IMPACTS OF PRECIPITATION VARIABILITY ON PLANT COMMUNITIES By Todd Robinson A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Plant Biology Ecology, Evolutionary Biology & Behavior – Dual Major 2011 1 ABSTRACT IMPACTS OF PRECIPITATION VARIABILITY ON PLANT COMMUNITIES By Todd Robinson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his work would not have been completed without the support of the KBS community as a whole, but in particular staff including Greg Parker, Joe Simmons, and the FSC crew along with Stu Basset and the grounds crew provided important logistical support in the field. I also would not have been able to complete my experiments without the help of Harry Barton, Ana Begej, Nathan Galbreath, Mahala Clayton, Justin Rensch, and Liana Nicols, I appreciate all of the long hours you spent assisting me. Less obvious, but still important contributions came from the computer and administrative support staff, especially John Gorentz, Stefan Ozminski, Sally Shaw, Jenny Smith, and Rene Wilson. While there are many people who discussed science with me over the years, I am especially grateful for the friendship of Sigrid Smith, Sara (Parr) Syswerda, Anne Royer, Brook Wilke, Abbie Schrotenboer, and Mridul Thomas. You helped me refine my ideas and provided companionship during good and bad times. Additional feedback on ideas and thoughtful discussion from other members of the Gross lab helped shape my dissertation, especially frm Wendy Mahaney, Greg Houseman, Chad Brassil, Don Schoolmaster, Tony Golubski, and Alex Eilts. I’d especially like to thank Emily Grman, my lab mate for all seven years for all of the brainstorming, discussion, help, and listening to me grumble that helped me get through my dissertation. I would also like to thank my advisor Kay Gross, who has read more versions of this dissertation than I would have thought possible and helped improve my writing and presenting ability tremendously during my time in graduate school. My other committee member Gary iv Mittelbach, Rich Kobe, and Christopher Klausmeier have also contributed greatly to my development as a scientist. I would also like to acknowledge the funding received that made this work possible, including two LTER student awards, the Don Hall Endowed Scholarship, the G.H. Lauff Research Award, and a Graduate School Professional Development Award, along with fellowship support from the Plant Science program and the College of Natural Science. Finally, my fiancé Ellen Holste has provided me with an enormous amount of support including field assistance, reading over papers, and emotional support when the weather refused to cooperate. I am incredibly grateful to have you in my life and appreciate all that you have done that has helped me complete this dissertation. v TABLE OF CONTENTS LIST OF TABLES vii LIST OF FIGURES viii CHAPTER ONE: INTRODUCTION 1 CHAPTER TWO: ALTERED EMERGENCE OF AGRICULTURAL WEEDS IN RESPONSE TO PRECIPITATION VARIABILITY 8 Abstract 8 Introduction 9 Methods 10 Results 13 Discussion 18 CHAPTER THREE: THE IMPACT OF ALTERED PRECIPITATION VARIABILITY ON ANNUAL WEED SPECIES GERMINATION AND GROWTH Abstract Introduction Methods Results Discussion 21 21 22 23 28 30 CHAPTER FOUR: IMPLICATIONS OF ALTERED PRECIPITATION VARIABILITY ON ANNUAL PLANT PRODUCTIVITY AND STRUCTURE Abstract Introduction Methods Results Discussion 34 34 35 37 45 60 CHAPTER FIVE: PRECIPITATION VARIABILITY AND PLANTS: EMPIRICAL AND MODELING APPROACHES TO UNDERSTANDING THE EFFECTS OF CLIMATE CHANGE ON SPECIES GROWTH AND COMPETITION Abstract Introduction Methods Results Discussion 65 65 66 68 74 79 CHAPTER SIX: CONCLUSIONS 84 REFERENCES 88 vi LIST OF TABLES Table 4.1: Mean and range proportion of total biomass for each of the six species analyzed in the LTER data across the 15 years of our data set. 40 Table 4.2: Differences in AICc (! AICc) between models using different periods of precipitation or the date of disturbance to explain community biomass across the 15 years in our LTER data set. 47 Table 4.3: Differences in AICc (! AICc) between models explaining final biomass or richness using precipitation (P) during the first week of the growing season and temperature during that week. 48 2 Table 4.4: Summary of R and significance for the final models from the model selection using the long-term data. 52 Table 4.5: Seedling emergence as a function of precipitation (P) and warming (W) treatments. 52 Table 4.6: Biomass and richness as a function of precipitation (P) and temperature (W) treatments. 56 Table 4.7: Intra and inter-specific affects of seedling density on final density and biomass. 57 Table 5.1 List of equations used in the mathematical model. 73 Table 5.2 List of model variables and parameters. 73 vii LIST OF FIGURES Figure 1.1 Predicted responses of species and communities to variability in precipitation in a) xeric (water limited) and b) mesic habitats (after Knapp et al 2008). 3 Figure 1.2 Frequency distribution of intervals between rainfall events (precipitation >0.254mm) during the growing season (May-August) at the KBS LTER site. Figure 2.1: Seedling emergence through time in response to watering treatments. 5 15 Figure 2.2: Effect of watering treatments on relative abundance of monocots (proportion of total seedlings that emerged) in each watering treatment. 