CHAPTER Life Cycles, Limiting Factors, and Behavioral Ecology of Four Loricariid Catfishes in a Panamanian Stream Mary E. Power* The Loricariidae, a Neotropical family of armored catfish, is the fifth most diverse fish family in the world (Nelson, 1994), with more than 600 species (possibly more than 800 species; Scott Schaefer, pers. comm.). Loricariids occur in small streams to large rivers from southern South America north to Panama (Burgess, 1989; Buck and Sazima, 1995). Loricariid populations have also naturalized in a few warm-water Habitats in North America where they have been introduced by humans (e.g., Barron 1964; Nico and Martin 2001). Ichthyologists and aquarists have long been fascinated by this group but relatively few long-term studies have documented their lives in nature. In this chapter, I summarize field observations on the behavior and natural history of loricariids near the northern boundary of the family?s *Department of Integrative Biology, University of California, Berkeley, Berkeley, CA 94720- 3140, USA. 582 Catfishes natural distribution in the Rio Frijoles (9'9' N, 79'44'W). This stream drains secondary tropical rain forest of the Parque National Soberania in central Panama. The Rio Frijoles was once a tributary of the Chagres River but after construction of the Panama Canal, now empties into Lake Gatun. Four loricariids occur in streams of the Parque National Soberania: Ancistrus spinosus (Eigenmann et Eigenmann) , Hypostomus (Plecostomus) plecostornus (Linnaeus) , Rineloricuriu (Loricaria) urucunathu Kner et Steindachner, and Chaetostornus fischeri Steindachner. Most of my observations are on Ancistrus spinosus, the most common loricariid in deeper stream pools of the Rio Frijoles. Here, I review field observations from a 28-month field study, describing loricariid responses to variation in their algal food, threats from various types of predators, physical stress from deposited fine sediments, and seasonal expansion and contraction of their habitat. The factors that limit these catfish vary as fish grow from eggs to juveniles to subadults and reproductive adults, and also seasonally, with hydrologic contractions and expansions of their habitat. Their responses to heterogeneity and change in their environments are constrained by their adaptations and may well influence their fates as Neotropical streams change under increasing impact from human land use. I I I , I MORPHOLOGICAL EQUIPMENT Loricariids are not only speciose, but remarkable for their aspect diversity (sensu Rand, 1967). Body forms range from stout and compact to thin and including bizarre, often sexually dimorphic spines, tentacles, and fin filaments, whose functions are largely unknown. Nonetheless, loricariids share characters that bespeak their benthic habits. Dermal bony plates armor most of the body except for the ventral belly. Swim bladders are comblike tooth plates with which they scrape food from surfaces. They also cling with their mouths, resting upside down under ledges or logs, or locomoting up waterfalls (Gradwell, 197 1; Grzimek, 1984). Opercles of loricariids are small and entirely ventral (Fig. 19. I), protecting the Gills from nipping characins or poeciliids, but possibly exposing them to fouling or abrasion on sandy or silty surfaces. Several species of loricariids can use their digestive tracts as accessory air-breathing organs under hypoxia I I elongate. Certain species have striking color patterns and appendages reduced or absent. Their entirely ventral suctorial mouths have four i Maw E. Power 583 - Fig. 19.1 Small ventral opercle slit of a male Ancistrus spinosus (Gradwell, 107 I; Gee, 1976; Kranier and Graham 1976), or in acidic or H;S-rich waters (Brauncr ct al., 1995; Armbrustcr, 1998). REPRODUCTION Many loricariids exhihit inale parental care of eggs and carlv fry. While males of some species carry eggs under large flaps of thcir lower lip, iiiost loricariid fathers guard eggs and hatchlings in protected nest cavities. Young Rineloricaricz urucmtfiu aiid Ancistrus spitwstts bcgin their livcs as eggs in their father's nests (Fig. 19.2; Power, 1981; Moodie and Power, 1982). Suitable nest sites appear to be in short supply for both species arid may limit recruitment in rhe Rio Frijoles. Most of the Ancistrus nests 1 observed were in hollow logs, although I found one male Ancistrus, snialler than other breeding males, tending eggs in a narrow hole in a limestone bedrock. He occupied this cavity when it was only IO ctii beneath the water surface and abandoned it as the water level fell during thc dry 584 Catfishes - Fig. 19.2 Life cycle of Ancistrus spinosus. Top: The brooding male is larger than drawn here and usually braced by his pectoral spines so as to block access to the nest. Middle: Young of the year that escape from pool predators after fledgings move into nursery riffles. Bottom: When they attain Standard Lengths of about 4-5 cm, young fish move back into deeper pool Habitats. season. The hollow logs in which Ancistrus nest must have openings close in diameter to the width of the anterior body of the breeding male. The