Limnol. Oceanogr., 38(4), 1993,753-x74 0 1993, by the American Society of Limnology and Oceanography, Inc.

Landscape linkages between geothermal activity and solute composition and ecological response in surface waters draining the Atlantic slope of Costa Rica

Catherine M. Pringle’ Section of Ecology and Systematics and Center for the Environment, Cornell University, Ithaca, New York 14853

Gary L. Rowe Department of Geosciences, Pennsylvania State University, University Park 16802

Frank J. Triska U.S. Geological Survey, Water Resources Division, 345 Middlefield Road, Menlo Park, California 94025

Jose F. Fernandez Institut0 Costarricense de Electricidad, San Jose, Costa Rica

John West Department of Plant Biology, University of California, Berkeley 94720

Abstract

Surface waters draining three different volcanoes in Costa Rica, ranging from dormant to moderately active to explosive, have a wide range of solute compositions that partly reflects the contribution of different types of solute-rich, geothermal waters. Three major physical transport vectors affect flows of geothermally derived solutes: thermally driven convection of volcanic gases and geothermal fluids; lateral and gravity-driven downward transport of geothermal fluids; and wind dispersion of ash, gases, and acid rain. Specific vector combinations interact to determine landscape patterns in solute chemistry and biota: indicator taxa of algae and bacteria rcflcct factors such as high temperature, wind-driven or hydrologically transported acidity, high concentrations of various solutes, and chemical precipitation reactions. Many streams receiving geothermally derived solutes have high levels of soluble reactive phosphorus (SRP) (up to 400 pg liter-l), a nutrient that is typically not measured in geochemical studies of geothermal waters. Regional differences in levels of SRP and other solutes among volcanoes were typically not significant due to high local variation in solute levels among geothermally modified streams and between geothermally modified and unmodified streams on each volcano. Geothermal activity along the volcanic spine of Costa Rica provides a natural source of phosphorus, silica, and other solutes and plays an important role in determining emergent landscape patterns in the solute chemistry of surface waters and aquatic biota.

I Present address: Institute of Ecology, University of Georgia, Athens 30602.

Acknowledgments This paper is dedicated to the memory of Peter Kilham

whose intense interest in linkages between geochemistry and ecological processes has been an inspiration to many of his students and colleagues.

We are especially grateful to Mark Williams and John Melack for their advice in revising the manuscript and to James Affolter, Susan Brantley, Owen Bricker, Bruce Haines, Robert Hecky, William Melson, Karen Rice, and Fred Scateena for their comments. We thank Rodolfo Var- gas-Ramirez, Ruth Tiffer-Sotomayor, and Victor Hugo- Perez for their help in sampling streams and in nutrient analyses and Gary Zellweger for aluminum, chloride, and sulfate analysis of water samples. We also thank Pat Char- ley and David Jones for their drawings and the Organi- zation for Tropical Studies for providing logistic support.

This study was supported by National Science Foun- dation grant BSR 87-17746 and a National Geographic Society Grant to C. M. Pringle. In addition, the writing of

The solute composition of stream waters may be altered by natural processes occurring aut- side of the surficial catchment, such as geo- thermal activity- the convective circulation of hot fluids through upper portions of the Earth’s crust. For instance, solute-rich waters can’be created by degassing of hot magma followed by condensation and absorption of magmatic gases into groundwater and subsequent fluid- rock interactions. Stream waters that have been modified by volcanic and geothermal process- es are characteristically dominated by solutes

this paper was partially supported by BSR 9 l-07772 to C. M. Pringle and F. J. Triska. G. L. Rowe acknowledges NSF grant EAR 86-57868, the David and Lucille Packard Foundation, and the Penn State Earth Systems Science Center.

753

754 Pringle et al.

that are a distinctive geochemical signature of the processes involved and a source of poten- tial chemical and biotic pattern in the land- scape.

Although geochemists have documented the occurrence of unusual water chemistries in geothermal areas (e.g. White 1957; Henley and Ellis 1983; Henley 1985) and biologists have conducted studies of the biota of extreme types of geothermal waters (e.g. Stockner 1967, 1968; Castenholz 1969; Ashton and Schoeman 1984), the linkage between geothermal processes and emergent regional biotic patterns remains little explored.

High temperature is not necessarily a char- acteristic of geothermal waters. Although high temperatures are involved in the formation of geothermal fluids, geothermally modified wa- ters can be identified after they have cooled by their distinctive solute chemistries. Geochem- ical studies of geothermal waters rarely include measurements of phosphorus (but see Giggen- bath 1974), although it is an important nutri- ent for primary production. Likewise, few bi- ological studies have analyzed effects of high concentrations of geothermally derived solutes such as phosphorus on nutrient cycling (but see Pringle and Triska 199 1) or identified re- gional chemical and biotic patterns in geo- thermal surface waters.

Geothermal systems are common in tecton- ically active zones of the Earth’s crust. Al- though the volcanic history of Central Amer- ican landscapes offers dramatic illustrations of the relationships among past geomorphic de- velopment, current geothermal activity, and surface water composition, no studies have ex- amined regional landscape patterns in the chemistry of geothermal streams in Central America. Highly studied, active volcanoes in Costa Rica provide the opportunity to exam- ine how volcanism and associated geothermal activity can affect patterns in the composition of surface waters draining volcanic landscapes. Results can then be applied to older and less- studied, dormant volcanoes where effects of underlying geothermal activity may be less ap- parent.

This work is an outgrowth of studies on the nutrient dynamics of low-order streams locat- ed at the base of a dormant volcano (Barva) in Costa Rica’s Cordillera Central (Pringle and Triska 199 1). This and other work (Pringle et al. 1986, 1990; Pringle 1991) documents ex-

trcme variations in nutrient concentrations in closely adjacent streams. Algal growth in these streams was affected by stream-water phos- phorus content, which is largely controlled by the presence or absence of solute-rich ground- water inflows rich in P; algal growth in tribu- taries fed by P-rich inflows was P sufficient while algal growth in tributaries without P-rich inflows was P limited (Pringle and Triska 199 1). Similarities in the chemical features of P-rich springs and streams at the base of Barva Vol- cano with the chemical features of known geo- thermal waters present in other volcanic and geothermal systems suggest that Barva’s P-rich springs are affected by geothermal activity (Pringle et al. 1990).

The regional focus of this study precludes in-depth study of any one stream and is de- signed to place previous studies (e.g. Pringle and Triska 199 1) in the context of the regional volcanic landscape. The new discipline of landscape ecology (Forman and Godron 1986; Risser 1987), which focuses on the inftuence of landscape heterogeneity on abiotic and bi- otic processes, has illustrated that temporal and spatial scales of many ecological studies are too small. In this study we apply concepts of landscape ecology to examine how underlying geological processes affect landscape patterns in stream solute composition and produce a biological response.

Our objectives are to identify landscape pat- terns in the occurrence of geothermal waters and the solute composition of these waters within volcanic landscapes that exhibit vary- ing types of volcanic and geothermal activity; to compare these landscape patterns with the distribution of aquatic biota (periphyton and bacteria) at both regional and local scales (be- tween volcanoes and between watersheds of the same volcano); to establish the linkage be- tween landscape patterns in the composition of geothermal waters, physical transport vec- tors associated with volcanic and geothermal activity, and emergent biotic patterns; and to discuss regional implications of volcanic and geothermal activity on stream solute compo- sitions from the perspective of landscape ecol- WY-

Description of study site: Geological setting

Three volcanoes are examined (Fig. 1). Bar- va, a dormant volcano with its last magmatic

Landscape linkages and streams 755

eruption in 1867, is in the Cordillera Central; Poas, a moderately active volcano with its last magmatic eruption between 1953 and 1959, is in the same mountain range, directly adjacent to Barva to the northwest; Arenal, an explo- sively active volcano currently in an active phase of magmatic eruption, is in the Cordille- ra de Tilaran, -60 km to the northwest of Poas.

Recent work on the age of Plinian tephras of Costa Rican volcanoes has established the timing and intensity of eruptions (Melson and Saenz 1973; Melson 1984; Borgia et al. 1988). The edifices of these three volcanoes are con- structed of alternating sequences of lava flows and pyroclastic deposits (ashflows, lahars, and ignimbrites) classified as basaltic and andesitic in composition. These lavas and pyroclastic materials display silica contents in the range of 45-60 wt%, much of which is contained in easily weathered volcanic glass. Primary crys- tal phases of these rocks are plagioclasc and pyroxene with lesser amounts of olivine, am- phibole, and Fe-Ti oxides. P contents of these rocks range from 0.15 up to 0.50 wt% P,O,. Representative basalts and andesites collected at Poas display an average P content of 0.22 wt% P205 (Prosser and Caq 1987; G. Rowe unpubl. data). The climate of the Cordillera de Tilaran and Cordillera Central is wet (3-5 m yr- l), with most rain falling from May to No- vember.

Barva (2,906 m asl), the largest and least studied of the three volcanoes (Fig. 2), consists mainly of deposits older than 300 yr. Its rugged terrain is covered with dense vegetation, rang- ing from tropical wet forest in the Altantic lowlands (35-250 m asl) to montane forest at its summit (2,500-2,906 m asl) (Pringle et al. 1985, 1990). We confined our studies to an- thropogenically undisturbed watersheds on the Atlantic (north) side of Barva, primarily within Braulio Carrillo National Park and La Selva Biological Reserve (Fig. 2). The La Selva- Braulio Car&lo land corridor is the last tran- sect of intact rain forest spanning elevations from 35 to 2,900 m as1 left on the Atlantic slope of Central America (Pringle ct al. 1985).

Poas (Fig. 3) is a composite volcano (2,708 m asl) that has a long history of eruptions af- fecting areas near the crater. Peas has dis- played several types of volcanic activity since first observed in 1828, including frequent phreatic (geyserlike) eruptions of steam, water,

and mud through a hot crater lake which occur over time frames of minutes to years; rare phreato-magmatic eruptions involving ejec- tion ofash and lava(1834, 1910, 1953-1954); and nearly continuous fumarolic activity char- acterized by release of high-temperature steam and volcanic gasses (Raccichini and Bennett 1979; Casertano et al. 1983). Water levels in the hot crater lake (Laguna Caliente) are af- fected by the magnitude of underlying volcanic activity; the lake has disappeared several times prior to renewed eruptive activity. Its most recent disappearance occurred in April 1989 just before this study and was followed by phreatic activity which ejected mud, sulfur, and old lake sediments to heights approaching 2 km, with ashfall affecting the western flank of the volcano (Rowe et al. 1992). Conden- sation of acidic volcanic gases (S02, H2S, HCl) released by summit fumarolic activity results in severe acid rain which, due to prevailing wind conditions, primarily affects the western flank of the volcano (Rosario-Alfaro et al. 1986; Rowe et al. 1992, in press).