17 Figure 3.1: Biomass and germination responses of S. faberi and C. album to variation in precipitation amount and interval between watering events. 29( K$8+#)(L;M(<''+".(-"#$"%$&'($'("-)#"8)(%&%".(,&55+'$%0(/$&5"**(H*&.$6(/.",1(.$')I("'6( *4),$)*(#$,7')**(H8#"0(.$')I(%7#&+87(%$5)($'(%7)(EFG(@ABC("''+"..0(%$..)6(5$,#&4.&%*;((((((((LN( K$8+#)(L;O(C)."%$-)("/+'6"',)(&3(%7)(%&4(P(*4),$)*(HMQQR=OSSTI;(((((((((((((((((((((((( ( ((((((((LN K$8+#)(L;R(C)."%$&'*7$4(/)%2))'($'$%$".(4#),$4$%"%$&'(H3$#*%(2))1("3%)#(%$.."8)I("'6(0.254mm) during the growing season (May-August) at the KBS LTER site. Data are combined for a nineyear period (1996-2004). Arrows indicate interval treatments used in the greenhouse and field experiments. 5 In Chapter two I examined the impact of altered precipitation regimes on seedling emergence from a natural seedbank (soil collected from an annually tilled community on the KBS LTER). I monitored the timing and total emergence of seedlings of seedlings and compared the responses of two broad functional groups, monocots and dicots, to treatments that manipulated both total precipitation and intra-annual variability . Emergence increased for both groups with higher variability under dry conditions, but not wetter conditions, and the relative timing of emergence between the groups also changed with precipitation variability. Chapter three summarizes the results of three greenhouse experiments designed to follow up on the results of Chapter two to examine species responses to precipitation variability at emergence and initial growth. As in Chapter two, I varied the frequency and total amount of precipitation. In these experiments I both followed emergence from a natural seed bank (experiment 1) and compared the responses of two common species in this community (Setaria faberi and Chenopodium album) in experiments initiated with field-collected seeds of each species. In these experiments I was able to document differences in species responses to precipitation variability in germination, growth, and survival. Chapter four combines an analysis of a 16 yr data set from the KBS Long Term Ecological Research (LTER) on how climate variability impacted productivity, species richness and abundance in an annual weed community and an experimental field manipulation of early season precipitation and temperature on this community. The historical analysis showed that early precipitation and temperature (one week following tillage) was a stronger predictor of community biomass and species richness than average seasonal precipitation and species differed in their response to these initial climatic conditions. The field experiment generally supported the results of the analysis of the long-term data and revealed some interesting differences in 6 species responses to these variables that suggest that climatic variation may change the effect of competitive interactions. In Chapter five I report the results of a field experiment where I examined how precipitation variability affects the growth of two dominant species in this community (S. faberi and C. album) in monocultures and two species mixtures. These species had shown different responses to precipitation variability in the greenhouse experiments (Chapter 3) and differed in rooting structure/morphology and other traits likely to affect their water use and acquisition. I found that the growth of both species declined with increased variability in monoculture, but only C. album declined when the two species competed against each other. These results motivated the development of a simple mathematical model that considered how a tradeoff in water acquisition traits might impact competition between species for water. The final chapter provides a general summary of the dissertation and discusses ideas for future research motivated by these results. 7 CHAPTER TWO ALTERED EMERGENCE OF AGRICULTURAL WEEDS IN RESPONSE TO PRECIPITATION VARIABILITY Abstract Precipitation variability is a known driver of plant community dynamics in arid and semiarid ecosystems, but a growing body of evidence suggests that other ecosystems are also responsive to precipitation variability. The composition and abundance of annual plant communities, including the weeds that occur in row crops, can be shaped by events that control seedling emergence. Predicted changes in the variability (amount and frequency) of precipitation events early in the growing season are thus likely to be important factors influencing the composition of weed communities in row crops. I explored the impact of withinseason precipitation variability on the emergence of weed seedlings from the soil seed bank in a greenhouse experiment. I varied the interval between precipitation events under average and reduced precipitation levels, mimicking predicted changes in precipitation variation in the Upper Midwestern USA. Soils collected from the Kellogg Biological Station Long Term Experimental Research site were the source of the seed bank for these experiments. The experiment was run for 39 days and I recorded the total number of seedlings, date they emerged, and whether they were monocots or dicots. Longer intervals between