SEED PREDATION BY OWYHEE HARVESTER ANTS AND THE POTENTIAL OF SEED INTRODUCTIONS IN RECOVERY EFFORTS FOR SLICKSPOT PEPPERGRASS

..................................................................................................................... vii LIST OF TABLES ............................................................................................................. xi LIST OF FIGURES .......................................................................................................... xii CHAPTER ONE: CONSEQUENCES OF OWYHEE HARVESTER ANT (POGONOMYRMEX SALINUS) FORAGING BEHAVIOR ON SEED RECRUITMENT IN SLICKSPOT PEPPERGRASS (LEPIDIUM PAPILLIFERUM) ...................................1 Abstract ....................................................................................................................1...................................................................................................................

xiii The thick horizontal line within the box represents the median. While the boxplot reflects the distribution of the data prior to log transformation, the letters above the bars indicate significant differences among the groups based on a means comparison of log-transformed values. The effect of barrier treatment was significant (p <0.0001) where the exposure of L. papilliferum to foraging ants had a significant effect on the remaining number of seeds indicating that P. salinus is an influential seed predator of the rare plants.   (Fletcher et al. 2001a,b, Kettenring et al. 2009, Vergeer and Kunin 2011, Martin and Meinke 2012, Sharp Bowman et al. 2017. Measuring the extent of herbivory, and its fitness consequences to individual plants is a critical first step in the assessment of how herbivory impacts plant populations. These impacts may vary widely in response to the unique physical, physiological, and/or behavioral characteristics associated with each herbivore and plant species, as well as the environmental conditions present in a particular ecosystem. In some instances, herbivory appears to enhance plant fitness by stimulating growth and reproduction (Inouye 1982, Abhilasha andJoshi 2009), inducing defense mechanisms (McArt et al. 2013) and increasing the rates of seed dispersal (Wilson et al. 2012). In most instances, however, herbivory has unfavorable effects on growth, survival, dispersal, and fitness of plants (Bruelheide and Scheidel 1999, Rand 2002, Poveda et al. 2003, Barber et al. 2011. In addition to the costs generally associated with herbivory, (i.e., loss of photosynthetic area), herbivory on seeds (commonly referred to as seed predation or granivory) can adversely affect individual plants and their populations because seed predation imposes an immediate cost on offspring recruitment (Crawley 1989, Weppler and Stocklin 2006, Kolb et al 2007, Burgos et al 2008.
The consequences of herbivory may be particularly severe for rare plant species, where any adverse effects of herbivory on survival and recruitment could limit or prevent population maintenance or recovery (Kettenring et al. 2009, Ancheta and Heard 2011, Martin and Meinke 2012, Leonard and Auken 2013. Herbivory that results in damage or loss of seeds is a major contributing factor to low recruitment of individuals into rare plant populations (Crawley 2000, Méndez et al. 2004, Albert et al. 2005, Raju et al. 2009). Thus, rare species, which by definition generally experience limited abundance, limited habitat availability, and restricted geographic ranges (Rabinowitz 1981), are at a greater risk of population decline and extinction in the face of herbivory than are more common species (Gaston 1994, Johnson 1998, Crawley 2000, Matthies et al. 2004, Ancheta and Heard 2011 (Brassicaceae), a rare and threatened plant endemic to southwestern Idaho.
Harvester ants in the genus Pogonomyrmex are voracious seed predators that have the capacity to remove large numbers of seeds from their environment (Tschinkel 1999, MacMahon et al. 2000. While some Pogonomyrmex species forage individually with little or no coordination among nestmates (Gordon 1991, Gordon 1995, MacMahon et al. 2000, others, including P. salinus, forage collectively along trunk trails that lead to food patches (Janzen 1971, Hölldobler and Wilson 1990, Taber 1998, Hölldobler et al. 2001).