The upper watersheds of rivers draining the densely vegetated slopes are within Volcan Poas National Park. The lower and intermediate slopes are in agriculture. We focused our stud- ies on the northern slope, where watersheds

756 Pringle et al.

Fig. 2. Location of geothermally modified streams (0) zncountered in the Volcan Barva study area.

are relatively undisturbed by anthropogenic activities.

Arena1 (1,633 m asl) is the youngest and most active of the three volcanoes (Fig. 4). In its 4,000-yr lifespan it has experienced at least nine cataclysmic eruptions, making it Costa Rica’s most explosive volcano (Melson 1984). Arena1 has been erupting continuously since 28 July 1968, after 400 yr of dormancy. Ac- tivity is primarily explosive Strombolian erup- tions with associated block lava flows, rare py- roclastic flows, and lahar deposits. Arena1 is

Fig. 3. Location of geothermally modified streams (0) draining Volcan Poas. Those streams sampled that ap- peared to be unmodified by geothermal activity are also indicated (0). Site numbers are referred to in Fig. 5.

also affected by acid rain, but it is not nearly $s severe as that observed at Poas (W. Melson pers. comm.). Large areas of undisturbed pre- montane wet forest exist on its southern slopes. Extensive portions of the northern slopes are covered with exposed lava flows and tephras. Erosion of unvegetated slopes is rapid and large mudflows are common (Funk and Melson 1989). Flows originating from 1968 to present have built an immense lava field on the west slope (Fig. 4). Arena1 thus provides the youn- gest volcanic terrain of the three study areas.

Methods Waters of first-third order streams draining

Barva (n = 1 O-3 l), Poas (n = 14), and Arena1 (n = 9) were sampled for major nutrients [NOa- + N03--N, NH4+-N, soluble reactive P (SRP), total P (TP)], major solutes and trace metals (Al, Ba, Be, Cd, Ca, Cl-, Co, Cu, Fe, Pb, Li, Mg, Mn, MO, Na, Si, SOd2-, Sr, V, Zn), and other chemical and physical parameters (con- ductance, pH, alkalinity, and temperature). Waters containing solutes introduced by vol- canic or geothermal activity or those affected by mixing with geothermal fluids were iden- tified by their solute compositions and tem-

Landscape linkages and streams 757

Fig. 4. Location of geothermally modified streams (0) draining Volcan Arenal. Those streams sampled that ap- pcared to be unmodified by geothermal activity are also indicated (0).

perature and were classified according to a modified version of the general classification system for geothermal waters presented by White (1957) and Henley (1985). In this sys- tem, waters are classified according to their pH and the dominant anions and cations present (Table 1; e.g. Na-Cl, Na-HCO,, and acid-SO,). The origins of the various water types have been related to the specific geothermal pro- cesses and environments (Henley 1985) dis- cussed below.

Low-order streams were sampled once in each of the volcanic study areas during periods of minimal rainfall in July 1989. Additionally, an expedition sponsored by the National Geo- graphic Society in the dry season of 1986 al- lowed us to sample high-elevation areas on the north slope of Barva, accounting for the vari- able sample size in Table 2 for some param- cters. We also sampled geothermal hot springs that arise in the Rio San Rafael drainage at the northern base of Platanar Volcano (2,183 m) in the Cordillera Central to the northwest of Poas (Fig. 1) for purposes of comparison with other geothermal waters in our study sites.

Two replicate filtered (0.45-pm Milliporc) and two replicate unfiltered samples were col- lected in acid-cleaned polyethylene bottles at each site for chemical analyses. Samples were transported on ice to La Selva Biological Sta-

Table 1. General classification system for geothermal waters used in this study (modified from White 1957; Henley 1985).

Water type

Main water types Na-Cl

Na-HCO,

Principal pH range anions

4-9 Cl

5-7 HC03

Principal cations

Na, lesser Ca

Na, lesser Ca

Steam-heated waters Acid-SO, o-3 so, H+ , lesser

Na, Ca, Mg, signif- icant Al, Fe

Na-SO,-HC03 4.5-7 S04, lesser Na, lesser HCO, Ca, trace

Mg, Fe, Al

Mixed water types Acid-SO&l o-5 SO& Cl H -+ , lesser

Na, Ca, Mg, signif- icant Al, Fe

Na-Cl-HCO, 7-8.5 Cl, lesser Na, lesser HCOJ Ca

Dilute Cl 6.5-7.5 Cl, lesser Na, lesser HCO, Ca

tion where filtered samples were analyzed for SRP (Strickland and Parsons 1972), NH4+-N (Solorzano 1969), and N02- + NO,--N (Kamphake et al. 1967). TP was measured in unfiltered samples with the molybdenum blue technique (Strickland and Parsons 1972) pre- ceded by acid hydrolysis. pH was measured with an Orion electrode, and alkalinity was determined by potentiometric titration (Am. Public Health Assoc. 1985). Conductance was measured with a conductivity meter at 25°C.

One filtered sample was preserved with ni- tric acid (Ultrex) for major cation and trace metal analysis at the central laboratory of the U.S. Geological Survey (Denver, Colorado) by inductively coupled plasma spectroscopy. Ad- ditional filtered samples were collected for analysis of Al which was measured via atomic- absorption flame spectrophotometry, and Cl- and SOd2- which were measured via ion chro- matography. A Student-Newman-ISeuls test (Sokal and Rohlf 198 1) compared means for key elements between geothermal waters and

758 Pringle et al.

Table 2. Means (X), standard deviations (k SD), ranges (min-max), and sample size (n) for temperature, conduc- tivity, pH, and various solutes in first-third-order streams draining Barva, Poas, and Arenal.

Barva Poas Arena1

Tcmp (“C)

Cond. ($S cm-‘)

HCO,- (mg liter-‘)

PH

NO,- + NO,--N (pg liter I)

NH,-N (pg liter-l)

SRP (pg liter-‘)

DIN : SRP

TP (pg liter - I)

Al (gg liter-‘)

Ba (pg liter--l)

Ca (mg liter I)

Cl - (mg liter-l)

Fc (mg liter- I)

Li (mg liter-‘)

Mg (mg liter I)

Mn (pg liter *)

Na (mg liter- I)

Si (mg liter-‘)

x range n x range n R range n range n x range n x range n x range n range n .z range n x range n x range n x range n x range n x range n R range n R range n x range n x range n 2 range n

20.2k5.8 12.0-25.5

37 280.2k 153.6

11 J-425.0 37

85.4k 111.8 0.0-440.0

27 4.6-7.1

37 122.61k64.9

0.0-39 1.2 27

2O.Ok21.8 0.0-102.0

27 75.3k94.5 <5.0-301.2

27 0.2-l 00.3

26 89.1+110.1 < 5.0-405.0

28 143.lk69.4

65.5-223.0 4

19.Okll.4 3.0-41 .o

10 2.4k4.3 0.0-19.0

31 5.8k6.6 1.8-28.5

31 33.9k23.2 12.0-80.0

10 0.2kO.7 0.0-2.1

10 3.3k6.3 0.0-25.3

31 20.8k42.8

0.0-l 40.0 10

7.7+ 10.2 0.6-44.0

31 2l.lkl9.5

4.4-71-o 10

16.61k 1.7 26.7k5.9 12.5-19.0 22.0-36.0

14 9 5 16.2+ 1,493.5 1,033.5*1,166.4

47.0-5,700.o 75.0-3,000.0 14 9

21.5+ 13.5 351.2-1213.1 0.0-45.0 55.3-611.8

14 9 2.3-7.3 6.9-8.3

14 9 60.7k69.2 229.2k273.5

0.0-285.5 28.3-950.0 14 10

34.1 k37.0 82.1 f 82.0 12.9-l 57.8 16.8-268.4

14 10 55.7k 120.6 88.4f77.3 10.1-471.2 7.9-234.1

14 10 1.4-29.6 3.4-54.7

14 9 57.2_+ 123.0 102.11k84.8 10.9-480.6 9.1-279.1

14 10 42.4k58.9 26.2k23.8

0.7-l 54.0 0.0-62.9 14 9

11.9fl7.0 64.5k76.8 2.2-66.0 4.0-213.9

14 9 15.7k20.3 60.2k55.7

3.2-84.0 8.9-144.9 14 9

34.31k 106.3 44.3k53.3 1.4-403.0 0.3-l 32.3

14 10 21.8224.6 17.3k36.4

2.7-76.0 4.0-56.0 14 9

4.2k7.5 18.2k22.1 0.0-30.0 0.0-64.0

14 9 7.5+: 16.6 35.6k38.3 0.7-65.0 I .7-100.0

14 9 229.6f752.9 917.2+1,328.4

0.6-2,840.O 1 .O-2,809.O 14 9

7.6+- 13.7 77.6k97.2 0.6-55.0 4.3-240.0

14 10 40.7k20.1 73.7k23.4

6.3-95.0 35.0-l 10.0 14 9

Landscape linkages and streams 759

Table 2. Continued.

SO,*- (mg liter-‘)

Sr (pg liter-‘)

V (pg liter- ‘)

Zn (pg liter- I)

x 10.4k24.0 range 0.3-71.9 n 31 x 32.2k21.3 range 6.0-79.0 n 10 x 1.4k2.9 range 0.0-9.0 n 10 x 13.7f 11.9 range 0.0-40.0 n 10

109.11322.0 44.01-67.5 1.8-l ,225.0 1.2-221.2

14 10 131.8xkl97.2 484.7t-465.4

25.0-800.0 68.0-l ,236.4 14 9

21.3k67.6 13.9+ 17.6 ’ 0.0-256.0 0.0-53.0

14 9 19.3k57.2 5.8k7.1

0.0-217.0 0.0-20.4 14 9

those unmodified by geothermally introduced solutes and between different types of geo- thermal waters. A one-way ANOVA examined variation of SRP between and among the three volcanoes. Calculations were made with the geochemical model program, WATEQ4F (Ball et al. 1987).

Precipitation collectors were placed in the lower watersheds of the Anonas, Desaguc, Agrio, Gata, Lata, and Pozo Azul Rivers, on the northwestern flank of Poas. Precipitation collectors were located at sample sites on these rivers as indicated in Fig. 3. One sample col- lector was placed at each of the six sites men- tioned above for a 12-h period on each of four sampling dates. Precipitation was analyzed for pH, Cl-, and SOd2- according to methods dc- scribed above.

