Proc. Natl. Acad. Sci. USA Vol. 80, pp. 4766-4769, August 1983 Ecology

Ornitholimnology: Effects of grazing by the Andean flamingo (Phoenicoparrus andinus) (predation/Surirella/lacustrine microbenthos) STUART H. HURLBERT AND CECILY C. Y. CHANG Department of Biology, San Diego State University, San Diego, California 92182 Communicated by G. Evelyn Hutchinson, May 6, 1983

ABSTRACT Experimental exclusion of the Andean flamingo (Phoenicoparrus andinus) from shallow water areas of a salt lake in the Bolivian Andes caused large increases in the biomass of mi- croorganisms inhabiting the surface sediments, especially a large diatom (Surirella wetzeli), amebas, ciliates, and nematodes. This is a conservative demonstration of the influences that water birds in general exert on the structure of aquatic ecosystems. Of the -8,900 species of birds in the world roughly 9% feed entirely or predominantly on aquatic organisms. Their diets range over the entire spectrum of available foods, from fish to in- vertebrates, from vascular plants to microscopic algae. The more abundant of these birds may strongly affect the dynamics of their prey populations and, indirectly, the dynamics of entire aquatic ecosystems. However, few experimental studies of these interactions have been attempted (1-4), and we scarcely have an inkling of how the structure and functioning of any aquatic ecosystem would be altered by the disappearance of one or more of the bird spe- cies feeding in it. In this report we document the strong influence exerted by the Andean flamingo (Phoenicoparrus andinus) on lacustrine microbenthos, the assemblage of microscopic organisms in- habiting the surface sediments of a lake bottom. P. andinus is a.common inhabitant of numerous shallow, mostly saline lakes of the altiplano region of the central Andes of South America (5-7). Two otherflamingos, James' (Phoenicoparrusjamesi) and the Chilean (Phoenicopterus chilensis), also occur abundantly on many of these same lakes. Other flamingo species are dom- inant components of salt lake and coastal lagoon ecosystems in Africa and Eurasia, and the fossil record indicates that, in the past, still other flamingos occupied similar environments in North America and Australia (8-10). The role that flamingos play in the functioning of these lake ecosystems is thus of general in- terest. We suspect that this role is great in consequence of flamingos' high feeding rates and the large size of their feeding FIG. 1. Location of study site in Laguna Puripica Chico, Bolivia, flocks. and arrangement of control (F = flamingos) and exclosure (NF = no flamingos) areas. STUDY AREA AND METHODS To test this suspicion we conducted an exclosure experiment more numerous than the other two species; this pattern was during January and February 1979 in Laguna Puripica Chico, maintained throughout our study (Table 1). On earlier visits we a small (1.0 km2), shallow (maximal depth, ca. 50 cm), slightly had found even larger numbers of flamingos here. Approxi- saline (8.2 g/liter), high-altitude (4,393 m) lake in the desertic mately 3,000 (66% P. jamesi, 29% P. andinus, and 5% P. chi- altiplano of the southwesternmost tip (22031' S, 67°30' W) of lensis) were observed on the lake in December 1975 (6) and Bolivia; this is one of several lakes scattered around the margin about 5,700 (65% P.jamesi, 33% P. andinus, and 2% P. chilensis) of a salt-encrusted basin called Salar de Chalviri (Fig. 1). On in November 1977 (7). By way of comparison, the lesser flamin- visits in late December 1978 and early January 1979 we found go (Phoeniconais minor) occurs on some African lakes at 10 times flamingos to be abundant on this lake and P. andinus to be much these densities (11). On January 2 we selected a mudflat that was being used as a P. but not the other The publication costs of this article were defrayed in part bypage charge feeding ground by andinus by species. payment. This article must therefore be hereby marked "advertise- ment" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviations: F, flamingo(s); NF, no flamingo(s). 