Extended seed rain period of Adenostoma fasciculatum impacts diverse seed predators

Aims The principal chaparral species in California, Adenostoma fasciculatum, an evergreen, sclerophyllous shrub, is broadly distributed and provides habitat and food resources for a large and diverse animal community. The effects of climate change, including elevated temperatures, fire frequency and severity, along with increased urban encroachment, have placed pressure on chaparral habitats in California. Our goal is to investigate aspects of reproductive ecology as a measure of the potential resiliency of A. fasciculatum. We focus on seed rain (all seed falling into the seed traps regardless of origin) and seed banks in the context of plant-animal interactions and regeneration. Methods Stand recovery following disturbance is achieved through both resprouting and germination from established persistent soil seed banks. In this study we focus on seed ecology using a series of experiments to document the length and quantity of seed rain, seed predation, parsing the importance of the community of granivores, and evaluating the connection between stand age and germination rate from soil seed banks. Important findings Our research documented an 8-month seed rain duration with over 1 million seeds per m2, multiple seed predators including passerines (songbirds) and rodents, and points to the possibility of native ants playing a role in the seed dispersal process. This is important given the recent advancement of the invasive Argentine ant (Linepthema humile) into Californian chaparral. This research demonstrates a clear relationship between A. fasciculatum and both resident and migratory granivores in the chaparral. We documented that a 39-year-old stand had higher germination rates than those which were 16, 20, 41 and 71 years old and how seed banks play a major role in assuring resiliency following fire. These findings are important for wildland managers to assure the continued resiliency of A. fasciculatum.