The success of collective foraging relies upon the ability of workers to recruit nestmates to profitable food patches. Using pheromones laid along trunk trails by returning foragers, as well as chemical cues produced by seeds in the vicinity of nests, harvester ants actively recruit nestmates to increase the number of foragers exploiting patches of food (Brown et al. 1979, Hölldobler 1976, Hölldobler et al. 2001, Johnson 2000, Johnson 2001, Greene et al. 2013. Mobilization of foragers to food patches decrease mean individual search times and increase the rate of harvest as well as the cumulative number of seeds collected from within a patch (Brown et al. 1979, Hölldobler 1976, Hölldobler et al. 2001, Johnson 2000, Johnson 2001, Greene et al. 2013. Although the proportion of seeds removed from patches by harvester ants may vary for a variety of reasons (i.e. seed species, seed abundance, alternative seed availability, nutritional requirements of colony, etc.), the removal of preferentially harvested seed species can be as high as 100%, and may lead to reductions in plant abundance and shifts in plant distributions (Anderson and Ashton 1985, Hobbs 1985, Crist and MacMahon 1992, Ireland and Andrew 1995. Lepidium papilliferum is a rare plant endemic to sagebrush-steppe habitat in southwestern Idaho. Within sagebrush-steppe, the plant is restricted to microsites known as "slick spots" ) -shallow depressions of natric soils characterized by distinct clay layers and surface water retention that is higher than that of surrounding areas . The unique habitat requirements of L. papilliferum, along with the plant's limited range and declining numbers (see , Sullivan and Nations 2009 has raised concern among land managers and conservationists regarding the species' long-term viability. Range-wide declines in L. papilliferum have been largely attributed to the loss of suitable habitat as a result of urbanization, agriculture, livestock grazing, the spread of invasive species, and an increase in wildfire frequency (Moseley 1994, United States Fish andWildlife Service 2016). More recently, seed predation by Owyhee harvester ants has been identified as a potentially significant source of seed loss to L. papilliferum Robertson 2009a, Robertson 2015). Seeds of L. papilliferum are preferentially harvested by P. salinus (Schmasow and Robertson 2016), and in a preliminary analysis of seed loss to harvester ants, White and Robertson (2009a) reported that ants can remove >40% of mature fruit directly from plants and up to 90% of seeds experimentally placed on the ground beneath plants. However, this latter statistic was based on the removal of only a small number of seeds (N=10) placed beneath plants. The extent to which harvester ants collect and consume total seed output of individual plants and within slick spots remains unknown.
For many plant species, the production of large seed crops may ensure the survival of enough seeds to sustain a population despite intense seed predation and/or extreme environmental conditions. While high-density resource patches have the potential to attract a greater number of predators (i.e., a positive numerical response), predation risk to individual seeds in high-density patches may be offset through a dilution effect (Lehtonen andJaatinen 2016, Wenninger et al. 2016), particularly if seed availability increases above a consumption threshold for colonies (Janzen 1971, Kelly 1994) and there is limited capacity for a numerical response. In Owyhee harvester ants, foraging ranges of neighboring colonies do not overlap (Howell and Robertson 2015).
This lack of overlap means that any numerical response to increased food availability will be limited to an individual colony's ability to recruit foragers. It has been estimated that individual colonies of Pogonomyrmex ants collect and consume 50,000 to 81,000 seeds in a given foraging season MacMahon 1992, Pirk andLopez de Casenave 2006), although based on the seed intake rate of P. salinus reported by Schmasow (2015), Robertson and Jeffries (2016) suggest that the total number of seeds collected by colonies may be considerably higher when small seeds, such as those produced by L. papilliferum, are prevalent in diet. However, the density-dependent effects of seed predation on L.
papilliferum individual fitness has not been explored.
I conducted a manipulative field experiment to quantify seed predation by Owyhee harvester ants on individual L. papilliferum and an observational study of ant foraging behavior. Also, I investigated density-dependent effects of seed predation on L.
papilliferum fitness by examining whether the proportion of seeds collected from individual plants was inversely related to the total number of flowering plants within their respective slick spots. I predicted that the proportion of seeds being harvested would decrease as the number of L. papilliferum plants increased within a slickspot, at least after a threshold of available seed numbers was met.

Study Area
The study was conducted from June-August in 2012 at a population of Lepidium  (Blom et al. 1991), although densities below 40 colonies per hectare are more typical (Porter and Jorgensen 1988, Blom et al. 1991. A mature P. salinus colony typically consists of 5,000 to 10,000 workers (MacKay 1981, Johnson 2000 and may survive about 20 years Jorgensen 1988, MacMahon et al. 2000) as long as the founding queen survives and continues to lay eggs (Gordon 1991).