Qualitative samples of epilithic algae were collected by scraping five rocks per site with a small brush and razor blade. Scrapings were preserved with 5% Formalin and algal taxa were later identified with phase and bright- field microscopy. Dominant algal taxa were determined through direct counts on a Sedg- wick-Rafter cell. Algal standing crop was vi- sually estimated as sparse (little to no visible algal growth), moderate (a visible algal coating but < 1 mm thick), or abundant (a thick algal coating > 1 mm thick). Where macroscopic growths of bacteria were present, they were collected and identified based on macroscopic characteristics and light microscopy.

study areas (Table 2). A one-way ANOVA in- dicated that regional differences between vol- canoes in concentrations of most stream sol- utes (Table 2) were not significant (P > 0.05). General trends were evident. Streams draining Barva had the lowest mean conductance and lowest mean concentrations of most solutes (c.g. Ca, Cl-, Li, Mg, Mn, Si, SOd2-, Sr, and V). Surface waters of Arena1 had the highest mean conductance, highest mean concentra- tions of solutes (Ba, Ca, Cl, Li, Mg, Mn, Na, Si, Sr), and highest concentrations of major nutrients (SRP, TP, N02- + N03--N, and NH,+-N; Table 2). Concentrations of Be, Cd, Co, Cu, Pb, and MO were below detection for most of the streams sampled (Be < 0.5 E.cg liter-l; Cd < 1.0 pg liter-I; Co < 3 hg liter-l; Cu < 3 pg liter- ‘; Pb < 10 hg liter-r; MO < 10 pg liter-l).

P levels (SRP, TP) averaged for streams on each volcano were moderately high (Table 2); however, wide variability existed among streams draining a given volcano. Poas had the widest range in SRP levels, 1 O-47 1 pg SRP liter-‘.

Results

Landscape patterns in stream solute com- position -Two types of geothermal waters were identified from Barva (Table 3): dilute, Na- Cl-HC03 waters with neutral pH and signifi- cantly higher (P < 0.01) conductance, HCO,, Ca, Cl-, Fe, Mg, Na, SRP, TP, Si, and SOd2- than solute-poor waters that are unmodified by geothermal activity; and acid-SO, waters characterized by low pH and high levels of SOd2-, Si, SRP, and Ca (Table 3).

Surface waters displayed a large variation in solute composition within each of the three

With one exception, all Na-Cl-HCO, waters we encountered came from low elevations (be-

760 Pringle et al.

Table 3. Comparison of chemical properties of first-order streams draining the Atlantic slope of Barva that are unmodified (solute-poor) by geothermal activity vs. geothermal waters (Na-Cl-HCO, and acid-SO,). Means @*SD), ranges (min-max), and sample sizes (n) are indicated. The elevation or elevational range (m asl) at which these water types were encountered is indicated (modified from Pringle 199 1).

Geothermal

Unmodified (35-2,900 m)

Na-CI-HCO, (35-350 rnj

Acid-SO, (2.000 rnb

PH Cond. (@ cm - *)

HCO,- (mg liter-l)

Ca (mg liter-‘)

Cl- (mg liter- ‘)

Fe (mg liter-l)

Mg (mg liter-‘)

Na (mg liter-‘)

NO2 + NO,--N (pg liter- I)

SRP (pg liter-l)

TP (pg liter- I)

Si (mg liter-l)

SO,*- (mg liter-‘)

n

range x range R range x range x range x range x range x range R range x range x range x range x range

6.9-7.1 15.1 k3.8 11 .O-23.0 5.6k3.2 0.0-10.0 0.2kO.2 0.0-0.7 2.1 kO.2 1.9-2.6

6.Ok5.0 3.0-10.0 0.3kO.2 0.0-0.5 1.3kO.4 0.6-l .7

138.01~75.1 69.0-266.0 8.9k5.6

<5.0-l 5.0 13.4k8.9 <5.0-25.0

6.Ok2.8 4.4-9.0 0.4&O. 1 0.3-0.6

8

6.8-7.1 280.8& 106.0 150.0-425.0

253.0-c 109.2 148.0-440.0

8.4f6.0 3.0-19.0

15.5k8.1 8.4-28.5

43.0-c 13.2 26.0-50.0 12.5k8.8

4.9-25.3 22.8k 12.2 11.9-44.0

111.31k63.4 30.0-l 73.0

221.7t-66.2 101.0-290.0

257.3k90.1 142.0-405.0 48.3+ 12.0 38.0-54.2 5.4k4.1 2.4-l 2.8

8

4.6 400.0

0.0

17.0

5.6

80.0

5.4

6.5

204.0

112.2

115.6

71.0

71.9

tween 35 and 150 m asl) at the base of the north slope where the foothills of the Cordille- ra Central merge with the coastal plain (Fig. 2). The one exception came from a lava dike at 350 m asl. The only acid-SO, stream iden- tified was the Azufre spring which has its origin at -2,000 m as1 on the north slope (southern- most geothermal site indicated in Fig. 2). The smell of H2S is very pronounced in the area near the Azufre’s origin during the dry season (pers. obs.).

Landscape patterns in the occurrence and solute composition of geothermal waters draining Poas were identified (Table 4, Figs. 3, 5): acid-SO,-Cl waters with lower pH and higher conductance, Ca, Cl-, Fe, Mg, Na, N02- + N03--N, SRP, TP, Si, and SOd2- than all other streams sampled on Poas; and slightly acid to neutral, solute-rich waters with lower

pH and significantly higher (P < 0.05) con- ductivity, Ca, Cl-, Mg, Na, and SOd2- than streams unmodified by geothermal activity.

The Rio Agrio, which has its source on the northwest flank of the active Poas crater (Fig. 3), is the only acid-SO,-Cl stream we encoun- tered and is the most solute-rich stream we sampled (Table 4). Slightly acid to neutral, sol- ute-rich waters occur on the northwestern flank (Anonas, Desague, Gata, Lata), while streams unmodified by geothermally derived solutes are located on the northeastern flanks (Table 4; Figs. 3, 5).

Cl- and SOd2- concentrations in acid to ncu- tral, solute-rich streams on the western flank exhibit a pattern of diminishing concentra- tions to the north and south of Rio Agrio, while pH increases (Figs. 3, 5). Mean Cl- and SOd2- concentrations in surface waters draining the

Landscape linkages and streams 761

Table 4. Comparison of chemical properties of surface waters draining Poas. Unmodified (solute-poor) refers to waters that are not contaminated by geothermally derived solutes; geothermal refers to waters that have been affected by geothermal processes. Means (X&SD), ranges (min-max), and sample sizes (n) are indicated. Two types ofgeothermal waters are compared: slightly acid, solute-rich waters and acid-SO&l waters.

PH Cond. ($S cm- I)

HCO, - (mg liter-l)

Ca (mg liter -I)

Cl- (mg liter-‘)

Fe (mg liter-l)

Mg (mg liter-l)

Na (mg liter-l)

NO,- + NO,--N (pg liter- I)

SRP (pg liter- ‘)

TP (pg liter-l)

Si (mg liter-l)

Sod2 (mg liter-‘)

n

range x range x range ji- range x range x range x range w range R range x range R range x range x range

Unmodified (NE flank)

6.7-7.1 82.5k27.7 48.0-l 20.0 29.Ok6.8 20.0-32.5 8.0f2.6 4.6-10.9 2.2kO.9 1.4-3.6

9.7Ik9.3 0.7-26.2 2.6kO.8 1.6-3.9

3.7+: 1.3 2.1-5.4

58.9& 19.2 . 39.5-94.1 29.9zk 14.6 15.2-45.8

30.3+ 15.4 14.0-50.2

35.9k8.9 23.7-45.4 10.3k8.3

1.8-20.6

7

Slightly acid (NW flank)

4.8-6.7 192.5k73.5 120.0-295.0 14.5kl2.3

0.0-30.0 16.6f3.9 12.2-21.4 13.1 k8.4

6.6-25.0 81.1k109.5

3.0-240.8 4.4* 1 .o 3.2-5.6 5.541.2 4.9-7.1

17.4412.9 0.0-28.7

12.1t-1.9 10.7-14.7 12.9k2.4 10.9-16.4

43.6k4.9 36.7-48.5 5 l.Ok26.7 30.9-89.3

4

Geothermal

Acid-SO,43

Agrio Laguna* (NW flank) (summit)

2.3 0.0 5,700.o -

0.0 -

84.0 2,480.O

403.0 32,200.O

52.4 1,080.O

65.0 265.0

55.0 550.0

285.5 -

471.2 -

480.6 33,600t

95.0 200.0

1,225.O 68,300.O

1 1 * Solute chemistry data of Laguna (=Laguna Caliente) collected in January 1987 (Rowe et al. in press). t Values from Rowe (199 1).

northwestern side of the volcano are signifi- cantly greater than in all drainages on the east- ern side (Fig. 5). In a southerly direction away from Rio Agrio (Fig. 3), sites 2 and 1 have lower Cl- and Sod*- concentrations of 25 and 8 mg Cl- liter-’ and 89 and 35 mg SO,*- liter- I. In a northerly direction away from the Agrio, sites 5 and 6 exhibit lower Cl- and SO,*- concentrations of 12.8 and 6.6 mg Cl- liter-’ and 48.8 and 30.9 mg SO,*- liter-‘. In con- trast, on the eastern side of the volcano, Cl- and Sod*- concentrations are low, ranging from 1.4 to 3.6 mg Cl- liter-’ and 1.8 to 23.8 mg Sod*- liter-l (stream sites 7-13; Figs. 3, 5). With the exception of the Rio Agrio, SRP lev- els are significantly lower on the northwestern flank of the volcano (range, 10.7-14.7 bg SRP

liter-l) relative to the northeastern flank (15.2- 57.8 pg SRP liter-l).

Precipitation collected at the Rio Desague (site 2; Fig. 3) had the lowest pH relative to precipitation collected at other streams on the northwestern flank (Table 5).

Waters of both the Quebrada Gata and the Rio Dcsague, which occur in parallel drainages to the north and south of the Rio Agrio (Fig. 3), were milky blue. This color results from refractory properties of a suspended white pre- cipitate that also covered rocks in the Gata in a dense floc (pers. obs.).

Thermal streams and springs classified as neutral, Na-Cl-HCO, and Na-Cl-SO,-HCO, waters (Table 6) issue from the base of the massive field of coalescing lava flows that cov-

762 Pringle et al.

ACID RAIN

vt +t 7-

Table 5. pH, Cl-, and Sod2 (mg liter I) measured in precipitation collected in selected drainages on the western flank of Poas.