Introduction For plants to persist in particular habitats, they must be able to survive the spatial and temporal variation in habitat stresses. Reducing risk at different stages of plant life histories is a critical product of natural selection interacting with the environment of the plant [1]. California chaparral is a semi-arid vegetation that provides multiple biotic and abiotic constraints, dominant plants of which may exhibit variation among life history stages. Covering a broad geographic expanse and climates within California (approximately 23% of land area), chaparral includes nearly a quarter of the plant diversity of the California Floristic Province [2,3]. The Mediterranean-climate of California chaparral hosts multiple stresses for plants, from the characteristic summer rainless periods and subsequent wildfires to variation in geology and animal communities of mutualists and antagonists. Multiple life history combinations of plants in this vegetation illustrate multiple solutions to these stresses. Among the woody dominants are three ways plants cope with drought and fire, from deep-rooted, obligate sprouters; relatively deep-rooted in the healthcare sector as a consultant. My work as a healthcare consultant through my private consulting business, Garaventa Consulting, provided support in the form of salaries for the author [JG], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. All research conducted was selffunded. No funds or support from any outside entity or academic scholarship was received. As my major professor, graduate advisor and second author, Dr. V. Thomas Parker has provided input, guidance, and contributions to this manuscript. The specific roles of these authors are articulated in the 'author contributions' section." The name of my private business is and has been Garaventa Consulting. The name of this business was amended to "Garaventa Consulting and Research" for the purposes of this submission. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Competing interests: Joanna M. Garaventa, the primary researcher in this matter, has a commercial affiliation with Garaventa Consulting and Research. Garaventa Consulting and Research did not play a role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript and only provided financial support in the form of the authors' salary and/or research materials. V. Thomas Parker has no affiliation with this organization. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
How Adenostoma disperses is not yet clear. These dry achenes exhibit no adaptations for wind dispersal [4]. Currently no published quantitative data exists on the duration of seed rain in A. fasciculatum stands as well as the timing of dispersal, nor how these patterns may relate to potential chaparral granivores, their behaviors, and phenology. The relationship between some chaparral plants and granivores through secondary dispersal (diplochory) has been well documented and previous chaparral studies have recorded seed preferences of rodents and their role as secondary dispersers through scatter hoarding and seed caches [21]. Several studies have documented ant-mediated seed dispersal in grasslands and other Mediterranean habitats [22][23][24][25] through caching of seeds within or near nests. This appears to be particularly significant in semi-arid habitats [4]. Whether ants play a role in A. fasciculatum seed dispersal is unknown. The behavior of many ant species to clear areas surrounding their nests may lead to soil seed bank deposition. Depending on the depth of deposition, such caches will contribute to germination.
Adenostoma fasciculatum stands are generally long-lived; documented stand age based on fire intervals has exceeded 80 years [26][27][28][29]. As a stand ages, the overall biomass of the stand eventually decreases over time and seed deposition rates decline [16]. Seed banks of A. fasciculatum are maintained with a significant seed rain; one study determined seed bank densities ranging from 2000 to 21,000 seeds per m 2 depending on the age of the stand [8]. Its' seeds can germinate lower in the soil profile than those of desert or grassland seeds-more than 5 cm depth-thus ensuring that some seeds remain protected from high soil temperatures generated during fires [4,8,30]. Consequently, post-fire seedling densities can be considerable in the first year [14,31,32]. Post-fire germination and seedling densities seem to fall into two patterns. In some cases, older stands have higher germination rates than younger ones simply due to a larger soil seed bank [8,[33][34][35][36]. In contrast, the most productive age for seed germination may lie somewhere between 30 and 40 years of age, declining thereafter [8]; for example, germination rates of younger and older stands show far lower germination rates than mid-aged stands (~30-40 yrs.) [8,34,35]. Mid-range maximum germination rates occur in other plant species (Genista monspessulana) [36]. If mid-aged stands have higher overall germination rates, it may be due to their continued ability to deposit large amounts of seeds to the seed bank despite granivory and loss of viability in older seed cohorts. Because the soil seedbank is continually fluctuating with new seeds being deposited balanced against seed consumption and viability loss in older seeds, declining seed production with age would result in lower postfire germination rates in older stands. soil seed bank [7,10].
Adenostoma fasciculatum has a complex life history that undoubtedly underlies its dominance in chaparral throughout the California Floristic Province. Changing climate in California indicates increased temperatures and a longer rainless period, potentially modifying chaparral fire regimes by increasing fire frequency and severity [37,38]. Anthropomorphic manipulation of chaparral has led to attempts for restoration at some sites [33,39]. While life history data on Adenostoma is sufficient to provide some estimate of these impacts on recovery after fire, much less is known about its animal interactions and their resiliency to change, especially in the context of evolving fire regimes. Our overall objectives are to test some of the past patterns seen with respect to seed production and seed banks, while expanding on other less studied dimensions. Theoretical and empirical studies already have indicated that there can be various trait dimensions to persistence of a plant population in a habitat [40][41][42][43]. That has led to studies investigating the extent to which population persistence may reflect trade-offs between, for example, dispersal and seed dormancy, depending on the predictability of successful seed germination and seedling establishment [44][45][46]. These trade-offs have been called bet hedging, and while bet hedging at the seed bank stage is well acknowledged [35,47,48], other life history stages have also exhibited diverse responses to environmental stresses [49][50][51]. Recognition of their significance is reflected in a series of recent attempts to set up frameworks for dispersal, seed banks, and seed dormancy [52][53][54][55], for example, describe aspects of seed-trait functional ecology which incorporates categories of dispersal, persistence, germination timing and establishment. Patterns found in Adenostoma may expand on the predictions of these frameworks due to its multiple potential responses. Consequently, our specific study aims are fourfold: to clarify seed production and dispersal patterns by determining the rate, length, and peak seed deposition period of a mid-aged A. fasciculatum stand; to determine whether a correlation exits between stand age and seed bank seed germination across multiple stands; to document the community of animals which feed on A. fasciculatum seed using motion-activated wildlife cameras; and to determine whether there is a differential preference among insects, birds or rodents in seed consumption. Field site access for all sites was secured prior to sampling being conducted by permits or permission issued by the following staff and entities: East Bay Regional Park District (EBRPD), Del

