Spatial and temporal patterns across an ecological boundary: Allochthonous effects of a young saltwater lake on a desert ecosystem

Journal of Arid Environments 73 (2009) 811–820

Contents lists available at ScienceDirect

Journal of Arid Environments journal homepage: www.elsevier.com/locate/jaridenv

Spatial and temporal patterns across an ecological boundary: Allochthonous effects of a young saltwater lake on a desert ecosystem C.S. Brehme a, *, W.I. Boarman a,1, S.A. Hathaway a, A. Herring a, L. Lyren a, M. Mendelsohn a, K. Pease a, M. Rahn b, C. Rochester a, D. Stokes a, 2, G. Turschak a, R.N. Fisher a a b

U.S. Geological Survey, San Diego Field Station, 4165 Spruance Road, Suite 200, San Diego, CA 92101, USA Field Station Programs, College of Sciences, San Diego State University, 5500 Campanile Drive, San Diego, CA 92182, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 28 February 2008 Received in revised form 10 March 2009 Accepted 14 March 2009 Available online 8 May 2009

We documented changes in the abundance and composition of terrestrial flora and fauna with respect to distance from the sea edge and timing of large allochthonous inputs from the Salton Sea, California. We found significant effects that were most pronounced within 300 m of the shore, but extended 3 km inland via coyote scat deposition. The zone within 300 m of the sea had a higher density of vegetation with a distinctly different plant composition. The denser vegetation supported higher abundances of birds and reptiles. Coyotes exhibited spatial and temporal responses to marine subsidies of fish, while birds were likely subsidized by aquatic aerial insects. Top-down control, as well as dietary and habitat preferences, may have resulted in reduced number of ants, beetles, and small mammals near the sea. Species responses to the habitat edge appeared to be associated with life history, as the near shore habitat favored habitat generalists and shore specialists, while inland desert habitat favored many sand and open desert specialists. Ecosystem responses support current theories of allochthonous spatial subsidies and consumer-resource dynamics but were limited in scope, magnitude, and distance. Published by Elsevier Ltd.

Keywords: Desert ecology Ecotone Edge effect Landscape ecology Salton Sea Trophic interaction

1. Introduction Although terrestrial and aquatic environments are often considered distinct systems, they are intimately linked by a dynamic exchange of water, nutrients, and organisms. Nutrients in the form of plant detritus, associated microbes, and insects can flow from terrestrial to aquatic systems via leaf litter, wind, runoff, and inflow from streams (Bardgett et al., 2001; Minshall, 1967; Mulholland, 1981; Odum and Heald, 1975). Similarly, water and nutrients from aquatic plant and animal biota can flow from aquatic systems to terrestrial systems via episodic flooding and transport through the terrestrial food web (Anderson and Polis, 1998; Harding and Stevens, 2001; Loucks, 1975; Murakami and Nakano, 2002). Overall, information on allochthonous effects of aquatic systems on adjacent terrestrial systems is limited. The majority of work has

* Corresponding author. Tel.: þ1 6192256427; fax: þ1 6192256436. E-mail addresses: [email protected], rfi[email protected] (C.S. Brehme). 1 Present address: Conservation Science Research and Consulting, 2522 Ledgeview Place, Spring Valley, CA 91977, USA. 2 Present address: San Diego Natural History Museum, P.O. Box 121390, San Diego, CA 92112, USA. 0140-1963/$ – see front matter Published by Elsevier Ltd. doi:10.1016/j.jaridenv.2009.03.002

been primarily focused on single taxonomic groups in riparian systems (Briers et al., 2005; Henschel et al., 2001; Murakami and Nakano, 2002). In addition, the prevailing theories on the dynamics of allochthonous subsidization of terrestrial biota have largely come from research on islands in the Gulf of California (Garcia and Whalen, 2003; Polis and Hurd, 1996a,b; Polis et al., 1997; SanchezPinero and Polis, 2000). In the gulf system, the difference in productivity between the sea and desert is extreme, resulting in increases in the abundance of terrestrial consumers (arthropods, lizards, rodents, and coyotes) at the mainland shoreline and on small islands. Overall, increased abundance of subsidized terrestrial consumers is predicted to be greatest in systems with large doses of aquatic input and low in situ production (Polis and Hurd, 1996b; Polis et al., 1997). The relative magnitude and distance inland of these increases in consumer abundance are predicted to be greater in systems with smaller shore length-to-area ratios (i.e. smaller islands). However, the inclusion of predator–prey flow dynamics theory predicted that subsidized generalist predators may depress number of local prey, often referred to as ‘‘spillover predation’’ (Polis et al., 1997). Similar to islands in the Gulf of California, our study takes place at a sea–land interface. However, rather than desert islands surrounded by ocean, we studied the Sonoran Desert system