watering events increased emergence for both groups at reduced precipitation levels, but had no effect (dicots) or slightly decreased emergence (monocots) under average precipitation levels. The interval between events also affected the timing of germination for monocots and dicots; longer intervals delayed germination for both groups, but dicots germinated later than monocots at short and long intervals. My results add to the growing body of literature suggesting that predicted changes in precipitation 8 variability due to climate change are likely to affect plant communities across a broad range of systems. In particular, my results suggest that monocot weeds may become less abundant during dry years under more variable precipitation regimes. Introduction Climate change models predict widespread changes in the frequency and intensity of precipitation events across ecosystems (Easterling et al., 2000, IPCC 2007, Weltzin et al., 2003), altering water availability to plants. Most of the research on precipitation impacts on plant communities in annuals has focused in arid and semi-arid environments and has shown that variation in precipitation may impact germination, growth, and potentially coexistence of plant species (Clauss and Venable, 2000, Novoplansky and Goldberg, 2001, Pake and Venable, 1995, Schwinning and Sala, 2004, Sher et al., 2004). However, there is growing evidence that plants that occur in mesic environments also respond to precipitation variability (Fay and Schultz, 2009, Knapp et al., 2002, Lundholm and Larson, 2004) and that they may respond differently, even in the opposite direction, compared to species in arid systems (Heisler-White et al., 2009, Knapp et al., 2008). Annual species may be particularly sensitive to variation in precipitation because soil moisture can be an important cue controlling germination and seedling emergence (Baskin and Baskin, 2001, Clauss and Venable, 2000). Several studies have shown that precipitation affects the abundance and composition of the emerging seedling community (Baskin and Baskin, 2001, Lundholm and Larson, 2004). Variation in emergence time can affect seedling growth rate (Ross and Harper, 1972), competitive ability (Rice and Dyers, 2001), and may be an important determinant of population growth rate in annuals (Levine et al., 2008). Predicted changes in 9 precipitation due to climate change may therefore have large effects on annual plant community composition and abundance. In this study, I examined whether precipitation variability, in the form of changing intervals between precipitation events, interacted with total precipitation to control the number and composition of seedlings emerging from the soil seedbank. I used an annual weed community as a model system to explore this question because weed management is an important component of agriculture and changes in the abundance and composition of weeds in row crops likely will have economic consequences (Pimentel et al., 2000). Methods Experimental setup Soil for this experiment was collected from successional plots (~1 ha) established at the Kellogg Biological Station (KBS) Long Term Ecological Research (LTER) site in southwest Michigan (http://lter.kbs.msu.edu/about/experimental_design.php). Within these successional old-fields a set of annually tilled, but uncropped, microplots (20m by 30m) that are dominated by annual weeds, were established in 1989. Weed species occurring in these plots are typical of those in the surrounding row-crops (Davis et al., 2005). Surface soil samples (~5 cm depth) were taken from six replicate plots in late May 2007, less than a week after tillage. Soil from all of the plots was mixed and put through a Lindig Soil Shredder to break up large soil aggregates and remove large stones and clumps of plant material (1.5cm sieve). After processing, the soil was divided among 48 trays (25 by 25 cm, 5 cm depth) that were filled to a depth of 4.25-5 cm and placed on a grated bench in a heated greenhouse. Greenhouse temperatures were set to maintain a maximum of 29.5° C and a minimum of 15.6° C which is within the range temperatures seeds might experience in late May and June, and no supplemental lighting was 10 used. Trays were unwatered until the experiment was initiated (Day 0 of the experiment, June th 15 ) 20 days after collection. On Day 0, all of the trays received 500mL of tap water in the morning and an additional 250mL in the afternoon to simulate wet spring conditions and stimulate germination. Four days later the experimental watering treatments were established; seedlings that emerged following the initial watering and before the first experimental watering were not included in the subsequent analyses as they did not germinate in response to the experimental manipulations. Experimental design The experimental design was a two by three factorial manipulation, with two levels of total water (average and reduced) and three intervals between watering events (4, 8 and 12 days). To control for variation in greenhouse conditions, trays were grouped into eight blocks on the greenhouse bench and all six treatment combinations were represented once in each block. The average water treatment was based on a 10-yr record of growing