Harvester ants forage diurnally from spring to autumn whenever surface temperatures are sufficiently warm (MacKay 1981, Hobbs 1985, Crist and MacMahon 1991, Taber 1998. Daily foraging activity usually occurs in the morning and late afternoon, with periods of inactivity during the hottest portions of the day (Whitford et al. 1976, Hobbs 1985, Crist and MacMahon 1991. Pogonomyrmex ants are single-load, central place foragers (Brown et al. 1979, Stephens et al. 2007. They forage up to 20 m from their nest, with a majority of foraging occurring within 12 m (MacMahon et al.
2000, Burris 2004, White andRobertson 2009a). When nests are in close proximity to one another (i.e., <20 m), neighboring colonies share non-overlapping boundaries in the areas between their nests (Howell and . Although Pogonomyrmex ants collect seeds from a wide variety of plant species, they tend to specialize on abundant, small-seeded species (Crist and MacMahon 1992, MacMahon et al. 2000, Pirk et al. 2009, Pirk and Lopez de Casenave 2011, Ostoja et al. 2013, Schmasow and Robertson 2016. Many species also incorporate arthropods (living and dead), fungi, feces, and assorted vegetation into their diets (Hölldobler and Wilson 1990, Taber 1998, Belchior et al. 2012. Owyhee harvester ants are known to forage on seeds from a variety of plant species that include, but are not limited to, Poa secunda May through July, with seed drop occurring from mid-June through August. Seed production is positively correlated with plant size. An average-sized biennial produces about 8,000 seeds, whereas annuals typically have seed sets at or below 215 seeds (Schmasow 2015). Seeds that drop to the ground become part of a persistent seed bank that in some years may represent the majority of the population . Current estimates are that L. papilliferum seeds remain viable in the seed bank for about 10 years , although in laboratory studies most seeds remain viable for at least double that time (I.C. Robertson, personal communication).

Seed Predation Experiment
In June 2012, I initiated a study to quantify the amount of L. papilliferum seeds collected by harvester ants from individual plants. I selected 20 slick spots that were occupied by flowering L. papilliferum and located within 8 m of an active P. salinus colony. This distance is well within the 12-m foraging range typical of harvester ants Porter 1982, MacMahon et al. 2000, personal observations). Throughout this study, I considered slick spots as the experimental unit. Aspirated ants were immediately placed in glass vials (one vial per colony) along with any items they were carrying (note: when aspirated, harvester ants steadfastly hold onto food in their mandibles). The samples were returned to the laboratory and viewed under 10x magnification. To confirm the occurrence of seed predation by harvester ants on L.
papilliferum in the experimental slick spots, I recorded the number and identity of seeds present in each sample, as well as the number of A. tridentata leaves, arthropods, and unidentified organic fragments that were present.

Statistical Analyses
All statistical analyses were conducted in R version 3.1.3 (R Development Core Team 2013). The confidence interval was set at 95% and the results were considered statistically significant with a p ≤0.05.
Data from the seed predation experiment were analyzed using a likelihood ratio test with the R packages 'car' and 'lme4'. The data were fit into a general linear mixed model with the "number of seeds in soil" as a function of barrier treatment (i.e., access by ants; ants excluded; seed drift). I considered experimental slick spots which were associated with an individual ant colony as a random effect and plant height and overhead flowering area as fixed effects. A planned comparison of the least square means of the three treatment levels was performed using a Tukey's HSD test in the R package 'lsmeans'. Prior to analysis, the number of seeds counted in the soil samples were logtransformed to achieve a normal distribution of the residuals and homogeneity of variance.
As plant   I found no evidence of clipping on plants within sealed barriers, which confirms that the control barriers were effective at preventing seed predation by harvester ants. The use of barriers to exclude harvester ants from access to L. papilliferum had a statistically significant effect on the number of seeds remaining on the soil surface (Figure 1.7; χ 2 [df 2] = 31.53, p <0.0001). Specifically, significantly more seeds were present in the top layer of soil beneath plants that had ants excluded than those that were exposed to ants (means comparison, p <0.0001) resulting in an average seed removal of 73.2% (N=20, range = 0-97.7%) from plants with raised barriers. Likewise, significantly more seeds were present in the top layer of soil beneath plants in the seed drift treatment (i.e., plants with a raised inner barrier and sealed outer barrier) than those that were exposed to ants (p = 0.0002). By contrast, there was no significant difference in the number of L.

