Ln\tionnient,il Btolo2y of f-ihes Vol. 9. No 2, pp 103 115. 19x3 1 ( 19x3. Ilr W ItmL Publi\her~. The I4,igue I I ! Grazing responses of tropical freshwater fishes to different scales of variation in their food K e yw o rd s : A 1 g i vor y , H a b i t ;I t c I1 o ice. Lori c;i r i ids , N eo t r o p ica 1 s t rea ni s . Pe r i ph y t on , Sea son a I i t y . I d ea I free distribution Synopsis Grazing fishes in fleotropical streams confront variation in their attached algal food that ranges in scale from differences in quality among algal cells to diffcrences in the primary prodiictii4ty of habitats available to the fishes. Fishes may respond to this \xriation on some scales but not others. For example. loricariid catfish in a Panamanian stream tracked variation in algal productivity among pool habitats very closely. In sunny pools where algae grew about seven times Lister than in shaded pools, loricariids were six to seven times denser. Consequently. growth rates of pre-reproductive Arici.s[iws spiriosirs (the most common species in pools) were similar in pools of different canopies. corresponding to predictions from the 'ideal free distribution' hypothesis. But on a smaller scale. within pools, avoidance of avian and terrestrial predators outweighed foraging considerations. Larger species and size classes avoided water shallower than 20 cm, where (as a result) the only standing crops of attached algae large enough to be measurable by scraping occurred. During the dry season when food was. most limiting. loricariids overlapped more in their substrate use as different species sought cover in common refuges such as logs and root tangles in pools. Seasonal variation in growth rates of pool-dwelling loricariids reflect these constraints. Introduction Fishes foraging in neotropical streams confront considerable temporal and spatial variation in their food. How do they respond to this variation, and to what extent do their feeding responses account for their numbers and distributions in streams? In this paper. I address these qiiestions for fishes that graze periphyton, or attached algae, from stream sub- strates. Here I consider grazers to be animals that eat foods that are small. sessile and line-grained (MncArthur & Levins 1964) in their distribution relative to their consumer. Distinctions between grazing and browsing herbivorous fishes made by Hiatt & Strnsbourg (1000) and Kccnleysidc (1070) are followed here: grazers crop their food so closely that they often ingest bits of the substrate, while browsers nip off vegetation further from the sub- strate and do not ingest appreciable amounts of it. Following Wetzel(l975, p. 390), I use periphyton to refer to microfloral growth on various aquatic substrates. This usage overlaps with the term 'auf- wuchs' as used by some authors (Ruttner 1952, Fryer & lles 1972). Variation in the avaiI:ibility of periphyton for grazing stream fishes occiirs on small. intermediate and large spatial scales. On ii small scale. thcre arc differences in the nutritional value of organisms with- in the periphyton community. On an intermediate scale. within pool or ril.lle habitats of strca111s. a I pi vor o ii s tis h e s e nco 11 n t er both con t i n 11 ou s and discontinuous spatial variation in their food. Gra- dients in periphyton standing crops and community compositions occur with depth. At a given depth', periph1,ton varies on different substrates such as logs, cobbles, sand or mud (Bluni 1956, Round 1964. Hjmes 1970. Whitton 1975). Grazers them- selves create mosaics of patches, even on initially homogeneous substrates. by leaving different sites in various stages of recovery. On a large scale, differences in the density of forest canopy over a stream cause large differences in the primary production of periphyton below. Variation in streams on ii \.et larger scale occurs from the headLvaters to the ri\.er mouths (Cummins 1977, Vannote et al. 1980), but this scale will not be considered here. Heterogeneity from all sources affects the rates and efficiencies with which algi- vor ou s lis hes harvest food s . In this paper, 1 discuss adaptations and con- straints of grazing fishes that affect their responses to small. intermediate and large scale variation in periph1.ton food. For large and intermediate scales, 1 emphasize observations on armored catfish (Lori- cariidae) that graze periph1.ton in ;t Panamanian stream. Small scale variation Periph!.ton communities in srreams commonly in- clude diatoms and encrustink or filamentous blue- green, green and red algae (Round 1965, Hynes 1970). Our knowledge of the nutritional qualities of freshwater algae for fishes is still sketchy, but considerable variation has been uncovered in studies of algal grazing by invertebrates (Porter 1977). Herbi\ores are often limited by the quality, in particular the protein content, of their food (Westoby 1974. White 197s). The protein content of thirteen species of freshwater algae ranged from IO to 46x, of the dry weight, with the highest protein concentrations occurring in €irgIiw/ and three bluegreen algae, Attdwotzu, Mic*roc:i:sri.v and Ai'hrrrii_orrtc.rior, (Boyd 1973). Interestingly. the protein content 01' these algae (42 to 46:,,) roughly coincides with optima I prot ei n concen t rii t ions for catfish juveniles grown on artificial diets. Dupree 81 Sneed ( 1966) grew channel catfish, lc/c/Ii/ri/s pi~ric~~urus, on diets of constant caloric content. but with protein levels ranging from 12 to 52",,. Catfish gained weight faster as diet protein increased until it was about 40",,, after which weight gain decreased. Digestion of protein entails higher metabolic costs than digestion of fats or carbohydrates, and the decreased growth on extremely rich protein diets in processing these foods (discussed in Boyd & Goodyear 197 I ). But high protein content in bluegreen algae may be offset by toxins (Gentile 1971, Willorighby 1977) or by cell walls that resist digestion (Fish 1951, Porter 1977, but see Moriarty & Moriarty 1973). Dussault 81 Kramer (198 1 ) reared guppies, Poccilirr r-rticwk(itn, on pure diets of the green alga Ch/or~~- couwit and the filamentous bluegreen, Oc~logo- /iii/m. The fish grew and matured on the first diet. but not on the second. Bluegreen algae have also been found inferior to other types of algae as foods for a number of invertebrates (Calow 1975, Porter 1977). Algivorous fishes can break down cell walls in three ways: mechanical grindins, acid lysis or cellu- lase enzymes derived from gut microflora. Mullets (Miigil spp.) and some herbivorous reef fishes have thick-walled. gizzard-like portions of their stom- achs where they use ingested mineral particles to c grind cell walls of algae and bacteria (Hiatt 81 Strasbourg 1960, Odum 1970, Payne -I 978, Ogden & Lobel 197%). Tilupin nilotiru and Hrrp1oc~lir.oiiti.v nigripirinis, cichlids that feed on bluegreen algae, can lower the pH of their stomachs to about 1.4, at which point bluegreen cell walls are completely lysed (Moriarty & Moriarty 1973). Acid secretion in the stomachs of these cichlids, however, begins only when ingestion starts at dawn. Stomach pH is not at its minimum irntil late morning, so four to six hours of feeding elapse before assimilation of the ingested algae is up to its maximum efficiency of 70 to go(:,, (Moriarty & Moriarty 1973). Cellulnsc enzymes which lyse bluegreen cell walls have never was presumably due to increased met, 'I b 0 I' IC costs been found in \.crtebrates unless associated with microbes or in\pertebrates in the gut (Pre.is & Blaszczyk 1977). Cellulase activity, however, has been found in the gilts of' several fishes (Stickncy & Shumwoy 1974. Prejs & Blnszczyk 1977, Niedcr- holzer & Hofer 1977). Channel catfish, /ctdirrirs prricfufiis. had cellulase activities in their gilts until they were exposed to streptomycin, indicating that enzymes were derived from gut microbes (Stickncy & Shumway 1974). A congener. /. richrlosirs, per- haps assisted by gut niicrollora, was able to digest and assimilate the bluegt-een alga Aric~hrrc~tr ,flo.v- rrqircrc~ with 67.5"~,, efficiency, while the green alga Spirogj*r~r was only 23.7",, assimilated (Gunn et ill. 1977). The high xssiniilation of the protein rich bluegreen alga suggests that bluegreens are poten- tially important foods for these catfish, as they make up large portions of the gut contents of catfish from natural habitats (Gunn et al. 1977). Striped mullet, Mirgil wp11~rlir.v. have long coiled intestines up to five tinies the length oftheir bodies. which contain flagellates that may digest cellulose (Odum 1970). Relatively long intestines are also found in loricariids. I ti loricariid feces. intact