Journal of Arid Environments 88 (2013) 165e174

Contents lists available at SciVerse ScienceDirect

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

Seed removal patterns in burned and unburned desert habitats: Implications for ecological restoration A.A. Suazo a, *, D.J. Craig a, C.H. Vanier b, S.R. Abella a a b

Department of Environmental and Occupational Health, University of Nevada Las Vegas, Las Vegas, NV 89154-3064, USA Division of Educational Outreach, University of Nevada Las Vegas, Las Vegas, NV 89154-1019, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 24 March 2012 Received in revised form 25 August 2012 Accepted 28 August 2012 Available online 16 October 2012

In desert ecosystems, selective foraging by seed consumers affects distribution and composition of soil seed banks, influencing plant population dynamics. However, the roles of consumers in burned habitats where direct seeding is used during ecological restoration to replenish depleted seed banks have not been well established. We evaluated seed removal for nine seed species over 12 months in burned and unburned shrublands in the Mojave Desert, USA. Percentage of total seed removed was highest during spring (16% of offered seed) and summer (21%). Rodents removed much of the large-seeded Coleogyne ramosissima in both burned and unburned habitats, while seed removal of this species by ants was low in burned and moderate in unburned habitats. Ants removed the greatest amount of small-seeded species (Penstemon bicolor, Encelia farinosa, and Sphaeralcea ambigua) in unburned habitat, indicating that ants can exploit different seed masses. Seed removal imposes limitations on seed availability, particularly for large-seeded species, as both rodents and ants selected seeds of C. ramosissima. Successful restoration seeding projects in arid lands may require protecting seed from granivore pressure, and seed species selection and season of seeding warrant consideration to reduce seed loss. Ó 2012 Elsevier Ltd. All rights reserved.

Keywords: Exotic grasses Granivory Plant establishment Recruitment Seeding Wildfire

1. Introduction Invasion by exotic annual grasses has increased the amount and continuity of fine fuels in southwestern North American deserts, corresponding with increased frequency and intensity of wildfires (Brooks and Chambers, 2011). Repeated, high-intensity fires can diminish native plant abundance by killing seeds stored in soil seed banks (Esque et al., 2010). To mitigate negative effects of wildfires on late-successional desert plant communities, land managers often use direct seeding to reestablish seed banks and accelerate native plant establishment (James et al., 2011). Seeding may be the only practical approach to reintroduce a native seed source to large burned areas in steep and rocky terrain. However, success of seeding projects is low because broadcast seeds are exposed to desiccation, wind, and seed consumers (Bainbridge, 2007). Nevertheless, manipulating site conditions (Montalvo et al., 2002), seed species mixtures (Abella et al., 2009), seeding rates, and granivore pressure (Orrock et al., 2009) can improve seeding success. Seed-

* Corresponding author. Present address: College of Natural Resources, University of Idaho, PO Box 441133, Moscow, ID 83844-1133, USA. Tel.: þ1 208 885 6538. E-mail address: [email protected] (A.A. Suazo). 0140-1963/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jaridenv.2012.08.016

eating rodents and ants are common in deserts and can alter plant recruitment and composition through seed predation and dispersal (Price and Jenkins, 1986); thus, it is particularly important to understand how seed consumers affect broadcast seeds. The effects of granivores on density and species composition of soil seed banks, which might affect post-fire plant recovery, are well established in many desert ecosystems throughout the world (Bricker et al., 2010; Hulme, 1998). In the Mojave Desert, for example, granivorous rodents harvest much of the annual seed production of large-seeded species (Soholt, 1973). Results from long-term exclusion of granivorous rodents in the Chihuahan Desert indicate that density of large-seeded species doubles in the absence of rodents (Brown et al., 1979). Rodents selectively forage on large seeds for many possible reasons. Large seeds are conspicuous and remain longer on the soil surface than small seeds. Large seeds have difficulty entering the soil seed bank; consequently, large seeds are exposed to granivorous rodents over a longer period than small seeds (Bekker et al., 1998). In addition, optimal foraging theory predicts that it might be more efficient for a seed predator to forage for energy and nutrient-rich large seeds (Schoener, 1971). Granivorous rodents can also locate seeds buried in the soil by memory, tactile cues, and olfaction (Vander Wall, 1998). Thus, these animals are well equipped for securing seed

166

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

resources to fulfill their dietary needs. Specific features of seed resources can cause selective foraging by granivorous rodents, which could dramatically influence structure of desert plant communities. Harvester ants are also major seed consumers in arid environments (Brown et al., 1979). Seed selection is due partially to morphological constraints of fit between seeds and ant mandibles (Davidson, 1977; Pirk and Lopez de Casenave, 2010). As a result, small-bodied ant species prefer small seeds while large-bodied ant species prefer large seeds (Pirk and Lopez de Casenave, 2010). This suggests that ants can exploit seeds of a wide range of seed masses. In the central Monte Desert, Argentina, seed mass was an important factor determining seed preference and diet of three species of harvester ants (Pirk and Lopez de Casenave, 2010). Ants likely selected large seeds to maximize their energy intake per time and energy spent harvesting seeds, as predicted by optimal foraging theory (Schoener, 1971), but preference for large seeds was weaker in ant species with morphological limitation in mandible length. These ants were unable to carry large seeds back to their nest (Pirk and Lopez de Casenave, 2010). Similar seed-preference patterns by harvester ants on the basis of seed size occur in North American deserts (Davidson, 1977). Seed attributes are rarely considered when developing seed mixes for seeding disturbed arid lands (Chambers and MacMahon, 1994). Understanding how different seed consumers select seeds is important in the success of restoration efforts, because granivore pressure is not random and selective seed harvesting can lead to severe seed loss. Consequently, granivores could be important to successional trajectories of plant communities undergoing restoration efforts (Brown and Heske, 1990). Seed loss by seed consumers can be influenced by vegetation cover and habitat conditions, such as post-fire environments (Zwolak et al., 2010), and the type of consumer (i.e., rodents or ants). Some species of desert rodents respond positively to the open habitat created after fire (Horn et al., 2012; Vamstad and Rotenberry, 2010). For example, bipedal species, such as kangaroo rats (Dipodomys merriami), increase in abundance after desert fires (Horn et al., 2012; Vamstad and Rotenberry, 2010). Other species, many of which are quadrupedal, mostly forage in under-shrub microhabitat to minimize predation risk (Sivy et al., 2011). Fire initially reduces shelter available to rodent granivores, which may lead to either lower granivory or riskier foraging behavior in burned areas. However, post-fire foraging patterns (i.e., seed removal) have not been well documented. Ant foraging activity is not necessarily affected by fire (Zimmer and Parmenter, 1998), but high rates of seed removal have occurred after fire in sclerophyllous (Andersen, 1988; O’Dowd and Gill, 1984) and savanna habitats in Australia (Parr et al., 2007). Interactions among seed consumers and habitat conditions are of particular interest in the context of using revegetation seeding techniques, as seed consumers may remove substantial seed used in restoration efforts. The seed resources available to a granivore provide the context in which an individual seeks to optimize its net energy gain. When a variety of seeds are abundant, granivores can prefer seeds that provide the most energy per time and effort spent in harvesting them (Schoener, 1971). Seeding failure is likely a result of multiple interacting biotic and abiotic factors; however, studies that evaluate the relative influences of seed traits (e.g., seed mass, length, morphological features, and nutrient content) and seed consumers on plant establishment are lacking in many arid ecosystems, including the Mojave Desert (Chambers and MacMahon, 1994). Here, we used experimental exclosures and plant species commonly used in direct seeding to evaluate seed removal patterns of nine Mojave Desert perennial species in burned and unburned