season precipitation (MayAugust, 1996-2005) from the KBS LTER weather station. I used the median growing season precipitation (314mm) over this period for the average treatment and set the reduced water treatment at 50% of the average amount, in line with more extreme model predictions (Kling et al., 2003). The interval treatments varied the amount of time between watering events while keeping total water constant. Because I was interested in effects on emergence, I used early season (May) data from the KBS LTER to determine typical intervals between rainfall events. From 19962005, the longest dry period in May ranged from 4 to 10 days, with an average of 6 days. Over the entire growing season (May to August), the maximum dry period observed was 15 days during this period. Based on these data, I established dry interval lengths for the experiment of 11 4, 8 and 12 days, corresponding to events at average levels of water of approximately 1.2, 2.4, and 3.6cm. These intervals fall within the historical range of variation in rainfall observed over the entire growing season in this area (SW Michigan), although 12-day intervals occur primarily later in the growing season. I included treatments at the limits of existing variability because climate change may lead to scenarios outside the current natural variation (Nippert et al., 2006). To avoid water ponding on the soil surface or excessive drainage from trays and to more closely simulate the time span over which larger (and longer) rain events occur I split watering applications over the day and only added a maximum of 500mL of water to a tray at any one time. Watering treatments were initiated on Day 4; the last water addition was done on Day 36. rd Seedlings were censused every 4 days starting on the 3 day of the experiment and th continued for 39 days; the final census was taken on July 24 , three days after the final water addition treatment. Any seedlings germinating after this point will likely contribute less to the final community having a much shorter growing season and high temperatures may cause secondary dormancy. At each census I recorded the numbers of monocot and dicots seedlings. Newly emerged seedlings were marked with colored plastic toothpicks to identify census date and to facilitate species identification before being removed by clipping. High post germination mortality early in the experiment, particularly in the 12-day interval treatment, limited my ability to identify seedlings to species. However, all seedlings were identified as monocots or dicots, so analyses of the responses of these two functional groups and total seedling emergence to my precipitation treatments are reported here. Data Analysis I analyzed the data using JMP 5.01. I examined the total number of seedlings, the number of monocots and dicots, and the time until 50% emergence per tray), and the difference 12 between monocot and dicot 50% emergence date. I used total water added, the interval between watering events, and their interaction as fixed effects and block as a random effect. The number of seedlings that emerged prior to the experimental treatments being imposed (between day 0 and 4) was also evaluated as a fixed effect to account for any differences between trays in the initial seedbank. An interval-squared term was used to remove curvature in residuals. Tukey’s HSD was employed to compare treatment means. Results Emergence Of the ~ 7,000 seedlings that emerged during the experiment, over 2,000 emerged in response to the initial watering before the initiation of interval treatments. These initial seedlings were excluded from the final analyses so that the response to interval treatments could be assessed. Analyses that included initial emergence showed qualitatively similar results to the results reported here. Seedling emergence depended on both water amount and interval length. The total number of seedlings that emerged was highest at longer intervals and reduced total water, but interval had no effect on total emergence at average levels of total water (amount*interval F1,37 = 66.08, p<0.0001, Figure 2.1a, b). The emergence of monocots and dicots both depended on the interaction between amount and interval (F1,37 = 27.92, p<0.0001 and F1,37 = 27.6 p<0.0001 respectively, Figure 2.1c-f). At reduced water amounts, monocot emergence increased with longer intervals, but at average water, emergence decreased as interval lengthened (Figure 2.1c,d). Dicot emergence also increased with longer intervals at reduced water (Figure 2.1e), but was unaffected by interval at average levels of total water (Figure 2.1f). The proportion of overall seedlings that were monocots also depended on the interaction of total water and interval 13 Figure 2.1 Seedling emergence through time in response to watering treatments (excluding initial seedlings counted on day 4). Total number of seedlings emerging at (a) reduced and b) average precipitation treatments at different interval lengths; (c, d) number of dicot seedlings and (e, f) number of monocot seedlings emerging in response to these treatments. Values are means with standard errors; error bars that are not apparent are included within the symbol. Arrows indicate the average number of days to 50% emergence for each treatment. 