Stream Date PH Cl SO,,’

Pozo Azul 16 Nov 89 5.0 2.3 1.5 30 May 90 - 0.8 2.0

Anonas

Dcsaguc

Agrio

Gata

Lata

16 Nov 89 30 May 90 29 Jul 90

9 act 90 16 Nov 89 30 May 90 29 Jul 90

9 Ott 90 16 Nov 89 30 May 90 29 Jul 90

9 Ott 90 16 Nov 89 30 May 90 29 Jul90

9 Ott 90 16 Nov 89 30 May 90 29 Jul90

9 Ott 90

4.4 1.8 0.5 4.0 0.8 2.4 4.0 - 5.8 1.7 I.0 3.9 1.6 0.5 3.9 1.2 2.9 3.6 3.0 2.9 5.1 1.1 0.7 4.0 0.9 0.5 4.3 1.0 1.2 4.1 3.0 2.0 5.8 1.0 2.3 4.3 1.2 0.5 - - -

- - - 4.2 1.7 0.3 - - - -

6.5 -

4.3 -

0.7

1225

- A

29 Jul90 - - I

v- 9 Ott 90 6.2 1.0 0.7 L a,

z lOO*

0 E

cr the northwestern flank of Arenal. These flows

- 501 and tephra units (Fig. 4) were deposited during

B the current eruptive cycle which began in 1968. Both types of geothermal waters have signifi-

12 4 5 6 7 8 9 10 11 12 13 cantly greater (P < 0.05) conductance, alka- I 8 linity, Ca, Cl-, Mg, Na, Si, and SOd2- than NORTHWESTRANK NCXII-EAST FLANK streams draining older flows on the north-

4 eastern and southwestern flanks of the volcano

ACID BRINE (Table 6). Distribution of dominant algal taxa - Dom-

Y

inant algal taxa in selected streams draining Barva, Poas, Arenal, and Platanar are pre- sented in Table 7. Of the 27 streams sampled for both water chemistry and algae, 12 were selected to represent a range of geothermal wa- ter types. Specifically, for each volcano, rivers were selected to represent each type of geo- thermal water type occurring on that particular volcano. One representative river on each vol- cano was selected that is unaffected by geo- thermal activity.

Poas Drainage Fig. 5. Landscape patterns in pH, chloride (mg Cl-

liter- I), and sulfate (mg SOd2- liter I) in second- and third- order streams draining the wcstcrn and eastern flanks of Poas, which we hypothesize are influenced by wind-dis- persed acid rain (black bars), subsurface volcanic brines, and acid rain (striped bars), and those that appear to be little influenced by these factors (white bars). Arrows in- dicate which of these agents are hypothesized to dominate chemical composition. Poas drainage numbers on the x-axis correspond to clockwise arrangement of stream sampling sites illustrated in Fig. 3. (Sample site 3 is omitted because it is a first-order seep fed by local groundwater that feeds into sample site 2 which is a third-order stream; see Fig. 3.)

Diatoms (Bacillariophyta) were the domi- nant group of algae in all localities except for thermal springs (Tucaron, Tabacon 1 and 2,

Landscape linkages and streams 763

Table 6. Comparison of chemical properties of surface waters draining Arenal. Means (X-SD), ranges (min-max), and sample sizes (n) are indicated. Unmodified (solute-poor) refers to waters that arc not contaminated by geothermally derived solutes; geothermal refers to waters that have been affected by geothermal processes: Na-Cl-HCO,, and Na- Cl-SO,-HCO, waters.

Temp. (“C) PH Cond. (@ cm-‘)

HCO, - (mg liter-l)

Ca (mg liter-‘)

Cl (mg liter-l)

Fe (mg liter-‘)

Mg (mg liter-‘)

Na (mg liter-‘)

NO, + NO, -N (pg liter- ‘)

SRP (pg liter-‘)

TP (pg liter-‘)

Si (mg liter-l)

SOdz- (mg liter- ‘)

n

range range x range x range

x range x range

x range R range K range R range x range K range x range x range

Unmodified (SW Rank)

22.0-23.0 7.2-8.3

227.52 133.3 75.0-400.0

127.7k90.9 55.3-281.2 16.5t-9.7

8.9-30.0 3.9k3.2 0.3-8.2

16.2k9.3 4.0-24.0 7. I k8.1 1.7-19.0

8.6k5.7 4.3-l 7.0

148.61k89.1 28.3-236.3 55.6k30.5 28.3-99.2 87.4k52.9 45.9-100.5 56.2& 16.2 35.0-74.0 3.2k2.9 1.2-7.5

5

Geothermal (NW Rank)

Na-CI-HCO, Na-Cl-SO.,-HCO,

24.0-35.0 32.0-36.0 7.8-7.9 7.3-8.0

1,7oo.o-t141.4 2,900.0+ 141.4 1,600.0-l ,800.O ,2,800.0-3,000.0

498.5k20.5 606.1k8.1 484.0-5 13.0 600.4-6 11 A 86.9t-7.2 142.5-13.5 81.8-92.0 140.0-144.9

100.7k44.6 106.7k2.2 69. I-l 32.2 105.1-108.2 12.1k6.4 5.1 kO.2

7.6-l 6.6 5.0-5.2 50.3_+ 10.5 93.5k9.2 42.9-57.7 87.0-100.0 98.4k61.9 230.3+ 13.8 54.6-l 42.2 220.5-240.0

231.9&204.5 520.4k607.6 87.3-376.5 90.7-950.0 60.7k37.6 104.4* 109.7 87.3-234.1 26.8-181.9

182.7k 136.3 135.7k84.5 86.3-279.1 75.9-l 95.4 88.4f9.5 lOl.lk12.7 8 1.7-95.1 92. l-l 10.0 35.3k7.3 153.7k95.5 30.1-40.4 86.1-221.2

2 2

Arena1 1), acidic environments (Agrio and Azufre), and streams affected by heavy mineral precipitation (Gata, Desague). Euglena mu- tabilis Schmitz (Euglenophyta) was observed at only one sampling site in acid waters (pH 2.3) of the Rio Agrio. Various green algae (Chlorophyta) were diverse in distribution at the sites studied. None were reproductive; con- sequently, more detailed identification was not possible. Micruspora was dominant in acidic waters of the Rio Azufre (pH 4.6) and present in acid springs (pH 4.1) at the summit of Poas (site 14, Fig. 4). It was also dominant in the neutral, solute-rich waters of the Guacimo spring (pH 6.7). All of these waters were rich in P. Ulothrix was present in acid waters of the Agrio and the Desague (Table 7).

(Cyanophyta) in thermal areas, the generic di- versity we encountered was relatively low. Os- cillatoria and Lyngbya were the most com- mon, followed by Dermocarpa, Pleurocapsa, Phormidium, and Scytonema. Albrightia tor- tuosa Copeland and Thalpophila imperialis Copeland were identified from the Tucaron thermal springs (62°C).

Red algae (Rhodophyta) were infrequent in stream samples. The most conspicuous was Compsopogon coerulens (I. Ag.) Kuetz which grew abundantly in the Rio San Rafael at the base of Volcan Platanar, in the Tabacon hot springs at the base of Arenal, and in the Salto at the base of Barva-all sites receiving geo- thermal inputs. Two species of Audouinella were found on all three volcanoes.

Given the ubiquitous nature of blue-greens Filamentous photoheterotrophic bacteria of

764 Pringle et al.

Table 7. Selected physical [temperature (“C); pH; alkalinity (mg HCO,- liter-l); presence (+) or absence (0) of visible chemical precipitate], chemical (Cl, SOd2-, SRP, NO,--N), and biological parameters [dominant algal taxa listed in order of decreasing dominance; relative algal standing crop (S-none to sparse; M-moderate; A-abundant)] in selcctcd geothermally modified streams draining each volcano.

Stream

Cl SO,’ SRP NO,-N Dom. taxa and Stand- Pre-

Temp. pH macroscopic

Alk. tip. (mg liter ‘) ing

(jig liter ‘) bacteria croo

Barva Azufre 13.0 4.6 nd 0 5.6 71.9 112.0 204.0 Microspora, sulfur-oxi- S dizing bacteria

Guacimo Spring 23.0 6.7 440 0 28.5 12.8 301 .O 30.0 Microspora, Fragilaria A Salt0 24.5 6.8 256 0 10.9 4.0 56.6 84.1 M

Poas

Pantano

Desague Agrio

24.5 7.0 30 0 3.3 1.0 14.9 43.6

17.5 4.8 nd + 25.0 89.3 11.8 15.5 19.0 2.3 nd 0 403.0 1,225.0 471.2 285.5

Lyngbya, Dermocarpa, Compsopogon, Audoui- nella, Achnanthes, Terpsinoe

Navicula S

S M

Gata 17.0 6.3 15 -k 12.8 48.8 11.3 25.4

Angel 18.0 6.8 20 0 2.4

Arena1 Tabacon 1, Tabacon 2

36.0 8.0 611 0

20.6 23.7

221.2 181.9

48.6

105.1 950.0

Arena1 1 35.0 7.8 459 0 69.1 40.4 234.0 376.0 Guillermina 13.0 6.9 68 0 3.6 1.9 7.9 34.5

Platanar Tucaron 62.0 7.0 912 0 X38.5 62.7 261.9 8.9

Ulothrix Euglena mutabilis, Ulo-

thrix, Erustulia rhom- boides, Cocconeis, Pin- n&aria

Lyngbya, Oscillatoria, Scytonema, Pleurocap- sa, Surirella, Melosira, Cocconeis, Chlamy- domonas

Frustulia, Melosira, Coc- coneis, Surirella

Phormidium, Oedogoni- urn, Oscillatoria, flexibacteria

Phormidium, Pinnularia Navicula, Synedra, Melo-

sira, Oscillatoria

Oscillatoria, Albrightia tortuosa, Thalpophila imperialis, flexibacteria

S

S

A

A S

A

the genus Chlorojlexus were observed as a ge- latinous, orange undermat beneath the surface few millimeters of Phormidium in the Tucaron hot springs at the base of Platanar and in hot seepages feeding into the Tabacon of Arena1 (Table 7). Sulfur-oxidizing bacteria were pres- ent as filamentous, white growths where the Azufre spring emerged at 2,000 m as1 on Barva.

Discussion Geothermal processes and associated phys-

ical transport vectors are discussed below as a prerequisite to distinguishing and interpreting the landscape pattern observed in regional and local stream solute composition and examin-

ing emergent biotic patterns and their regional ecological significance.

Geothermalprocesses and the role ofphysical transport vectors - From a landscape ecology perspective, vectors are defined as transport mechanisms or forces that drive flows of ma- terials between elements of the landscape (For- man and Godron 1986). Physical transport vectors impose nonrandom directional fluxes of materials on the general background of re- source concentration gradients and passive dif- fusion (Wiens et al. 1985) and thus play an integral role in determining landscape pat- terns. Here we use the phrase “physical trans- port vectors” to refer to forces that affect the flow of materials within landscapes.

Landscape linkages and streams 765

In areas of active volcanism, heat to drive the subsurface convective circulation of hot fluids is supplied by near-surface magma bod- ies. The fluids involved are either seawater or deeply penetrating groundwaters of meteoric origin. The composition of geothermal fluids is determined by many processes including ad- dition of volatiles and metals from crystalliz- ing and degassing magmas, the boiling of fluids as they rise to the surface, water-rock reac- tions, and mineral dissolution and precipita- tion reactions (Ellis and Mahon 1964, 1967) along subsurface and surface flow paths.