habitats. In addition, we conducted a field seedling emergence study to explore effects of seed consumers on plant recruitment. Specifically, we evaluated whether 1) seed removal patterns differed between rodents and ants, and if 2) temporal variability and seed mass influenced seed removal in burned and unburned habitats. We expected the amount of seed removed by ants and rodents in burned habitat to be larger than that removed in unburned habitat because foraging efficiency would increase with reduced vegetation cover in the post-fire environment (Horn et al., 2012; Vamstad and Rotenberry, 2010). We also expected seed mass to explain patterns of seed removal, as seed consumers would select heavy over light seeds because heavy seeds have a higher energetic content (Schoener, 1971). Foraging activity of ants and rodents diminishes with declining temperatures (Pol et al., 2011); thus, we expected temporal variation to reveal low amounts of seed removed during winter. 2. Materials and methods 2.1. Study area The study area was 65 km south of Las Vegas, Nevada, USA (35 860 76.6300 N, 115 450 72.7400 W) at an elevation of 1260 m in the Mojave Desert. Precipitation in 2008, the year we began the experiment, was 12 cm according to a weather station 2 km northwest of the study sites (Western Regional Climate Center, Reno, NV). Soil temperature, at a 5-cm depth in interspaces between shrubs (Onset HOBO temperature data loggers, Cape Cod, MA, USA), ranged from 4  C in February to 25  C in June. We studied seed removal and seedling recruitment patterns in the 13,585-ha Goodsprings Fire, which was ignited by lightning in June 2005. In fall 2008, we established two study sites: one in burned habitat and one in unburned habitat. Each site was more than 150 m from burn edges to minimize edge effects, and within the same soil type and topography. Vegetation cover in burned areas was sharply reduced post-fire. Only standing burned skeletons of dominant plant species, such as Larrea tridentata, remained (Fig. 1). 2.2. Seed consumers Data from a night of trapping in April 2010 from trap lines set in burned and unburned habitat revealed rodent activity in both types of habitat. Four species of seed-eating rodents were caught in the unburned site: Ammospermophilus leucurus (white-tailed antelope squirrel), D. merriami (Merriams’ kangaroo rat), Chaetodipus formosus (long-tailed pocket mouse), and Dipodomys microps (chisel-toothed kangaroo rat). Two species inhabited the burned site: D. merriami and Peromyscus maniculatus (deer mouse). D. merriami dominated both habitats, composing 60% (9/ 15) of captures in burned and 69% (9/13) of captures in unburned habitat. While we did not measure ant species composition, granivorous ant species likely to inhabit the sites belong to the genera Pogonomyrmex, Pheidole, and Messor, which are widely distributed across Mojave Desert habitats (Rundel and Gibson, 1996). 2.3. Experimental cages Experimental cages to manipulate granivore access were cylindrical in shape, constructed of hardware cloth (1.3 cm mesh size), and were 30 cm high and 45 cm in diameter. All cages were buried to a depth of 10 cm to deter rodents from digging under cages and included tops to exclude birds. Different cage types allowed access to: 1) ants only, 2) rodents only, 3) ants and rodents, or 4) neither

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

167

2.4. Seed removal experiments Seed removal patterns were evaluated in six blocks of three cage types: 1) ant access, 2) rodent access, and 3) no access (control) to experimental seeds. Seed removal trials were conducted at monthly intervals, starting in October 2008 and ending in October 2009. Removal trials were not conducted in December 2008 due to snow cover at the study sites. In each seed removal trial, we used seeds of eight (1st trial) or nine (other trials) native species (Table 1). In each cage within each block, a Petri dish (120 mm  15 mm) containing 20 seeds of each species was established. After four nights, seeds remaining in Petri dishes were sorted and counted. Seeds were handled with long-point tweezers to eliminate confounding factors (e.g., human scent) that might bias seed removal rates (Duncan et al., 2002). To relate seed traits to amount of seed removed, 20 randomly selected seeds per species were weighed on an analytical balance (Sartorius, NY, USA) and maximum length was measured using digital calipers (Mitutoyo Corp., Chicago, IL, USA). Seeds were obtained from local collections from the U.S. Bureau of Land Management (Southern Nevada District, Las Vegas, NV, USA). 2.5. Seedling recruitment

Fig. 1. Burned (top) and unburned (bottom) habitats with experimental cages used to evaluate seed removal and seedling establishment of nine plant species native to the Mojave Desert, USA.

ants nor rodents. Cages which allowed rodent access had a single hole (6 cm  6 cm) cut in each cardinal direction to allow small mammal seed removal. Cages which excluded ants were surrounded by a 2-cm exposed area of vinyl flashing coated with fluon (BioQuip, CA, USA) to create a slippery surface and prevent ants from entering cages.