812

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

surrounding the land-locked Salton Sea of southern California. During summer, the Salton Sea experiences large-scale fish and bird die-offs which contribute to periods of increased allochthonous subsidies to the surrounding desert system. We sampled multiple taxonomic groups, including vegetation, ants, beetles, flies, birds, reptiles, small mammals, bats, and coyotes (Canis latrans), at varying distances from the sea edge. We hypothesized that the spatial flow of nutrients and detritus from the highly productive Salton Sea would subsidize the productivity of the immediate adjacent Sonoran Desert ecosystem, resulting in bottom-up trophic effects. We expected that water and nutrient subsidies would increase the primary productivity of plants near the sea (as measured by cover) and result in increased abundance of herbivore consumers (small mammals, ants), and that detritus and carrion subsidies would result in increased abundance of detritivores (beetles, flies) and their secondary consumers (lizards, bats, birds), as well as increased activity of large scavengers carnivores (coyotes). We expected that the effects would be greatest during the season of increased allochthonous input (aquatic insect emergence and fish die-offs). Because the total amount of allochthonous input is significantly less than reported for the Gulf of California, we also expected the magnitude of effects to be lower than reported in those studies. In addition, we predicted that the effect distance would be narrow, since our system has an infinitely low shore length-to-terrestrial area ratio in comparison to island systems. Finally, we examined compositional trends of faunal groups in relation to distance from the shore and predicted that individual species would exhibit different responses according to their habitat and dietary preferences. 2. Methods 2.1. Study site The Salton Sea is a large (974 km2), young, eutrophic, and highly saline lake formed in 1907 as a result of an accidental two-year diversion of Colorado River water. It is now fed primarily by municipal and domestic wastewater, streams, and agricultural runoff. Its salinity level is 26% greater than ocean water and continues to increase as soluble salts accumulate due to evaporative water loss (Watts et al., 2001). The sea is inhabited by primarily marine fauna that were purposefully (Carpelan and Linsley, 1961; Walker et al., 1961) or accidentally (Costa-Pierce and Doyle, 1997) introduced from the Pacific Ocean and the Gulf of California. It is a highly productive system, supporting abundant and rich communities of phytoplankton, zooplankton, benthic invertebrates, and fish (Detwiler et al., 2002; Lange and Tiffany, 2002; Riedel et al., 2002; Tiffany et al., 2002). It has also become a very important stopover and breeding ground for many species of migratory and resident waterbirds (Patten et al., 2003). During the summer, large algal blooms create anoxic conditions in large portions of the lake that are often associated with massive fish and bird die-offs (Watts et al., 2001). As a result, large volumes of dead fish, birds, and barnacles wash ashore along the perimeter of the lake. During the years of this study, an estimated average of >2000 g m1 yr1 (dry mass) of dead fish biomass alone washed onto the shoreline (Salton Sea Authority, unpublished data). The surrounding desert is primarily composed of creosote bush (Larrea tridentata) and desert saltbush (Atriplex polycarpa and A. canescens) habitat on alluvial soils and aeolian dunes with an estimated dry biomass of approximately 130 g m2 yr1 (Chew and Chew, 1965). At our primary study site, shoreline aquatic material was primarily composed of dead fish, with a lesser amount of barnacles and algae distributed within 30 m of the sea edge. Here, the sand was moist

with pronounced waves of salt deposits from the gradually receding shoreline. By 50 m, there was not a measurable amount of aquatic detritus. Vegetation was dominated by alkali sink plant species. A gradual or distinct change to primarily open creosote bush scrub with intermittent sand dunes was noted up to 500 m from the shoreline. 2.2. Field sampling As part of a broader study and inventory of the Salton Sea terrestrial ecosystem, we sampled floral and faunal communities in linear gradients from the sea edge over a period of a year. We sampled vegetation, ants, beetles, birds, bats, flies, reptiles, and small mammals at varying distances from the sea at each of 35 sampling nodes (seven transects located at 50, 300, 550, 800, and 1050 m from the edge of the sea) established within the U.S. Navy’s former Salton Sea Test Base, now managed by the U.S. Bureau of Land Management (Fig. 1). Coyote scat was sampled up to 1 km from the sea edge. All sampling was conducted between April 2001 and December 2002. 2.2.1. Vegetation We surveyed the perennial vegetation along a 50 m transect centered at each node (Sawyer and Keeler-Wolf, 1995). At 0.5 m intervals, we recorded plant species, canopy height, leaf litter depth, and substrate type, for a total of 100 data points per node. 2.2.2. Ants We installed five ant pitfall traps at each node (Fig. 2). The traps were 50 mL centrifuge tubes containing 25 mL of propylene glycolbased antifreeze buried level with the ground (Suarez et al., 1998). Traps were open for 10 consecutive days during each of five sampling sessions conducted from April to December 2001. Winged queens and males were not used in the analyses since they may have originated from outside the site. 2.2.3. Beetles We installed two transects of five pitfall traps (19 cm diameter plastic cereal bowls buried level with the ground) 15 and 25 m south of each node to sample ground-obligate arthropods (Fig. 2). One transect was placed in open substrate, while the second was placed next to vegetation. Twelve 3-day sampling sessions were conducted from March 2001 to February 2002. 2.2.4. Flies We sampled flies at four nodes 50 m from the sea and four nodes 1050 m inland (transect lines 1, 4, 5, and 7) using two sheets of sticky flypaper attached to a vertical polyvinyl chloride pipe 1.5 m above ground. The traps were set for two 4-day sampling sessions in May and June of 2001. We counted flies on each paper as an index of abundance, but did not identify individuals to species. 2.2.5. Birds We conducted 5-min point counts 50 m from each node (Mendelsohn et al., 2007). We counted landbirds and shorebirds seen or heard within 100 m of each node from 0.5 to 1 h before, to 4 h after sunrise. Because we were interested in impacts of the sea on terrestrial fauna, we did not include waterbirds observed on the immediate shoreline or on the sea. We conducted 15 complete surveys from March 2001 to February 2002. 2.2.6. Bats We surveyed bats by slowly walking each transect from the sea edge to the inland node between sunset and 4 h after sunset while recording bat vocalizations with an Anabat II bat detector (Titley Electronics, New South Wales, Australia). Calls were assigned to the

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

813

Fig. 1. The location of the Salton Sea in southwestern California and configuration of our sampling nodes along the southwestern portion of the sea for vegetation, ants, beetles, reptiles, birds, bats, and small mammals.

nearest node. We conducted five complete surveys from April 2001 to February 2002. 2.2.7. Reptiles We installed pitfall arrays of seven 6-gallon buckets connected by shade cloth drift-fences at each node (Fisher et al., 2008). Three 1.0 m long funnel traps were placed along the drift fence for capturing large snakes and lizards (Fig. 2). Animals were identified to species and released each morning. We sampled for four consecutive days every 4–6 weeks from March 2001 to April 2002 for a total of 12 sampling sessions. 2.2.8. Small mammals We sampled small mammals at each node for 2–3 consecutive nights. Sampling was conducted using 18 paired live traps per node

(small 7.6  9.5  30.5 cm and large 10.2  11.4  38.1 cm, H.B. Sherman Traps, Tallahassee, FL; Fig. 2). Traps were baited with a mixture of wild birdseed and rolled oats. Animals were identified to species and released each morning. We conducted three sampling sessions from March 2001 to February 2002. 2.2.9. Coyotes We added two additional study areas to survey coyotes (Mecca Beach within the Salton Sea State Recreation Area and the Wister Unit of the Imperial Wildlife Management Area) through the establishment of multiple independent transects. We established transects parallel to the shoreline within five zones at increasing distances from the shore: 0–250 m, 251–750 m, 751–1750 m, 1751–3750 m, and >10 km (Fig. 1b). Transects were 2-km long and 6-m wide and were cleared of scat and dead fish 2 and 4