father served as a kind of "armored cork", plugging the entrance against potential predators of his young, such as characin fishes (e.g., the active and ubiquitous Astyanax ruberrimus) and freshwater crabs (Potamocarcinus richmondi). Crabs are found in hollow nest logs when these are not occupied by breeding male Ancistrus. The size at which the& exposed dorsal armor is hard enough to resist puncture by large crabs may set the lower size crabs eati Hypos tomi Ancisi fairly long in early M 3 km stuc downstrea in an undf was alway over the shades of attached i two brood was seen f occupied 1 nest, whic tended th, away in a had trave riffles to 01 been capt Male at first rep Ancistrus : their snoi cutaneous larger ma forked. Bc periodical hormonal1 Function 1 nest log, b their una1 outside. 1 pelvic fins stagnant r Carter an( r than drawn nest. Middle: ursery riffles. we back into nings close male. The ice against the active mocarcinus ;e are not [r exposed lay set the Maw E. Power 587 - the urid em. nest !!z by the n of itral tion ural cles tion for also bels iale kely t of hile lost .eed des. lost rere ests ung wge . In I in :ion rus, Ller, ged low n at jily occupied and reproduced in 2.54 cm diameter PVC pipes that Moodie deployed in the Rio Frijoles. Mature male Rineloricaria have wider, rounder snouts than females and, unlike females, develop jaw bristles along their lateral edges. Moodie also noted a second dimorphism: males could swing their pectoral spines forward to lie nearly parallel with their jaws, while females could only swing their pectoral spines far enough forward to project out from the body at a 90" angle (Moodie and Power, 1982: fig. 1). This posture potentially created a grasping organ for the males (by pinching objects between the jaw bristles and the pectoral spine), which Moodie thought might be used to manipulate females in the nest. It is possible (although not an interpretation favored by Moodie) that males might use their grasping ability to prevent superfluous females from entering nests and injuring or dislodging older eggs, once nests were filled. Multiple females laid eggs in his artificial nests and like Ancistrus, Rineloricaria males would stay with the nests even after they had been withdrawn from the water, opened to inspect eggs, closed and returned to the stream. Like Ancistrus, Rineloricaria males that have obtained a nest will brood several successive clutches of eggs (up to five for Rineloricaria (Moodie and Power 1982)). I never observed nests of Hypostomus or Chaetostoma in the Rio Frijoles. These species may have nested deep under rock ledges or in holes that appeared in clay banks. Hypostomus plecostomus occupied burrows dug in firm clay banks. I observed a large individual rubbing the rough leading edge of its pectoral spine against a clay bank near a group of other holes. This motion looked like digging behavior. The holes could have been made by prawns (Mucrobrachium sp.) that occurred in the stream, however, or by tree roots projecting into the flow that had rotted away. Neither Chaetostoma nor Hypostomus showed any apparent sexual dimorphism. Hypostomus (Lowe-McConnell 1967) and possibly Chaetostomu rear their young in bank holes. ONTOGENETIC HABITAT SHIFTS (FIG. 19.21 When young loricariids are 18 mm long SL (Ancistrus) or 12 mm SL (Rineloricaria), they leave the father's nest. While I have never witnessed a fledging episode, these must be times of great mortality, as soft, weakly swimming young catfish are readily eaten by common fishes in the stream (Astyunax, Brycommericanus, Piabucimj . Young loricariids must exit from a pool that is full of swimming predators, many of which are active 24 t 588 Catfishes hours a day. Those that survive move into shallow riffles, where they can take cover in interstices among cobbles. I rarely observed armored catfish less than 3 cm SL in pools, or in water > 30 cm deep. On the few occasions when I witnessed a young of the year (yoy) Ancistrus (ca. 2 cm SL) traversing pools, it would edge across the stream margin near the surface, harassed by characins that struck at it in rapid forays from deeper water. Young loricariids reared in shallow riffles where they grazed algae. (Juvenile Rineloricaria include insect larvae in their diets and become more exclusively algivorous with size or age (Pineda, 1975). Young riffle- dwelling loricariids did not appear to be food limited, as algal standing crops in riffles were higher than in pools (Power, 1984a). Primary productivity was also probably elevated by the greater flux of dissolved nutrients in flowing water (Whitford and Schumacher, 1964) and increased insolation relative to beds in deeper pools. Although their armor was not yet hardened, predation pressure on young loricariids was probably reduced by their habitat selection. The Gape-limited aquatic predators to which they were most vulnerable avoided water shallower than 20 cm, where these characins, gobies, and other larger fishes were themselves at risk from diving and wading birds that frequented the river (Power 1984a, 1987). Loricariids smaller than 3-4 cm could