems. For 392 example, landscapes with high densities and clumped resources, rather than distributed, 393 may be characterized by high growth variance and greater degrees of density-dependent 394 emigration and mortality. Hence the degree of spatial resource clumping may modify the 395 strength of the negative power curve of growth versus density. 396 While we found positive instead of negative density-dependent growth, such a 397 finding is common within experimental systems where food is given proportional to 398 density (Brown, Brown & Srivastava 1992). Within natural systems, site-resource levels 399 are assumed to be constant across densities (Kaspersson, Höjesjö & Pedersen 2010) 400 where low-density sites have a greater amount of resource relative to competitor number 401 than high density (Noël, Grant & Carrigan 2005; Ward et al. 2007). Should we have 402 provided resource levels more in line with natural systems, it is conceivable that we 18 403 would have also found negative density dependence with variance remaining higher at 404 high densities. 405 Our findings show the potential for the spatial structure of resource patches, 406 irrespective of other potentially varying characteristics (i.e. resource quality, patch size), 407 to influence population growth through changing the effective number of competitors 408 within the system, and therefore, the degree to which competitive ability (phenotype) 409 confers enhanced fitness. Future work should seek to understand how the spatial 410 distribution of patches varying in quality may interact to influence within-population 411 growth variance and the consequences of such variance for the regulation of natural 412 populations, potentially allowing for further insight as to the decisions, pressures and 413 structuring factors, which underlie patterns of individual distribution within landscapes. 414 415 Acknowledgements 416 We would like to thank the animal care staff at Concordia University who cared for 417 housed fish on a day-to-day basis and István Imre as well as an anonymous referee for 418 comments on the manuscript. This research was supported by an NSERC (Natural 419 Sciences and Engineering Research Council of Canada) grant to PPN. 420 421 Data Accessibility 422 Data available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.7nb71 423 (Jacobson, Grant & Peres-Neto 2015). 424 425 References 19 426 Abrahams, M.V. (1986) Patch choice under perceptual constraints: a cause for departures 427 from an ideal free distribution. Behavioral Ecology and Sociobiology, 19, 409- 428 415. 429 Anderson, W.G., McKinley, R.S. & Colavecchia, M. (1997) The use of clove oil as an 430 anesthetic for rainbow trout and its effects on swimming performance. North 431 American Journal of Fisheries Management, 17, 301-307. 432 Araújo, M.S., Guimarães, P.R.Jr., Svanbäck, R., Pinheiro, A., Guimarães, P., Dos Reis, 433 S.F. & Bolnick, D.I. (2008) Network analysis reveals contrasting effects of 434 intraspecific competition on individual vs. population diets. Ecology, 89, 1981- 435 1993. 436 Bolnick, D.I. (2004) Can intraspecific competition drive disruptive selection? An 437 experimental test in natural populations of sticklebacks. Evolution, 58, 608-618. 438 Borcherding, J. & Magnhagen, C. (2008) Food abundance affects both morphology and 439 440 behaviour of juvenile perch. Ecology of Freshwater Fish, 17, 207-218. Boujard, T., Labbé, L. & Aupérin, B. (2002) Feeding behaviour, energy expenditure and 441 growth of rainbow trout in relation to sotcking density and food accessibility. 442 Aquaculture Research, 33, 1233-1242. 