In the geothermally active landscapes of Barva, Poas, and Arena1 we identify three ma- jor physical transport vectors (Table 8) that determine geographic patterns of geothermal waters: thermally driven convection of vol- canic gases and geothermal fluids; lateral trans- port of geothermal fluids downslope along flow paths determined by hydraulic gradients and local hydrogeologic structures (faults, perme- able stratagraphic units, etc.); and wind-di- rected transport of ash, gases, and acid rain generated by fumarolic or eruptive activity. Effects of physical transport vectors (Table 8) are summarized below with respect to each volcano.

Geographic patterns in the composition of types of geothermal waters sampled on the northern flank of Barva (Figs. 2, 6A) corre- spond to physical transport vectors acting sin- gly or in combination: upward convection of low temperature, magmatic vapors rich in H$; and lateral and downward transport of geo- thermal fluids generated in the summit hydro- thermal system (Fig. 6A). These processes were inferred from the two types of geothermal wa- ters identified at Barva (acid-SO, waters pro- duced by the interaction of H,S-rich steam with near-surface groundwater, and Na-Cl- HCO, waters rich in P which outcrop near the base of the volcano; Table 3).

Low pH (4.5), high SOd2- (72 mg liter-l), low Cl - (5.6 mg liter-‘), and elevated cation contents of the Azufre spring (Table 3) are characteristic features of acid-SO, waters (Ta- ble 1) which can form from several processes. The most common origin is condensation of H,S-rich steam produced in a subsurface boil- ing zone into zones of cool, oxygenated groundwater. Hence, acid-SO, waters are often

Table 8. General characteristics of physical transport vectors associated with volcanic activity that determine the solute composition of geothermal waters draining Bar- va, Poas, and Arenal.

Vector A-thermally driven upward convection A, = H,S A, = H,S, HCl, and SO, A, = HCl, CO,

Vector B-lateral and downward transpart B, = long flow path (> 10 km) Bz = short flow path (< 2 km)

Vector C-wind-directed transport C, = acid rain CL = ashfall

__-

called steam-heated waters. f Jcat provided by steam condensation and the oxidation of H,S by oxygenated groundwater results in forma- tion of H,SO, which vigorously attacks sur- rounding wallrocks, resulting in high concen- trations of rock-forming elements such as Na. Ca, Mg, Al, Fe, and Si:

H,S + 20, = 2H” + SO.& (1)

NaAlSi,O, + 4H,O + 4H+ = Na+ + A13+ + 3H,SiO * 49 (2)

Fe,O, + 6H’ + 2Fe3+ + 3H20. (3) The slightly thermal nature of acid-SO, wa-

ters of the Azufre (temperatures are consis- tently l-2°C higher than that of a nearby stream; unpubl. data), combined with the strong odor of H,S which emanates from the spring area during the dry season, is strong evidence that the solute composition of Azufre spring water directly reflects the absorption of H,S-rich steam into near-surface oxygenated groundwaters. Further evidence for summit geothermal activity includes observations of’ low temperature (SSOC) fumarole vents lined with native sulfur and hot springs with tem- peratures of -6OOC in an area of precipitous terrain located -8 km north of the summit (Arce et al. 1980). The presence of native sulfur indicates emission of H,S, suggesting that the hot spring activity reported by Arce et al. ( 1980) is probably the acid-SO, type.

Solute ratios (Ca : Cl, Mg : Cl, Cl : S04, Cl HCOJ of solute-rich waters surfacing at the base of Barva are very similar to ratios of the

Pringle et al.

cold r&ml 0odlurncMoriQ blcabonolo qxlngs

B. Poas

Neutral rodlum-chloriL-blmrbawrte~ hot springs

Fig. 6. General structural models illustrating how geo- thermal processes affect spatial patterns in the chemistry of surface waters draining the volcanoes studied. Physical transport vectors are indicated where wavy arrows indicate upward convection of magmatic vapors, black arrows in- dicate subsurface lateral flow of volcanic waters, and white arrows indicate wind dispersion of acid rain and (or) ash. A. Barva. H,S-rich vapors (I) from a deep, actively cooling magma body are absorbed into surficial groundwaters. H,S oxidation occurs when waters are exposed to the atmo- sphere, emerging (2) as acid-SO, waters of the Azufre spring (2,000 m). Extensive, subsurface, lateral flow of acid geo- thermal waters occurs (3), during which CO, is incorpo- rated. Neutral, Na-Cl-HCO, springs emerge at the base (4) of Barva (35-300 m). B. Poas (modified from White 1957; Henley and Ellis 1983; Brantley et al. 1987). Fu- marolic vapors (1) rich in Cl and S condense in Laguna Caliente (2), an acid-SO&l lake. Downward seepage of brines from the lake (3) enters subsurface groundwater with direct inputs into Rio Agrio. Fumarolic vapors (4) are condensed by precipitation. Acid rain (5) is dispersed by northeasterly winds affecting rivers on the northwest side, with effects concentrated on the Rio Desague with diminishing effects on rivers in drainages to the north (Gata) and south (Anonos) of the Desague. C. Arena1

Table 9. Comparison of mean ratios of major solutes in seawater with stream waters unmodified by geothermal activity (solute-poor streams on Barva and neutral-pH streams draining Poas) and with geothermal waters (sol- ute-rich streams on both Barva and Arenal). Sample size-n.

Unmodified Geothermal

Barva Poas Barva Arenal (solute- (neutral (solute- (solute-

Seawater Poor) pH)* rich) rich)

Na:Cl 0.57 0.62 1.21 1.47 0.46 Ca:Cl 0.02 0.10 3.25 0.51 0.81 Mg:Cl 0.07 0.14 1.31 0.75 0.55 Cl- : so,2- 6.92 7.00 0.30 2.87 2.54 Cl- : HCO,- 1,360 0.75 0.16 0.12 0.18 n 1 8 16 8 1

* Average of 16 neutral-pH streams draining the eastern, western, and south- cm flanks of Poas (Rowe 199 1).

more concentrated, geothermal waters of the Na-Cl-HCO, type that surface at the base of Arena1 (Table 9), further supporting our con- clusion that the former waters are geothermal and of the Na-Cl-HCO, type. Solute compo- sitions and ratios in neutral pH rivers draining Peas; produced by normal carbonic acid weathering of flank material unaffected by geo- thermal activity, are dissimilar to those of sol- ute-rich streams at the base of Barva (Table 9). Data presented in Table 9 from neutral pH rivers of Poas were sampled during the dry season when most river water is of direct groundwater origin and thus representative of the compositional range that would be pro- duced during normal soil-rock weathering in volcanic watersheds unaffected by geothermal activity (e.g. Garrels and Mackenzie 1967). Data presented in Table 9 also illustrate that solute-poor waters surfacing at the base of Bar- va have solute ratios similar to that of sea- water. In contrast to solute-rich waters con- taminated by geothermal fluids, solute-poor waters are composed of rainwater that has been only slightly modified by chemical weathering processes.

Na-Cl waters characterize the deeper por-

t- (modified from Henley and Ellis 1983). Vapors arising from shallow subsurface magma bodies (1) interact with groundwaters arising as neutral, Na-Cl-HCO, (and Na- Cl-SO,-HCO,) hot springs (2) near the terminus of recent lava flows on the north side.

Landscape linkages and streams 767

tions of geothermal systems associated with active volcanoes and are generally produced by incorporation of magmatic fluids rich in chloride and alkali metals. Ascending Na-Cl waters may be further concentrated by near- surface boiling which removes water and other volatile species such as CO2 and H$, leaving behind a residual fluid with an enhanced chlo- ride content. Na-HCO, waters are produced by incorporation of COZ of volcanic origin into circulating groundwaters. The CO2 is due to degassing from a nearby magma body or may be derived from condensation of CO,-rich steam produced during boiling. Incorporation of CO, results in formation of carbonic acid which reacts with the surrounding wallrock to produce cations and bicarbonate. Mixing and dilution of different types of geothermal water can result in a variety of “mixed waters” such as Na-Cl-HC03 (Table 1)

It is common for Na-Cl waters to be dis- charged at the base of a volcano at great dis- tances from upflow areas where acid-SO, wa- ters may occur (Henley 1985), particularly in regions of local high relief and lower perma- nent water tables typical in tropical areas (Hen- ley and Ellis 1983). This discharge explains the relatively localized occurrence of dilute Na- Cl-HC03 waters at the base of Barva (Fig. 2), -30 km downslope of the Azufre at the ter- minus of Pleistocene lava flows. Most Na-Cl- HC03 waters that have been identified on Bar- va issue near the terminus of the most recent of two main lava flows known for lowland ar- eas. Discharge points of geothermal waters ap- pear to be determined by the shape and hy- drogeological properties of the flows and possible faulting (Pringle et al. 1990; Pringle 1991).

Spatial patterns in pH, and levels of Cl- and SOd2- in geothermally modified streams draining the northwest flank of Poas (Fig. 5) are determined by upward convection of mag- matic vapors laden with H2S, HCI, and SO,; subsurface Aow of volcanically derived brines; and wind-dispersed, aerial inputs of acid rain.

Waters of the Agrio exhibit the most anom- alous solute composition on Poas. Brantley et al. (1987) and Rowe et al. (1989) indicated that the acid-SO,-Cl waters of the Agrio are due to subsurface leakage of highly acid-SO&l brine formed in the summit crater lake (La-

guna Caliente, Table 4). The crater lake is the surface expression of a small summit hydro- thermal system where condensation of as- cending fumaroles on the lake bottom pro- vides inputs of heat (lake temperature, 40- 90°C) and acidic volatiles (mainly S02, H2S, and I-IQ vector A, Table 8; Fig. 6B) that result in formation of an extremely acidic (pH - 0), SO,-Cl brine (Brantley et al. 1987). Dissolu- tion and leaching of volcanic materials result in elevated cation and trace metal concentra- tions in the lake (Table 4).

The hydrology of the summit plays an im- portant role in the location of acid brine leak- age to surface waters on the flanks. Source springs for the Agrio watershed occur 3-5 km from the active crater, along a lava-lahar con- tact (Rowe et al. 1989). Because old cones dominate the topography surrounding the ac- tive crater on its southern and southeastern sides, it is probable that the water-table divide is lowest to the west and northwest so that brine discharge would be in the direction of the Rio Agrio watershed (Sanford et al. in press).