hide under cobbles from these birds. When they outgrew this cover, at 4-5 cm SL, they also began to outgrow their vulnerability to most of the Gape-limited swimming predators in pools, and juveniles at this size (probably within their first year of life) moved into deeper pools. In pools, subadult loricariids entered a crowded environment in which they became limited by food for most of the year. Standing crops of algae in pools were so scant that on substrates > 20 cm deep, algae were not macroscopically detectable (substrates did not even feel slippery). Microscopic sampling revealed only scant cover of very small, adnate diatoms (primarily Achnanthes sp. (Power, 198413)). Visible accumulations of algae occurred, however, along pool margins on substrates c 20 cm deep, which grazing catfish appear to avoid due to their vulnerability in shallow water to wading and diving birds (Power, 1984a, 1987; Power et al., 1989). These birds fish most commonly and effectively in water < 20 cm deep, possibly because fish there have too little time to escape between the surface splash and the strike (Kramer, 1983; Power, 1984a). Along the Rio Frijoles, these shallow substrates were largely avoided even at night (Fig. 19.3), although there was some tendency for loricariids, particularly the smaller size classes and more cryptic species (Rineloricariu) to shift to slightly shallower Ha Herons occurred alc have been found wit 1965). The avoidant rings" of algae to ac in the dry season, w SEASONAL CHI Pool- dwelling loricai season (Power, l! - Fig. 19.3 a. Depth quG of a given species and : classes of armored catf snorkeling censuses of rainy season (August, ( months (January and JI then loricariid numbers following week. b. Grow errors. ED = early dry s and April); ER = early through November). R only records that fell 1 1981). There they can -mored catfish few occasions :a. 2 cm SL) ar the surface, deeper water. grazed algae. become more Young riffle- algal standing 34a). Primary x of dissolved r, 1964) and lthough their loricariids was rnited aquatic iter shallower er fishes were nted the river Id hide under -5 cm SL, they Gape -limited obably within nent in which crops of algae lgae were not eel slippery). small, adnate iccumulations ates < 20 cm ulnerability in 987; Power et in water < 20 ;cape between la). Along the even at night s, particularly via) to shift to i Mary E. Power 589 slightly shallower Habitats in darkness (Power, 1984a). Night-fishing tiger Herons occurred along the Rio Frijoles. In other Habitats, these Herons have been found with "fair-sized, bony fish" in their stomachs (Wetmore, 1965). The avoidance of shallow water by grazing catfish allow "bathtub rings" of algae to accrue along shallow pool margins, which persist even in the dry season, when larger catfish in pools are most food limited. SEASONAL CHANGES Pool-dwelling loricariid catfish showed zero or negative Growth in the dry season (Power, 1984a). These weight measurements may have I <,Ot I II , . , , * , , EDLDERLR EDLDERLR 4 cm nil" SIOC Rlvw 11.0. Season - Fig. 19.3 a. Depth quartiles (shallowest depth such that at least 25% of the indivic-als of a given species and size class occur at that depth or less) for four species and six size classes of armored catfish. Quartiles estimated from data taken over two years during six snorkeling censuses of 180 m2 quadrats distributed over a 3 km reach. Twice during the rainy season (August, October), the dry season (March and April), and during transitional months (January and June), physical conditions within each quadrat were measured, and then loricariid numbers in quadrats were counted on two days and two nights during the following week. b. Growth rates of Ancisfrus of different length intervals, with two standard errors. ED = early dry season (December through February); LD = late dry season (March and April); ER = early rainy season (May and June) and LR = Late rainy season (July through November). Records are from marked, recaptured, and reweighed individuals; only records that fell within the seasonal intervals delimited were used. (After Power, 1981). 590 Catfishes underestimated nutritional deprivation during the dry season, as a subsequent study (Power, 1984c) suggested that starving Ancistms could substitute water for fat in tissues. With the onset of the rainy season, as stream flow deepened, the catfish were able to move into areas formerly too shallow to graze and showed a Growth pulse (Fig. 19.3b). Growth slowed after they had depleted the higher standing crops of algae formerly refuged in shallow water but remained positive throughout the rainy season. Seasonal changes in base flow discharge in the Rio Frijoles were modest and the channel habitat did not obviously expand, because much of the channel was fairly trapezoidal in cross section. What did expand by a factor of three, however, was the amount of substrate under > 20 cm of water. The expansion of this "safe" habitat was tracked by a corresponding 3-fold decrease in the density of loricariids. During my first field season, I was surprised after the transition from