814

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

Ant pitfall trap

Herpetofauna pitfall trap array with drift fence Funnel (snake) trap

15 m

Paired small mammal Sherman live-traps

We used Bray–Curtis (Czenkanowski’s) coefficient of similarity to compare the species composition at all distances from the sea to the composition at the most inland sites (1050 m from shore). This coefficient is calculated by dividing the sum of differences between species abundances across samples by the sum of species abundances across samples and subtracting from one, taking both the presence/absence and relative abundance of species into account (Kent and Coker, 1992). The coefficient ranges between zero and one (zero indicating complete dissimilarity and one indicating complete similarity). It is well suited to species abundance data because it ignores joint absences and is largely driven by the most dominant species in the community (Krebs, 1989). For flies, where only a subset of data was collected, and various follow-up analyses, we performed t-tests. SAS 9.1 statistical software (SAS Institute, Inc., Cary, NC) was used for all analyses. 3. Results

Ground obligate arthropod (beetle) pitfall trap

Fig. 2. A schematic diagram of a trapping array at a sampling node.

weeks prior to sampling. We counted scat and dead fish to quantify coyote activity and to estimate available fish biomass. We also collected and analyzed scat for fish contents (Crooks, 1999). Four complete sampling sessions were conducted from July 2001 to December 2002. 2.3. Analyses Because the allochthonous input from the sea is largely seasonal, we divided the data into two seasons (die-off season: June through October and nondie-off season: November through May) according to Salton Sea fish and bird die-off records (Salton Sea Authority, 2006). For each taxonomic group, we averaged counts across the sampling sessions to derive a mean abundance per sampling node. Non-normally distributed variables were square-root or log-transformed prior to analysis. For model selection, we used backward and forward elimination multiple regression (BEMR and FEMR) to determine if distance (from sea edge), the presence of dunes, and/or the abundance of other taxonomic groups accounted for a significant amount of variation in each of the models. The square of distance (distance2) was included to account for nonlinear responses to the sea’s edge. Cover was used as a descriptive indicator of vegetation structure, since it is frequently correlated with above ground biomass (Hafner, 1977). Taxonomic groups were tested in the model if they had a known direct trophic or habitat relationship with the response group. In these procedures, variables are automatically either added to the null model (FEMR) or removed from the full model (BEMR). Partial F-tests are conducted between models to determine the contribution of each variable to model fit. The final models contain only variables with significant explanatory power (Kleinbaum et al., 1998). Because we had a large number of possible explanatory and response variables, these procedures were most efficient and thorough for our use in multiple model selection with the different taxonomic groups. A cutoff of P ¼ 0.10 was used as the acceptance criteria for all variables in each model. Although these procedures usually result in the same final model, BEMR has more of a tendency to overfit the data. Therefore, if a variable was included in BEMR, but not FEMR, the most conservative model with the least number of independent variables was chosen.

Floral and faunal groups exhibited different patterns in abundance with respect to distance from the sea (Fig. 3). Final regression models accounted for 0–46% of the total variance in floral and faunal abundances (Table 1). Most groups exhibited evidence of compositional changes next to the sea (Table 2, Fig. 4). Results of these analyses are presented by taxonomic group below. 3.1. Vegetation We identified 13 plant species across all sample nodes. Plant cover ranged from 1 to 77% among sampling nodes, averaging 13.5% over the study area. There was a strong nonlinear negative response of plant cover to distance from the sea (Table 1, Fig. 3a). Within 300 m of the sea, total cover was 1.9–4.6 times greater than at distances of 550 m or greater. The increase in cover was primarily due to a change in species composition. Plants associated with the creosote bush and desert saltbush communities were fairly evenly distributed throughout the study area. Plants associated with alkali sink vegetation (Allenrolfea occidentalis and Distichlis spicata) were abundant only within 300 m of the sea (Table 2, Figs. 3a and 4). Follow-up regression analyses revealed that distance was a significant predictor only of the density of alkaline sink vegetation but not of the density of the creosote bush/desert saltbush communities (Table 1). The presence of salt-cedar (Tamarix spp.) was largely associated with a sandy wash which ran through a single transect (A) at the 50 m and 1050 m nodes. 3.2. Ants We captured 4414 ants representing 15 species. Ant abundance showed a strong linear positive response to distance from the sea (Table 1, Fig. 3b), averaging 12 times greater inland versus next to the sea. Ants were negatively associated with the presence of sand dunes (P ¼ 0.022) and were 1.6 times more abundant during the summer die-off season. In addition, the composition of ant species near the sea was different compared with the inland community (Table 2, Fig. 4). For example, a member of the Dolicheroderus family (Forelius pruinosus) was very dominant at inland sites (58.8% of total), but only a small component of the community near the shore (1.0% of total). The native southern fire ant (Solenopsis xyloni) was 6.2 times more abundant inland, but was the most abundant species by the shore. In contrast, the harvester ant (Pogonomyrmex californicus) was evenly distributed throughout the study area and was in similar densities (averaging 7.4 versus 7.3 ants/session) next to both the shore and inland. Alkali sink vegetation was a negative predictor (P ¼ 0.087) of ant abundance.

Vegetation Cover

20

b

AS

15

Mean abundance

Percent cover

a

10 CBS 5

Ants

120 100 80

DO

60 40 NDO

20 0

0 50

300

550

800

50

1050

300

Distance from sea (m) Beetles

7

d

5 4

NDO

3

DO

2 1

Flies

40 30 25 20 15

DO

10 0

50

300

550

800

1050

50

Distance from sea (m)

4

f Mean abundance

Mean abundance

Birds

5 DO

3 2 NDO

1

Bats

1.5 1.0

NDO DO

0.5 0.0 -0.5

50

300

550

800

1050

50

300

Reptiles

1.6

550

800

1050

Distance from sea (m)

Distance from sea (m)

h

Small Mammals

5

1.2 DO

1.0 0.8 0.6

NDO

0.4

Mean abundance

1.4

Mean abundance

1050

Distance from sea (m)

0

g

1050

5

0

e

800

35

6

Mean abundance

Mean abundance

c

550

Distance from sea (m)

NDO

4 3 2

DO 1

0.2 0

0.0 50

300

550

800

50

1050

Distance from sea (m)

i

300

550

800

1050

Distance from sea (m)

Coyotes

16.0

Mean abundance

14.0 12.0 10.0 8.0

DO

6.0 4.0 2.0

NDO

0.0 0

2000

4000

6000

8000

10000

Distance from sea (m) Fig. 3. Relationship between floral and faunal abundances to distance from the sea. Results are presented for vegetation (a), comparing alkaline sink (AS) and creosote bush scrub (CBS), and for the different faunal groups (b–h), comparing mean abundance during the nondie-off season (NDO) to the die-off season (DO). Die-off season (June–October) corresponds to summer months with algal blooms, fish and bird die-offs, and aquatic insect emergence.