Seedling recruitment and survival of nine native species (Table 1) were evaluated in six blocks of four cage types: 1) ant access, 2) rodent access, 3) ant and rodent access, and 4) no access (control). In November 2008, we hand broadcasted 20 seeds per species in each of the four cages within each block (seed removal and seedling recruitment experiments were conducted in different cages). Seeds were patted down to ensure contact with the soil. We inventoried cages for seedling emergence at monthly intervals, and emerged seedlings of sown species were tagged and counted. We recorded seedling survival of tagged individuals during the first five months after emergence and conducted a final assessment 17 months after emergence to evaluate effects of granivory and habitat on survival. While seed of some desert species require dormancybreaking mechanisms and exhibit different germination rates in the field than in the laboratory, we wished to ensure that at least a portion of the experimental seed was readily germinable. We assessed germination by germinating 120 seeds per species wrapped in moistened paper towels in a laboratory (Table 1). 2.6. Data analysis The proportion (transformed as Oarcsin) of seeds removed for each plant species was analyzed as a partially hierarchical model (Quinn and Keough, 2002) which included: experimental block as a random subject effect, burned and unburned habitats as

Table 1 Seed characteristics of species used to evaluate patterns of seed removal by granivorous rodents and ants in burned and unburned Mojave Desert scrub habitat. Mass is reported as mean  SE. Life form follows the U. S. Department of Agriculture PLANTS Database (http://plants.usda.gov/). Species

Species code

Family

Life form

Seed mass (mg)

Ambrosia dumosa Baileya multiradiata Coleogyne ramosissima Encelia farinosa Eriogonum fasciculatum Hymenoclea salsola Larrea tridentata Penstemon bicolor Sphaeralcea ambigua

AMDU BAMU CORA ENFA ERFA HYSA LATR PEBI SPAM

Asteraceae Asteraceae Rosaceae Asteraceae Polygonaceae Asteraceae Zygophyllaceae Scrophulariaceae Malvaceae

Subshrub/shrub Forb Shrub Subshrub/shrub Subshrub/shrub Subshrub Shrub Forb Subshrub/forb

4.7 0.5 15.8 1.0 1.0 3.9 4.7 0.5 1.0

NA ¼ test not performed.

        

0.71 0.05 0.61 0.10 0.10 0.38 0.36 0.11 0.08

Germination (%) 4 64 NA 8 9 35 38 NA 26

168

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

between-subject effects (tested over blocks within habitats), and seed removal treatment and date as within-subject effects, along with all interactions (tested over their respective interactions with blocks within habitats) as fixed effects. The data would ideally have been analyzed in a generalized linear model with a binomial error including the effect of plant species and its interactions with other fixed effects. However, variability in seed removal differed sufficiently among species such that there was no stable model which could incorporate the species together in the same analysis. Thus, models were fit for each species independently and residuals met the normality assumption after transformation. Multiple comparisons were made using a Bonferroni adjustment for habitat and treatment combinations within dates for each species (a ¼ 0.05). Analyses were conducted in SAS software (version 9.1, Cary, NC, USA). Relationships between seed mass and average probability of seed removal were analyzed using model II ordinary least squares regression. All variables were log10 transformed to meet assumptions of normality prior to analysis, and 95% confidence intervals for slopes were created from 1000 permutations (Legendre and Legendre, 1998). Slopes which did not include zero in their 95% confidence interval were considered significant. The analysis was implemented in R 2.10.1 (http://www.R-project.org). Effects of burned and unburned habitat and seed removal treatment on seedling recruitment were evaluated using a generalized linear model with a binomial error term. In the partially hierarchical model, experimental blocks were the subject effect, burned or unburned habitat was the between-subject effect (tested over blocks within habitat), ant and rodent exclusion were separate within-subject effects, and all interactions of fixed effects (tested over each effect’s interaction with habitat within blocks) were included. Mean values were back-transformed prior to reporting, and multiple comparisons were Tukey-adjusted. Pairwise comparisons were performed when the ANOVA p-value was significant (main effect p < 0.05 and interaction p < 0.10). Analyses were conducted in SAS software (version 9.1, Cary, NC, USA). Seedling survival probabilities among burned and unburned habitat and granivory treatments were estimated using the Cox proportional hazard regression model conducted in R 2.10.1 (http:// www.R-project.org). 3. Results 3.1. Seed removal patterns The amount of seeds removed by rodents and ants differed by seed species, burned and unburned habitat, and time seed trials were conducted (Table 2; Fig. 2). Seed removal for Ambrosia dumosa, Hymenoclea salsola, and L. tridentata was relatively low across all seed trials dates. Rodents removed between 20 and 30% of A. dumosa, H. salsola, and L. tridentata seeds in burned and unburned habitat across multiple seed trials (Fig. 2). The

percentage of seeds removed by ants was statistically significant only for L. tridentata and only once throughout the study (Fig. 2). Thus, rodents were the main seed removers of A. dumosa, H. salsola, and L. tridentata seeds. Patterns of seed removal for Baileya multiradiata, Coleogyne ramosissima, Encelia farinosa, Eriogonum fasciculatum, Penstemon bicolor, and Sphaeralcea ambigua were more variable through time and space, ranging from 0 to 100% seed removal (Fig. 2). Both rodents and ants removed appreciable amounts of seeds of these species (Fig. 2). For example, rodents removed 100% of C. ramosissima seed in four (out of 11 seed trials) seed removal trials while ants removed between 50 and 90% of B. multiradiata, E. farinosa, E. fasciculatum, P. bicolor, and S. ambigua seeds at least once (out of 12 seed trials) during the study (Fig. 2). Seed removal patterns for the latter species were more structured and clearer as a function of burned and unburned habitat and the time (i.e., seasonality) seed trials were conducted. Seed removal by ants was sharply lower in burned compared to unburned habitat for C. ramosissima, E. farinosa, E. fasciculatum, and S. ambigua seeds (Fig. 2). Ants removed between 50 and 90% of offered seeds in unburned habitat, and the zenith of seed removal took place during the month of July (Fig. 2). The proportion of seeds removed consistently decreased through the months of August and September and was zero in October 2009 (Fig. 2). In burned habitat, seed removal by ants peaked in May when ants removed 30e40% of B. multiradiata, E. farinosa, E. fasciculatum, P. bicolor, and S. ambigua seeds (seed removal for C. ramosissima was about 10% in May), and seed removal never reached the high proportions observed in unburned habitat (Fig. 2). Patterns of seed removal by rodents were also influenced by burned and unburned habitat and the month seed trials were conducted. Unlike ants, seed removal was less consistent across species and habitat, but several patterns were evident. In burned habitat, rodents removed seeds of A. dumosa, C. ramosissima, H. salsola, and L. tridentata, while in unburned habitat rodents removed seeds of B. multiradiata, C. ramosissima, E. farinosa, E. fasciculatum, L. tridentata, P. bicolor, and S. ambigua (Fig. 2). Overall, proportions of seed removed in burned and unburned habitats peaked from 20 to 40% for most species, with C. ramosissima being the exception (Fig. 2). Rodents removed between 90 and 100% of C. ramosissima seed in burned habitat, and 40e50% in unburned habitat. In burned habitat, seed removal for C. ramosissima peaked in July, and remained high (100% seed removal) during the months of August, September, and October (Fig. 2). In unburned habitat, high percentages (w50%) of seed removal occurred in March and April (Fig. 2). Furthermore, removal by rodents for C. ramosissima and L. tridentata seeds shifted from a winter (November) or spring (March and April) peak in seed removal in unburned habitat to a summer (July) peak in burned habitat (Fig. 2). In October 2009, rodents substantially removed seeds of A. dumosa, H. salsola, and L. tridentata, but not in October 2008.