the dry to the rainy season to find loricariids suddenly sparser. I searched the entire river for migrating groups of loricariids and investigated episodic predation (by migrating cormorants or aquarium collectors) as a possible cause for drop in loricariid density, mistaking it for a drop in abundance. It was only after constructing bathymetric maps of the 3-km study reach, drawn from detailed surveys that enabled resolution of 10-cm depth contours in the channel habitat, and learning of the depth tolerances of loricariids, that I appreciated that seasonal expansion of this critical habitat stratum was the cause of the density changes. This interpretation was confirmed during the next year's transition from the dry to the rainy season, when the first flood elevated stream stage. After this flood, the water table in the surrounding watershed remained low, so stream discharge subsided within days. Censuses of quadrats marked on the bed showed that catfish spread out with the high flow, becoming 3-fold less dense, then reconcentrate with the abating flow, regaining their preflood density within a week, i.e., within such a short period that migration or recruitment could not have accounted for the observed density fluctuation (Power, 1981). SENESCENCE Another indication that their conservative space use protects larger Ancistms from predation is that a few older individuals (large males) grew increasingly easy to capture over time, as if they were becoming diseased or senescent. Some of these individuals showed aberrations of the eyes- the iris which normally projects down to make a horseshoe-shaped pupil Fig. 19.4 Senescei was eroded and : that also appear infirmities, these environments, si avoided shallow HABITAT PA1 Loricariids are tl Frijoles; macroir deeper Habitats. stream margins ( characoids brow: larger loricariids the fact that the could infer that substrate. Accol species would competition. Mary E. Power 591 Fig. 19.4 Senescent Ancistrus with eroded iris and cloudy pupil was eroded and hence these older individuals had raggcd, circular pupils that also appeared cloudy (Fig. 19.4). Despite their blindness or other infirmities, these older individuals persisted many months in their natural environments, suggesting that they remained safe from predators if they avoided shallow waters where mammals and birds fished. HA6 ITAT PART IT I 0 N I N G Loricariids are the only major grazers of algae in deeper pools of thc Rio Frijoles; macroinvertebrate scrapers were sparse and virtually absent in deeper Habitats. Frog tadpoles occurred seasonally but only along shallow stream margins (depths generally < 15 cm). Some poeciliids, cichlids, and characoids browsed algal filaments but did not scrape surfaces. Given that larger loricariids were food limited for much of the year (as evidenced by the fact that they grew at less than their potential maximum rates), one could infer that they compete with each other for limited food or grazing substrate. According to classical niche theory, one might expect that species would partition grazing substrate to reduce interspecific competition. 592 Catfishes There were some differences among the species in tendency to use particular substrates. Ancistrus was common on large submerged logs, which they grazed and used for cover (nimbly circling around logs to keep out of my reach, for example). They also grazed bedrock, large boulders, cobbles, and consolidated clay banks. These substrates were also commonly used by Hypostomus, which were particularly abundant on the clay banks. Rineloricaria, with its narrower mouth and more cryptic body, was abundant on clay and bedrock and also on cobbles and pebbles in the shallower pool heads and tails, and in shallower mid-pool environments. Chaetostoma fischeri, the one species in the Rio Frijoles that could not use its stomach as an accessory air-breathing organ, was restricted to constricted reaches where the flow was both fast and deep. There was some rough color matching among the species with the substrates- Chaetostoma was dark brown with gold glints and often matched the dark rocks in the rapids where they occurred, while Hypostomus and Rineloricaria were both dun colored, matched to clay and mudstone substrates. Black Ancistrus was the color of submerged logs that retained their dark bark. I was not overly impressed with substrate specialization or partitioning among the species, however. In the dry season, as food became more limited, species overlapped more in the types of substrate they used (Power, 1981, 1983), as documented by Lowe-McConnell (1975) for Neotropical fish stranded in isolated pool "lifeboats" in northern South America. These observations contradict theoretical predictions of increased niche separation to reduce competition during periods of resource limitation (Zaret and Rand, 1971). RESPONSES TO FOOD QUALITY AND RENEWAL Armored catfish generally live within bounded pool Habitats and eventually regraze sites on pool substrates. If an individual returns to a previously grazed site too early, scant algae will have regrown. Buck and Sazima (1995) observed