High concentrations of Si (95 mg SiO, li- ter-‘) and SRP (471 pg liter-‘) in the Agrio (Table 4) are probably due to extensive dis- solution and leaching of volcanic rock and glass by acidic brines and thermal waters. Chemical modeling indicates that Si contents are regu- lated by equilibrium with amorphous Si, while maximum P concentrations may be controlled by precipitation of an amorphous form of the ferric phosphate mineral, strengite (Rowe 1991):

Fe3+ + H2P04- = FePO 4(3) + X-I+. (4) Rio Agrio waters are also in equilibrium with amorphous ferric hydroxide: basaltic lavas collected at the mouths of acid-Cl-SO, source springs of the Agrio are composed almost en- tirely of amorphous silica and ferric oxyhy- droxides. P contents (as P2O3) of the altered lavas are 1.27 wt% (vs. 0.16 wt% for unaltered samples of the same lava), indicating a nearly eightfold enrichment in P during the alteration process. The enrichment could be due to pre- cipitation of amorphous ferric phosphate; however, adsorption of phosphate onto charged Fe-oxide surfaces is an equally valid mecha- nism for the phosphate enrichment observed

768 Pringle et al.

in these rocks. Adsorption of SRP onto sus- pended Fe-hydroxide particles, with the sub- sequent precipitation of amorphous Fe-phos- phate layers on the oxide particle surfaces, has been shown to be an important mechanism for regulating SRP concentrations in low-alkalin- ity rivers and streams (Fox 1989).

Mixing of acidic geothermal fluids rich in Fe and Al with waters of near-neutral pH re- sults in precipitation of Fe and Al hydroxides when pH is between 3.5 and 5 (Rowe 1991). Neutralization of acid fluids, followed by pre- cipitation of metal hydroxides capable of ad- sorbing large amounts of phosphate, will sharply reduce SRP concentrations of geo- thermally modified streams and rivers. The Gata which is a river with near-neutral pH, and whose coloration and water chemistry in- dicate acid inputs in the upper watershed, has one of the lowest SRP concentrations of any river draining the flanks of Poas (Table 7). Thus, among the study streams draining Poas, elevated SRP concentrations provided by the interaction of acidic geothermal fluids and vol- canic rocks occurs only under highly acidic conditions.

Stream solute chemistry indicates that the Agrio is the only river draining Poas that is directly affected by subsurface brine inputs, while acid rain that is derived from summit fumarolic activity and distributed by wind af- fects other rivers on the northwestern flank. Both Cl- and S042- have been measured at several hundred milligrams per liter-l in pre- cipitation collected at the summit (Rowe 199 1). This acid, solute-rich precipitation is dispersed by prevailing northeasterly winds (Fig. 6B) that are deflected to the west by large cones just to the south of the active crater. Correspondingly, spatial patterns in the pH of precipitation sam- ples collected in stream drainages on the north- western slope (Table 5) correspond to spatial patterns in pH, Cl-, and S042- (Fig. 5).

Of the rivers draining the northwestern flank of Poas, the chemical composition of the De- sague (site 2; Fig. 3) appears to be most affected by acid rain and ashfall. The upper drainage of the Desague receives the most acidic rain (Table 5) and abundant ashfall because of its location immediately to the west of the large cones, just south of the active crater, which deflect wind in the direction of the Desague watershed (Rowe et al. 1989). The headwaters

of the Desague are directly adjacent to the ac- tive crater and have pH values near 3 and gradually increasing to 4.8 at the base of the volcano (at our sampling site), as waters are progressively diluted by lower elevation, near- neutral springs and streams (Rowe 1991).

Relatively high solute concentrations in thermal streams draining the flanks of Arena1 reflect effects of degassing from the subsurface magma body and recently deposited lava, ash, and pyroclastic flows. Easily leached ash and fresh lava that are constantly added to the western flank of Arena1 results in enhanced fluxes of rock-forming elements released dur- ing weathering, accounting for the highly con- centrated nature of thermal waters (both Na- Cl-HCO, and Na-Cl-SO,-HCO, types) that issue from the outer edges of the recently de- posited lava field on the west flank. The fresh lava of Arena1 is in marked contrast with the older, more weathered terrains of Barva and Poas where easily leachable deposits have al- ready been weathered to a high degree. Since Arena1 has been in a period of constant effusive activity over the past 20 yr, its Na-Cl-HCO, waters (Table 6) are much more concentrated than those observed on Barva. In contrast to the deep and only weakly degassing magma of Barva, the shallow magma body of Arena1 is actively degassing large quantities of acidic gases (C02, SO,; Fig. 6C) that generate slightly acidic waters which, in turn, leach more recent lava and ash (Bigot and Barquero 1986). Shal- low magmatic vapors are continuously rising and interacting with groundwater on the north- west slope accounting for the constant warm temperatures of three hot springs that emerge in this area (pers. obs.). Cooling lava flows have gas compositions that are enriched in HCl and CO2 relative to sulfur gasses (Stoiber and Rose 1970) so that waters draining the flows are typically of the Cl-HCO, type.

Older thermal springs arising from subsur- face aquifers in the Central Valley, on the op- posite side of the southern Cordillera Central, are also neutral Cl-HC03 waters (Paniagua and VanderBilt 1979), containing higher concen- trations of Cl and lower concentrations of Na, Si, and Mg than thermal streams draining Are- nal. P levels in thermal streams of Arena1 and solute-rich seeps of Barva are similar to levels measured in thermal springs of southwestern Africa and the northwestern portion of the Cape

Landscape linkages and streams 769

Table 10. Comparison of P levels and molar DIN: SRP ratios in streams draining landscapes with underlying geothermal activity with streams draining geothermally inactive landscapes. Means @&SD), ranges (min-max), and sample sizes (n) arc indicated.

Costa Rica Barva

Poas

Arena1

Africa

Costa Rica Old Pliocene lavas

Subarctic (taiga)

Temperate

Wet tropical

SRP TP

(pg her ‘) DIN : SRP Reference

Geothermally active landscapes

x 75.3f94.5 89.1kllO.l This paper range <5.0-301.2 < 5.0-405.0 0.2-100.3 n 27 28 26

x 55.7f 120.6 57.2* 123.0 This paper range 10.1-471.2 10.9-480.6 1.4-29.6 n 14 14 14

x 88.4k77.3 102.lk84.8 This paper range 7.9-234.1 9.1-279.1 3.4-54.7 n 10 10 9

range - 100.0-200.0 - Ashton and Schoeman (1984)

Geotheramlly inactive landscapes

R 0.8+ 1.2 5.2k5.2 Pringle et al. (1990) range 0.0-3.0 0.0-5.9 56.7-391.8 n 9 9 7

x 9.0 - Meybeck (1982)

x 11.0 - Meybeck (1982)

x 12.0 - Meybeck (1982)

Province, South Africa (Ashton and Shoeman 1984; Table 10).

In highly alkaline waters such as the thermal Na-HCO, waters of Arenal, SRP concentra- tions may be controlled by equilibrium with hydroxyapatite:

5Ca2+ + 3HPOd2- + 40H- = Ca,(PO4)3(OH)c,p,tite) + 3H20. (5)

Geochemical modeling indicates that the Na-HCO, waters of Arena1 are saturated to slightly supersaturated with respect to hy- droxyapatite. In contrast, the solute-rich, al- kaline waters of Barva are greatly undersatu- rated with respect to all apatite phases.

Ecological response to geothermal activity- Urban et al. (1987) emphasized that the agents of pattern formation are interwoven into the development of the landscape. In our study, our understanding of landscape pattern in the solute composition of geothermal waters has emerged after identification and consideration of regional geothermal processes and physical transport vectors. Similarly, an emergent bi- ological response would be expected due to

physical-chemical properties of the solute pat- tern, especially with respect to microbial com- munities which constitute the link between aquatic chemistry and higher trophic levels.

Kilham and Hecky (1973) similarly make the linkage between the solute composition of surface waters and the geographic distribution of biota. They found the range of fluoride con- centrations in surface waters of East Africa to be the greatest recorded globally, in this case resulting from the natural weathering of fluo- ride-rich volcanic rocks common in the region rather than from geothermal activity. Kilham and Hecky presented evidence that high flu- oride concentrations may influence the distri- bution of not only zooplankton, phytoplank- ton, and higher aquatic plants but also higher trophic levels such as man and livestock.

Many excellent studies have described the biota of high-temperature geothermal waters (e.g. Brues 1927; Copeland 1936; Castenholz 1969), and experimental work has been done on the biochemical ecology of hot springs (Brock and Brock 1966; Stockner 1968). The biota of geothermal waters spanning a range

770 Pringle et al.

in physical-chemical parameters has also been described for different regions (e.g. Ashton and Schoeman 1984; Cassie and Cooper 1989). It is clear that a number of factors interact to determine the distribution of algae in thermal streams, and important patterns have emerged regarding maximum temperature tolerances and temperature ranges (e.g. Stockner 1967), effects of acidity, and other factors (e.g. Cas- tenholz and Wickstrom 1975). Similarities have been found in the algal and bacterial flora of geothermal environments located in geo- graphically distant areas of the world, includ- ing South Africa, North America, Iceland, and New Zealand (e.g. Ashton and Schoeman 1984; Castenholz 1973).

Procaryotic, blue-greens (Cyanophyta), which are tolerant of high temperatures and known to predominate in thermal waters (e.g. Castenholz and Wickstrom 1975) were dom- inant in hot springs draining both Arena1 and Platanar. Luxuriant growths of a thermophilic species of Phormidium were observed in the Tabacon and hot springs (35-36°C) draining Arena1 (Table 7). Some of the many species of Phormidium found in the U.S. are found in hot springs (Prescott 1962). The blue-green A. tortuosa and T. imperialis collected in the Tu- caron springs (62°C) at the base of Platanar are tolerant of high temperatures and have been recorded from thermal springs in Yellowstone National Park by Copeland (1936).

The diatom Pinnularia, also present in the Arena1 Na-Cl-HCO, hot springs, is one of the few species of eucaryotic algae known to occur in nonacid thermal springs in the 35-45”C range (Stockner 1967; Kullberg 197 1).

A prominent characteristic of thermal, al- kaline streams is the presence of the filamen- tous, photoheterotrophic flexibacterium, Chlorojlexus (Castenholz and Wickstrom 1975). This bacterium appears to have a worldwide distribution despite its restriction to thermal environments with temperatures between 40 and 70°C (Castenholz 1973). Chlo- roflexus was observed in thick mats under growths of blue-greens (cyanobacteria) in al- kaline thermal springs (62°C) draining Platan- ar and in hot seepages (40-50°C) feeding into the Tabacon streams of Arena1 (Table 7).

The Guillermina typifies a stream that is unmodified by geothermal inputs, exhibiting low ambient temperatures, alkalinity, and SRP

and dominated by a more ubiquitous diatom assemblage (e.g. Navicula, Synedra, and Me- losira).