816

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

Table 1 Regression models, testing effects of distance from the Salton Sea and trophic interactions on floral and faunal abundances. Dependant variables

Whole model

Parameters in model

R2

Parameter estimates

P

Vegetation coverb Distance Distance2 Alkali sink vegetationb Distance Distance2 Creosote bush n.sc

0.318

0.010

0.957

Ant abundance Season Distance Dune presence Alkali sink vegetation cover

0.409

<0.0001

Beetle abundance Dune presence

0.120

Bird abundance Vegetation cover Season

0.274

0.318

Parameters not retained in final modela P

Std. coefficient

0.002 2.460 0.311

0.013 0.051

1.978 1.527

0.889 0.120

0.005 0.019

2.264 1.866

0.685 0.171 3.930 0.234

<0.0001 0.005 0.022 0.087

0.484 0.341 0.251 0.184

Dunes

0.002

dist2, beetles, reptiles, birds

0.003 0.576

0.003

0.346

dist, dist2, season, cover, ants, reptiles, birds, bats

0.150 0.232

<0.0001 0.047

0.479 0.211

dist, dist2, dunes, ants, beetles,

<0.0001

Bat abundance n.sc

dist, dist2, season, cover, beetles, birds

Reptile abundance Vegetation cover Season Ant abundance

0.305

Small mammal abundance Season Reptiles Coyote abundance (n ¼ 15) Distance Distance2 Season Fish abundance

0.126

0.466

<0.0001 0.040 0.154 0.083

0.006 0.008 0.044

0.298 0.325 0.247

0.390 0.700

0.006 0.019

0.361 0.307

0.452 0.085 0.176 0.024

0.042 0.014 0.052 0.021

1.900 2.177 0.279 0.484

dist, dist2, dunes, beetles, birds

0.011 dist, dist2, cover, ants, beetles, birds

0.001

a

Abbreviations used: distance (dist), distance squared (dist2), total vegetation cover (cover). Total cover and alkali sink vegetation cover were highly correlated. Therefore, where cover was significant, both total cover and alkali sink vegetation cover were tested separately to determine which variable was most predictive in the model. c No variables were significant in the model. b

3.3. Beetles and flies We captured 830 beetles representing 20 species. Beetles in the Tenebrionidae family comprised 85.9% of captures. There were no significant trends with respect to distance from the sea (Fig. 3c). The presence of dunes was the only significant predictor of beetle abundance (P ¼ 0.003, Table 1), accounting for 12.0% of the overall variation. Beetles were an average of 1.8 times more abundant when sand dunes were present. The composition of the beetle community remained relatively stable (Table 2, Fig. 2) with the Tenebrionid beetle, Edrotes ventricosus, as the most dominant species over the entire study area. This species is a habitat generalist that occupies stable and unstable dune habitat (Barrows, 2000; Rahn, 2000). The second most abundant species differed in relation to distance from the sea. Next to the sea, Cryptoglossa verrucosus, a generalist that avoids dune habitat, replaces Cryptoglossa laevis, a generalist associated with dune habitat (Table 2; Barrows, 2000). Flies were 2.5 times more abundant, 50 m from the sea compared to 1050 m inland (t ¼ 4.384, P ¼ 0.005; Fig. 3d). 3.4. Birds We observed 534 birds representing 45 species. Of these, 94.3% were landbirds, while the remaining 6.7% were shorebirds observed within 300 m of the shore. The final regression model accounted for 27.4% of the total variation in bird abundance. Birds

exhibited a strong positive response to vegetation cover (P ¼ 0.0026) and were more numerous (50% increase, P ¼ 0.047) during the die-off season than the rest of the year (Table 1). The composition of the bird community was similar at distances within 800 m of the sea, but differed from the most inland site (Fig. 3e). A small resident passerine, the verdin (Auriparus flaviceps), was a dominant component of the bird community throughout the area. However, many water associated birds such as the black-necked stilt (Himantopus mexicanus), snowy plover (Charadrius alexandrinus), common yellowthroat (Geothlypis trichas), and red-winged blackbird (Agelaius phoeniceus) were primarily found nearer to the sea, but not inland. Similarly, many desert and open habitat associated birds such as the lesser nighthawk (Chordeiles acutipennis), horned lark (Eremophila alpestris), and common poorwill (Phalaenoptilus nuttallii) were commonly found at the most inland sites (Table 2). 3.5. Bats We detected 67 bats representing seven species. Neither distance, vegetation cover, nor the abundance of other taxa helped to predict bat activity (Table 1). Furthermore, there was no difference in abundance between seasons (Fig. 3f). The bat species most frequently recorded were the pipistrelle (Pipistrellus hesperus), pallid bat (Antrozous pallidus), and western mastiff bat (Eumops perotis; Table 2). The pipistrelle and western mastiff bat are aerial

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

817

Table 2 Floral and faunal species composition in relation to distance from the Sea. Group

Near Sea (50 m)

Inland (1050 m) Habitat affiliationa

Most abundant species

Mean

Percent of total

Most abundant species

Mean

Percent of total

Vegetation

Allenrolfea occidentalis Isocoma acradenia Larrea tridenta

13.7 4.6 4.4

55.1 18.4 17.8

Larrea tridenta Tamarix spp. Isocoma acradenia

4.7 2.6 1.4

43.4 23.7 13.2

Ants

Solenopsis xyloni P. californicusb Camponotus festinatusc

2.5 1.2 0.3

53.9 27.2 6.3

Forelius pruinosus Solenopsis xyloni P. californicusb

41.2 8.2 1.2

78.1 15.5 2.2

Beetles

Edrotes ventricosus Cryptoglossa verrucosus Metaponium sp.