Table 2 Analysis of variance (ANOVA) results for the effects of burned or unburned habitat, date, and seed consumer on seed removal of nine Mojave Desert perennial plant species. Numerator (Num df) and denominator degrees of freedom (Den df) are shown as: df for species other than CORA (df for CORA). F values are shown for each term in the model; bold denotes p  0.05. Species abbreviations follow Table 1. Effect

Num df

Den df

AMDU

BAMU

CORA

ENFA

ERFA

HYSA

LATR

PEBI

SPAM

Habitat (H) Date (D) Treatment (T) HD HT DT HDT

1 11 2 11 2 22 22

10 110 20 110 20 220 220

0.15 9.31 6.81 2.36 4.29 1.19 1.00

1.48 11.46 7.14 1.75 0.09 1.99 1.74

2.97 4.62 53.61 5.95 24.48 3.53 7.21

8.67 9.86 6.96 2.36 0.98 3.81 2.53

3.89 19.59 9.17 1.33 3.27 2.75 1.60

0.01 15.10 4.32 1.33 0.49 0.86 1.08

1.34 6.14 14.28 1.35 0.16 1.54 1.83

5.78 14.50 10.03 3.12 0.36 2.46 2.01

0.63 19.83 10.90 0.85 0.71 4.01 1.53

(10) (10) (20) (20)

(100) (100) (200) (200)

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

169

1.0 Ambrosia dumosa

Baileya multiradiata Burned-ant access Burned-no access Burned-rodent access Unburned-ant access Unburned-no access Unburned-rodent access

0.8 0.6 0.4

*

*

*

0.2

*

*

*

0.0 1.0

Coleogyne ramosissima

*

0.8

*

*

*

Encelia farinosa

*

* *

0.6

*

*

*

0.4

*

*

*

*

**

*

Seed removed (proportion)

0.2 0.0 Hymenoclea salsola

Eriogonum fasciculatum

*

0.8

*

0.6

*

0.4

*

* *

0.2 0.0

Penstemon bicolor

Larrea tridentata 0.8

*

0.6 0.4

*

*

*

*

*

0.2

*

* * *

*

0.0

Oct 09

Sep 09

Aug 09

Jul 09

Jun 09

May 09

Apr 09

0.4

Mar 09

*

Feb 09

* 0.6

Jan 09

0.8

Nov 08

Oct 08

*

Sphaeralcea ambigua

*

* *

0.2

0.0

Oct 09

Sep 09

Aug 09

Jul 09

Jun 09

May 09

Apr 09

Mar 09

Feb 09

Jan 09

Nov 08

Oct 08

Fig. 2. Proportion of seeds removed for nine Mojave Desert perennial plant species in burned and unburned habitat in 2008 and 2009. ‘*’ denotes a significant pairwise comparison with at least one other data point within species.

3.2. Seed mass Proportion of seed removed was related to seed mass in 6 out of 12 sampling dates, and the relationships also depended on burned and unburned habitats and seed consumer (Fig. 3).

Overall, ants removed lighter seeds (Table 1) of B. multiradiata, E. farinosa, Eriogonum fasiculatum, P. bicolor, and S. ambigua more than heavier seeds of A. dumosa, C. ramosissima, H. salsola, and L. tridentata (Fig. 3a and d). Rodents removed heavy C. ramosissima seeds more than lighter seeds of A. dumosa, H. salsola, L. tridentata,

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

log10 (proportion seeds removed + 1)

0.06

a

Unburned-ant Slope = -0.031 (-0.058, -0.005)

Oct 2008 PEBI

0.05

0.04 ERFA

0.03

ENFA

0.02

BAMU

SPAM HYSA

0.01

LATR AMDU

0.00 -0.4

0.06

log10 (proportion seeds removed + 1)

170

b

Nov 2008

0.05

0.04

0.03

0.02 ENFA SPAM PEBI BAMU ERFA

0.01

0.0

0.2

0.4

0.6

0.8

0.0

0.2

0.4

log10 seed mass (mg) 0.18

c

Apr 2009 0.16 Unburned-rodent Slope = 0.070 (0.020, 0.120) 0.14

0.10 0.08 ENFA

0.04

LATR HYSA AMDU

BAMU ERFA PEBISPAM

0.02

d

CORA

0.12

0.06

0.00

0.16

May 2009

CORA

1.0

1.2

1.4

Burned-ant Slope = -0.050 (-0.077, -0.022)

SPAM ENFA PEBI

0.14

BAMU

0.12 ERFA

HYSA

0.10

LATR AMDU CORA

0.08

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0

0.2

log10 seed mass (mg)

e

0.35 Sept 2009

0.30 Burned-rodent Slope = 0.125 (0.016, 0.234)

CORA

0.25 0.20 0.15 0.10 LATR

0.05

PEBIERFA BAMUENFA SPAM

0.00 0.0

0.2

0.4

0.4

0.6

0.8

1.0

1.2

1.4

log10 seed mass (mg) log10 (proportion seeds removed + 1)

log10 (proportion seeds removed + 1)

0.8

0.06 0.0

0.35

0.6

log10 seed mass (mg)

log10 (proprtion seeds removed)

log10 (proportion seeds removed + 1)

0.18

LATR HYSA AMDU

0.00 -0.2

Burned-control Slope = -0.009 (-0.016,-0.002)

HYSA AMDU

f

Oct 2009 0.30 Burned-rodent Slope = 0.116 (0.025, 0.208)

CORA

0.25 0.20 0.15

HYSA ENFA LATR

0.10 PEBIERFA BAMU SPAM

0.05

AMDU

0.00 0.6

0.8

1.0

1.2

1.4

log10 seed mass (mg)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

log10 seed mass (mg)

Fig. 3. Relationships between proportion of seeds removed and seed mass (mg). Sample size was 9 for all monthly trials except October 2008 (n ¼ 8). Slopes for the regression lines for each granivore treatment were statistically different from 0, and permuted 95% confidence intervals are reported. Species abbreviations follow Table 1.

and all lighter seeds weighting 1.0 mg (Fig. 3c, e, and f). The slopes of these relationships were still significant when the heavy seeded C. ramosissima was excluded from the regression analysis (data not shown).