that sites cleared by loricariids in the Ribeirao da Serra, a small, clearwater forest stream in Southeastern Brazil, presented new algal Growth in about 24 h. Average return times of armored catfish in sunny pools of the Rio Frijoles, where the loricariids were most dense, were about 9-10 h (Power, 1984b), much shorter than the potential doubling time of diatoms even in productive environments. These short return times, along with the scant algal standing crops observed, suggested that Rio Frijoles substrates were heavily overgrazed. Undergrazing might also be returnec deposit€ in part I Sea: are dep season, stream downstx water c 1981). floods v reduced the bed flow pel the watl by Anci. sometin passed : that the in fact, energy 1 of clear unknov Venezu anothei comm.) differen not prl import2 signific: Silt species initiate grazing away frc were re 1984c; Mary E. Power 593 endency to use ubmerged logs, ind logs to keep large boulders, 3tes were also bundant on the re cryptic body, rl pebbles in the . environments. it could not use s restricted to eep. There was le substrates- atched the dark ypostomus and and mudstone gs that retained ;pecialization or as food became mate they used ne11 (1975) for northern South predictions of ring periods of WAL I Habitats and ual returns to a cown. Buck and the Ribeirao da hazil, presented armored catfish 'ere most dense, n the potential its. These short Erved, suggested lergrazing might also be disadvantageous for loricariid catfish, however. If loricariids returned after long time intervals, substrates would become silted. The deposited sediment could be either a stress or a food for grazers, depending in part on its organic content. Seasonality influences the quality and persistence of sediments that are deposited on the substrates grazed by armored catfish. In the rainy season, these sediments are derived from clay-rich soils that wash into the stream during floods. Suspended sediments are deposited, or flushed downstream, depending on the concentration of suspended load in the water column on the recession limb of the flood hydrograph (Power, 1981). With several floods a month, the probability is high that large floods would scour away deposited sediments. Also, sediment accrual is reduced during the rainy season because even at base flow, currents over the bed are faster and boundary layers thinner. During the dry season iow- flow period, sediment sifts down more continuously as organic matter in the water column settles onto substrates in still pools. This silt is displaced by Ancistrus and Hypostomus with head-shaking behavior with which they sometimes initiate grazing on heavily silted patches. It is also ingested and passed as consolidated feces. A feeding experiment in Panama suggested that the sediment was not a food for prereproductive (6 cm)Ancistrus and, in fact, imposed energetic costs estimated at up to 20% of their daily energy budgets (Power, 1984~). This cost may have been due to the costs of clearing sediment from their small, entirely ventral Gills (Fig. 19.1) by unknown mechanisms, possibly mucus sloughing or coughing. In Venezuela, however, Alex Flecker and colleagues have observed that another species, Ancistrus trirudiatus, could grow on silt (A. Flecker, pers. comm.) . Whether this disparity is due to difference in sediment quality or different adaptations of the Panamanian versus Venezuelan Ancistrus is not presently known, but merits further investigation, given the importance of identifying potential energy sources for these ecologically significant fish. Siltation is a stress for Panamanian and apparently also for four species of Brazilian loricariids, which Buck and Sazima (1995) observed to initiate similar wiggling head-down movements when they commenced grazing on heavily silted substrates, movements which wafted sediment away from the area to be grazed. In both Habitats, previously cleared spots were regrazed by the same individual or other catfish (Fig. 19.5; Power, 1984c; Buck and Sazima, 1995). I often observed smaller loricariids 40 .. ... o *. ,' ,/` 8 I I/ % Substrate Area Cleared ! I I Selectivity of Ancistrus for cleared substrate in enclosures. Points are the % of g&ng bouts initiated on a given day on cleared substrate, plotted against the % of the substrate in an enclosure cleared on that day. If grazing bouts were initiated at random with respect to sediment, equal numbers of points should fall above and below the line X=Y (adapted from Power, 1981 and 1990). regrazing sites opened in silted areas by larger loricariids. To test whether small fish would select such sites, I built a choice cage out of coarse plastic screen (6-mm mesh Vexar) with three chambers (Power, 1984~). The central chamber housed a small test Ancistrus (4-7 cm SL) and one of the side chambers housed a large individual (10-14 cm SL). Holes in the chamber partitions allowed the small but not the large catfish to move among chambers. When the cage was placed in heavily sedimented-areas, the small individual moved into the chamber with the large individual more frequently than would be expected