14 (F1,37 = 19.33, p<0.0001, Figure 2.2). Monocots made up less than 50% of all seedlings in all treatments except the 4-day, reduced total water where they made up over 60% of the seedlings that emerged (Figure 2.2). This was driven by the very low germination of dicot seedlings at this treatment (Figure 2.1e) Emergence timing Total water and interval length also interacted to alter the speed of total seedling emergence (amount*interval, F1,37 = 19.3, p<0.0001). Longer intervals increased the number of days until 50% emergence while increased total water delayed the date of 50% emergence at the 8 and 12-day intervals (Figure 2.1). Longer intervals between watering events delayed emergence for both monocots (~8 days, Figure 2.1e-f, F1,36 =352.36, p<0.0001) and dicots (~5.5 days, Figure 2.1c-d, F1,36 = 56.21, p<0.0001). In addition, both groups responded to the square of longer intervals (monocots F1,36 = 6.26, p=0.0171, dicots F1,36 = 9.91, p=0.0033). Increased total water also delayed emergence for monocots (<1 day, Figure 2.1e-f, F1,36 = 6.73, p=0.0136) and dicots (~2 days, Figure 2.1c-d, F1,36 = 13.08, p=0.0009) and there was no interaction with interval length. Overall, dicots emerged more slowly than monocots but this delay depended on interval (F1,36 = 11.09, p=0.002, Figure 2.1c-f) and interval squared (F1,36 = 18.52, p=0.0001). Dicots and monocots reached 50% emergence at about the same time in the 8 and 12-day interval treatments; however dicots emerged more slowly than monocots under the 4-day interval treatment (Figure 2.1c-f). The difference between monocot and dicot emergence was greater at 15 Figure 2.2 Effect of watering treatments on relative abundance of monocots (proportion of total seedlings that emerged) in each watering treatment. Bars are means ± SE; n=8. Light grey bars are reduced total water, dark grey bars are average total water. 16 average versus reduced water but this difference was marginally significant (F1,36 = 3.77, p=0.06, Figure 2.1c-f). Discussion My results show that changes in precipitation amount and the interval between precipitation events are both likely to impact emergence and composition of weed communities that are common in row crops in the Midwestern USA. Importantly, I found that changes in the total amount of water and the interval between watering events interacted to alter the total number of seedlings that emerged from the soil seed bank and the relative abundance of monocot versus dicot weeds that emerged from the seed bank. This supports results from other studies (e.g. Fay and Schultz, 2009, Knapp et al., 2002) that have found that precipitation amount and variability can significantly impact mesic plant community composition. The interactive effect of total water and interval on seedling emergence in this study may be due to the larger pulses of water that were delivered on each date at the longer intervals. Larger pulses of water during a single event may affect germination if species have moisture dependent thresholds where germination is stimulated (or inhibited, Baskin & Baskin, 1998). This may explain the differences in the effect of interval on seedling emergence at the reduced and average water treatments that I observed. The total amount of water that was delivered at all intervals at average water may exceed the minimum level of moisture necessary for germination, whereas at the reduced amounts of water, only the longer intervals delivered enough water at one time to exceed this threshold. Results from greenhouse and laboratory studies on germination responses to desiccation provide some insights into why dicots in my study responded strongly to longer intervals (extended dry periods) at the reduced water treatment. For example, Bouwmeester (1990) found 17 that Chenopodium album, one of the common dicots at my study site (Davis et al., 2005) shows increased germination after a desiccation event. This suggests that germination of C. album might be stimulated following the dry down that occurs when intervals between precipitation events are extended, perhaps through increased sensitivity to germination factors like gibberellic acid (Bouwmeester 1990). The higher total number of dicot seedlings I observed with longer intervals at the reduced water amount is consistent with this observation. Germination of C. album, and potentially other dicots, might be enhanced with increased precipitation variability if this results in longer intervals between large rain events in the spring when these weed species typically emerge in row crops. The timing of seedling emergence can have important effects on their subsequent survival, competitive interactions, and reproduction (Rice and Dyers, 2001, Ross and Harper, 1972, Turkington et al., 2005). While the delay in emergence of both monocot and dicot with longer intervals was not surprising (Figure 2.1), monocots emerged earlier than dicots when the interval between precipitation events was short (every 4 days), at both average and reduced water levels. This might translate into a competitive advantage for monocots when intervals are shorter that could translate into changes in weed species composition in the field. The importance depends