3.3. Seedling recruitment Seedling emergence was observed for only one species, C. ramosissima, during the study. The proportion of seedlings

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

emerging for this species in burned (46 of 480 sown seeds) and unburned (44 of 480 sown seeds) habitats was not statistically different (F1, 10 ¼ 0.53, p ¼ 0.48). Main effects of ant access (F1, 10 ¼ 0.02, p ¼ 0.89) and ant and rodent access (F1, 10 ¼ 0.05, p ¼ 0.82) to seeds did not affect seedling emergence, and interactions between ant access and burned and unburned habitat (F1, 10 ¼ 0.44, p ¼ 0.52) or ant and rodent access and burned and unburned habitat (F1, 10 ¼ 0.02, p ¼ 0.89) did not influence seedling emergence. However, rodent access to seeds had a significant effect on seedling emergence (F1, 10 ¼ 9.76, p ¼ 0.01). The interaction between burned and unburned habitat and rodent access to seeds was marginal (F1, 10 ¼ 3.64, p ¼ 0.08), and analysis of mean differences indicated that rodents significantly reduced seedling emergence in unburned habitat (p ¼ 0.03, Fig. 4). Seedling survival was affected by burned and unburned habitat and seed consumers. Survival curves showed that ant access to seeds decreased survival synergistically when rodents also had access to seeds in burned and unburned habitat. Rodent access alone to seeds reduced survival in burned habitat (Fig. 5). At 17 months (August 2010) after seedling emergence, total seedling survival in burned habitat was 21% compared to 7% in unburned habitat.

1.0

0.4

Survival (proportion)

0.2

1.0

Fig. 5. Coleogyne ramosissima seedling survival in a) burned and b) unburned shrublands in the Mojave Desert, USA.

80% of offered seeds during 2 out of 11 seed trials (Fig. 2). However, ants also removed high proportions of light-weight seed species in multiple seed trials, indicating that ants can exploit a diverse range of seed masses. The association between seeds removed by ants and seed size becomes clearer when the proportion of seed removed was modeled as a function of seed mass. Our linear regression analysis indicates that ants selectively remove seeds of

Seedling emergence (proportion)

Seedling emergence (proportion)

Aug 10

Jul 10

Jun 10

May 10

Apr 10

Mar 10

Feb 10

Jan 10

Dec 09

Nov 09

Oct 09

Sep 09

Ants allowed

Jul 09

Ants excluded

Jun 09

0.02

May 09

0.04

Apr 09

a

0.06

0.0

Mar 09

a a

0.08

Unburned habitat

0.6

Granivory is common in desert ecosystems, and many studies have illustrated that seed consumers can be major sources of seed loss (Gordon, 1980; Hulme, 1998; Soholt, 1973). Seed consumers can therefore influence establishment of seed introduced by broadcast seeding during ecological restoration efforts. Yet, potential effects of consumers often are not considered when planning restoration projects. Our study documents considerable differences in proportion of seeds removed by granivorous rodents and ants in burned and unburned habitats, and the observed seed removal patterns were linked to seed mass and varied in time and space. Thus, seed consumers can be important sources of seed loss. Seed predation risk is related to seed size, with heavy seeds being preferred by seed consumers because these seeds provide the greatest amount of energy (Kelrick et al., 1986; Schoener, 1971). Ant seed removal patterns indicate that ants were attracted to heavy seeds of C. ramosissima, thus following predictions of optimal foraging theory (Schoener, 1971). Ants removed between 70 and

a

b

0.8

0.2

0.10

Ant access Ant and rodent access No access Rodent access

0.0

4.1. Seed removal

0.12

Burned habitat

0.6

0.4

Burned Unburned

a

0.8

4. Discussion

0.14

171

0.18 Burned Unburned

0.16 0.14 b

0.12 0.10

ab

ab

0.08 0.06

a

0.04 0.02

Rodents excluded

Rodents allowed

Fig. 4. Mean (SE) proportion of Coleogyne ramosissima seeds emerging in cages that allowed or denied access to seed-eating ants (left panel) or rodents (right panel) in burned and unburned Mojave Desert habitat. Letters denote significant differences within habitats.

172

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

light-weight species (i.e., seeds weighting 1.0 mg). This indicates that ants preferred to forage for small over large seeds. The preference for small seeds could be explained by morphological limitations of an ant’s mandible to carry heavy seeds (Davidson, 1977; Pirk and Lopez de Casenave, 2010). Thus it appears that the ant communities at our study sites could be mainly composed of smallbody ant workers. The size of an ant worker is highly correlated with size of preferred forage items (Davidson, 1977; Pirk and Lopez de Casenave, 2010). Consequently the size of a seed harvested by ants could be partly determined by the body size distribution of workers in the ant community. Smaller workers likely prefer to carry small seeds and large workers likely prefer to carry large seeds back to the nest for consumption. Seed mixes composed of a wide variety of seed sizes may provide some relief from ant seed predation. In this study, we used different-sized seeds from nine species actively used in restoration of burned desert habitat (Table 1). Proportions of seed removed by ants for A. dumosa and H. salsola (seed mass: 4.7 mg and 3.9 mg, respectively) were never significant during 12 months of seed trials (Fig. 2), suggesting that these two species mostly escaped ant seed predation. Additionally, escaping ant seed predation could provide a window of opportunity for these species to incorporate into the soil seed bank. However, other seed traits such as seed coat thickness and morphological characteristics (e.g., spines) require consideration when developing seed mixes. Heavy seeds are often removed by rodents more than light seeds, highlighting the complexity and importance of seed size in desert rodent seed selection (Brown and Heske, 1990). The association between seed size and rodent seed selection is exemplified by the patterns and amounts of seeds removed for C. ramosissima. Rodents removed 100% of the offered seed during 4 out of 12 seed trials (Fig. 2), suggesting a strong preference for this large-seeded species (seed mass 15.8 mg). We found a positive relationship between proportion of seed removed and seed mass (Fig. 3), supporting the idea that rodents preferred removing C. ramosissima seeds over light-weight seeds. Other studies have reported similar conclusions. Long-term exclusion of rodents substantially increases the density of heavy seeds stored in soil seed banks (Brown and Heske, 1990; Brown et al., 1979). Apparently, heavy seeds are an attractive resource for rodents, likely because of higher energetic content of heavy compared to light seeds. Preference for heavy seeds has been correlated to greater carbohydrate content (Kelrick et al., 1986) and removing heavy seeds could indicate preference for the most nutritious seeds. This concurs with optimal foraging theory (Schoener, 1971), and studies in other arid systems (e.g., Gordon, 1980). Granivore preference for large-seeded species may have implications for plant establishment. High granivore pressure on preferred species can drastically reduce the number of plants that are able to establish and may move the plant community toward dominance by small-seeded species (Brown, 1986; Brown and Heske, 1990). In our seedling establishment study, C. ramosissima emergence was influenced by rodent seed predation, as preventing rodent access to seeds significantly increased seedling recruitment in unburned habitat. This finding supports the idea that plant community establishment is influenced by seed consumers (Bricker et al., 2010; Brown, 1986; Orrock et al., 2009). Moreover, seedling survival probabilities within the first 2 months after emergence were low in cages where both rodents and ants could access seeds, suggesting that seed consumers influenced seedling establishment not only through granivore pressure, but also possibly through herbivory. Effects of herbivory on seedling establishment have been documented elsewhere for this long-lived desert shrub (Meyer and Pendleton, 2005). Establishment of late-successional shrubs such as C. ramosissima is important because these shrubs create fertile soils