Seed rain
Annual seed rain was monitored on a bi-weekly basis within a single site at the Mt. Diablo location on a south-facing slope at 829 m elevation from August 2016 to March 2017 (Fig 1). The annual rainfall during this time was 675 mm with average annual rainfall being 515 mm per year. Twelve individual seed rain stations were selected along an approximately 250 m transect paralleling the slope using a nested sample design. Individual station locations were determined using a random number table. Each collecting station contained two 25 x 25 cm trays (0.0625 m 2 ): one was covered with hardware cloth (0.69 cm openings) and the other uncovered. Each tray was secured to the ground using metal U-shaped landscape staples to prevent movement and predation of seeds within the covered trays.
Seeds from each of the twelve covered trays were collected, bagged, labeled, and brought back to the lab on a bi-weekly basis. Seeds within the uncovered trays remained in place until the end of the study in March 2017 at which time they were bagged, labeled, and transported back to the lab for counting. Seed rain samples were sieved to eliminate detritus and then counted. Only whole seeds and those which were less than 50% consumed were included in the count.

Seed predation by granivores
Motion-activated wildlife cameras were placed at six of the twelve seed rain collection stations within the Mount Diablo study site. Data from motion-activated wildlife cameras were collected from SIM (Subscriber Identity Module) cards and analyzed on a bi-weekly basis. Video data was reviewed to determine frequency of visits by type of species.

Differential predation testing
Four additional seed station locations were established within the same A. fasciculatum stand at the Mount Diablo site at the culmination of the seed rain study. Each seed station location contained three 0.0160 m 2 (20 x 8 cm) area containers providing varying degrees of access for granivores to seed: one uncovered station to allow full access to seeds by either rodents or birds; the second which was partially covered by hardware cloth screen (0.69 x 0.69 cm openings) elevated approximately 2.5-3.0 cm above the station to exclude birds but permit access by rodents and insects; and, the third station completely covered by the hardware cloth screen to permit access to only insects.
Motion-activated wildlife cameras were set-up at each of the four sites and a total of 100 seeds were added to each container. Seed stations were checked each week for two weeks to assure seed offering was still adequate. Video data from SIM cards were collected from cameras each week.

Seed bank germination rate by stand age
Twelve soil seed bank samples were collected from the five separate sites for estimating seed bank sizes. Each soil seed bank sample was collected in two parts: the upper 2 cm and the lower 3 cm collected separately. The age of each stand was determined using documentation supplied by park management and CalFire data except for the Del Valle site. The age of this stand was determined by cutting 3 primary branch sections from individual random obligate seeding shrubs within the stand, then cutting a 2.0 cm thick section from each, counting growth rings under a dissection microscope, and averaging the results.
Seedbank samples were collected at random points along a transect bisecting each site using a random number table. A 0.0625 m 2 area quadrat frame (25 x 25 cm) was utilized for collection of the upper 5 cm of soil. Soil seed bank samples were collected in a stratified manner with the upper 2 cm of soil being collected and bagged separately from the lower 2-5 cm of soil. Samples were labeled and transported back to the lab following collection. A total of 120 separate samples were collected.
Preparation of each sample included: air drying in an open container and sieving through a 0.69 cm mesh to eliminate large rocks, twigs, and other non-soil debris. Samples were dried in a 71˚C oven for 30 minutes to mimic a fire heat pulse, cooled, and transported to the greenhouse, and spread on 2 cm of clean sand in individual 25 x 25 cm trays. Samples were watered weekly with a dilute mixture of pyroligneous acid (~1:1000) and gibberellic acid (0.01 ppm) to facilitate germination and monitored to assure adequate moisture. After several weeks, samples were transported from the greenhouse to a netted outdoor area to provide natural ambient temperatures.
Potential differences among sites and between layers was tested by ANOVA. Linear and polynomial regression models were used to fit to the seed bank data.