by chance. When the choice cage was placed in sediment-free areas in the stream, there was no effect of the larger individual on the position of the small catfish (Power, 1984~). In another experiment, catfish densities were reduced to ca. 0.1 their ambient densities in 4 m2 enclosures built in a sunny pool that had previously supported up to 6 10-g individuals m-'. Catfish regrazed small sites opened within enclosures significantly more than sites that had not been opened by grazing (Fig 19.5; Power, 1990). At these low densities, catfish grazing significantly reduced sediment standing crops and enhanced epilithic algal productivity and, surprisingly, algal abundance. Attached algal productivity and abundance were lower in the open habitat under the more intense ambient grazing regimen but still higher than in t fish-free e weeks (Pc "op timizec allowing r "overgrazt mechanist "prudent 1 AGNOS` Given the algae that might exl observed stream PO in the dry using thes Interspeci Hypos tom and were interactio drifted a] between E bouts ob: (Power, 1 grazing ii opercular individua were rare out of si platform. frequent at trac tint One each otb skirmish. twelve c Ancistrus marked a ;ures. Points are the % of tted against the % of the re initiated at random with ! and below the line X=Y iids. To test whether : out of coarse plastic :Power, 1984~). The n SL) and one of the n SL). Holes in the irge catfish to move ily sedimented-areas, the large individual Uhen the choice cage : was no effect of the L (Power, 1984~). luced to ca. 0.1 their inny pool that had atfish regrazed small m sites that had not these low densities, ,tanding crops and :ly, algal abundance. :r in the open habitat t still higher than in Mary E. Power 595 fish-free exclosures, in which thick carpets of silt accrued after several weeks (Power, 1990). This result suggested that loricariids could have "optimized" their harvests while grazing by foraging less frequently, and allowing more algal recovery between grazing bouts. That they in fact "overgrazed" in the open habitat suggested that they lacked a social mechanism, like territoriality, that could have increased the likelihood of "prudent grazing" by catfish. AGNOSTIC INTERACTIONS Given the intense intraspecific and interspecific competition for attached algae that Rio Frijoles loricariids experience for much of the year, one might expect to see interference as well as exploitative competition. I observed the four species grazing on gridded bedrock platforms in four stream pools by day and night (using dim underwater lights at night) and in the dry and rainy seasons over a two-year period. Most of the loricariids using these substrates were Ancistrus, with Hypostomus next in abundance. Interspecific interactions were relatively rare. In the few that I observed, Hypostomus and Rineloricaria rarely initiated interactions with Ancistrus and were almost always displaced by them. Hypostomus intraspecific interactions involved prolonged, gentle body contact and ended when fish drifted apart. The most common and intense interactions occurred between grazingAncistrus. Out of a total of 647 5-minute Ancistrus grazing bouts observed, 11 1 bouts were interrupted by agnostic interactions (Power, 198413). These occurred when two individuals of similar size grazing in close proximity bumped into each other and flared their opercular spines. In the most intense interactions that I observed, individuals pushed against each other with these spines. Such skirmishes were rare, brief, and immediately followed by flight of both participants out of sight to shelter under a rock ledge adjacent to their grazing platform. This behavior suggested that grazing catfish avoided more frequent or prolonged interference behavior because of the risk of attracting attention from aerial predators. One observation of marked individuals suggested that Ancistrus knew each other as individuals. Two marked 6-7 cm SL Ancistrus hid after a skirmish. One returned and resumed grazing in the middle of a group of twelve conspecifics of similar size. Several minutes later the other Ancistrus returned and worked its way through this group, ignoring 596 Catfishes first catfish and resume the skirmish. Olfactory individual recognition in ictalurids has been demonstrated experimentally (Todd et al. 1967). The frequency of skirmishes did not influence Ancistrus Growth. Although interference episodes were 2-4 times more frequent in crowded groups of Ancistrus in sunny, productive pools than in sparse groups in dark pools, the average Growth rates of individuals in all of these pools was statistically indistinguishable. Similar somatic Growth rates, particularly of prereproductive individuals, suggested that the interference behavior imposed insignificant energetic costs (Power, 1984.b; Oksanen et al., 1995) and did not impede individuals from closely tracking variation in algal Growth rates on the pool-to-pool scale over kilometers of stream reach. IDEAL FREE DISTRIBUTION AND PRODUCTIVITY TRACKING Because they are not apparently constrained by social dominance or pool- dwelling