1.8 0.4 0.3

50.4 12.2 8.9

Edrotes ventricosus Cryptoglossa laevis Eleodes armata

1.9 0.8 0.4

45.5 20.0 9.0

Birds

Verdin Yellow-rumped warbler Black-necked stilt Loggerhead shrike

0.4 0.3 0.3 0.2

11.4 7.6 7.6 7.1

Blue-gray gnatcatcher Verdin Lesser nighthawk Yellow-rumped warbler

0.3 0.2 0.2 0.2

14.2 12.3 12.3 10.4

Bats

Antrozous pallidus Eumops perotis Pipistrellus hesperus

0.14 0.11 0.09

29.4 23.5 17.6

Eumops perotis Pipistrellus hesperus Antrozous pallidus

0.11 0.06 0.06

30.8 15.4 15.4

Reptiles

Aspidoscelis tigris Uta stansburiana Callisaurus draconoides

0.40 0.32 0.10

39.1 31.0 9.2

Aspidoscelis tigris Chionactis occipitalis Coleonyx variegatus

0.23 0.19 0.10

31.1 26.2 13.1

1.50 0.64 0.27 0.16

55.9 24.0 10.0 6.0

Dipodomys merriami Chaetodipus penicillatus Dipodomys deserti

1.83 0.27 0.21

73.9 10.9 8.7

S

Small mammals Dipodomys merriami Chaetodipus penicillatus Peromyscus maniculatus Mus musculus

c

D

S

G

D

S ¼ shore/water associate, G ¼ generalist, D ¼ desert specialist. Pogonomyrmex californicus. (Buckley).

insectivores, while the pallid bad primarily feeds on terrestrial arthropods (Orr, 1954). 3.6. Reptiles We captured 331 reptiles representing 16 species. The final regression model explained 30.5% of the variation in reptile abundance. Reptiles were more abundant in areas with greater plant coverage, greater ant densities, and during the summer die-off season (90% increase; Table 1, Fig. 3g). The composition of reptile species differed somewhat from the sea to inland areas (Table 2, Fig. 4). Although the western whiptail (Aspidoscelis tigris) was the most abundant species across the study area, the side-blotched lizard (Uta stansburiana) and the zebra-tailed lizard (Callisaurus draconoides) were co-dominants by the sea, while the western shovel-nosed snake (Chionactis occipitalis), western banded gecko (Coleonyx variegates), and flat-tailed horned lizard (Phrynosoma mcallii) were co-dominants inland.

mammals by the sea decreased by 38–50% during the die-off season (two-sample t ¼ 2.584, P ¼ 0.023). The abundance of reptiles was also a significant positive predictor of small mammal abundance (Table 1). Species composition remained relatively stable 1.0

0.8

Similarity Coefficient

a b

G

Habitat affiliationa

0.6

0.4

3.7. Small mammals 0.2

We captured 229 small mammals representing 7 species. The final regression model accounted for 12.6% of variation in small mammal abundance (Table 1). Season was the most significant predictor (P ¼ 0.006) of small mammal abundance, which was 43% lower during the summer die-off period than during the nondie-off period. This effect appeared to be largely driven by differences in seasonal abundance closest to the shore (50 m and 300 m, Fig. 3h). In comparison to the relatively stable inland sample nodes, which varied 6–25% between seasons, the average abundance of small

plants

bats

ants

reptiles

beetles

small mammals

birds 0.0 50

300

550

800

1050

Distance from sea (m) Fig. 4. Change in community composition in relation to distance from the sea using Bray–Curtis coefficient of similarity. All comparisons are to the most inland distance (1050 m). Values can range from 0 (very dissimilar) to 1 (very similar).

818

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

throughout the site. The two most abundant species were heteromyid rodents, Merriam’s kangaroo rat (Dipodomys merriami) and the desert pocket mouse (Chaetodipus penicillatus; Table 2, Fig. 4). 3.8. Coyotes Coyote activity increased during the die-off season and with the availability of dead fish. Activity also increased in a nonlinear fashion with proximity to the sea (Table 1, Fig. 3i). We collected a total of 1630 and 635 dead fish during the die-off and nondie-off seasons, respectively. All dead fish were collected along the transect closest to the shore (0–250 m). Coyote activity increased by an average of 80% during the fish die-off season (two-sample t ¼ 2.903, P ¼ 0.011) and they consumed an average of 390% more fish (twosample t ¼ 2.726, P ¼ 0.016). Coyote scat deposition increased the most (90–190%) at the two transects located between 250 m and 1750 m from the shore. Fish parts were found in coyote scat up to 3 km from the shoreline. 4. Discussion 4.1. Spatial trends among taxa The effect of the Salton Sea on the terrestrial ecosystem was evident in the plant community. The input of nutrients, excess water, and salts created a zone of alkaline sink vegetation that was twice as dense as that of the adjacent creosote bush and desert saltbush habitats. A concurrent study on soil chemistry found that chloride, sulphide and sodium ion concentrations were significantly higher at sites nearest the sea (Alonso et al., 2005). Accordingly, plant communities appeared to largely follow a gradient of moisture and salinity. The dominant plant species within 300 m of the sea are physiologically adapted to mesic conditions and able to tolerate very high concentrations of salts, while most of the predominant species inland are adapted to more xeric conditions and tolerate lower concentrations of salts (Holland and Keil, 1995; Mozingo, 1987). Birds, small mammals, reptiles, and coyotes all exhibited positive trends in density with increasing proximity to the sea. Regression analysis revealed that birds and reptiles were more abundant in association with higher density of plant cover rather than with distance from the sea. Small mammals and bats also showed positive trends in abundance in relation to cover, although not significant. The relationship between animal abundance, richness, and plant cover in desert habitats is well established (i.e. Cornett, 1987; Polis et al., 1997; Sewell, 2001). Within desert environments, grasses, herbs, shrubs, and trees create nutrient-rich microhabitats via accumulation of leaf litter and active recycling of dead plant material (Vasek and Lund, 1980). In turn, seed and leaf production support many insect herbivores and their secondary consumers while providing cover from direct sun, wind, and terrestrial and aerial predators. Large increases in terrestrial arthropod densities near shore have been documented in other studies (Barrett et al., 2005; Polis and Hurd, 1996b; Sanchez-Pinero and Polis, 2000). Surprisingly, we found either a negative association (ants) or no association (beetles) with distance from shore. Since ants were significantly and negatively associated with alkali sink vegetation, we believe the lower densities of ants closer to the sea were largely due to lower food resource availability associated with this vegetation. The most abundant ant species found throughout the study area, F. pruinosus, was largely absent from the shore. This species is known to preferentially feed on insects and honeydew, a liquid excreted from various herbivorous insects (Ho¨lldobler and Wilson, 1990; Wheeler, 1910; Wheeler and Wheeler, 1973). Both of these