blue- green and green algal filaments that appear \,iable are common, while nearly all diatom frustrules ;ire empty, suggesting that the latter ;ire more digestible for loricariids (personal observations). In summary. nutritional quality of algae for algivorous fishes depends both on properties of the algae (including the medium in which they have grown, Spoehr & Milner 1949, Gerloff & Skoog 19541, and on adaptations of the fishes. Sufficient variation in nutritional values of components of periphyton exists to suggest that selective ingestion might be advantageous for algae-grazing fishes. On the other hand. algivores like other herbivores might depend on different components of their diets for different nutrients (Westoby 1974) or need to restrict their intake of different toxins in particular foods ( Freeland & Janzen 1974), and therefore might require ;I mixed diet. The ad win t age for a 1g;ie-gt-m i n g li shes i n d i scI i m i - nuting among the foods they inscst is ;in issue scpara~c l'ium their ability to do so. Some herbi- vo 1-0 i I s d ;i m sc I I'i she s brow sc m ;I c 1-c~ I pic I'rom s 11 b- strutcs on marine recl.s selectively (Lassuy 1980) while others ti-ed iinselcctively (Montgomery 1980). I3ut selective ingestion 01' items sccms less li.;isiblc lor fishes grazing niicroalgx only IO'S or IOO's 01' microns in diiitiietcr, given that the lishes' mo1;ths ;ire scvcral niillimetct-x or ccntimctcrs u idc. One 01' the ni ore delicately hro\\,si ii~ ;i I2 i \,oroiis I'i shcx. Poccilirr ivticii/(r/ci. was observed by Ilussault Kr Kratner (1981), who used cinematogriiphy. With each bite. the guppies ate 3mni2, an area equi- va!ent to about IO4 diatoms of' typical size lain flat in a layer one cell thick. Selective ingestion of organisms I'roni algal mats is probably more difficult for grazers than for browsers. Mouths of many bottom grazing fishes have thick, fleshy, often suctorial lips, iinsuited for selective nipping. Such mouths occur in loricariid catfish, prochilodontid characins (Roberts 1973). grazing cyprinids such 11s the African Lrrhco, and C;!.ririoc.liciliis. the mountain carp of Borneo and Thailand (Hora 1933). In addition. the mot-pholo_ey of loricarids is such that they cannot see the sub- strate on which they are grazing (Fig. I ). The ability of algivorous fishes to select certain components of periphyton must depend on the spatial distribution of these components. Hfrplo- cAror1lis guoritheri, an African 'mbuna' cichlid from Lake Malawi, selectively nips off' long filaments of green and bluegreen algae which protrude from the algal felt on rock substrates (Fryer 1959). Snails. Plunorbis coriforfirs, feeding with radulas selec- tively ingest bacteria from detritus on stones (Calow 1974, 1975). The ability of the snails to select bacteria may depend both on clumping by the bacteria and on chemical sensing of' these aggregates by the snails (Calow 1974). Physical en vi ron men t a I fact or s ni a y in 11 LI en ce c I 11 in pi n g by algae. In culture, ii bluegreen, ,.lricih(rorirr (:\,/i/ic/i.i(u. develops gelatinous walls which caiise I'ilamcnts to clump when the ;ilgu is incubated in \\:iter with sodium: calcium ratios similar to those in ;I Kenyan I'ishes could more li-asibly sclcc~ or reject this alga in such cn\,ironnicnts. I I' a tt iichcd iilg:il communi t ics are st r;i t il'icd. sod:\ l;ikc ( Fry~t- 6i IICS 1972). Perhiil?< grazing 107 the depth at which l'ish penetrate while feeding ('grazing severity'. sensu Alcock 1964) would tiffect the composition 01' their diet. Gastropods grazing ni ;I r i ne ni ic r oa I gac se Icc t i ve I y re ni o ve t h osc diatoms n.hich are more loosely adhered and superficially positioned in the algal mat (Nicotri 1977). Con- ceivubly, il' grazing severity were increased during periods 01' relative food shortage. grazer diets might inclii'de ii wider range of niicrollora. This would be consistent with predictions from the foraging theory developed for animals that conscime coarse-grained prey (MacArthur & Pianka 1966, Eiiilen 1966, Schoener 197 I ). Because of the information just presented it is likely that a grazing fish's diet depends largely on its choice of feeding sites. because the composition of at t ached a Iga I com ni ii nit ies cha nges wit h posit ion and substrate in streams. Some examples of varia- tion