Foraging Observations and Seed Predation
papilliferum seeds present in the top layer of soil beneath plants where ants were excluded from those used in the seed drift experiment (p = 0.47). This latter result indicates that seed drift from beyond the perimeter of barriers does not explain the low number of seeds found on the soil surface when individual L. papilliferum were exposed to ants. Combined with my foraging observations, the seed predation experiment indicates that seed predation by harvester ants, and not seed drift, is responsible for the reductions in seed number beneath plants exposed to ants.
There were no statistically significant differences in the flowering area ( respectively; Wilcoxon rank sum test, the mean ranks of high density and low-density treatments were 6.0 and 12.43 respectively, z = -2.23, p = 0.024).

Discussion
Harvester ants are important seed consumers in the arid and semi-arid grasslands of North America (MacMahon et al. 2000), and at the landscape scale, their foraging activities have the potential to influence plant communities and the population dynamics of the species they contain (Reichman 1979, Inouye et al. 1980, MacMahon et al. 2000. Compared to common plant species, rare species are particularly vulnerable to herbivory and seed predation because of their already low abundance and susceptibility to population decline (Crawley 2000, Ancheta and Heard 2011, Inouye et al. 1980, Combs et al. 2011 nests each year. This number is considerably higher than earlier estimates of annual seed intake by harvester ants. For example, Crist and MacMahon (1992) and Pirk and Lopez de Casenave (2006) estimated that P. occidentalis and P. rostratus collected ~60,000 and ~81,000 seeds, respectively, per season. However, the seeds harvested by ants in these earlier studies were larger than those typically collected by P. salinus. Larger seeds are associated with increased handling time, which decreases the rate of return to the nest (Schmasow 2015). Moreover, because larger seeds are generally higher in energy content than smaller seeds (Kelrick et al. 1986), fewer are required to support a colony. Thus, estimates of annual seed collection in Pogonomyrmex may be influenced by the size of seeds in the diet. Assuming the value Schmasow's (2015) estimate is a reasonable approximation for annual seed collection by P. salinus, slick spots that produce seed numbers in excess of 400,000 will exceed the maximum capacity for intake by ants (note: the actual number is undoubtedly smaller because ants do not consume L. papilliferum seeds exclusively). While this may seem like a large number of seeds, Robertson and Jeffries (2016) estimated that some slick spots at this field site produce in excess of 2 million seeds, and in a few cases up to 8 million. Thus, conditions exist where L.
papilliferum seed production greatly exceeds the capacity for ants to collect and consume them.
Lepidium papilliferum numbers fluctuate widely year to year, largely in response to the amount and timing of precipitation during the previous winter . for L. papilliferum are analogous in their effect to mast seeding, the synchronous production of seeds by a population of perennial plants. Mast seeding is viewed as a reproductive strategy to reduce the impact of seed predation by overwhelming the predator's ability to harvest seed due to a satiation threshold (Janzen 1971, Kelly 1994).
While L. papilliferum does not exhibit seed masting per se, stochastic events that produce favorable conditions for growth can result in greater numbers of flowering plants with a higher than average seed production . These favorable reproductive seasons may result in a predator satiation effect allowing a higher proportion of seeds to survive and replenish the seed bank.
While the data I collected during this study indicate that seed losses were proportionally lower when there was a higher number of plants in a slick spot, several improvements to the experimental design should be considered in future studies. For example, while the number of flowering plants present in a given slick spot allows for a rough approximation of seed availability, the number of seeds produced by individual plants varies considerably as a function of the plant's size (Schmasow 2015). Thus, measures that estimate seed production within a slick spot more directly are needed.
Likewise, the technique I used to quantify seed numbers in soil samples has a seed recovery rate that is lower than expected given our understanding of seed production by individual slickspot peppergrass plants (Schmasow 2015). While this technique allowed for a reasonable quantification of seed loss due to predation, new techniques should be explored in the interest of obtaining more accurate estimates.