resources are likely in short supply near the sea, as most herbivorous insects cannot feed on the dominant alkali sink plant, iodine bush, due to high salt concentrations in its tissues (Mozingo, 1987; Ungar, 1991). In contrast, generalist and seed opportunist ant species (S. xyloni, Pogonomyrmex spp.) were found in lower densities, but more evenly distributed throughout the study area. For beetles, the lower apparent abundance by the sea edge was associated with the absence of active sand dunes. We captured primarily Tenebrionid beetle species that are often restricted to or prefer active dune habitat and may be rare or absent from cooler or moister sites (Barrows, 2000; Rahn, 2000). It is also possible that lower arthropod densities next to the Salton Sea may be in part due to top-down trophic controls. Generalist small mammal species, such as deer mice (Peromyscus maniculatus), which were only found at the closest distances to the sea, have been suggested to suppress Tenebrionid beetle densities elsewhere (Stapp and Polis, 2003). 4.2. Temporal trends among taxa Allochthonous inputs from the Salton Sea to the surrounding terrestrial ecosystem can be highly variable in time. Algal blooms, fish die-offs, bird die-offs, and aquatic insect emergence typically occur in the spring and summer months or ‘‘die-off season’’ (Salton Sea Authority, unpublished data). Increased seed fall and temperatures may also influence the activity of animal species during this time. Ants, birds, reptiles, and coyotes were all at higher densities during the summer die-off season. Of these, only birds and coyotes exhibited increased densities nearer to the sea. Both are highly vagile and can thus move large distances within and among habitats to exploit the seasonal and spatial distribution of resources (Polis et al., 1997). Although we lack direct proof, we hypothesize that higher densities of birds near the sea during the die-off season may be in part a response to the emergence of brine flies (Ephydra hians) in great densities at the sea (Jehl, 1988). Many birds will actively forage on emergent aquatic insects, and we expect that brine flies are a significant resource to aerial insectivores around the Salton Sea. Although we did not document such a response in bats, this may have been because too few bats were recorded during our surveys to discern a difference. The low numbers may have been partly due to insufficient survey effort or lack of suitable roosting sites such as caves, rock crevices, and artificial structures (Racey and Entwistle, 2003) within 20 km of our study site. Coyotes and small mammals had differing patterns of abundance in relation to distance from the sea during the die-off season. Coyote activity increased two to threefold during the die-off season (within 250–1750 m of the shore), and coyote fish consumption increased fivefold. Furthermore, coyote scats containing fish were found up to 3 km from the sea, confirming that coyotes exploit marine subsidies and transport aquatic nutrients to inland areas (Rose and Polis, 1998). In contrast, densities of small mammals decreased near the sea during the die-off season. Since lower densities were found only within 300 m of the sea, we do not believe decreases in activity or torpor explain this trend. One possible explanation is spillover predation from coyotes. Predators that exploit temporary availability of allochthonous prey can also dramatically depress native local (in situ) prey items (Henschel et al., 2001; Murakami and Nakano, 2002; Polis et al., 1997). Similar patterns have been documented with coyote and coastal rodent populations along the Gulf of California (Rose and Polis, 1998). Ant and reptile densities were significantly greater during the summer die-off period. The increases were largely at inland sites, indicating an increased use of in situ rather than aquatic-based resources. Sabo and Power (2002) and Barrett et al. (2005) found that western fence lizards (Sceloporus occidentalis) and side-

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

blotched lizards, respectively, exhibited increased densities in response to availability of aquatic-based or subsidized terrestrial prey; however, we did not find any increases for insect prey or reptiles in our system. 4.3. Community composition Most faunal groups exhibited marked changes in community composition relative to distance from the sea. It has been widely reported that habitat generalists and edge specialists benefit from habitat edges while habitat specialists suffer (Bentley et al., 2000; Laurance and Yensen, 1991; Lidicker, 1999; Lidicker and Koenig, 1996). We found similar faunal patterns in relation to the edge of the Salton Sea. Habitat generalists, such as the Tenebrionid beetles (E. ventricosus and Eleodes armata), western whiptail, side-blotched lizard, coyote, and species associated with wetlands (i.e. yellowrumped warbler (Dendroica coronata) and black-necked stilt) were only present at the sea or were more predominant near the sea. In addition, two generalist small mammal species, the deer mouse and the house mouse (Mus musculus) were only captured at the trap arrays closest to the sea. On the other hand, species with specialized adaptations to loose sand or open desert habitat, such as the Tenebrionid beetle (C. laevis), western shovel nosed snake, flat tailed horned lizard, desert kangaroo rat (Dipodomys deserti), lesser nighthawk, and horned lark were found more frequently inland. 4.4. Comparison to Gulf of California system Our results largely support those expected from the theories of spatial subsidies as proposed by Polis and Hurd (1996b). Although profound, the presumed effects of allochthonous inputs from the Salton Sea on adjacent Sonoran Desert productivity appeared to be markedly reduced in comparison to those reported in nearby Gulf of California island and peninsular desert systems. We found faunal densities from 0.1 to 2.9 times greater on the shoreline in comparison to the remarkable 3–24 (and 560) fold increases for the same taxa reported along the Gulf of California (Polis and Hurd, 1996a,b; Sanchez-Pinero and Polis, 2000). The lesser response was largely expected, as the total allochthonous input from the Salton Sea is likely several orders of magnitude lower than the Gulf. The majority of biomass from the Salton Sea is composed of dead fish and birds that are actively removed from the shoreline due to wildlife disease and recreational concerns. This differs from the rich intertidal communities and algal seaweed detritus found in the Gulf. In addition, rather than concentrating on small islands with high perimeter–area ratios, the subsidies and associated responses in flora and fauna of the Salton sea rapidly diffuse into the surrounding desert ecosystem. Finally, the Salton Sea is a relatively young 100-year old saline lake. Therefore, specialized adaptations of Colorado desert terrestrial flora and fauna to its non-native marine faunal inputs have had little time to develop. We expect this explains why positive shoreline responses were largely limited to highly vagile animals (birds and coyotes) and habitat and dietary generalists. 5. Conclusions Our findings should serve to expand the understanding of allochthonous dynamics and spatial subsidies across multiple spatial scales and highly distinctive aquatic systems. They also contribute to our knowledge of species response patterns to ecotonal boundaries. We recognize that our study largely uses indirect evidence for spatial subsidies. Future studies at the sea should include isotope analyses to directly quantify the magnitude of these subsidy effects on the terrestrial ecosystem.