predators, all pool-dwelling loricariids could presumably enhance their fitness by closely tracking large-scale (pool-to-pool) variation in the availability of their algal food. In fact, loricariids tracked algal productivity quantitatively in pools distributed over a 3 km reach of the Rio Frijoles. Algal Growth rates varied 16-fold among these pools, because of variation in forest canopy shading stream substrates. I censused 16 pools distributed over the 3 km study reach of the river for 12 consecutive months by snorkeling through them and counting loricariids, including individuals hiding under ledges, which I detected with a small underwater flashlight. Catfish densities (both in terms of individuals and biomass) scaled to the area of grazeable (flat bedrock) substrate was linearly related to the relative productivity of a given pool. Measurements of algal Growth and depletion by collective loricariid grazing suggested that these rates were similarly balanced in the 6 dark, 6 half-shaded, 6 moderately sunny, and 3 very sunny pools. As Fretwell and Lucas (1970) predicted for animals able to achieve an Ideal Free Distribution, standing crops of algal food, somatic Growth rates of prereproductiveAncistms, and survivorship of all Ancistrus were similar among the 16 sunny, half- shaded, and dark pools censused monthly for one year (Power, 1984b; Oksanen et al., 1995). PROSPEl UNDER In the Neo natural veg excessive 1 and wider breeding s embedded and other streams in the most SI and adult by young c loricariids human PI nocturnal In the loricariids students a numbers : Marias a1 abundanc Paradon triradiatus (Flecker, around th did not o swimmink in shallou Habit stream fis reduc tior complex The ripa particulai long hori streams continue 598 Catfishes growth adaptation produces cave-like undercuts that extend several meters under the bank, which are excellent Habitats and breeding sites for larger fish. Like many fishes, loricariid populations may be potentially quite resilient. Adults are long lived (several decades) with moderate lifetime fecundity. Many species have adaptations to cope with stresses such as transient hypoxia and intermittent substrate sedimentation. Long-lived loricariids, can therefore, persist in unsuitable Habitats with many years of successive recruitment failure, via the "storage effect" (Warner and Chesson, 1985). As environmental stresses grow more prolonged or chronic, however, habitat loss will constrict the life history bottlenecks of loricariid catfish past the point of recovery. We are at risk of losing many loricariid species before we have even begun to appreciate their spectacular diversity and the highly evolved morphological, behavioral, and ecological adaptations that mold their lives in nature. References Armbruster, J.W. 1998. Modifications of the digestive tract for holding air in loricariid and scoloplacid catfishes. Copeia: 663-675 Barron, J.C. 1964. Reproduction and apparent over-winter survival of suckermouth armored catfishes Plecostomus sp. in headwaters of San Antonio River. Texas I. Sci. 16: 449. Brauner, C.J., C.L. Ballantyne, D.J. Randall and A.L. Val. 1995. Air-breathing in the armored catfish (Hoplosternum littorale) as an adaptation to hypoxic, acidic, and hydrogen-sulfide rich waters. Can. J. Zool. 73: 739-744. Buck, S. and I. Sazima. 1995. 1995. An assemblage of mailed catfishes (Loricariidae) in southeastern Brazil: distribution, activity, and feeding. Ichthyol. Explor. Freshw. 6: 325-332. Burgerss, W.E. 1989. An Atlas of Freshwater and Marine Catfish. A Preliminary Survey of the Silurifionnes. TFH Publications, Neptune City, PA. Carter, G.S. 1934. Results of the Cambridge Expedition to British Guiana, 1933. The freshwaters of the rain-forest areas of British Guiana. J. Linn. SOC. Zool. 39: 147-193. Carter, G.S., and L.C. Beadle. 1930. The fauna of the swamps of the Paraguayan Chaco in relation to its environment. I. Physicochemical nature of the environment. Zool. 1. Linn. SOC. 37: 205-258. Flecker, AS. 1984. The effects of predation and detritus on the structure of a stream community: a field test. Oecologia 64: 306-313. Flecker, AS. 1992. Fish trophic guilds and the structure of a tropical stream: weak direct vs. strong indirect effects. Ecology 73: 927-940. Fretwell, S.D. a habitat di: 36. Gee, J.H. 197t reduced (Trichomy 1030- 103 Gradwell, N. 1 Grzimek, B. 19 York, NY. Hawkes, J.W. 1 Kramer, D.L. I relation ti Kramer, D.L., oxygen cc 7: 47-55. Lowe-McConi waters. A -1975. Fish ( -1979. Ecolo Lond. 44: -1987. Ecolc CambridE Moodie, G.E.1 Loricaria Nelson, J.S. 15 Nico, L.g. anc P terygopli fish intro Oksanen, T., consumei Ono, R.D. 19f taste-bud J. Morphc Pineda, E. 19 (Chupapi, Panama. Power, M.E. Panaman -1983. Graz variation -1984a. DeF Ecology 6 147-158. i -- 7- Mary E. Power 599 Fretwell, S.D. and H.L. Lucas. 1970. On territorial behavior and other factors influencing habitat distribution in birds. I. Theoretical development. Acta Bzotheoretica 19: 16- Gee, J.H. 1976. Buoyancy and aerial respiration: factors influencing the evolution of iotentially quite reduced swimbladder volume of some Central American catfishes noderate lifetime (Trichomycteridae, Calhchthyidae, Loricariidae, Astroblepidae) . Can. 1. ZOO^. 