(i.e., islands of fertility) that improve formation of seed banks and establishment of other plant species (Thompson et al., 2005). Thus, selective removal of C. ramosissima seeds by granivores might influence restoration efforts by limiting ecosystem processes that affect long-term plant population dynamics. Our results suggest that granivore pressure on selected seeds might be reduced if a highly nutritious seed (i.e., decoy seed) species was added in seed mixes to free up target species from consumer pressure (Longland and Bateman, 1998). This restoration strategy has been implemented in semi-arid habitats in the Great Basin, USA and has increased establishment of a native grass whose seeds are primarily dispersed by granivorous rodents (Longland and Ostoja, in press). 4.2. Burned and unburned habitats Changes in vegetation structure caused by fire can also influence a seed consumer’s ability to harvest seeds. The two granivorous taxa showed strong differences in seed removal patterns, suggesting that habitat characteristics may have had differential effects on seed consumers. Based on high proportions of C. ramosissima seeds removed by rodents in burned habitat, it appears that rodent activity was high, suggesting that rodents can potentially consume large amounts of seeds and maintain stable populations in burned habitats. Our small mammal trapping data indicate that Merriam’s kangaroo rat (D. merriami) was a dominant species in burned habitat, corresponding with other longer term studies in post-fire habitats in the Mojave Desert (Horn et al., 2012; Vamstad and Rotenberry, 2010). Heteromyid rodent abundance, particularly Merriam’s kangaroo rat, is maintained or enhanced following wildfire (Horn et al., 2012; Vamstad and Rotenberry, 2010), and this high abundance may result from improved foraging efficiency in a structurally simplified habitat (e.g., more expansive and numerous open spaces, Fig. 1) (Horn et al., 2012; Vamstad and Rotenberry, 2010). Similar patterns of post-fire seed predation by rodents have been documented in other ecosystems (Broncano et al., 2008; Denham, 2008; Zwolak et al., 2010). By three years after our study fire, it appears that seed predation by rodents can be a main source of seed loss in burned habitats, and also on older burns because abundance of seed-eating rodents appears to increase after fire (Horn et al., 2012; Vamstad and Rotenberry, 2010). Consequently, reseeding efforts may need to be conducted shortly after fire when rodent abundance may be low due to direct and indirect fire effects (Esque et al., 2003). On the other hand, low proportions of seed removed by ants in burned habitat suggest that fire may have negatively affected their foraging behavior. More importantly, it appears that seeds may be escaping predation by ants. This is particularly evident for large seeds of C. ramosissima. Ants removed between zero and 10% of offered seeds throughout the study. The low number of seeds removed is surprising because ant seed removal has increased after fire in Mediterranean sclerophyll vegetation (Andersen, 1988), and fire appears not to affect ant abundance through direct mortality (Zimmer and Parmenter, 1998). However, effects of wildfire on harvester ants have not been evaluated in the Mojave Desert. Fire size and intensity are likely to influence ant community composition, nest density, and competition for resources (Andersen, 1988). Seed removal was more frequent for the smaller seed species e E. farinosa, E. fasciculatum, P. bicolor, and S. ambigua. Therefore, the observed seed removal patterns for these species suggest that offered seeds were near ant nests (Crist and MacMahon, 1992), or that ants did not have to travel far to remove seeds, thus preferentially removing small seeds. At greater distances from the nests, energy-rich food becomes more valuable and ants would be expected to remove large seeds (Crist and MacMahon, 1992; Schoener, 1971). Alternatively, small ants may be more abundant in