819

Acknowledgements We thank USGS scientists; Dino Barnham, Steve Carroll, Sara Schuster, Sierra Hayden, Chris Haas, Allan Hebbert, Tritia Matsuda, Zsolt Kahancza, and Elizabeth Gallegos for assisting with field data collection, Andrea Atkinson and Julie Yee for statistical advice, Doug Chamblin for editorial support, and several anonymous reviewers provided helpful comments on earlier drafts of this manuscript. We thank Dr. Andrew Suarez and Dr. Phil Ward at the Universities of Illinois and California, Davis for assisting with ant identification. We also thank Douglas Barnum, USGS Salton Sea Science Office, and Cheryl Rodriguez, Bureau of Reclamation, for providing invaluable logistical support and Cameron Barrows, Center for Natural Lands Management, and Rich Rust, University of Nevada, Reno for assisting in identification of beetles. This research was funded in part by the Salton Sea Science Office, Bureau of Reclamation and conducted on U.S. Bureau of Land Management, California State Parks, and the California Department of Fish and Game lands. The use of trade, product, or firm names in this publication does not imply endorsement by the U.S. Government. References Alonso, R., Bytnerowicz, A., Yee, J.L., Boarman, W.I., 2005. Atmospheric dry deposition in the vicinity of the Salton Sea, California – II: measurement and effects of an enhanced evaporation system. Atmospheric Environment 39, 4681–4689. Anderson, W.B., Polis, G.A., 1998. Marine subsidies of island communities in the Gulf of California: evidence from stable carbon and nitrogen isotopes. Oikos 81, 75–80. Bardgett, R.D., Anderson, J.M., Behan-pelletier, V., Brussard, L., Coleman, D.C., Ettema, C., Moldenke, A., Schimel, J.P., Wall, D.H., 2001. The influence of soil biodiversity on hydrological pathways and the transfer of materials between terrestrial and aquatic ecosystems. Ecosystems 4, 421–429. Barrett, K., Anderson, W.B., Wait, D.A., Grismer, L.L., Polis, G.A., Rose, M.D., 2005. Marine subsidies alter the diet and abundance of insular and coastal lizard populations. Oikos 109, 145–153. Barrows, C.W., 2000. Tenebrionid species richness and distribution in the Coachella Valley sand dunes (Coleoptera: Tenebrionidae). Southwestern Naturalist 45, 306–312. Bentley, J.M., Catterall, C.P., Smith, G.C., 2000. Effects of fragmentation of Araucarian vine forest on small mammal communities. Conservation Biology 14, 1075–1087. Briers, R.A., Cariss, H.M., Geoghegan, R., Gee, J.H.R., 2005. The lateral extent of the subsidy from an upland stream to riparian lycosid spiders. Ecography 28, 165–170. Carpelan, L.H., Linsley, R.H., 1961. The pile worm, Neanthes succinea (Frey and Leuckart). In: Walker, B.W. (Ed.), The Ecology of the Salton Sea California, in Relation to the Sportfishery. Fish Bulletin 113. California Department of Fish and Game, Sacramento, California, pp. 49–62. Chew, R.M., Chew, A.E., 1965. The primary productivity of a desert shrub (Larrea tridentata) community. Ecological Monographs 35, 355–375. Cornett, J., 1987. Wildlife of the North American Deserts. Nature Trails Press, Palm Springs, California. Costa-Pierce, B., Doyle, R., 1997. Genetic identification and status of tilapia regional strains in southern California. In: Costa-Pierce, B.A., Rakocy, J. (Eds.), Tilapia Aquaculture in the Americas, vol. 1. The World Aquaculture Society, Baton Rouge, Louisiana, pp. 1–20. Crooks, K.R., 1999. Mammalian carnivores, mesopredator release, and avifaunal extinctions in a fragmented system. Ph.D. thesis, University of California, Santa Cruz, California. Detwiler, P.M., Coe, M.F., Dexter, D.M., 2002. The benthic invertebrates of the Salton Sea: distribution and seasonal dynamics. Hydrobiologia 473, 139–160. Fisher, R.N., Stokes, D., Rochester, C., Brehme, C., Hathaway, S.A., Case, T.J., 2008. Herpetological monitoring using a pitfall trapping design in southern California. In: U.S. Geological Survey Techniques and Methods, 2-A5. U.S. Geological Survey, pp. 1–44. Garcia, A., Whalen, D.M., 2003. Lizard community response to a desert shrubland– intertidal transition zone on the coast of Sonora, Mexico. Journal of Herpetology 37, 378–382. Hafner, M.S., 1977. Density and diversity in Mojave Desert rodent and shrub communities. Journal of Animal Ecology 46, 925–938. Harding, E.K., Stevens, E., 2001. Using stable isotopes to assess seasonal patterns of avian predation across a terrestrial-marine landscape. Oecologia 129, 436–444. Henschel, J.R., Mahsberg, D., Stumpf, H., 2001. Allochthonous aquatic insects increase predation and decrease herbivory in river shore food webs. Oikos 93, 429–438. Holland, V.L., Keil, D.J., 1995. California Vegetation. Kendall/Hunt Publishing Company.