4766 Downloaded by guest on September 28, 2021 Ecology: Hurlbert and Chang Proc. Natl. Acad. Sci. USA 80 (1983) 4767 Table 1. Effects of the Andean flamingo (P. andinus) on the microbenthos of Laguna Puripica Chico, Bolivia Sampling date Taxon or feature January 3a January 19b February 4 February 24c Time since exclosures established, days 1 17 31 51 Water depth, cm 0-1 2-4 9-11 3-5 Number of flamingos on lake P. andinus 654 716 1,159 1,000 P.jamesi 81 9 0 12 P. chilensis 29 75 68 30 F NF F NF F NF F NF Individuals, no. per mm2 of sediment surface Diatoms, totald 4,117 2,895 890 859 1,816 631 7,538 12,112 Surirella wetzelie 32 30 37 54 27 12 52 156* Navicula spp.' 3,764 2,733 795 761 1,613 594 7,077 10,631 Amphora spp. 17 33 0.3 0.8 0.0 0.7 280 553 Nitzschia spp. 78 58 12 13 46 5 76.5 330 Protozoa, total 4.0 5.9 6.7 31.2+ 2.8 7.6 4.7 52.2t Amebasg 2.8 1.8 6.4 29.2 2.5 6.5 2.8 43.8t Ciliatesh 1.6 4.0 0.6 2.0 0.1 1.0 1.4 9.0O Nematodes' 0.59 0.66 0.82 0.44 0.30 1.31 0.50 1.96X Biomass, tug per mm2 of sediment surface Blue-green algae, totali 0.146 0.103 0.027 0.360 0.009 0.140 1.54 3.16 Diatoms, totald 17.5 13.7 15.3 25.7 12.8 6.1 30.9 73.3+ Surirella wetzelie 13.0 10.0 13.8 23.3 10.1 4.8 20.0 58.0T Navicula spp.f 2.48 2.20 0.69 1.22 1.02 9.56 3.77 7.58* Amphora spp. 0.20 0.19 0.005 0.015 0.00 0.01 2.8 3.5 Nitzschia spp. 0.19 0.24 0.04 0.03 0.12 0.001 0.19 0.91 Protozoa, total 0.80 2.24 0.61 3.62t 0.24 1.66 1.36 8.98T Amebasg 0.20 0.13 0.48 2.20t 0.18 0.49 0.21 3.29t Ciliatesh 0.62 2.10 0.23 1.32t 0.04 1.16 1.15 5.28+ Nematodes' 0.15 0.48 0.48 0.14 0.14 0.22 0.26 1.05T Total microbenthos 18.7 16.2 16.5 30.1 13.3 8.1 35.1 86.2t Tabulated values for the microbenthos are "hybrid" means, each the arithmetic mean of: (i) the median of and (ii) the arithmetic mean of the 4 or 10 replicate values. Statistical significance of each difference between NF and F areas was assessed by the Mann-Whitney U test and is denoted, by symbols above the NF means, as follows: no symbol, P > 0.10; *, 0.10 - P > 0.05; t, 0.05 > P > 0.01; and T, P < 0.01. a Census data are for January 4. bCensus data are for January 31; values do not include 150 flamingos unidentified to species. CCensus data are for February 18; values do not include 200 flamingos unidentified to species. dThirteen genera of diatoms were encountered, the great majority contributing negligibly to total diatom abundance. eAveraged 135 ,um in length. An individual that size has volume of -373,000 ,m3 and occupies about 9,100 /Im2 of sediment surface. fAt least 10 species were present, but 3 small (mean valve lengths, 10-30 ,um), chain-forming (chain ranging up to at least 250 ,um in length) species strongly dominated both numerically and in terms of biomass. An undetermined species, usually contracted into an approximate spherical shape about 30 ,um in diameter. hFour undetermined hypotrich and holotrich species, ranging approximately from 40 to 130 ,um in length. 'Taxonomic study of 7 specimens found two genera to be represented: Ethmolaimus and Monohystera (E. M. Noffsinger, personal communication). Mean length for all 131 nematodes measured was 793 ,um (range: 147-3,330 ,tm). Includes only filamentous genera (Lyngbya, Oscillatoria, Schizothrix, and Anabaena); unicellular forms probably were present but contributed negligibly to total blue-green biomass. This was located about 70 m off the eastern shore, had a thin region of each area and combined to vield a single composite film of water over it, and consisted of 10-15 cm of soft sedi- sample for that area. The only locations avoided in our sampling ments overlying a firm sandy bottom. In a row along this flat were where the sediment was covered bv excrement or had a and spaced at about 15-m