Combined with the ongoing threats of habitat degradation caused primarily by invasive plants and increased frequency of wildfire , seed predation by harvester ants represents a potentially serious threat to the long-term survival of slickspot peppergrass. Specifically, when L. papilliferum numbers are low, seed predation by harvester ants may be very costly to offspring recruitment. By contrast, when favorable conditions produce large numbers of L. papilliferum, the plant likely is able to replenish its seed bank, at least to some extent. The ability to periodically replenish seed banks likely serves as a buffer to years when seed predation by ants severely limits individual reproductive success. In the long term, survival of L. papilliferum will depend largely on the abiotic conditions that influence reproductive success. Conditions that adversely affect L. papilliferum productivity will make the plants more vulnerable to the effects of seed predation by harvester ants. This possibility is a concern given the ongoing degradation of L.
papilliferum habitat and the potential consequences of climate change on precipitation patterns that influence L. papilliferum populations. Given these expected changes, it is important to continue assessing the role that seed predation by harvester ants has on L.           generally require at least one overwintering period to break dormancy, seeds introduced in the spring were first subjected to a laboratory protocol to break dormancy. A total of nine plants germinated from the 19,800 seeds released in this study (three reproducing annuals, three reproducing biennials, and three rosettes with potential for flowering in the subsequent spring). While the number of germinating plants tallied in this study was very low, this study confirmed that L. papilliferum can successfully germinate, flower, and fruit when seeds are introduced into unoccupied slick spots. Efforts to improve the success of seed introductions are needed before seed introductions can be considered as a viable approach to the conservation and management of L. papilliferum populations.

Introduction
Biodiversity is a driving force of ecosystem function, and it sustains versatility which promotes stability in natural systems (Lefcheck et al. 2015, Weisser et al. 2017. Climate change, land use, and other anthropogenic influences have resulted in degradation, fragmentation, and loss of habitat in many ecosystems, threatening their biodiversity (Brooks et al. 2002, Millennium Ecosystem Assessment 2005, Giam et al. 2010, Krauss et al. 2010, Ashraf et al. 2012, Alofs et al. 2014, Staude et al. 2018). These effects have also been felt throughout sagebrush-steppe habitat in the western United States and Canada, where anthropogenic disturbance (e.g., irrigated agriculture, livestock grazing, off-road vehicles, and military training), altered wildfire cycles, and exotic species invasions are causing many sagebrush-steppe communities to lose native diversity and be replaced by grasslands (Rosentreter 1992, Watts 1998, Hilty et al. 2003, Yeo 2005, Huntly et al. 2011, Mitchell et al. 2017, Seipel et al. 2018). In the process, native forb species have experienced severe declines (Creutzburg et al. 2015, Mitchell et al. 2017).
Macbr.] (Brassicaceae), a rare mustard endemic to southwestern Idaho, is an example of a sagebrush-steppe species that has declined in abundance as a result of habitat degradation and fragmentation . Currently, there are 91 sites known to contain L.
papilliferum ( have yielded unverifiable results because the introductions were made within existing populations, thereby making it impossible to determine whether observed seedlings originated from introduced seeds or the existing seed bank (C.W. Baun, personal communication). The primary goal of the present study was to introduce L. papilliferum seeds in habitat not currently occupied by L. papilliferum to assess whether these introductions have the potential to assist in recovery efforts for the species.
Within sagebrush-steppe habitat, L. papilliferum grows within patchily distributed microsites known as "slick spots" -shallow depressions of soil characterized by higher levels of water accumulation, sodium content, and clay content relative to surrounding soil , Quinney 1998. The plant exhibits two main life history trajectories -annual and biennial. Annuals germinate, flower, set seed and die within a single growing season, whereas biennials exist as vegetative rosettes in the first year and then flower, set seed and die in their second year (Quinney 1998). The plant's small, white flowers, which grow on multi-flowered inflorescences that typically bloom from late May to late June, attract a wide variety of insect pollinators Klemash 2003, Robertson andLeavitt 2011). Mature seeds dehisce from their fruits in late summer, at which point they enter the soil seed bank and remain dormant and viable for up to 11 years , Meyer et al. 2006, perhaps longer (I.C. Robertson, personal communication). Only a subset of seeds in the seed bank germinate in a given year, even when conditions seem ideal .
Slickspot peppergrass has the highest extirpation rate among rare plant taxa in Idaho ) and is currently listed as a threatened species under the Endangered Species Act (United States Fish and Wildlife Service 2016). Habitat degradation and fragmentation as a consequence of urbanization, agriculture, livestock grazing, invasion of exotic plant species, and increased frequency of wildfires, are thought to be major contributors to population declines . Intense seed predation by Owyhee harvester ants, Pogonomyrmex salinus (Olsen) (Hymenoptera: Formicidae), may also contribute to the species' vulnerability and decline (Chapter 1, White and Robertson 2009).