443 444 445 Brown, G.E. & Brown, J.A. (1993) Do kin always make better neighbours?: the effects of territory quality. Behavioral Ecology and Sociobiology, 33, 225-231. Brown, G.E., Brown, J.A. & Srivastava, R.K. (1992) The effect of stocking density on 446 the behaviour of Arctic charr (Salvelinus alpinus L.). Journal of Fish Biology, 41, 447 955-963. 448 Currens, K.P., Sharpe, C.S., Hjort, R., Schreck, C.B. & Li, H.W. (1989) Effects of 20 449 different feeding regimes on the morphometrics of chinook salmon 450 (Oncorhynchus tshawytscha) and rainbow trout (O. mykiss). Copeia, 3, 689-695. 451 Drucker, E.G. & Lauder, G.V. (2003) Function of pectoral fins in rainbow trout: 452 behavioral repertoire and hydrodynamic forces. The Journal of Experimental 453 Biology, 206, 813-826. 454 Edelaar, P., Siepielski, A.M. & Clobert, J. (2008) Matching habitat choice causes directed 455 gene flow: a neglected dimension in evolution and ecology. Evolution, 62, 2462- 456 2472. 457 458 459 460 461 462 Fausch, K.D. (1984) Profitable stream positions for salmonids: relating specific growth rate to net energy gain. Canadian Journal of Zoology, 62, 441-451. Fretwell, S.D. & Lucas, H.L.Jr. (1970) On territorial behavior and other factors influencing habitat distribution in birds. Acta Biotheoretica, 19, 16-36. Gatz, A.J.Jr. (1979) Ecological morphology of freshwater stream fishes. Tulane Studies in Zoology and Botany, 21, 91-124. 463 Gotceitas, V. & Godin, J-G.J. (1992) Effects of location on food delivery and social 464 status on foraging-site selection by juvenile Atlantic salmon. Environmental 465 Biology of Fishes, 35, 291-300. 466 Grand, T.C. & Grant, J.W.A. (1994) Spatial predictability of resources and the ideal free 467 distribution in convict cichlids, Cichlasoma nigrofasciatum. Animal Behaviour, 468 48, 909-919. 469 470 471 Grant, J.W.A. (1993) Whether or not to defend? The influence of resource distribution. Marine Behaviour and Physiology, 23, 137-153. Grant, J.W.A. & Guha, R.T. (1993) Spatial clumping of food increases its 21 472 monopolization and defense by convict cichlids, Cichlasoma nigrofasciatum. 473 Behavioral Ecology, 4, 293-296. 474 475 476 Grant, J.W.A. & Imre, I. (2005) Patterns of density-dependent growth in juvenile streamdwelling salmonids. Journal of Fish Biology, 67, 100-110. Grant, J.W.A. & Kramer, D.L. (1990) Territory size as a predictor of the upper limit to 477 population density of juvenile salmonids in streams. Canadian Journal of 478 Fisheries and Aquatic Sciences, 47, 1724-1737. 479 Grant, J.W.A., Gaboury, C.L. & Levitt, H.L. (2000) Competitor-to-resource ratio, a 480 general formulation of operational sex ratio, as a predictor of competitive 481 aggression in Japanese medaka (Pisces: Oryziidae). Behavioral Ecology, 11, 670- 482 675. 483 Imre, I., Grant, J.W.A. & Cunjak, R.A. (2005) Density-dependent growth of young-of- 484 the-year Atlantic salmon Salmo salar in Catamaran Brook, New Brunswick. 485 Journal of Animal Ecology, 74, 508-516. 486 Imre, I., Grant, J.W.A. & Cunjak, R.A. (2010) Density-dependent growth of young-of- 487 the-year Atlantic salmon (Salmo salar) revisited. Ecology of Freshwater Fish, 19, 488 1-6. 489 Jacobson, B., Grant, J.W.A. & Peres-Neto, P.R. (2015) Data from: The interaction 490 between the spatial distribution of resource patches and population density: 491 consequences for intra-specific growth and morphology. Dryad Digital 492 Repository http://dx.doi.org/10.5061/dryad.7nb71. 493 Jenkins, T.M.Jr., Diehl, S., Kratz, K.W. & Cooper, S.D. (1999) Effects of population 494 density on individual growth of brown trout in streams. Ecology, 80, 941-956. 22 495 Jobling, M. & Baardvik, B.M. (1994) The influence of environmental manipulations on 496 inter- and intra-individual variation in food acquisition and growth performance 497 of Arctic charr, Salvelinus alpinus. Jounal of Fish Biology, 44, 1069-1087. 