of periphyton on this intermediate spatial scale are considered next. Intermediate scale variation Different physical and chemical conditions t"i\or different algae. Therefore, the compositions of' periphyton communities in streams vary with depth, exposure to current and substrate (Bliim 1956. Round 1964, Hynes 1970, Whitton 1975). But to algivorous fishes. the physical features that govern their access to periphyton in various microhabitats may be more important than periphyton composi- tion. These factors include substrate rugosity, sedi- mentation or currents that affect energy expended by fish while grazing, and cover available from predators at or near the grazing site. (By cover, I mean any factor that reduces vulnerability to pre- dators, for example, physical shelter or sufficient depth of water to impede terrestrial or avian pre- dators.) The richest standing crops of periphyton often occiir at the shallow murgins of streams and in shallow riflles. where exposure to light and nutrient iliixes are maximal, and where habitats first inter- cept nutrients wnshed in 1.1-om the land. But algi- voroi~s fishes, pal-ticlllurly after they have groun large. do not occiir in higher densities in shallow w a t e r. w lie re food a vii i I ;i bi I i t y is p re sii ni ;i bl y greater. Larger lislies generally stay in deeper water (Hellier 1962. Bowen 1979. [>e Silva & Silva 1979, Power 1983a). probably to avoid avian and terrestrial predators. For example, algae-grazing catfish (Loricariidne) in the Rio Frijoles of Central Panama are vulnerable to herons and kingfishers that commonly fish in water less than 20 cni deep. After loricariids grow too large to hide under cobbles in riffles, they almost never occiir in water shallower than 20 cm. although they do niove into shallower water by night (Power 1983a). To qiiantify periphyton availability as ;I i'unction of depth, I scraped samples from cobbles collected during March, near the end of the dry season. In water less than 10 cm deep, the median standing crop was 0.68 mg ash free dry weight (afdw) cni-' (range = 0.27 to 1.50. N = 17). In water 10 to 20 cni deep, the median standing crop w;is 0.18 nig ai'dw cmP2 (range = 0 to 0.26, N = 9). In water deeper than 20 cni. I could not scrape siifficient periphyton from anywhere in the stream channel within a 3 km reach to be measurable ivithin the errors of niy technique. Larger loricariids were clearly food-limited during the dry season, when they stopped growing and sometimes lost weight (Power 1981, and see below). But they did not venture into water shallower than 20 cm to graze, even durins the end of the dry season Lvlien food was most limiting in deeper water. Roberts (1972) has pointed out that algivorous fishes, whose gut contents appear indistinguishable, may still be feeding from different substrates or in different microhabitats. To study the distri- bution of Rio Frijoles loricariids over various siibstrates and depth intervals, I mapped a 3 km reach of the stream. During the dry seiison, this reach was surveyed with ;I hand level. In the ii ps t rea ni k i Io ni e t er . cr oss-s t ream t 1-21 n sect s were nindc at roughly 9 m intervals. 01- wherever it niiijor change in the stream's path or depth occiii-red. A 1 os meter tape \\:IS siretched ;ICI-OSS the stream and sul'ficient depth measurements made to constriict ;I bathymetric map with IO ctii contour inter\.aIs. The stream banks were also mapped up to contour lines 40 cm abo\ze the dry season water lei-el so that the map could be used during the higher stream stages of the rainy season. The lower 7 km were mapped in less detail. by sur\.eyins up the thalweg (i.e. deepest part) of the stream and noting only the positions of the water`s edge and the points 40 cm abo\,e dry season water level. After a base niap n-as dra\\ n from these measurements, details were sketched in from spot me;isiirenietits made in topographically complex areas. Throughout the 3 km reach, sub- strate type w;is also mapped. Within the mapped 3 kni reach. 150 quadrats. 