Analyses used in study
Seed rain. Cumulative seed rain count over test period. Determination of peak seed rain period. Determination of maximum bi-weekly seed rain. Wilcoxon Sign-Rank test and paired t-test of data to determine significance between open vs. closed trays.
Seed predation by granivores. Analysis of motion-activated wildlife camera data to determine timing and frequency of visits to seed traps by granivores.
Compilation of a list of granivores visiting traps. Differential predation testing. Analysis of differential between initial seed offering and seeds remaining at the end of the test period.
Analysis of motion-activated wildlife camera data to determine types of granivores visiting traps.
Seed bank germination rate by stand. Best-fit analysis using a linear-regression model.

Seed rain
Seed rain occurred continuously from August 2016 through March 2017 (Fig 1A). Peak seed rain occurred from the end of September to mid-October 2016 with a second smaller peak noted in early December 2016. During the peak period, mean maximum bi-weekly seed rain was 44,288 seeds m -2 . During the secondary peak period mean maximum bi-weekly seed rain was 28,416 seeds m -2 . Cumulative seed rain over the duration of the study (August 2016 to March 2017) was calculated to be more than 1 million seeds m -2 (Fig 1B).

Seed predation
Seed predation in open trays was substantially higher than screened (closed) trays, with open trays having approximately 75% of their seeds consumed using closed trays as the baseline. A Wilcoxon Sign-rank test of open vs. closed mean seed tray counts yielded a P-value of 0.0004883 and a paired t-test p = 3.413e-06. Both tests showed a statistically significant difference between the two treatments of open vs. closed seed traps (Fig 2A and 2B). Granivore species consisted primarily of rodents and passerines. Peromyscus maniculatus (Deer Mouse) and Neotoma cinerea (Bushy Tailed Woodrat) were the primary rodent granivores and Catharus guttatus (Hermit Thrush) and Junco hyemalus (Dark-eyed Junco) the primary passerine seed predators. Rodent predation seemed to follow the initial fall peak seed rain while passerine predation was highest during the winter and early spring months. Migratory species of the bird species Zonotrichia atricapilla, (Golden-Crowned Sparrow) and Spinus psaltria, (Western Goldfinch) were also found to browse on A. fasciculatum seed during the fall and winter months. This is of interest as it suggests that migratory bird species may utilize A. fasciulatum seed as a food source during their over wintering in California. Resident birds were also observed eating at the seed trays: Melazone crisalis (California Towhee), Sitta carolinensus (White Breasted Nuthatch), and Pecile rufensens (Chestnut-backed Chickadee) and the majority of observed predation was by resident passerines (Table 1). Insect consumption of seeds was documented during two time periods: a) August to September 5-9% consumption; and b) October to November when 2-4% of all seeds showed signs of insect predation. Insect predation was not documented during the remainder of the study period. Insects found in seed rain samples included ants Prenolepis imparis (winter ant) and Veromessor sp. (Smooth harvester ant) and Dirt-colored seed bugs (Rhyparochromidae) (Fig 3A).

Differential seed predation
In the differential seed predation experiment, the largest number of seeds consumed across all treatments during the three-week test period occurred in open trays with 57% of seeds removed (341 of the 600 seeds). Elevated screens and closed (screened) trays had approximately the same amount of total seed predation during the same period (105 and 93 seed lost,  Fig 2B). Motion-activated wildlife camera observations were made of rodents in both the open and partially covered trays and ants were observed only in the covered trays.