Seed predation by harvester ants represents a potentially serious impediment to the successful use of seed introductions to augment or establish L. papilliferum populations. Harvester ants, which frequently nest within L. papilliferum populations , regularly incorporate L. papilliferum seeds into their diet (Schmasow and Robertson 2016) and have the capacity to remove as much as 90% of the fruits and seeds produced by individual plants (White and Robertson 2009, I.C. Robertson unpublished data). Because seed predation by ants could hamper the establishment of L.
papilliferum populations in otherwise favorable habitat, the timing of seed introductions may be critical to the success of this recovery measure. Seed introductions that coincide with the timing of natural seed deposition for L. papilliferum may be exposed to higher levels of seed predation than seeds released in late fall, once ants have ceased foraging for the year, or seeds released in early spring before ants become active. However, it is unknown whether fall or spring introductions would adversely affect germination success. Therefore, the second goal of my study was to determine whether the timing of seed introduction influences the success of introduction efforts. I predicted that, because of the risk of seed predation by harvester ants, seeds introduced during the fall and spring would have a higher likelihood of surviving to germinate than seeds introduced the previous summer. Note that because L. papilliferum seeds are dormant upon release from the parent plant and require at least one overwintering period to break dormancy, seeds introduced in the spring were first subjected to an established laboratory protocol to break dormancy.

Study Site
The study was conducted from Therefore, for the purposes of this analysis, I focus exclusively on germination success in the field. Although this measure cannot account for seeds that remained dormant in the soil, I assumed that the proportion of seeds that remained dormant was spread evenly across treatments.
I selected a total of 22 slick spots located within 15 m of an active harvester ant colony (Figure 2.2). The experimental design within each slick spot included three seed introduction times (summer, fall, and spring) and three treatments related to the risk of seed predation (protected from ants, access by ants and small mammals, and access by ants only). The small mammal treatment was included to account for the possibility that rodents may also contribute to the predation of L. papilliferum seeds (e.g., Anderson and MacMahon 2001).
The summer introduction time coincided with the timing of natural seed deposition by L. papilliferum, and thus the seeds were exposed to an extended period of foraging by harvester ants. By contrast, the fall and spring introductions were intended to reduce the exposure of seeds to ant predation since harvester ants are usually inactive at these times. Seeds released in the summer and fall received no preparation because I assumed that the natural overwintering period would be sufficient to break their dormancy. Seeds released in the spring were subjected to a protocol to break dormancy.
Following Billinge and Robertson (2008), seeds intended for spring release were stored at room temperature in the dark for three months. I then scarified the seeds by rubbing them gently between two sheets of 320 grit sandpaper, imbibed them with deionized water for 24 h on filter paper in Petri dishes, and placed them in cold stratification at 4 o C for 8 weeks. A portion of the seeds (N = 1,000) was retained in the lab to test germination success. These seeds were distributed evenly across 10 Petri dishes lined with filter paper moistened with deionized water. The dishes were placed in a location with a natural photoperiod and temperatures that ranged from 21-23 o C. After 10 days (i.e., one week after the first seedlings were detected) I counted the total number of seeds that had germinated.
Small cages were used for all seed introduction treatments in the field. Each cage consisted of an 8-10 cm high plastic ring cut from a 15-cm diameter flowerpot, and the top of each cage was covered with 1 cm hardware cloth. The "access by ants and mammals" cages were elevated 4-5 cm from the ground using plastic rebar supports, while the "access by ants" cages were elevated 4-5 cm from the ground using a ring of 1.0 cm hardware cloth that prevented access by mammals. The "protected from ants" cages were sealed directly on the ground to prevent access by seed predators (Figure 2.3).
I scattered 100 seeds on the soil surface within each cage at the appropriate introduction

Results
Of the 1,000 seeds set aside from the scarification protocol to break dormancy, only 45 germinated within their Petri dishes. At the field site, a total of three L.

Discussion
Seed introductions offer a promising approach to the conservation and management of rare plant populations (Dalrymple et al. 2011, Godefroid et al. 2011, Guerrant 2012, Atondo-Bueno et al. 2016, Menges et al. 2016 is dependent for growth and survival , Traversa et al. 2013.