1 nil in area. \yere nailed into the streambed and located with respect to nearby permanent vegetation. Loricariids in these qiiadt-ats Lvere censused by day and night by snorkelling (Pou.er 198 1 ), and substrates on ivhich the \.arious species occurred were noted. Mud, sand. pebbles, cobbles, leaf mats. wood. submerged grasses. consolidated clay banks and bedrock platforms were available as substrates to loricariids in the Rio Frijoles. Bedrock and wwod substrates were typical in pools along the outer \valls of meander bends where the stream had been turned by bedrock formations or trees. Pebbles and cobbles were the predominant substrates in riffles. Leaf mats and mud accumulated in the deeper, slower uater during the dry season. but were Ilushed away ditring the rainy season. The relati\.e area of substrates in the 3 km reach of the Rio Frijoles during the dry and the rainy seasons are shown in Figure 7 along with the average propor- tions of loricariids of each of the four species sighted on ii gi\.en substrate. All size classes for a given species have been pooled. Substrate use by different loricariid size classes within a species will be discussed elsewhere (Power I98 1 and in prepara- tion). The most common pool-dwelling 1oric;iriid in the Rio ti-i-joles, ,4/1(~i~/vi~.s, strongly preferred wood ;is ;I substrate in both the dry and the rainy season (see ti$. 9). Their high proportional occiii-rence on wood I-clativc to its iivailability resulted in high densities 01' ,4//(,i.~/r//.s in root tangles anti deadfalls in thc stream. This species also commonly grazed on bedrock and clay substrates in pools. Alpe were relatively accessible to wide-mouthed grazers such 21s Aiicistr//.s on these flat substrates, which also offered solid support. energetically important to loricariids whose adaptations for benthic life in- cl tide reduced swim bladders. Auci,s/ri/.s commonly rested, but did not graze, on sandy floors of deep pools. In the dry season, their apparent preference for sand, coupled with increased avoidance of pebbles. resulted from the concentration of these fish in the deeper portions of pools as the water level dropped. Hjpo.sror~~i.s ( = Plwo.sto/ui/.s) rested by day in pools, under ledges or on the sandy bottoms. By night, they \,entiired out to graze cobble substrates in the deeper portions of riffles. In the dry season, 0.4 0.2 O.! 0.2 0.4 0.2 0.4 0.2 G WMCyLvB S PCb Dry season Rainy season O n 1 rnn G W M CyLvS B P Cb Substrate type Propor!iona' area of substrate Proportion of loricariid sightings [J Fll. somatic growth rates in dil'l'ercnt pools (Fig. 6, top). In the rainy season. these growth rates (g per 100 days) were (mean &SE, (N)) 2.7hO.X (16), 2.Xh0.3 (63) and 3.1 f0.7 ( 17) in shaded, half-shaded and sunny pools, respectively. In the dry season, growth rates of pre-reproditctive Aiicisrri~s were 0.2 * 0.2 (23). -0. I &0.2 (24), and 0.0&0.3 (19) in shaded, hu 11'-sh ad ed ;I nd sunny pools. Di fferences among growth rates of pre-reproductive A/icistri/.r in dit- ferent pools were not statistically distinguishable (Power 1983~). This patlern, in which food availability was simil'ir in sunny crowded habitats and in shaded, sparsely populated pools, corresponds to the 'ideal free distribution' predicted by Fretwell ( 1972) and others (Royama 1970) for animals that meet two assumptions. First, if animals have adequate knowledge of the relative qualities of habitats in their environments, and second, if they are free to settle in the best available at any time, then in a relatively empty environment they should settle in the best habitat first, until it is degraded by in- creased density. When the quality of this site decreases to that of the next best habitat, animals should begin to colonize the second site until it is no better than the third best. As density increases, all available habitats fill in proportion to some balance between their intrinsic quality (reflected by rates of periphyton production for Rio Frijoles loricariids) and degradation of habitat due to competition among residents. Individuals in different habitats should have equal litnesses, which may be indexed by some combination of growth rates, survivorship. and reproductive success. For pre-reproductive All- ci.sir~/.r, food availability may have been the major factor affecting fitness in different habitats, its siirvivorship was similar for fish in different pools (Power 1983~). Such ;I pattern could only be maintained in the Rio Frijoles if loricnriids moved among pools in response to changes in local food availability. The stream was a dynamic environment, and changes