Seed bank germination
Germination rate from soil seed bank collections by stand age was tested across five sites ranging in age from 16 to 71 years of age. Stand age was defined as time since last fire and germination rates by seedling emergence. Maximum germination rates occurred in soil from the 39-year-old stand, with the next highest germination rate by the 71-year-old stand. Germination rates in the lower 3 cm of soil were consistently higher in all five locations (F 9,110 = 3.286, p = 0.0014). Seed density in the upper 2 cm was about half (52.5%) that of the lower 3 cm, thus representing only 34.4% of the seed bank for the top 5 cm of soil depth. While total germination visually supports the mid-stand age having peak densities, no linear or non-linear regression model was a statistically significant fit, possibly due to the small number of stands and their variance; the best fit was a log-log linear model although it was not significant (

Seed rain
Adenostoma fasciculatum has been noted to have a summer seed dispersal period [2]. We noted a much more extensive seed dispersal period, in the range of 8-months, which in our study began in August and ended in March. To our knowledge, the lengthy seed rain of approximately 8 months duration was not previously documented for this species and is atypical of other chaparral shrubs which tend to have short bursts of seed rain which last only a few to multiple weeks [4]. Most taxa found with delayed dispersal like this are from desert plant communities [55][56][57][58][59], salt marshes [56], or Mediterranean-region shrub communities [60]. The lengthier seed rain suggests a risk-reducing response to extensive seed predation [50], providing sufficient seed for granivores, but extending dispersal past the activity periods of different types of seed predators, potentially allowing for some seed incorporation into the soil seed bank. Thus A. fasciculatum retains seed on its panicles as temporary aerial seed banks, incrementally depositing seeds from summer through early spring thus increasing the likelihood that seeds may bypass granivores and be preserved in the soil seed bank (Fig 3). The long dispersal phase is paired with high seed production. While others have documented relatively high seed production values in A. fasciculatum [8], we know of no studies with numbers as high as we found. The large number of seeds dispersed (in our sample year over 1 million m -2 ) attracted a broad community of granivores, similar to other arid and semi- arid communities [58,61,62]. Variation in seed bank density and depth distribution patterns overall suggests the interpretation that granivory has a major impact on soil seed banks in this species [6,62]. Finally, the type of granivore that dominated seed removal shifted during the long seed rain period of the stands.

Seed predation
Some seeds within the seed rain samples were observed to have potential pre-dispersal seed predation. Seeds collected during the fall period were eaten by insects as they contained insect bore holes and frass. The duration of this phenomenon was short (11 weeks) and did not reoccur in later samples. While it is not known for certain which insects caused this damage, several members of the Rhyparochromidae family (dirt colored seed bugs) were found in the seed samples. These species of insects are known seed eaters and their presence in the seed traps suggests they were among those that utilize A. fasciculatum as a food source. The animal community which utilizes A. fasciculatum seed following initial dispersal as a food source is quite diverse and includes several rodent, passerine, and insect species. The granivorous rodent and bird species are common in the region and known to be efficient seed predators in chaparral [6,7]. In arid areas of North America, these three groups commonly occur together as seed predators on the same species, generally with rodents having a greater impact, and ants and birds varying in impacts by species [17,18,61,[63][64][65]. The temporal patterns of foraging among these animals may have selected for the lengthy seed dispersal period.
Although not mentioned in previous research, ants were frequently found in seed trays and in bi-weekly samples. The area surrounding the site is filled with ant hills of Veromessor sp. (Smooth Harvester Ant) and several specimens of Prenolepsis imparis (False Honey Pot Ant) were found in seed tray samples. Red harvester ants (Pogonomyrmex barbatus) were frequently seen in stands of A. fasciculatum used in this study that were relatively open. Possibly these ant species are collecting and storing A. fasciculatum seeds in their burrows as a winter food source since ants prefer small seeds which are easy to carry [23]. Ant caching of seeds has been shown to be beneficial to some plants as seeds are shallowly buried thus making them more likely to benefit from phytochrome stimulation, water imbibition and ultimately germinate [25,60]. In the case of A. fasciculatum, ant dispersal could result in seeds being deposited into soil gaps below 5 cm which could result in germination following a fire. This process has been shown to be of significant benefit in grassland ecosystems as well as other Mediterranean habitats [22,66]. Further research which documents the potential role of ants as an A. fasciculatum seed disperser would be beneficial. Ants have been documented to play a dispersal role in only two other chaparral endemics, Dendromecon rigida (Bush Poppy) and Fremontodendron decumbens (Pine Hill Flannelbush) [64,65,67]. Ants may play a similar role regarding A. fasciculatum.
The role of ants and seed bugs as seed predators and possible seed dispersal agents was more extensive than expected. Ants as a possible secondary disperser has not been explored for this species and is of interest given the recent advancement of invasive species such as the Argentine ant (Linepithema humile) into Californian chaparral and the potential for displacement of native ant species [68,69]. Linepithema humile has been documented to have immigrated into California as far north as the San Francisco Bay Area. It displaces native ant species by occupying their nests and creating "super-colonies" which extend over a broad geographic area [68]. Being non-granivorous by nature, the influx of this species could significantly affect the role which native ants play in their relationship with A. fasciculatum through secondary dispersal and micro-caching. Without seed removal and caching by native ants, there could be increased opportunity for predation by other granivores thus decreasing the number of seeds that are annually deposited in the soil seed bank.
Temporal patterns between seed dispersal and granivore-types indicate a sequential shift of dominance among types of granivores. Analysis of seed predation by species type (Fig 4) portrays a temporal relationship between seed rain and three groups of animal seed predators: insects, rodents, and passerines. Insect predation, while limited to a 3-month period, coincides with peak seed rain deposition [67]. Rodent predation closely follows peak seed rain deposition periods. Passerine predation in A. fasciculatum stands seems to be unrelated to seed rain peaks or duration as it occurs primarily during winter months and perhaps is associated with migratory patterns or the lack of food availability elsewhere.
The high seed production and the lengthy dispersal period suggests strong selection by granivory [68]. Correlation between substantial seed production years and subsequent reproductive output of rodents and birds has been validated by previous studies that demonstrate a positive long-term effect on reproductive success of small mammals and birds [68,69]. Passerine predation may be connected to the need for increased food sources as nesting season approaches in early spring. In addition, migratory birds may utilize A. fasciculatum seeds as a food source during their overwintering in California or during their migratory passage south. Notably, A. fasciculatum stands generally follow the north-south passerine migratory trajectory along the California coast.