On Poas, the algal community of the re- ceiving waters of the Agrio is dominated by the known acidophile, E. mutabilis Schmitz and a blue-green-diatom assemblage. E. mu- tabilis is widespread in acidic (pH l-4) habi- tats, including peat bogs (Pentecost 1982), acid springs (Doemel and Brock 197 l), and acid coal-mine drainage (Hargreaves et al. 1975). It is also quite tolerant of heavy metals in a wide pH range (Hargreaves and Whitton 1976; Havas and Hutchinson 1983). Ulothrix is also found as a dominant alga in the Agrio. Gen- erally, green algae (Chlorophyta), such as Ulo- thrix, are tolerant of waters rich in trace ele- ments (Stokes 1983) such as those investigated here.

The flora of the Gata is influenced by a dense white coating of amorphous silica, Al-sulfate, and Al-hydroxide precipitates which retards or smothers growth of algal periphyton mats, re- sulting in very sparse algal growth (Tables 7, 11). Al-hydroxides scavenge P, resulting in rel- atively low levels of SRP (Tables 7, 11). The presence of this precipitate in combination with the sparse periphyton standing crop produces a distinct, milky blue water that readily iden- tifies one of the geochemical mechanisms con- trolling the ecology of these streams.

The Desaguc also was milky blue because of the same colloidally suspended precipitate, and significant amounts of volcanic ash were pres- ent in the drainage. The upper drainage was affected by ashfall associated with eruptive ac- tivity just before our sampling (pers. obs.). Ex- tensivc ash deposition and subsequent scour- ing appears to have removed attached algae and other biota, thereby acting as a major dis- turbance or “resetting mechanism.” Outside of a few dead strands of Uothrix on the down- stream faces of rocks, there was no visible algal standing crop.

An acid-tolerant species of Microspora and sulfur-oxidizing bacteria dominate the benthic community of the Azufre (Tables 7, 11) on Barva. Relative to thermal input and highly acid pH, the ecological consequences of geo- thermally induced pattern are more subtle in dilute, geothermal waters of the Na-Cl-HCO, type, which surface at low elevations at the base of Barva. P-rich, geothermal seeps have

Landscape linkages and streams 771

Table 11. Linkages between physical transport vectors and geothermal water types (and associated gcothcrmal modifications) with ecological rcsponsc in selected streams or lakes draining each of the three volcanoes. Vector definitions given in Table 8; not applicable-NA (e.g. stream not affected by geothermal transport vectors).

Stream/ lake

Transport vector(s)

Gcothcrmal water type

Ecological response Rererence

Barva Azufre

Salt0

Pantano

Poas Laguna Calien te

Agrio

Gata

Desague

Angel

Arena1 Tabacon 1 A,, J%

Guillermina NA

NA

Acid-SO,

Na-Cl-HCO,, high SRP

Unmodified, solute- poor, low SRP

Acid-SO,-Cl, high temp.

Acid-SO,-Cl, high SRP

Slightly acid, Si and Al-SO, precipitate, low SRP

Slightly acid, ash scour, colloidal Si, AI-SO, precipitate, low SRP

Unmodified, solutc- poor, low SRP

Na-Cl-SO,-HCO,, high temp.

Unmodified, solute- poor, low SRP

Acidophilic algae, sulfur bacteria

Micronutrient limi- tation of algal growth

P limitation of algal growth

No life

Acidopihlic algae, algae tolerant of heavy metals, moderately high algal standing crop

Sparse algal standing crop

No algal standing crop

Sparse algal growth, potential P limita- tion of algal growth

Thermophilic blue- greens, flexibacte- ria

Potential P limita- tion of algal growth

This paper

Pringle et al. (1986)

Pringle and Triska (1991)

Brantley et al. (1987)

This paper

This paper

This paper

This paper

This paper

-

been found to affect biotic nutrient cycling (Pringle and Triska 1991). As a consequence, nutrient limitation and its role in autochtho- nous stream production and biogeochemical cycling can vary significantly between streams in close proximity in the same watershed.

In situ nutrient bioassays of algal growth in dilute, Na-Cl-HCO, waters of the Salto River at the base of Barva (Fig. 2) have indicated that levels of major nutrients (N and P) are saturating to algal growth in the P-rich main channel, while P amendments stimulate algal growth in the P-poor, Pantano tributary where light is not limiting (Pringle and Triska 199 1). In the Salto, a micronutrient combination was found to limit algal growth under light-suffi- cient conditions (Pringlc et al. 1986).

Regional sign$cance of geothermal activity

on stream solute composition and ecology- Results of this study and others (Pringlc un- publ. data) suggest that predictable spatial pat- terns in nutrient chemistry can be expected to occur frequently along the Central American volcanic arc as a result of volcanic and asso- ciated geothermal activity. Anomalies in the solute composition of surface waters draining geothermal areas are not rare or isolated phe- nomena. Of the approximately 40 active vol- canoes in Central America, a fifth host acidic crater lakes formed by geothermal activity (Brantley et al. 1987). Similarities among some of the more well-studied crater lakes indicate that they evolved along the same chemical lines (Giggenbach 1974; Casadevall et al. 1984; Brantley et al. 1987). A further indication of the prevalence and pattern in the occurrence

772 Pringle et al.

of many of the geothermal stream types dis- cussed here can be found by examining maps of the volcanic mountain ranges of Central America (Pringle unpubl. data). Certain river names that are indicative of geothermal water types consistently recur. Examples include: Agrio (sour); Azufre (sulfur); Azul (blue); Cal- iente (hot); Gata (which refers to the milky blue of cat eyes); and Salitral (salty).

Comparisons of the high P levels measured in many geothermally modified streams of this study with the low, average levels of ortho- phosphate exhibited by relatively pristine riv- ers worldwide (Table 10) indicate that geo- thermal activity can represent a significant source of P in surface waters. Those few studies that have documented P concentrations in geothermal waters in other areas of the world also indicate high levels. Giggenbach (1974) measured concentrations of 2,400-l 2,700 pg PO,-P liter-l in a hot, acid-SO, crater lake in New Zealand. Ashton and Schoeman (1984) measured total P concentrations of 100-200 ,ug TP liter-l in geothermal springs of South Africa (Table 10).

Regional generalizations have often been made regarding nutrient properties of waters based on geology (e.g. Freeze and Cherry 1979). For instance, streams draining some geological areas are low in P, with relatively low N : P ratios (Table 10). Correspondingly, predic- tions of P limitation of primary production have been made for surface waters (e.g. Om- ernik 1977). Results presented here indicate that such regional generalizations and predic- tions regarding P or N limitation of stream primary producers are difficult to make in geo- thermally active volcanic landscapes. Geo- thermally active areas exhibit high local vari- ability in stream P concentrations and N : P ratios that range from low to high (Table 10) relative to Redfield’s ratio of 16 : 1 (Redfield et al. 1963).

In this study, differences in stream SRP con- centrations among volcanoes were not signif- icant due to high variation in SRP among streams draining individual volcanoes. High local variability in SRP among streams on Bar- va and Poas is due to differences in SRP levels between geothermally modified and unmodi- fied streams and among geothermally modified stream types. Na-Cl-HCO,, acid-SO,, and

acid-SO,-Cl waters on Barva and Poas, re- spectively, had significantly greater (P < 0.05) SRP than waters unmodified by geothermal activity. Slightly acid stream waters on Poas affected by geothermally derived acid rain had significantly less SRP than waters unmodified by geothermal activity due to precipitation re- actions of SRP with other elements. On Ar- enal, high variability in SRP among streams is not due to differences between geothermal and unmodified waters. Although geothermal waters generally contain higher SRP values than unmodified waters, differences are not significant. High levels of P occur in some un- modified waters as a result of the fresh ash and lava deposited daily by the active volcano.

Previous studies have indicated that growth of algal primary producers is light limited (Paaby-Hansen 1988) in P-rich streams of the Na-Cl-HCO, type that drain primary forest at the base of Barva. In light gaps where light is not a limiting factor, algal growth shows little response to N and P amendments (Pringle et al. 1986; Paaby-Hansen 1988; Pringle and Triska 199 1). We predict that the high levels of P found in some types of geothermal waters will translate into high productivity as the can- opy opens with increasing stream order and deforestation. In support of this hypothesis, high algal biomass and rates of Chl a accrual have been measured in P-rich streams of the Na-Cl-HCO, type that drain pasture at the base of Barva (Pringle unpubl. data). Much of the Central Valley and coastal area of Costa Rica has been deforested (e.g. Hartshorn et al. 1982), so geothermally introduced P may have significant regional effects on stream produc- tivity.

Given that many streams draining Central America have their headwaters on volcanic peaks, this study also raises the question of how geothermally derived solute inputs affect stream function as one moves from the vol- canic mountain range to the sea or into the Central Valley of Costa Rica. In active vol- canic landscapes, analogs to upstream-down- stream linkages by stream order (e.g. Vannote et al. 1980) should include chemical linkages related to geothermal inputs of various solutes, related shifts in pH at tributary junctions, and the resultant impacts on stream community structure and nutrient cycling.

Landscape linkages and streams 773

In conclusion, spatial patterns exist in the chemistry and biota of streams draining land- scapes affected by volcanic and geothermal ac- tivity. Landscape patterns in the occurrence of geothermal waters and in their solute com- positions are determined by the magnitude of underlying geothermal activity mediated by physical transport vectors (thermally driven upward convection, lateral and downward transport, and wind-directed transport). Phys- ical transport vectors interact to determine the flow of geothermally derived solutes within the landscape and resultant patterns in stream sol- ute chemistry and biota.

References AMERICANPUBLICHEALTHASSOCIATION. 1985. Standard

methods for the examination of water and wastewater, 15th ed. APHA.

ARCE, Y. H., AND OTHERS. 1980. Actividad residual en el Barva. Bol. Volcanol. 9: 5-l 1.

ASHTON, P. J., AND F. R. SCHOEMAN. 1984. A prelimi- nary limnological investigation oftwelve southern Af- rican geothermal waters. J. Limnol. Sot. S. Africa 10: 50-56.

BALL, J. W., D. K. NORDSTROM, AND D. W. ZACHMANN. 1987. WATEQ4F-a personal computer fortran translation of the geochemical model WATEQ2 with revised data base. U.S. Geol. Surv. Open-file Rep. 87- 50.

BIGOT, S.. AND J. BARQUERO. 1986. Estudio hidrogeo- quimico de1 Volcan Arenal: Contribution a la activ- idad eruptiva. Bol. Volcanol. 17: 7-14.

BORGIA, A., C. PORE, M. J. CARR, W. G. MELSON, AND G. F. ALVARADO. 1988. Structural, stratographic and petrologic aspects of the Arena1 Cerro Chato volcanic system: Evolution of a young stratovolcano complex. Bol. Volcano]. 50: 86-105.

BRANTLEY, S. L., A. BORGIA, G. ROWE, J. F. FERNANDEZ, AND J. R. REYNOLDS. 1987. Poas Volcano crater lake acts as a condenser for acid metal-rich brine. Nature 330: 47W72.

BROCK, T. D., AND E. BROCK. 1966. Temperature optima for alga1 development in Yellowstone and Iceland hot springs. Nature 209: 733-734.