were sometimes abrupt, ;IS when trees fell and opened canopies over pools. or when large amounts ol' sediment werc redistributed diiring Iloods and pools were created or filled. Loricariids responded quickly to such changes. For example, a new pool I13 created during ;I large llood was colonized by loricariids within ;I few weeks 01' its formation, and within two month5 it had loricariid densities typical of other pools with its canopy cover (Power 1983~). In addition to abrupt physical changes, more grad- ual demographic changes such as increased crowd- ing during the dry season occurred. Although most 01' the 1306 loricariids I marked over ;I two year period that were resighted were in their home pools, the fish apparentlj, sampled other pools sufliciently to take rapid ad\,antage of new feeding opportuni- ties. By redistributing themselves to track changing food availability among habitats, loricariids damped incipient pool-to-pool variation in peri- phyton standing crop as it arose. Conclusions Fishes grazing periphyton in neotropical streams respond to variation in their food on some scales but not on others. On a small scale, because periphyton is so fine-grained relative to the mouths of grazing fishes. it seems iinlikely that they could selectively ingest superior components unless these were clumped, stratified or otherwise spatially segre- gated. On an intermediate scale, within habitats, avoidance of predators appears to outweigh foraging considerations in the choice of depth strata by pool-dwelling loricariids in the Rio Frijoles. Similar- ly, large detritivorous cichlids. Tilcrpicr ( = Scrrotilero- dori) riio.s.vcrnihic.cr, in Lake Sibaya, South Africa also stay in deeper u'ater. where detritus is low in protein. Consequently, these cichlids are nutritio- nally stunted as adults, although not as juveniles when they feed in shallower areas where detritus is higher in protein (Bowen 1979). Proximity to cover also is an important factor in substrate choice by grazing loricariids. and during the dry season is more important than interspecific competition for food in the choice of grazing substrates by the two most common pool-dwelling species. On ii large scale. however, loricariids track pool to pool variation in periphyton availability very closely. Intrinsic rates of periphyton production in pools ;ire balanced by the collective depletion rates of resident loricariids so that food availability. h c ncc grow t h r ;i t es o 1' p re- re p rod 11 c t i \,e '-1 11 .i.s I 1.1 1,s. ;I rc si m i la r i n d i ffercn t poo I s. Dcspi t e de m ogrn p h i c, seasonal and occasional nearly catastrophic iluc- tuations in pool habitats, loricariids remain sul- I'ic i en t I y ;I w a re o f t heir en vi ro ti ni e n t to con t i ti it ;i I I y re-evaluate feeding opportunities in their home pools relative to others. By responding quickly. u,ithin weeks. to ynriation that arises on this large scale, these grazers act to respond rapidly so that the long-term a\xilability of food remains hirly constant from pool to pool in an otherwise rela- tively dynamic en\'ironnient. Acknowledgements I thank Don Kranier. Don Stewart, Bill Dietrich, Tom Zaret, Jim Karr and Paul Angernieiei- for their helpfill comments on various Lei-sions of this manuscript. I ani also grateful to Bob Paine and Gus van Vliet at the University of Washington for encouragement and useful discussions. This work n as supported by a Bacon Fellowship from the Smithsonian Institution and an NSF Doctoral Dissertation Improvement ALvard (6 1-7455). I thank the Srnithsonian Tropical Research Institute, Panama. for their generous support. References cited Alcock. M.B. 1964. The physiological significance of defoliation on the subsequent regrowth of grass-clowr mixtures and cereals. pp. 2541. hi: D.1. Crisp (ed.) Grazing in Terrestrial and Marine Environments. Blackwell. Oxford. Blum, .I.L. 1956. The ecology of river algae. Bot. Rev. 22: 29 1-34 1. Bowen. S.H. 1979. A nutritional constraint on detriti\.ory in fishes: the stunted populations of Srrr-otl~rrodo~i ~~~o,s.ra,,rhic~l~.s in Lake Sibaya. South Africa. Ecol. Monogr. 49: 17-31. Bo1.d. C.E. 1973. Amino acid conipobition of I'reshu;itcr alye. Arch. Hydrobiol. 73: 1-9. Boyd. C.F. & C.P. Goodyear. 1971. Nutritive quality of food in ecological systems. Arch. Hydrobiol. 