The low numbers of seedlings produced in my study may reflect a loss of seeds from slick spots, either through dispersal, predation, or death. However, seed dispersal and predation are unlikely explanations given the design of the experiment, and there is little reason to suspect seeds died in high numbers. Seed dormancy is a more likely explanation for the lack of germination, as L. papilliferum seeds can remain dormant in the soil seed bank for years Moseley 1998, Meyer et al. 2005). Seed dormancy and persistent seed banks are common in desert annuals (Went 1949) and are generally viewed as a bet-hedging strategy to decrease the risk of reproductive failure in highly variable and unpredictable environments (Philippi 1993, Tielbörger et al. 2012, Volis 2012. Indeed, the sagebrush-steppe ecosystem occupied by L. papilliferum is Given the persistent seed bank of L. papilliferum, and the large fluctuations in plant numbers that can occur across years , the success of seed introduction efforts should be assessed over longer periods than just one or two years. It is also important to consider germination success in a particular year relative to the success of the plant elsewhere. If germination rates are low in years when the plant is thriving range-wide, this may indicate that the introduction is not likely to be successful. Success may also be dependent on the number of seeds released; i.e., more seeds will likely improve the chances of success. A study subsequent to mine substantially increased the number of seeds introduced to unoccupied slick spots (5,000 seeds were released in each of 110 experimental slick spots); however, the success of these efforts has yet to be evaluated (I.C. Robertson, personal communication). Large-scale introductions of L. papilliferum seeds in recovery efforts should only be considered once the parameters for success have been established through carefully monitored experiments. Ideally, this assessment will include analysis of the soil chemistry needed to promote germination and growth within slick spots.
The use of protocols to break seed dormancy may increase the likelihood of seed germination following introduction. However, the scarification protocol I used, while moderately successful in a previous study (Billinge and Robertson 2008), was largely ineffective in my study (see also . Better success at breaking dormancy has been achieved by pricking the seed coat of individual seeds with a pin instead of using sandpaper to scarify a large number of seeds at once (Stillman 2006); however, this technique is labor intensive and would not be feasible on the scale needed for recovery efforts. Moreover, because desert annuals such a slickspot peppergrass rely upon seed banks and seed dormancy as a strategy for surviving unpredictable and ephemeral environmental conditions (Philippi 1993, Tielbörger et al. 2012, Volis 2012, the release of prepared seeds could result in complete failure if conditions for survival are unfavorable. Because harvester ants remove large quantities of L. papilliferum seed from slick spots (Chapter 1, , Schmasow and Robertson 2016, Robertson and Jeffries 2016, the presence of harvester ants is an important consideration in recovery efforts that involve seed introductions. Measures to address this problem include selecting sites without harvester ant colonies, eradicating ant colonies located near slick spots (Robertson et al. 2017, I.C. Robertson, unpublished data), using physical barriers to deny ants access to seeds (Chapter 1, Robertson and Jeffries 2016), and introducing seeds at times when ants are not active, such as late fall (this study, Robertson and Jeffries 2016).
Selecting sites without harvester ants and targeted eradication of ant colonies are the most feasible strategies for large-scale recovery efforts involving seed introductions.
Introduction of seeds late in the fall might preclude the initial loss of seeds to ants, but any seeds produced by plants in those slick spots would be vulnerable to predation.
Successful germination is an incomplete measure of success for any seed introduction study. For introductions to be relevant to conservation efforts, the plants must survive to reproduce. Slickspot peppergrass is reliant primarily on outcrossed pollination mediated by insects for reproduction (Robertson and Klemash 2003, Robertson and Ulappa 2004, Robertson and Leavitt 2011. In the present study, pollinator insects were observed on L. papilliferum flowers, and successful pollination and fruit production occurred in most cases of flowering despite the small number of plants available for insects to visit and cross-pollinate.
In summary, I have shown that the introduction of L. papilliferum seeds to unoccupied slick spots can result in successful germination, growth, and reproduction of the species. However, further research is needed to assess whether these successes can translate into an effective tool for recovery efforts. For seed introductions to be effective in recovery, it will be important to establish the appropriate number and distribution of seeds needed to maximize germination and outcrossing success; and to confirm the availability of pollinators at introduction sites prior to seed release. Research is also needed to determine whether differences in chemical profiles among slick spots influence germination, growth, and reproduction of L. papilliferum. Finally, the disruption caused by cattle trampling in my study highlights the need to select sites with little risk of physical disruption to slick spots, as disruption to the integrity of slick spots can