498 Kaspersson, R., Höjesjö, J. & Pedersen, S. (2010) Effects of density on foraging success 499 and aggression in age-structured groups of brown trout. Animal Behaviour, 79, 500 709-715. 501 502 503 Keeley, E.R. (2001) Demographic responses to food and space competition by juvenile steelhead trout. Ecology, 82, 1247-1259. Keene, J.L., Noakes, D.L.G., Moccia, R.D. & Soto, C.G. (1998) The efficacy of clove oil 504 as an anaesthetic for rainbow trout, Oncorhynchus mykiss (Walbaum). 505 Aquaculture Research, 29, 89-101. 506 Kennedy, M. & Gray, R.D. (1993) Can ecological theory predict the distribution of 507 foraging animals? A critical analysis of experiments on the Ideal Free 508 Distribution. Oikos, 68, 158-166. 509 Kim, J-W. & Grant, J.W.A. (2007) Effects of patch shape and group size on the 510 effectiveness of defence by juvenile convict cichlids. Animal Behaviour, 73, 275- 511 280. 512 Leavy, T.R. & Bonner, T.H. (2009) Relationships among swimming ability, current 513 velocity association, and morphology for freshwater lotic fishes. North American 514 Journal of Fisheries Management, 29, 72-83. 515 Li, H.W. & Brocksen, R.W. (1977) Approaches to the analysis of energetic costs of 516 intraspecific competition for space by rainbow trout (Salmo gairdneri). Journal of 517 Fisheries Biology, 11, 329-341. 23 518 519 520 521 522 Lobón-Cerviá, J. (2010) Density dependence contrains mean growth rate while enhancing individual size variation in stream salmonids. Oecologia, 164, 109-115. Lomnicki, A. (1988) Population ecology of individuals. Princeton University Press, New Jersey. Mason, N.W.H., Lanoiselée, C., Mouillot, D., Irz, P. & Argillier, C. (2007) Functional 523 characters combined with null models reveal inconsistency in mechanisms of 524 species trunover in lacustrine fish communities. Oecologia, 153, 441-452. 525 526 527 Newman, R.M. (1993) A conceptural model for examining density dependence in the growth of stream trout. Ecology of Freshwater Fish, 2, 121-131. Noël, M.V., Grant, J.W.A. & Carrigan, J.G. (2005) Effects of competitor-to-resource 528 ratio on aggression and size variation within groups of convict cichlids. Animal 529 Behaviour, 69, 1157-1163. 530 Ojanguren, A.F. & Braña, F. (2003) Effects of size and morphology on swimming 531 performance in juvenile brown trout (Salmo trutta L.). Ecology of Freshwater 532 Fish, 12, 241-246. 533 534 535 Olsson, J., Svanbäck, R. & Eklöv, P. (2007) Effects of resource level and habitat type on behavioral and morpholoigcal plasticity in Eurasian perch. Oecologia, 152, 48-56. Peres-Neto, P.R. & Magnan, P. (2004) The influence of swimming demand on 536 phenotypic plasticity and morphological integration: a comparison of two 537 polymorphic charr species. Oecologia, 140, 36-45. 538 539 540 Pinheiro, J.C. & Bates, D.M. (2000) Mixed-effects models in S and S-PLUS. Springer, Verlag New York, Inc. Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D. & the R Development Core Team (2012) 24 541 nlme: Linear and nonlinear mixed effects models. R package version 3.1-105. 542 Post, J.R., Parkinson, E.A. & Johnston, N.T. (1999) Density-dependent processes in 543 structured fish populations: interaction strengths in whole-lake experiments. 544 Ecological Monographs, 69, 155-175. 545 Robb, S.E. & Grant, J.W.A. (1998) Interactions between the spatial and temporal 546 clumping of food affect the intensity of aggression in Japanese medaka. Animal 547 Behaviour, 56, 29-34. 548 Rohlf, F.J. (2005a) tpsDIG2. Program for digitizing landmarks and outlines on image 549 files, scanner or video. [windows], version 2. Department of Ecology and 550 Evolution, State University of New York, Stony Brook (available from 551 http://life.bio.sunysb.edu/morph/). 