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

burned habitat and physically incapable of carrying large seeds (Davidson, 1977; Pirk and Lopez de Casenave, 2010). However, longer-term studies are needed to disentangle such fire-related effects. 4.3. Seasonality Air and soil temperatures are important abiotic factors influencing granivore foraging activity. Decreased foraging activity is associated with declining soil surface temperatures (Pol et al., 2011). Ants removed many seeds in unburned habitat during summer (i.e., JulyeAugust) while seed removal decreased considerably during winter (i.e., DecembereFebruary). A parallel result was observed in the pattern of seed removal by rodents. For example, rodents removed a high proportion of C. ramosissima seeds in unburned habitat during spring (i.e., March) and in burned habitat in summer (i.e., June). Overall, seed removal decreased considerably during winter (i.e., DecembereFebruary) in both burned and unburned habitats. This suggests that a shift over the cold/warm months influenced foraging behavior of both rodents and ants. In an applied restoration context, timing seeding to occur in early November may provide some relief from rodent and ant granivore pressure. Additionally, seeds could benefit from winter precipitation. High snow accumulation in December likely influenced C. ramosissima germination in this study, as seeds require cold stratification for germination (Meyer and Pendleton, 2005). Overall, our results suggest that utility of an optimal decoy seed species could depend on seed size, season or timing of seeding, and spatial and temporal distribution of seed consumers. Our study provides results on seed removal and plant establishment from an individual fire thus limiting the generality of the observed plant and seed removal patterns. Further work is necessary to evaluate generality of patterns identified by this study, including in deserts such the Mojave where fires historically are infrequent but are now a novel disturbance regime. 4.4. Secondary seed dispersal The slow rate of post-fire plant regeneration in desert habitats is likely influenced by lack of seed input from near seed sources due to limited dispersal ability of most desert plants (Venable et al., 2008). Our results suggest that intensive seed removal may further slow establishment of native vegetation after fire, if the seeds were consumed or lost (e.g., buried too deeply for seedlings to emerge) after removal. Consequently, results highlight the potential role that secondary seed dispersal (i.e., seed fate) may play in plant establishment in disturbed habitats, as granivorous rodents cache a large quantity of seed in or near their burrows (Vander Wall et al., 2005). For instance, seed caching by heteromyid rodents benefits establishment of a native grass (Longland et al., 2001) and rodent-mediated seed dispersal has been suggested for a native desert tree, Yucca brevofolia (Vander Wall et al., 2006). On the other hand, seed harvester ants may destroy gathered seeds because these ants are true granivores, not seed dispersers (Tschinkel, 1999). However, seeds rejected by ants could germinate from refuse piles (Rissing, 1986). Therefore, secondary seed dispersal, including subsequent seed bank formation (DeFalco et al., 2012), warrants further study in the context of revegetating disturbed habitats through seeding techniques. 4.5. Implications for ecological restoration Seed removal by desert seed consumers represents an important barrier to soil seed bank restoration in degraded habitats, and

173

protecting seeds from rodents and ants can greatly increase seedling establishment (DeFalco et al., 2012; Meyer and Pendleton, 2005; Orrock et al., 2009). Large-seeded species in our study experienced intense granivore pressure; therefore, selecting species for seeding based on seed characteristics that are not preferred by seed consumers may increase on-site seed retention. In addition, use of a ‘decoy’ seed with high preference value to seed consumers might alleviate granivore pressure on target species (Longland and Ostoja, in press), and future investigations should evaluate its potential effectiveness in an applied desert restoration setting. Seasonal foraging activities of seed consumers also warrant consideration when planning reseeding projects. Seeding in the Mojave Desert should be conducted in late October or early November to minimize seed predation. Our experimental manipulations have highlighted factors that influence seed loss in reseeding projects; thus experimental restoration can provide insights useful for developing strategies for ecologically and economically effective restoration of disturbed desert habitats. Acknowledgments We thank Toshi Yoshida for helping install granivory exclusion cages; Josh Hoines and the Southern Nevada Interagency Restoration Team and Christina Lund and the Bureau of Land Management for providing seeds; Tereza Jezkova for sharing the small mammal data; and John Fauth, Cayenne Engel, and Eugene Schupp for providing helpful comments on the manuscript. Comments made by two anonymous reviewers greatly improved the clarity of the final manuscript. This project was funded by the Joint Fire Science Program (grant 07-1-3-24) through a cooperative agreement between the National Park Service (Lake Mead National Recreation Area) and the University of Nevada Las Vegas. References Abella, S.R., Gunn, J.L., Daniels, M.L., Springer, J.D., Nyoka, S.E., 2009. Using a diverse seed mix to establish native plants on a Sonoran Desert burn. Native Plants Journal 10, 21e31. Andersen, A.N., 1988. Immediate and long-term effects of fire on seed predation by ants in sclerophyllous vegetation in south-eastern Australia. Australian Journal of Ecology 13, 285e293. Bainbridge, D.A., 2007. A Guide for Desert and Dryland Restoration. Island Press, Washington DC. Bekker, R.M., Bakker, J.P., Grandin, U., Kalamees, R., Milber, P., Poschlod, P., Thompson, K., Willems, J.H., 1998. Seed size, shape and vertical distribution in the soil: indicators of seed longevity. Functional Ecology 12, 834e842. Bricker, M., Pearson, D., Maron, J., 2010. Small-mammal seed predation limits the recruitment and abundance of two grassland forbs. Ecology 91, 85e92. Broncano, M.J., Rodrigo, A., Retana, J., 2008. Post-fire seed predation in Pinus halepensis and consequences on seedling establishment after fire. International Journal of Wildland Fire 17, 407e414. Brooks, M.L., Chambers, J.C., 2011. Resistance to invasion and resilience to fire in desert shrublands of North America. Rangeland Ecology and Management 64, 431e438. Brown, J.H., Heske, E.J., 1990. Control of a desert-grassland transition by a keystone rodent guild. Science 250, 1705e1707. Brown, J.H., Reichman, O.J., Davidson, D.W., 1979. Granivory in desert ecosystems. Annual Review of Ecology and Systematics 10, 201e227. Brown, J.H., 1986. The roles of vertebrates in desert ecosystems. In: Whitford, W.G. (Ed.), Pattern and Process in Desert Ecosystems. University of New Mexico Press, Albuquerque, New Mexico, pp. 51e71. Chambers, J.C., MacMahon, J.A., 1994. A day in the life of a seed: movements and fates of seeds and their implications for natural and managed systems. Annual Review of Ecology and Systematics 25, 263e292. Crist, T.O., MacMahon, J.A., 1992. Harvester ant foraging and shrub-steppe seeds: interactions of seed resources and seed use. Ecology 73, 1768e1779. Davidson, D.W., 1977. Species diversity and community organization in desert seedeating ants. Ecology 58, 711e724. DeFalco, L.A., Esque, T.C., Nicklas, M.B., Kane, J.M., 2012. Supplementing seed banks to rehabilitate disturbed Mojave Desert shrublands: where do all the seeds go? Restoration Ecology 20, 85e94. Denham, A.J., 2008. Seed predation limits post-fire recruitment in the waratah (Telopea speciosissima). Plant Ecology 199, 9e19.