820

C.S. Brehme et al. / Journal of Arid Environments 73 (2009) 811–820

Ho¨lldobler, B., Wilson, E.O., 1990. The Ants. Belknap Press of Harvard University Press, Cambridge, Massachusetts. Jehl Jr., J.R., 1988. Biology of the eared grebe and Wilson’s phalarope in the nonbreeding season: a study of adaptations to saline lakes. Studies in Avian Biology 12, 1–74. Kent, M., Coker, P.,1992. Vegetation Description and Analysis. Belhaven Press, London. Kleinbaum, D.G., Kupper, L.L., Muller, K.E., Nizam, A., 1998. Applied Regression Analysis and Multivariable Methods, third ed. Brooks/Cole Publishing Company, Pacific Grove, CA. Krebs, C.J., 1989. Ecological Methodology. Harper Collins Publishers, New York, 654 pp. Lange, C.B., Tiffany, M.A., 2002. The diatom flora of the Salton Sea, California. Hydrobiologia 473, 179–201. Laurance, W.F., Yensen, E., 1991. Predicting the impact of edge effects in fragmented habitats. Biological Conservation 55, 77–92. Lidicker, W.Z., 1999. Responses of mammals to habitat edges: an overview. Landscape Ecology 14, 333–343. Lidicker, W.Z., Koenig, W.D., 1996. Responses of terrestrial vertebrates to habitat edges and corridors. In: McCullough, D.R. (Ed.), Metapopulations and Wildlife Conservation. Island Press, Washington D.C. Loucks, O.L., 1975. Models linking land–water interactions around Lake Wingra, Wisconsin. In: Hasler, A.D. (Ed.), Coupling of Land and Water Systems. SpringerVerlag, New York, pp. 53–63. Mendelsohn, M.B., Boarman, W.L., Fisher, R.N., Deutschman, D.H., Hathaway, S.A., 2007. Salt sea spray and bottom-up dynamics in the terrestrial avifauna of Salton Sea. Journal of Arid Environments 68, 574–587. Minshall, G.W., 1967. Role of allochthonous detritus in the trophic structure of a woodland stream. Ecology 48, 139–149. Mozingo, H.N., 1987. Shrubs of the Great Basin: a Natural History. University of Nevada Press. Mulholland, P.J., 1981. Organic carbon flow in a swamp–stream ecosystem. Ecological Monographs 51, 307–322. Murakami, N., Nakano, S., 2002. Indirect effect of aquatic insect emergence on a terrestrial insect population through by birds predation. Ecology Letters 5, 333–337. Odum, W.E., Heald, E.J., 1975. Mangrove forests and aquatic productivity. In: Hasler, A.D. (Ed.), Coupling of Land and Water Systems. Springer-Verlag, New York, pp. 129–136. Orr, R.T., 1954. Natural history of the pallid bat (Antrozous pallidus). Proceedings of the California Academy of Sciences 28, 165–246. Patten, M.A., McCaskie, G., Unitt, P., 2003. Birds of the Salton Sea: Status, Biogeography, and Ecology. University of California Press, Berkeley. Polis, G.A., Hurd, S.D., 1996a. Allochthonous inputs across habitats, subsidized consumers, and apparent trophic cascades: examples from the ocean–land interface. In: Polis, G.A., Winemiller, K.O. (Eds.), Food Webs: Integration of Patterns and Dynamics. Chapman and Hall, pp. 275–285.

Polis, G.A., Hurd, S.D., 1996b. Linking marine and terrestrial allochthonous input from the ocean supports high secondary productivity on small islands and coastal land communities. American Naturalist 147, 396–423. Polis, G.A., Anderson, W.B., Holt, R.D., 1997. Toward an integration of landscape and food web ecology: the dynamics of spatially subsidized food webs. Annual Reviews of Ecology and Systematics 28, 289–316. Racey, P., Entwistle, A.C., 2003. Conservation ecology of bats. In: Kunz, T.H., Fenton, M.B. (Eds.), Bat Ecology. The University of Chicago Press, pp. 680–722. Rahn, M.E., 2000. Nested Coleoptera and Orthopters on sand dunes in the Basin and Range Province of western North America. Journal of Insect Conservation 4, 33–43. Riedel, R., Caskey, L., Costa-Pierce, B.A., 2002. Fish biology and fisheries ecology of the Salton Sea, California. Hydrobiologia 473, 229–244. Rose, M.D., Polis, G.A., 1998. The distribution and abundance of coyotes: the effects of allochthonous food subsidies from the sea. Ecology 79, 998–1007. Sabo, J.L., Power, M.E., 2002. Numerical response of lizards to aquatic insects and short-term consequences for terrestrial prey. Ecology 83, 3023–3036. Salton Sea Authority, 2006. Salton Sea Authority Plan for Multipurpose Project Available from: http://www.saltonsea.ca.gov (accessed 14.05.07.). Sanchez-Pinero, F., Polis, G.A., 2000. Bottom-up dynamics of allochthonous input: direct and indirect effects of seabirds on islands. Ecology 81, 3117–3132. Sawyer, J.O., Keeler-Wolf, T., 1995. A Manual of California Vegetation. Native Plant Society, Sacramento, California. Sewell, J., 2001. Desert Ecology: an Introduction to Life in the Arid Southwest. University of Utah Press, Salt Lake City. Stapp, P., Polis, G.A., 2003. Marine resources subsidize insular rodent populations in the Gulf of California, Mexico. Oecologia 134, 496–504. Suarez, A.V., Bolger, D.T., Case, T.J., 1998. Effects of fragmentation and invasion on native ant communities in coastal southern California. Ecology 79, 2041–2056. Tiffany, M.A., Swan, B.K., Watts, J.M., Hurlbert, S.H., 2002. Metazooplankton dynamics in the Salton Sea, California, 1997–1999. Hydrobiologia 473, 103–120. Ungar, I.A., 1991. Ecophysiology of Vascular Halophytes. CRC Press, Inc. Vasek, F.C., Lund, L.J., 1980. Soil characteristics associated with a primary plant succession on a Mojave desert dry lake. Ecology 61, 1013–1018. Walker, B., Whitney, R., Barlow, G., 1961. Fishes of the Salton Sea. In: Walker, B.W. (Ed.), The Ecology of the Salton Sea California, in Relation to the Sportfishery. Fish Bulletin 113. California Department of Fish and Game, Sacramento, California, pp. 49–62. Watts, J.M., Swan, B.K., Tiffany, M.A., Hurlbert, S.H., 2001. Thermal mixing, and oxygen regimes of the Salton Sea, California, 1997–1999. Hydrobiologia 466, 159–176. Wheeler, W.M., 1910. Ants: Their Structure, Development, and Behavior. Columbia University Press, New York. Wheeler, G.C., Wheeler, J., 1973. Ants of Deep Canyon. University of California, Riverside.