Seed bank germination
The seed bank germination study corroborates previously held hypotheses that seed bank density is related to stand age as the middle-aged stand (39 years old) had significantly higher rates than either younger or older stands [4,8,29,62]. The best fit model, however, was a linear model suggesting that seed banks may continue to increase in size through time and we just had too few sites to distinguish among hypotheses.
Of the 5 cm of soil collected for seed bank study, soil depth differed in seed density with the lower 3 cm of soil having significantly higher germination rates than the upper 2 cm in contrast to many species [69]. The dynamics of high predation rates at the surface and shallow depths also combined with non-refractory seeds not persisting, may explain the differential pattern of slow accumulation of seeds at depth. Few seedlings germinated within several of the five seedbank sites (16, 30 and 41 years since last fire). This may indicate younger stands have not had sufficient time to develop a substantial seed bank in the context of lower seed production and high seed predation rates [29,35]; while older stands may exhibit declining rates of seed production and potentially loss of viability in older seed cohorts [29]. Soil seed density may be affected by multiple factors including the resident animal community, the type of soil, amount of annual precipitation, pathogen attack, and whether the seed contains enough reserves to maintain the seed embryo. Given only five sites were tested in this study it is difficult to extrapolate more generally.
Adenostoma fasciculatum is the dominant species throughout most of the range of chaparral but climatic shifts place this species particularly at risk [38]. Chamise stands provide habitat and food for a wide variety of resident and migratory animals. The prodigious production of seeds combined with the extended seed rain period undoubtedly have contributed to its broad distribution range throughout California chaparral. This extended seed rain period accommodates a diversity of granivores including insects, but more work is required to understand the full breadth of plant and animal mutualisms and interactions within this community. The documentation of significant plant-insect interactions suggests future work could also focus on pre-dispersal seed predation and documentation of insects which may utilize A. fasciculatum flowers and dispersed seeds.