54: extend several I 36. breeding sites for stresses such as 1030- 103 7. ition. Long-lived ith many years of :t" (Warner and )re prolonged or )ry bottlenecks of sk of losing many appreciate their gicaI, behavioral, Gradwell, N. 1971. A muscular oral valve unique in fishes. Can. J. 2001. 49: 837-839. Grzimek, B. 1984. Grzimek`s Animal Life Encyclopedia. Van Nosttand Reinhold Co., New Hawkes, J.W. 1974. The structure offish skin. I. General organization. Cell Tis. Res. 169: Kramer, D.L. 1983. Aquatic surface respiration in the fishes of Panama: distribution in Kramer, D.L., D. Manley and R. Bourgeois. 1983. The effect of repiratory mode and oxygen concentration on the risk of aerial predation in fishes. Environ. Biol. Fish. Lowe-McConnell, R.H. 1967. Some factors affecting fish populations in amazonian -1975. Fish Communities in Tropical Freshwaters. Longman Inc., New York, NY. -1979. Ecological aspects of seasonality in fishes of tropical waters. Symp. 2001. SOC. -1987. Ecological Studies in Tropical Fish Communities. Cambridge Univ. Press, Moodie, G.E.E. and M. Power. 1982. The reproductive biologv of an armored catfish, Nelson, J.S. 1994. Ffihes of the World. Wiley Publ., New York, NY. Nico, Leg. and R.T. Martin. 2001. The South American suckermouth armored catfish, Pterygoplichthys anisitisi (Pisces : Loricariidae), in Texas, with comments on foreign fish introductions in the American southwest. Southwestern Naturalist 46: 98-104. Oksanen, T., M.E. Power and L. Oksaene. 1995. Ideal free habitat selection and consumer-resource dynamics. Amer. Natural. 146: 565-585. Ono, R.D. 1980. Fine-structure and distribution of epidermal projections associated with taste-buds on the oral papillae in some loricariid catfishes (Siluroidei, Loricariidae) . Pineda, E. 1975. Aspectos de la biologia, ecologiay distribuci6n de la familla Loricariidue (Chupapiedras) en la cuenca del Rio Bayano. Publ. Univ. Panama, Fac. Cienc. Farm., Panama. Power, M.E. 1981. The grazing ecology of armored catfish (Loricariidae) in a Panamanian stream. Unpub. Ph.D. Diss., Univ, Washington, Seattle, WA. -1983. Grazing responses of tropical freshwater fishes fishes to different scales of variation in their food. Environ. Biol. Fish. 9: 103-1 15. -1984a. Depth distributions of armored catfish: predator-induced resource avoidance? 1 ,York, NY. 147-158. I relation to risk of hypoxia. Enuiron. Biol. Fish. 8: 49-54. ure. 7: 47-55. waters. At. Simp. Biota Amazonica I: 177-186. Jding air in loricariid wal of suckermouth 110 River. Texas]. Sci. Cambridge. I Air-breathing in the hypoxic, acidic, and ishes (Loricariidae) in YO^. Explor. Freshw. 6: 5. Preliminary Survey of jh Guiana, 1933. The bc. 2001.39: 147-193. the Paraguayan Chaco he environment. ZOO^. Structure of a stream cal stream: weak direct Lad. 44: 219-241. Lorzcaria uracantha, from Central America. Enuiron. Biol. Fish. 7: 143-148. J. Moqhol. 164: 139-159. Ecology 65: 523-528. 600 Catfishes -1984b. Habitat quality and the distribution of algae-grazing catfish in a Panamanian stream. j. Anim. Ecol. 53: 357-374. -1984~. The importance of sediment in the feeding ecology and social interactions of an armored catfish, Ancistrus spinosus. Enuiron. Biol. Fish. 10: 173-181. -1987. Predator avoidance by grazing fishes in temperate and tropical streams: importance of stream depth and prey size. In: Predation: Direct and Indirect Impacts on Aquatic Communities. WC. Kerfoot and A. Sih (eds.). Univ. Press of New England, Hanover, NH, pp. 333-35 1. -1990. Resource enhancement by indirect effects of grazers: armored catfish, algae, and sediment. Ecology 71: 897-904. Power, M.E., T.L. Dudley, and S.D. Cooper. 1989. Grazing catfish, fishing birds, and attached algae in a Panamanian stream. Enuiron. Bid. Fish. 26: 285-294. Rand, AS. 1967. Predator-prey interactions and the evolution of aspect diversity. At. Simp. Biota Amazonica 5: 73-83. Schafer, S.A. and G.V Lauder. 1986. Historical transformation of functional design: Evolutionary morphology of feeding mechanisms in loricarioid catfishes. Syst. Zool. -1996. Testing historical hypotheses of morphological change: Biomechanical decoupling in loricarioid catfishes. Evolution 50: 1661-1675. Todd, J.H., J. Atena and J. E. Bardach. 1967. Chemical communication in the social behavior of a fish. Science 158: 672-673. Warner, R.R. and EL. Chesson. 1985. Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. Amer. Naturalist 125: 769-787. Wetmore, A. 1965. The birds of the Republic of Panama, Part 1. Smiths. Misc. Coll. 150: Whitford, L.A. and G.J. Schumacher. 1961. Effect of current on mineral uptake and Zaret, T.M. and AS. Rand. 1971. Competition in tropical fishes: support for the 35: 489-508. 78-1 19. respiration by a fresh water alga. Limnol. Oceanogr. 6: 423-425. competitive exclusion principle. Ecology 52: 336-342. The sub1 interconn large spac which m: (J u berthk developec the presei In co majority character tendency photoautc poor and except fc chemoaut *Departame Paulo, SP, I