BRUES, C. T. 1927. Animal life in hot springs. Q. Rev. Biol. 2: 181-201.

CASADEVALL, T. J., AND OTHERS. 1984. Crater lake and post-eruption hydrothermal activity, El Chichon Vol- cano, Mexico. J. Volcanol. Geotherm. Res. 23: 169- 191.

CASERTANO, L., A. BORGIA, AND C. CIGOLINI. 1983. El Volcan Poas, Costa Rica: Cronologia y caracteristicas de la actividad. Geofis. Int. 22: 2 15-236.

CASSIE, V., AND R. C. COOPER. 1989. Algae of New Zea- land thermal areas: A taxonomic guide to thermally associated algae (excluding diatoms) in New Zealand. Cramer.

CASTENHOLZ, R. W. 1969. Thermophilic blue-green al-

gae and the thermal environment. Bacterial. Rev. 33: 476-504.

-. 1973. Ecology of blue-green algae in hot springs, p. 379-414. In N. G. Carr and B. A. Whitton [eds.]. The biology of blue-green algae. Blackwell.

-, AND C. E. WICKSTROM. 1975. Thermal streams. p. 264-285. Zn B. A. Whitton [ed.], River ecology. Univ. Calif.

COPELAND, J. J. 1936. Yellowstone thermal Myxophy- ceae. Ann. NY Acad. Sci. 36: l-232.

DOEMEL, W. N., AND T. D. BROCK. 197 1. The physio- logical ecology of C.vanidium caldarium. J. Gen. Mi- crobiol. 67: 17-32.

ELLIS, A. J., AND W. A. MAHON. 1964. Natural hydro- thermal systems and experimental hot-water/rock in- teractions. Geochim. Cosmochim. Acta 28: 1323- 1357.

---, AND ---. 1967. Natural hydrothermal sys- tems and experimental hot water/rock interactions (Part 2). Geochim. Cosmochim. Acta 31: 5 19-538.

FORMAN, R. T., AND M. GODRON. 1986. Landscape ecol- ogy. Wiley.

Fox, L. 1989. A mode1 for the inorganic control of phos- phate concentrations in river waters. Geochim. Cos- mochim. Acta 53: 4 17-428.

FREEZE, R. A., AND J. A. CHERRY. 1979. Groundwater. Prentice Hall.

FUNK, V. A., AND W. G. MELSON. 1989. Volcanological and biological observatory at Arena1 Volcano (Ala- juela Province, Costa Rica). Biotropica 21: 95-96.

GARRELS, R. M., AND F. T. MACKENZIE. 1967. Origin of the chemical composition of some springs and lakes. p. 222-242. Zn Equilibrium concepts in natural water systems. Adv. Chem. Ser. 67.

GIGGENBACK, W. 1974. The chemistry of Crater Lake. Mt. Ruapehu (New Zealand) during and after the 197 I active period. N.Z. J. Sci. 17: 33-45.

HARGREAVES, J. W., E. J. LLOYD. AND B. A. W~IITTON. 1975. Chemistry and vegetation of highly acidic streams. Freshwater Biol. 5: 563-576.

~ AND B. WHITTON. 1976. Effect of pH on growth of’acid stream algae. Br. Phycol. J. 11: 2 15-223.

HARTSHORN,G.S.,ANDOTHERS. 1982. Costa Ricacoun- try environmental profile: A field study. Trop. Sci. Center, San Jose, Costa Rica.

HAVAS, M., AND T. HUTCHINSON. 1983. The smoking hills: Natural acidification of an aquatic ecosystem. Nature 301: 23-27.

HENLEY, R. W. 1985. The geothermal framework for epithermal deposits, p. l-24. In B. R. Berger and P. M. Bethke [eds.], Geology and geochemistry of epi- thermal systems. Sot. Econ. Geol. Rev. Econ. Geol. v. 2.

AND A. J. ELLIS. 1983. Geothermal systems, an- c&t and modem. Earth Sci. Rev. 19: l-50.

KAMPHAKE, L.J.,S. A. HANNAH,AND J.M. COHEN. 1967. An automated analysis for nitrate by hydrazine re- duction. Water Res. 1: 205-216.

KILHAM, P., AND R. E. HECKY. 1973. Fluoride: Geo- chemical and ecological significance in East African waters and sediments. Limnol. Oceanogr. 18: 932- 945.

KULLBERG, R. G. 197 1. Algal distribution in SIX thermal

774 Pringle et al.

spring effluents. Trans. Am. Microsc. Sot. 90: 412- 434.

MELSON, W. G. 1984. Prehistoric eruptions of Arena1 Volcano, Costa Rica. Vinculos (Muse0 Nacl., San Jose, Costa Rica) 10: 20-35.

7, -AND R. SAENZ. 1973. Volume, energy and cy- chcity of eruptions of Arena1 Volcano, Costa Rica. Bull. Volcanol. 37: 416-437.

MEYBECK, M. 1982. Carbon, nitrogen and phosphorus transport by world rivers. Am. J. Sci. 282: 401-450.

OMERNIK, J. M. 1977. Nonpoint. source stream nutrient level relationships: A nationwide survey. U.S. EPA- 600/3-77- 105, Ecol. Res. Ser.

PAABY-HANSEN, P. 1988. Light and nutrient limitation in a Costa Rican lowland stream. Ph.D. thesis, Univ. Calif., Davis. 194 p.

PANIAGUA, S. A., AND H. VANDERBILT. 1979. Geologia y geoquimica de las aguas termales valle central, Costa Rica. Cien. Tecnol. 3: 109-129.

PENTECOST, A. 1982. The distribution of Euglena mu- tabdis in Sphagnum with reference to the Malham Tam. North Fen. Field Stud. 5: 591-606.

PRESCO~, G. W. 1962. Algae of the western Great Lakes area. Brown.

PRINGLE, C. M. 199 1. Geothermally-modified waters surface at La Selva Biological Station, Costa Rica: Volcanic processes introduce chemical discontinui- ties into lowland tropical streams. Biotropica 23: 523- 529.

-, AND OTHERS. 1985. Natural history observations and ecological evaluation of the La Selva Protection Zone, Costa Rica. Brenesia 22: 189-206.

P. PAABY-HANSEN, P. D. VAUX, AND C. R. G;)LDMAN. 1986. In situ nutrient assays of periph- yton growth in a lowland Costa Rican stream. Hy- drobiologia 134: 207-2 13.

-, AND F. J. TRISKA. 199 1. Effects of geothermal groundwater on nutrient dynamics of a lowland, Cos- ta Rican stream. Ecology 72: 95 l-965.

- -, AND G. BROWDER. 1990. Spatial varia- tion in basic chemistry of streams draining a volcanic landscape on Costa Rica’s Caribbean slope. Hydro- biologia 206: 7 3-86.

PROSSER, J. T., AND M. CARR. 1987. Poas Volcano, Costa Rica: Geology of the region and spatial and temporal variations among the most recent lavas. J. Volcanol. Geotherm. Res. 33: 131-146.

RACCICHINI,~., AND F. D. BENNETT. 1979. Nucvas as- pectos de las erupciones de1 Volcan Poas, p. 148-l 62. Zn C. A. Vargas [ed.], Antologia El Volcan Poas, San Jose, Costa Rica.

REDFIELD, A. C., B. H. KETCHUM, AND F. A. RICHARDS. 1963. The influence oforganisms on the composition of sea-water, p. 26-77. Zn M. N. Hill [ed.], The sea. V. 2. Wiley-Interscience.

RISSER, P. G. 1987. Landscape ecology: State of the art, p. 3-14. Zn M. G. Turner [ed.], Landscape heteroge- neity and disturbance. Springer.

ROSARIO-ALFARO, M. E. FERNANLXZ, J. BARQUERO, J.J. RODRIGUEZ, ANDM. RODRIGUEZ. 1986. Lluviaacida de origen volcanica. Bol. Vulcanol. 16: 15-22.

ROWE, G. L. 1991. The acid crater lake system of Poas Volcano, Costa Rica: Geochemistry, hydrology and physical characteristics. Ph.D. thesis, Pennsylvania State Univ. 297 p.

-,S. L. BRANTLEY, A. BORGIA, J.F. FERNANDEZ,AND J. BARQUERO. 1989. Observaciones preliminares de1 sistema hidrotermal de1 volcan Poas Costa Rica. Bol. Vulcanol. 20: 23-3 1.

- - press.

J. F. FERNANDEZ, AND A. BORGIA. In Che’mical and hydrologic structure of Poas vol-

cano, Costa Rica. J. Volcanol. Geotherm. Res. -, AND OTHERS. 1992. Fluid-volcano interaction in

an active strato-volcano: The crater lake system of Poas Volcano, Costa Rica. J. Volcanol. Geotherm. Res. 49: 23-5 1.

SANFORD, W. E., L. F. KONIKOW, G. L. ROWE, AND S. L. BRANTLEY. In press. Groundwater transport of cra- ter lake brine at Poas volcano, Costa Rica. J. Volcanol. Geotherm. Res.

Sow, R. R., AND F. J. ROHLF. 198 1. Biometry, 2nd ed. Freeman.

SOL~RZANO, L. 1969. Determination of ammonia in nat- ural waters by the phenolhypochlorite method. Lim- nol. Oceanogr. 14: 799-80 1.

STOCKNER, J. G. 1967. Observations of thermophilic al- gal communities in Mount Rainier and Yellowstone National Parks. Limnol. Oceanogr. 12: 13-17.

1968. Algal growth and primary productivity in a thermal stream. J. Fish. Res. Bd. Can. 25: 2037- 2058.

STOIBER, R. E., AND W. I. ROSE. 1970. The geochemistry of Central American volcanic gas condensates. Geol. Sot. Am. Bull. 81: 2891-2912.

STOKES, P. 1983. Responses of freshwater algae to met- als. Prog. Phycol. 3: 87-l 12.

STRICKLAND, J. D. H., AND T. R. PARSONS. 1972. A practical handbook of seawater analysis, 2nd ed. Bull. Fish. Rcs. Bd. Can. 167.

URBAN, D. L., R. V. O’NEILL, AND H. H. SHUGART. 1987. Landscape ecology. Bioscience. 37: 119-l 27.

VANNOTE, R. L., G. W. MNSHALL, K. W. Cm~ms, J. R. SEDELL, AND C. E. GUSHING. 1980. The river con- tinuum concept. Can. J. Fish. Aquat. Sci. 37: 130- 137.

WHITE, D. E. 1957. Thermal waters of volcanic origin. Bull. Gcol. Sot. Am. 68: 1637-1658.

WIEMS, J. A., C. S. CRAWFORD, AND J. R. Gosz. 1985. Boundary dynamics: A conceptual framework for studying landscape ecosystems. Oikos 45: 42 l-427.