intervals from each other, we fenced deep depression (fresh flamingo track) in it. The larger more off four 4 x 3 m experimental (NF) areas, thereby excluding active organisms-e.g., nematodes-mav have been under- flamingos from them. Each fence consisted of four stakes pro- represented by this sampling method. Samples were preserved jecting about 1.2 m above the water, with two strands of nvlon in 5% formaldehyde and analvzed on return to San Diego. An twine connecting them. A total of 10 control (F) areas, two be- amount of sample equivalent to 0.015, 0.094, 0.17, or 3.5 mm2 tween each pair of NF areas and two at both ends of the row, of bottom sediment (depending on the taxon) was examined un- were delimited by imaginary lines (Fig. 1). der a compound microscope; linear dimensions were measured On January 3 and at 2- to 3-wk intervals over the following and biomass (strictly speaking, biovolume) was calculated for 2 months, the microbenthos was sampled with a microcorer in every organism estimated to have been alive at time of sam- each of the four NF and ten F areas. The microcorer took a pling. triangular plug with a surface area of 6.9 mm2 and a depth of During the experiment, P. andinus continued to feed in the 5 mm. Eight plugs were taken haphazardly from the central F areas but were completely excluded from the NF areas. P. Downloaded by guest on September 28, 2021 4768 Ecology: Hurlbert and Chang Proc. Natl. Acad. Sci. USA 80 (1983) jamesi and P. chilensis preferred to feed in other parts of the sells and S. linearis were found in the sediments of Laguna Puri- lake and were never observed closer than 50-100 m to our studv pica Chico, each in numbers 5-10% of those of live S. wetzeli. site. Light but frequent afternoon rains in January caused a The size structure and taxonomic composition of the micro- gradual rise in water level (Table 1), which later dropped when benthos was altered in a complex way. On the one hand, or- dry weather set in about January 28. As evidenced by their tracks, ganisms very much larger than Surirella (nematodes) and very P. andinus fed in the F areas much less (and possibly not at all) much smaller (amebas) both increased (F vs. NF areas, Feb- during the high water period (February 4) than during earlier ruary 24) proportionately more than did Surirella itself. On the and later periods. Water temperature probablv fluctuated dailv other hand, Surirella showed a greater response to flamingo from a low of 0-50C to a high of 12-220C, averaging cooler in exclusion than did any of the smaller diatoms, causing an up- February than in January. These estimates are based on several ward shift in the diatom size (length)-frequency distribution. series of temperature measurements made during this period For example, on February 24 diatoms <100 tum long accounted at another similarly situated shallow lake (Laguna Colorada, 4,278 for 15.5% of total diatom biomass in F areas but only 7.3% of m) 40 km to the NNW. it in NF areas (Mann-Whitney U test, P < 0.05), whereas ini- tially (January 3) they accounted for 22.2% and 25.4% (P > 0.25), RESULTS AND DISCUSSION respectively. Surirella by itself accounted, at the end of the By 17 days after establishment of the exclosures, a marked dif- experiment, for 81.5% of total diatom biomass in F areas but ference had developed between F and NF areas in the ap- 89.4% of it in NF areas (P > 0.05), whereas initially it ac- pearance of the sediment surface. Inside the fences, the sed- counted for 74.1% and 73.3% (P > 0.05), respectively. iments were smooth and more or less uniformly tan. Over the Alteration of the microbenthic assemblage by P. andinus may rest of the mudflat, including the F areas, the sediment surface have resulted not only from the flamingos' grazing but also from showed disturbance by flamingo beaks and feet and ranged in their concomitant defecation (once every few minutes when color