69: 256-270. Calow. P. 1974. E\ idence for bacterial feeding in f'/u/rorhi.\ um/or-fi/.\ Linn. Proc. rnalacol. Soc. Lond. 4: I45 156. Calow, P. 1975. The feeding strategies of two freshuater gastro- pods. hq'/lf.\ //rrrir/i/i.\ M iill and P/mor.hi.\ w///or/rr.\ Linn. (Pulnionata) in terms of ingestion riitcs and Libwrption eflicicncies. Oecologia 20: 33 49. 1 I4 Cumniins, K.W. 1977. From headwater streams to riterh,. Americon Biology Teacher 39: ?05-~3 12. De Silu. S.S. & E.I.L. Silva. 1979. Biology of young grey mullet. Mir,qi/ c.cy/itr/i,.\ L.. populations in a coastol lagoon in Sri Lankx. J. Fish Bid. I: 9 20. Dupree. H.K. & K.E. Sneed. 1966. Response of channel catfish fingerlings to different levels of major nutrients in purified diets. Tech. Paper 9 of the Bureau of Sport Fisheries and Wildlife. U.S. Gov. Printing Office: 1-2 I. Dussault, G.V. & D.L. Kramer. 19x1. Food and feeding be- haviour of the guppy. Poociliir wr;cdc~i[~ (Pisces: Poeciliidae). Can. J. Zool. 59: 684-701. Emlen. J.M. 1966. The role of time and energy in food preference. Amer. Natur. 100: 61 1417. Fish, G.R. 1951. Digestion in Tilupio cwdcwrrr. Nature 167: 900. Freeland. W.J. & D.H. Janzen. 1974. Strategies in herbivory by mammals: the role of plant secondary compounds. Amer. Natur. 108: 269-289. Fretwell. S.D. 1972. Populations in a seasonal en\ironment. Princeton University Press. Princeton. 21 7 pp. Fryer, G. 1959. The trophic interrelationships and ecology of some littoral communities of L. Nyasa \\ ith especial reference to the fishes. and a discussion of the e\olution of a group of rock-frequenting Cichlidae. Proc. 2001. SOC. Lond. 132: 153-28 I. Fryer, G. & T.D. Iles. 1972. The cichlid fishes of the Great Lakes of Africa. Oliver and Boyd, Edinburgh 641 pp. Gentile. J.H. 1971. Blue-green and green algal toxins pp. 27.~66. Iri: S. Kadis. A. Ciegler & S. Ajl (ed.) Microbial Toxins, 1.01. 7. Academic Press, New York. Gerloff, G.C. & F. Skoog. 1954. Cell contents of nitrogen and phosphorus as H measure of their availability for growth of M ic.rocp 1i.s ocwgi/ro.\o. Eco I ogy 3 5: 348-3 53. Goulding, M. 1980. The fishes and the forest. Univ. Calif. Press. Berkeley. 280 pp. Gunn. J.M., S.U. Qadri B D.C. Mortimer. 1977. Filamentous algae as a food soiirce for the brown bullhead. Imlurrrc /ich/o.ws. J. Fish. Res. Board Can. 34: 396401. Hellier. T.R. 1962. Fish production and biomass in relation to photosynthesis in the Laguna Madre of Texas. Pub. Inst. Marine Sci. liniv. Texas 8: 1-22. Hiatt. R.W. & D.W. Strasburg. 1960. Ecological relationships of the fish fauna on coral reefs of the Marshall Islands. Ecol. Monogr. 30: 65-127. Hora. S.L. 1933. Respiration in tishes. J. Bombay Natural History Society 36: 538--560. Hunter, R.F. 1964. Home range behavior in hill sheep. pp. 155-1 71. //I: D.J. Crisp (ed.) Grazing in Terrestrial and Marine Environments. Blackwell. Oxford. Hynes. H.B.N. 1970. The ecology of running waters. Uni\.. Torofito Press. Toronto. 555 pp. Kawanabe. H. 1969. The signilicance of social structure in production of the 'Ayii'. P/c~tri~lo.ssr~.s (i//iidi.s pp. 243-25 I . //I: T.G. Northcote (ed.) Symposium on Salmon and Trout in Streams, U.B.C. Fihh. Res. Inst., Vancouver. Keenleyside. M.H.A. 1979. Diversity and adaptation in lish behaviour. Springer-Vcrlag. New York. 20X pp. lirmer; l).L. 197X. ticprotlucti\e seasonallt) 111 the flshch 01. 21 tropical stream. Ecolopy 59: 976- 9x5. 1.a.ssuy. I). ti. l9XO. Elli.cts 01' 'farming' behat ior b> f'i/piI- ~~CC/II~K\ /iridm and ll~~iiii~~li~~~/~i~loil~i~i /'/[lgioriii,/ii/)i,rr on algal community strtic:trre. 13ti11. Mar. Sei. 30: 304 3 I?. Lowe-McConnell. R.H. 1964. The tishes of the Rupuni sa\nnna district of British Guiana. South America. I. Elcolopical groupings of the lish specie:, and the effects of the season;il Lowe-McConnell. R.H. 1967. Some factors aIf'ecting fish popu- lations in Amazonian waters. Atas Simp. ii Biota Amazonic;~ 7: 177-186. Lowe-McConnell. R.H. 1975. Fish commiinities in tropical freshwaters. Longman. London. 337 pp. MacArthur. R.H. S: R. Levins. 1964. Competition. habitat selection. and character displacement in ii patchy en\iron- ment. Proc. Nat. Aciid. Sci. USA 51: 1907-1210. MacArthur. K.H. & E.K. Pianka. 1966. On optimal usc of ;I patchy environment. Amer. Natiir. 100: 603-609. Montgomery. W.L. 1980. The impact ofnon-selective grazing by t he g ia n t bl tie damsel fish . Rl/c.i.o.