174

A.A. Suazo et al. / Journal of Arid Environments 88 (2013) 165e174

Duncan, R.S., Wenny, D.G., Spritzer, M.D., Whelan, C.J., 2002. Does human scent bias seed removal studies? Ecology 83, 2630e2636. Esque, T.C., Schwalbe, C.R., DeFalco, L.A., Duncan, R.B., Hughes, T.J., 2003. Effects of desert wildfires on desert tortoise (Gopherus agassizii) and other small vertebrates. The Southwestern Naturalist 48, 103e111. Esque, T.C., Young, J.A., Tracy, C.R., 2010. Short-term effects of experimental fires on a Mojave Desert seed bank. Journal of Arid Environments 74, 1302e1308. Gordon, S.A., 1980. Analysis of twelve Sonoran Desert seed species preferred by the desert harvester ant. Madroño 27, 68e78. Horn, K.J., McMillan, B.R., St.Clair, S.B., 2012. Expansive fire in Mojave Desert shrubland reduces abundance and species diversity of small mammals. Journal of Arid Environments 77, 54e58. Hulme, P.E., 1998. Post-dispersal seed predation: consequences for plant demography and evolution. Perspectives in Plant Ecology, Evolution and Systematics 1, 32e46. James, J.J., Svejcar, T.J., Rinella, M.J., 2011. Demographic processes limiting seedling recruitment in arid grassland restoration. Journal of Applied Ecology 48, 961e969. Kelrick, M.I., MacMahon, J.A., Parmenter, R.R., Sisson, D.V., 1986. Native seed preferences of shrub-steppe rodents, birds and ants: the relationships of seed attributes and seed use. Oecologia 68, 327e337. Legendre, P., Legendre, L., 1998. Numerical Ecology. Elsevier, Amsterdam. Longland, W.S., Bateman, S.L., 1998. Implications of desert rodent seed preferences for range remediation. Journal of Range Management 51, 679e684. Longland, W.S., Ostoja, S.M., Ecosystem services from keystone species: diversionary seeding and seed-caching desert rodents can enhance Indian ricegrass seedling establishment. Restoration Ecology, in press. Longland, W.S., Jenkins, S.H., Vander Wall, S.B., Veech, J.A., Pyare, S., 2001. Seedling recruitment in Oryzopsis hymenoides: are desert granivores mutualists or predators? Ecology 82, 3131e3148. Meyer, S.E., Pendleton, B.K., 2005. Factors affecting seed germination and seedling establishment of a long-lived desert shrub (Coleogyne ramosissima: Rosaceae). Plant Ecology 178, 171e187. Montalvo, A.M., McMillan, P.A., Allen, E.B., 2002. The relative importance of seeding method, soil ripping, and soil variables on seeding success. Restoration Ecology 10, 52e67. O’Dowd, D.J., Gill, A.M., 1984. Predator satiation and site alteration following fire: mass reproduction of alpine ash (Eucalyptus delegatensis) in southeastern Australia. Ecology 65, 1052e1066. Orrock, J.L., Witter, M.S., Reichman, O.J., 2009. Native consumers and seed limitation constrain the restoration of a native perennial grass in exotic habitats. Restoration Ecology 17, 148e157. Parr, C.L., Andersen, A.N., Chastagnol, C., Duffaud, C., 2007. Savanna fires increase rates and distances of seed dispersal by ants. Oecologia 151, 33e41. Pirk, G.I., Lopez de Casenave, J., 2010. Influence of seed size on feeding preference and diet composition of three sympatric harvester ants in the central Monte Desert, Argentina. Ecological Restoration 25, 439e445.

Pol, R.G., Casenave, J.L., Pirk, G.I., 2011. Influence of temporal fluctuations in seed abundance on the foraging behavior of harvester ants (Pogonomyrmex spp.) in the central Monte desert, Argentina. Austral Ecology 36, 320e328. Price, M.V., Jenkins, S.H., 1986. Rodents as seed consumers and dispersers. In: Murray, D.R. (Ed.), Seed Dispersal. Academy Press, Sydney, Australia, pp. 191e235. Quinn, G.P., Keough, M.J., 2002. Experimental Design and Data Analysis for Biologists. Cambridge University Press, New York. Rissing, S.W., 1986. Indirect effects of granivory by harvester ants: plant species composition and reproductive increase near ant nests. Oecologia 68, 231e234. Rundel, P.W., Gibson, A.C., 1996. Ecological Communities and Processes in a Mojave Desert Ecosystem: Rock Valley, Nevada. Cambridge University Press, New York. Schoener, T.W., 1971. Theory of feeding strategies. Annual Review of Ecology, Evolution, and Systematics 2, 369e404. Sivy, K.J., Ostoja, S.M., Schupp, E.W., Durham, S., 2011. Effects of rodent species, seed species, and predator cues on seed fate. Acta Oecologica 37, 321e328. Soholt, L.F., 1973. Consumption of primary production by a population of kangaroo rats (Dipodomys merriami) in the Mojave Desert. Ecological Monographs 43, 357e376. Thompson, D.B., Walker, L.R., Landau, F.H., Stark, L.R., 2005. The influence of elevation, shrub species, and biological soil crust on fertile islands in the Mojave Desert, USA. Journal of Arid Environments 61, 609e629. Tschinkel, W.R., 1999. Sociometry and sociogenesis of colonies of the harvester ant, Pogonomyrmex badius: distribution of workers, brood and seeds within the nest in relation to colony size and season. Ecological Entomology 24, 222e237. Vamstad, M.S., Rotenberry, J.T., 2010. Effects of fire on vegetation and small mammal communities in a Mojave Desert Joshua tree woodland. Journal of Arid Environments 74, 1309e1318. Vander Wall, S.B., 1998. Foraging success of granivorous rodents: effects of variation in seed and soil water on olfaction. Ecology 79, 233e241. Vander Wall, S.B., Kuhn, K.M., Beck, M.J., 2005. Seed removal, seed predation, and secondary dispersal. Ecology 86, 801e806. Vander Wall, S.B., Esque, T., Haines, D., Garnett, M., Waitman, B.A., 2006. Joshua tree (Yucca brevifolia) seeds are dispersed by seed-caching rodents. Ecoscience 13, 539e543. Venable, D.L., Flores-Martinez, A., Muller-Landau, H.C., Barron-Gafford, G., Becerra, J.X., 2008. Seed dispersal of desert annuals. Ecology 89, 2218e2227. Zimmer, K., Parmenter, R.R., 1998. Harvester ants and fire in a desert grassland: ecological responses of Pogonomyrmex rugosus (Hymenoptera: Formicidae) to experimental wildfires in central New Mexico. Environmental Entomology 27, 282e287. Zwolak, R., Pearson, D.E., Ortega, Y.K., Crone, E.E., 2010. Fire and mice: seed predation moderates fire’s influence on conifer recruitment. Ecology 91, 1124e1131.