552 Rohlf, F.J. (2005b) tpsRegr. Program for analyzing landmark data by thin-plate splines 553 using multivariate multiple regression of shape. [windows], version 1.31. 554 Department of Ecology and Evolution, State University of New York, Stony 555 Brook (available from http://life.bio.sunysb.edu/morph/). 556 Sheets, H.D. (2004a) CoordGen6. IMP-Integrated Morphometrics Package 557 IMPBasics.exe, version 6a. Coordinate Generating Utility, also file translator 558 to/from X1Y1, TPS file formats. Department of Physics, Casinus College, Buffalo, 559 New York (available from: http://www3.canisius.edu/~sheets/morphsoft.html). 560 Sheets, H.D. (2004b) PCAGen6. IMP-Integrated Morphometrics Package 561 IMPBasics.exe, version 6a. Principal components analysis based on partial warps 562 scores and principal axis scores. Department of Physics, Casinus College, 25 563 Buffalo, New York (available from: 564 http://www3.canisius.edu/~sheets/morphsoft.html). 565 Silver, P., Cooper, J.K., Palmer, M.A. & Davis, E.J. (2000) The arrangement of resources 566 in patchy landscapes: effects on distribution, survival, and resource acquisition of 567 chironomids. Oecologia, 124, 216-224. 568 Smith, T.B. & Skúlason, S. (1996) Evolutionary significance of resource polymorphisms 569 in fishes, amphibians, and birds. Annual Review of Ecology and Systematics, 27, 570 111-133. 571 Syarifuddin, S. & Kramer, D.L. (1996) The effect of group size on space use and 572 aggression at a concentrated food source in blue gouramis, Trichogaster 573 trichopterus (Pisces: Belontiidae). Environmental Biology of Fishes, 46, 289-296. 574 Tregenza, T., Hack, M.A. & Thompson, D.J. (1996) Relative competitive success of 575 unequal competitors changes with overall density. Oikos, 77, 158-162. 576 Ward, D.M., Nislow, K.H. & Folt, C.L. (2009) Increased population density and 577 suppressed prey biomass: relative impacts on juvenile Atlantic salmon growth. 578 Transactions of the American Fisheries Society, 138, 135-143. 579 580 Ward, A.J.W., Webster, M.M & Hart, P.J.B. (2006) Intraspecific food competition in fishes. Fish and Fisheries, 7, 231-261. 581 Ward, D.M., Nislow, K.H., Armstrong, J.D., Einum, S. & Folt, C.L. (2007) Is the shape 582 of the density-growth relationship for stream salmonids evidence for exploitative 583 rather than interference competition? Journal of Animal Ecology, 76, 135-138. 584 Webb, P.W. (1984) Body and fin form and strike tactics of four teleost predators 26 585 attacking fathead minnow (Pimephales promelas) prey. Canadian Journal of 586 Fisheries and Aquatic Sciences, 41, 157-165. 587 588 589 Weir, L.K. & Grant, J.W.A. (2004) The causes of resource monopolization: interaction between resource dispersion and mode of competition. Ethology, 110, 63-74. Wood, J.L.A., Grant, J.W.A. & Belanger, M.H. (2012) Population density and territory 590 size in juvenile rainbow trout, Oncorhynchus mykiss: implications for population 591 regulation. Canadian Journal of Fisheries and Aquatic Sciences, 69, 1121-1128. 592 593 594 Zar, J.H. (2010) Biostatistical analysis, 5th edn. Prentice-Hall, Upper Saddle River, New Jersey, USA. Zelditch, M.L., Swiderski, D.L., Sheets, H.D. & Fink, W.L. (2004) Geometric 595 morphometrics for biologists: a primer. Elsevier Academic Press, New York, 596 USA. 597 598 599 600 601 602 603 604 605 606 607 27 608 Figure legends 609 610 Figure 1. Shape analysis and spatial resource distributions. (a) Young-of-the-year 611 rainbow trout with 22 landmarks used within morphological analyses, arrows denote 612 locations of body and caudal width caliper measurements. (b) Photo and schematics of 613 spatial resource distributions tested: distributed (D), semi-clumped (S) and clumped (C). 