from tan to black, the latter color due to recently exposed feeding) and trampling of sediments. The separate influences sulfide-rich sediments. Ciliates and amebas had become sig- of these three activities cannot be assessed with our data. One nificantly more abundant in the NF areas, but no clear re- possible effect of sediment oxygenation by trampling is the in- sponses of other organisms were apparent. By the "high-water" hibition of nitrogen fixation by photosynthetic purple sulfur sampling date (February 4), there was no strong evidence of a bacteria and nonheterocystous blue-green algae, organisms often difference between F and NF areas for any taxon. But 3 wk extremely abundantjust beneath the surface sediments of lakes later, after water levels had dropped and P. andinus had re- in this region that are little utilized by flamingos (unpublished commenced heavy feeding in the F areas, all major taxa (phyla) data). except the blue-greens became markedly and significantly re- Several factors tended to minimize the impacts observed in duced in abundance in the F areas, as did total microbenthos our experiment. One was its brevity. Another was the un- biomass (Table 1). impeded exchange of water, and microorganisms in suspen- The abundance of the large, oval diatom Surirella wetzeli sion, between the F and NF areas; this occurred daily in re- (Fig. 2) at Laguna Puripica Chico and its dominance of the mi- sponse to the strong afternoon winds characteristic of the region. crobenthos were notable (Table 1). Because only about 110 S. A third was the temporary avoidance of the F areas by P. an- wetzeli cells are required to cover completely 1 mm2 of sedi- dinus when the water level rose (represented by February 4 ment surface, that diatom by itself was sufficiently abundant in sampling date). If flamingos had been excluded from the entire NF areas at the end of the experiment to form a continuous lake and for a long period of time, much more extensive changes layer almost 11/2 cells thick, on average, over the sediment sur- would likely have occurred. face; and even in the F areas, it covered about half of the sed- Though the experimental system was somewhat unusual in iment surface (Table 1). This abundance perhaps should have its details (flamingos on a high-elevation salt lake), we believe been predictable, given the density of P. andinus on the lake. our results are representative, in a conservative way, of the im- The genus Surirella is the principal food of P. andinus and is pact that water birds have on aquatic ecosystems. Their feeding a dominant component of the microbenthos in many altiplano activities can be intense in a great variety of aquatic environ- lakes (unpublished data). The correlation between their dis- ments, especially swamps, marshes, coastal lagoons, marine in- tributions is sufficiently strong that anyone wishing to collect tertidal. and shallow subtidal zones, rivers, and the ponds and quantities of live Surirella in the Central Andes could best lo- shallow lakes of prairies, tundras, deserts, coastal plains, and cate these diatoms by means of a birding telescope! The com- other landscapes. Any well-designed experiment that excludes monest Surirella species in altiplano lakes appear to be S. wetz- either a single abundant species of water bird or all water birds but no live of S. collectively from such a habitat is likely to show the birds to be eli, S. striatula, and S. sella. Frustules, cells, a dominant influence. Their exclusion will produce changes in the abundance and distribution not only of their prey but also of innumerable other species. Large changes in the taxonomic composition, size structure, and productivity of the aquatic community and even in various physicochemical properties of the system often will occur. Such influences are likely to be L o comparable in magnitude to those exerted by fish, which have been