\/)r/r/iotlo,i dorso/i.\ . on ;I lga I coninitinities in the Gulf of California. Mexico. Bull. Mar. Sci. 30: 290-303. Moriarty. D.J.W. 1973. The physiology of digestion of blue- green algae in the cichlid fish. Ti/qiiu ui/o/ii~/. J. Zool. Lond. 171: 25-39. Moriarty. D.J.W.. J.P.E.C. Darlington. I.G. Dum. C.M. Moriarty & M.P. Te\lin. 1973. Feeding and grazing in Lake George. Uganda. Proc. L. Soc. Lond. B. 184: 299-319. h,loriarty. D.J.W. & C.M. Moriarty. 1973. The assimilation of carbon from phytoplankton by t Wo herbivorous lishes. Tili//,ic/ /ri/o/icu and ~l[//~/iii./ir(jii/i.\ /ii~qiY/)iiii/is. J. Zool. Lond. 171: 41-55. Nicotri. M.E. 1977. Grazing effects of four marine intertidal herbivores on the microtlora. Ecology 58: 1020- 1032. Niederholzer, R. & R. Hofer. 1979. The adaptation ofdizestile enzymes to temperature. season and diet in the roach. R/f/i///.\ ru/i//is L. and rudd. S~.(/ri/i/iiu.s ~,/.!,//i/.[~/i/i//ii///ii//~ L. I. Cellu- lase. J. Fish Biol. 15: 41 1-416. Odum. W.E. 1970. Utiliziition of the direct grazing and plant detritus food chains by the striped mullet hfi/gi/ e,c/)/io///.\. pp. 222.~240. //I: J.H. Steele (ed.) Marine Food Chains. Oher and Boyd. Edinburg. Ogden, J.C. & P.S. Lobel. 197X. The role 01' herbivorous lishes and urchins in coral rcef communities. Env. I3iol. Fish. 3: 49 -63. Payne. A.I. 1978. Gut pH and digestive strategic> in cstuarine grey mullet (Mugilid;ic) ;ind tilapia (Cichlidae). .I. Fish Binl. 13: 627 630. Porter, K.G. 1977. The plant-animal intertiice in tl-eshwatcr ecosystems. Anier. Sci. 65: I59 170. Power. M.E. 19x1. Thc grazing ccology (11' armored catli.sh (Loricariidac) in ;I Pxiain:ini;in slrcam. P1i.D. Thcsis. Univ. W;ishington. Sc;ittle. IhX pp. Power. M.E. I9Xh. 1)cpth-distrihirtions of armored catfi\h: le on fish. J. Linn. Soc. Zool. 45: 103- 144. ! I IS predator-induced resource avoidance? Ecology (in print). Power, M.E. 1983b. The importance of sediment in the grazing ecology and social interactions of an armored catfish. Ami.c.irir.\ .~pi/~osr/s. Env. Biol. Fish. (in print). Power. M.E. 1.983~. Habitat quality and the distribution of algae-grazing catfish in a Panamanian stream. .I. Anim. Ecol. (in print). Prejs. A. & M. Blaszczyk. 1977. Relationship between food and cellulase activity in fresh\\ater fishes. I. Fish Bid. I I: 447 -452. Roberts. T.R. 1972. Ecology of tishes in the Amazon and Congo basins. Bull. Mus. Comp. Zool. Haw. 143: 117-147. Roberts. T.R. 1973. Osteology and relationships of the Prochilo- dontidae, a South American taniily of characoid lishes. Bull. Mus. Comp. Zool. Harv. 145: 213--?35. Round. F.E. 1964. The ecology of benthic algae. pp. 138-84. hr: D.F. Jackson (ed.) Algae and Man. Plenum Press. New York. Round, F.E. 1965. The biology of the algae. Arnold. London. 269 pp. Royama. T. 1970. Evolutionary significance of predators re- sponse to local differences in prey density: a theoretical study. Proc. Adv. Study Dynamics Numbers Popiil. (Oosterbeek): 344-357. Ruttner. F. 1957. Fundamentals of limnology. Uni\. Toronto Press, Toronto. 295 pp. Schoener. T.W. 1971. Theory of feeding strategies. Ann. Rev. Ecol. syst. 2: .x19 404. Spoehr. t1.A. B H.W. hlilncr. IV4V. The chemic;il composition of Clrlordlir: effect of environnicntiil conditions. Pliint Physiol. 34: 120 14Y. Stickney. K.R. & S.E. Shumway. 1974. Occui-1-encc ol'cellul~ix activity in the stomachs 01' fihhch. I. }Fish Biol. 447 452. Stinison. I. 1V73. The role ol'the territory iii the ecolog! 01' the intertidal limpet. f.oi!iu gi,yu/r/iw (Gray). Ecolofy 54: 1020 1030. Vannote. R.L.. G.U'. Minshall. K.W. Cummiti\. I.R. Scdell & C.E. Cushing. 1980. The ri\er contintitin1 concept. Can. I. Fish. Aquat. Sci. 37: 130- 1.37. Westoby. M. 1974, An anal!sis 01' diet selection by large generalist herbi\ores. Anicr. Natur. IOX: 290 -304. Wetzel. R.G. 1975. Limnolog. Saunders. Philadelphia. 743 pp. White. T.C.R. 1978. The importance of relative shortage of food in anininl ecolog!. Oecolosia 33: 71 XS. Whitton. B.A. 1975. Alfae. pp. XI- 105. I//: 13.A. Whitton (ed.) River Ecology. L nit. Calif. Press. Berkeley. Willoughby. L.G. IY77. FreshIbater Riolofy. Pica Press. New York. Zaret, T.M. 6: A.S. Rand. 1971. Competition in tropical stream fishes: support for the competetive exclusion principle. Ecolo_gy 52: 3-36-34?, I