614 Flow denotes the direction of water movement through channels, water depth was c. 615 18cm, resource patch depth 1.1cm. All measurements within figure are in centimetre. 616 617 Figure 2. Results from mixed-model ANOVAs of growth outcomes of trials. (a) Growth 618 (Gind) outcome for trials as a function of spatial and density treatments. (b) Variance in 619 growth (Gvar) outcome for trials as a function of spatial and density treatments. Open 620 circles denote high density, filled low-density trials, error bars are 95% confidence 621 intervals. 622 623 Figure 3. Spatial feeding dynamics and aggressive interactions observed within video 624 recordings. (a) Feeding dynamics of individuals throughout the first 15 min of feeding. 625 Shading is based on the number of seconds within 1 min that at least ½ body length of at 626 least one individual was within the resource patch. (b) Results from the mixed-model 627 ANOVA of the number of chases per fish within the first 15 min of feeding as a function 628 of spatial and density treatments. Open circles denote high density, filled low-density 629 trials, error bars are 95% confidence intervals. 28 630 Figure 4. Morphological differences between individuals with negative or positive (less 631 or greater) growth relative to their trial average (Grel) as a function of pre- or post-trial 632 morphology. In order to highlight differences variation is depicted at 10x the observed 633 range. 634 635 Figures 636 637 Figure 1. (a) (b) 28.6 18.5 18.5 57.25 28.6 7.6 37 18.5 S 28.6 74 flow 114.5 18.5 638 D C 28.6 639 640 641 642 29 643 Figure 2. 644 (a) 0.14 645 646 0.12 Growth 647 648 0.10 649 650 0.08 651 0.06 Distributed Distributed Semi-clumped Semi-clumped Clumped Clumped Spatial distribution (b) 652 653 0.10 654 Growth variance 0.08 655 656 0.06 657 0.04 658 659 0.02 660 0.00 Distributed Distributed Semi-clumped Semi-clumped Clumped Clumped 661 Spatial distribution 662 663 664 665 30 666 Figure 3. (a) Low density High density 0 1-15 16-30 31-45 46-60 5min 10min 15min 5min 10min 15min Chases per fish (b) Distributed Semi-clumped Clumped 667 668 669 670 31 671 Figure 4. Negative relativeGrowth growth Negative Relative Positive Positive relative Relative growth Growth Pre-trial Pre-trial Post-trial Post-trial 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 32 688 Tables 689 690 Table 1. F-values from mixed-model ANOVAs regarding the influence of treatment type 691 and morphology on within-trial growth, as well as treatment type on the number of chases 692 per fish observed within feeding recordings. Numbers in brackets are degrees of freedom Table 1 693 Factor Treatment Morphology Space (2) Density (1) SxD (2) Pre-trial (40) Post-trial (40) * p≤0.05 ** p≤0.001 Gind Response Gvar Chases 0.248 6.230* 0.952 9.524* 6.414* 9.959* 695 6.068* 44.841** 696 4.025 Gind Grel 697 2.744** 5.584** 1.628* 3.230** 698 694 699 700 701 702 703 704 705 706 707 708 709 33 710 Table 2. Correlation coefficients (r) from regressions of the relationship between 711 individual growth (Gind) and initial size variables for each spatial resource distribution: 712 distributed (D), semi-clumped (S) and clumped (C). Numbers in brackets are the 713 combined number of individuals in the replicates of that treatment type Table 2 Massi Lengthi BodyWi CaudalWi D (26) Low density S (24) C (22) D (77) 0.368 0.470* 0.323 0.309 0.105 0.169 0.188 0.098 0.746* 0.733 0.688 0.659 0.486* 0.565* 0.405 0.229 High density S (73) C (77) 0.299 0.391** 0.350* 0.262 -0.006* -0.022* -0.003 0.072 * p≤0.05 ** p≤0.001 714 Treatment Morphology Pos 34