demonstrated experimentally to be extensive (12-16). How can water birds do this, given their limited abilities to forage in deep water and the vagility of their flocks? Four fac- tors are important. First, in some habitats the fish community itself is strongly regulated by fish-eating birds (1). Second, birds have high metabolic and feeding rates, much higher, on a gram for gram basis, than those of fish. Third, these.rates are greatest FIG. 2. S. wetzeli (130 Atm long), dominant species in the micro- in cooler regions or seasons (or both), when their aquatic prey benthos of Laguna Puripica Chico. populations are growing and reproducing most slowly and Downloaded by guest on September 28, 2021 Ecology: Hurlbert and Chang Proc. NatL. Acad. Sci. USA 80 (1983) 4769 therefore least able to recover from predation. Finally, the mo- work was supported by grants from the National Geographic Society, mentary feeding or predation pressure exerted by a flock of National Science Foundation (DEB 76-02888), and San Diego State water birds can be very high, much higher than the momentary University Foundation. predation pressure exerted by the fish assemblage in a given 1. Elson, P. E. (1962) BulL Fish. Res. Board Can. 133, 1-87. water body is ever likely to be, because the momentary abun- 2. Goss-Custard, J. D. (1977)J. Appl Ecol 14, 721-739. dance of water birds, unlike that of fish, is not limited by the 3. Schneider, D. (1978) Nature (London) 271, 353-354. past food conditions of a given water body. Thus, a not inor- 4. Dodson, S. I. & Egger, D. L. (1980) Ecology 61, 755-763. dinately large flock of birds visiting a water body for only a few 5. Kahl, M. P. (1974) in Flamingos, eds. Kear, J. & Duplaix-Hall, N. days or weeks might have impacts on that ecosystem that per- (Poyser, Berkhamstead, Great Britain), pp. 93-102. 6. Hurlbert, S. H. & Keith, J. 0. (1979) Auk 96, 328-342. sist for a year or longer (4). 7. Hurlbert, S. H. (1978) Andean Lake and Flamingo Investigations, This last point has a significant consequence for methodol- Technical Report No. 1 (San Diego State Univ., San Diego, CA), ogy: purely descriptive studies comparing aquatic ecosystems pp. 1-16. with water birds present (at some given moment) to similar sys- 8. Miller, A. H. (1963) Condor 65, 289-299. tems without water birds are likely to prove uninformative and 9. Brodkorb, P. (1963) Bull Fla. State Mus. Biol Sci. 7, 180-293. misleading as to the influences exerted by the birds. This stands 10. Feduccia, A. (1978) Am. Sci. 66, 298-304. the early de- 11. Vareschi, E. (1978) Oecologia 32, 11-35. in contrast to the wealth of insight provided by 12. Hall, D., Cooper, W. & Werner, E. (1970) Limnol Oceanogr. 15, scriptive studies comparing lakes with and without fish (17-19). 839-928. Progress in ornitholimnology will be more dependent on care- 13. Hurlbert, S. H., Zedler, J. & Fairbanks, D. (1972) Science 175, filly controlled, large-scale, long-term exclosure experiments. 639-641. 14. Hurlbert, S. H. & Mulla, M. S. (1981) Hydrobiologia 83, 125-151. We thank G. Bejarano and Flora y Fauna Silvestre (Ministerio de 15. Lynch, M. (1979) Limnol Oceanogr. 24, 253-272. Agricultura, La Paz) for permission to study the flamingos; A. Barrero 16. Zaret, T. M. (1980) Predation and Freshwater Communities (Yale and the personnel of Mina Laguna Verde for hospitality and logistical Univ. Press, New Haven, CT). support; J. Keith, C. Mitchell, and T. Saire for field assistance; J. Ver- 17. Omer-Cooper, J. (1948) Proc. Egypt. Acad. Sci. 3, 1. faillie, K. English, and C. Nuss for assistance in data reduction; M. 18. Hrbacek, J. (1962) Rozpr. Cesk. Akad. Ved Rada Mat. Prir. Ved Noffsinger and J. Alexander for nematode identification; C. Land for 72, 1-116. figure preparation; and J. Keith and G. Cox for helpful comments. The 19. Brooks, J. L. & Dodson, S. I. (1965) Science 150, 28-35. Downloaded by guest on September 28, 2021