on whether the delay is associated with significant growth and reproductive costs for late emerging seedlings or if there are penalties for early emergence such as frost damage or tillage. Germination is a crucial step in the life history of an annual plant, but species can show plasticity in their germination fraction in response to environmental cues (Clauss & Venable, 2000). In this study, I found that in the reduced water treatment, dicot seedling emergence increased dramatically as interval length increased, but mortality in these treatments was quite 18 high. Without data on survival and growth it is impossible to predict the net effect of a particular germination response to precipitation variability. However my results suggest that changes in both the interval and amount of precipitation can significantly affect emergence of both moncot and dicot weed seeds in the Great Lakes region. Monocot and dicot and weeds are known to respond differently to management techniques such as herbicides (Bohan et al., 2005) and so understanding if there are broad differences in the responses of these two functional groups to precipitation variation may be informative for weed management. Further work is needed to determine if germination/emergence responses of weed seedlings to precipitation variation will result in compositional changes in the community or affect other stages of the life cycle (e.g. survival, growth and reproduction). However identifying how changes in precipitation affect weed species emergence is a first step in understanding how climate change might affect annual plant communities in mesic environments. 19 CHAPTER THREE THE IMPACT OF ALTERED PRECIPITATION VARIABILITY ON ANNUAL WEED SPECIES GERMINATION AND GROWTH Abstract Climate change models predict increasing variability in precipitation across the globe, with an increase in the incidence of large precipitation events, but decreasing overall event frequency. Research with annual species in arid and semi-arid ecosystems has demonstrated that precipitation variability can influence plant community dynamics, however less is known about the impact of precipitation variability in less water-limited ecosystems, including economically important agricultural systems. I conducted three greenhouse experiments to determine how variation in total precipitation and the interval between precipitation events affected emergence and growth of two common annual Midwestern weed species, Chenopodium album and Setaria faberi. Both species responded to precipitation variability however, the effect depended on life stage and precipitation amount, indicating that responses are highly context dependent. Emergence of both species increased with longer intervals between precipitation events at low total precipitation, but species responses varied under typical precipitation amounts. Individual seedling biomass of both species depended on interactions between total water and interval, but species responds differed; Setaria faberi biomass at was reduced with longer intervals, but Chenopodium album had either positive or no response. My results suggest that changes in precipitation variability likely will affect the composition and relative abundance of agriculturally important weeds. These results are important for understanding how changes in the temporal variability of precipitation due to global climate change could impact plants in non-arid communities. 20 Introduction Water availability affects plant community productivity (Laurenroth and Sala, 1992, Knapp and Smith 2001, Suttle et al., 2007) and species composition (Silvertown et al., 1999). Climate change models predict increasing variability in precipitation across the globe - both in annual amount and in temporal distribution, with less frequent, larger precipitation events expected to increase (Easterling et al., 2000, Weltzin et al., 2003, IPCC, 2007). Previous research has shown that precipitation variability plays an important role in structuring annual plant communities in arid and semi-arid ecosystems (Pake and Venable 1995, 1996, Clauss and Venable, 2000, Schwinning and Sala, 2004, Sher et al., 2004). However, less is known about how increased variability in precipitation will affect plant communities in less water-limited ecosystems (but see Knapp et al., 2002 and Fay et al., 2008). In addition, species in less waterlimited systems may respond differently to variability in precipitation delivery than arid systems, for example with variability increasing growth in arid ecosystems and decreasing it in less water limited systems. (Knapp et al., 2008, Heisler-White et al., 2009). Understanding how plant species respond to altered precipitation regimes will be important for predicting shifts in plant communities and managing species of conservation interest and pest species. Annual plant communities are likely to be strongly responsive to altered precipitation regimes because species composition and abundance are driven by germination/emergence dynamics that often depend on water availability (Baskin and Baskin, 1998, Lundholm and Larson, 2004). Events early in the growing season can have long-lasting impacts in annual communities (Ross and Harper, 1972, Levine et al., 2008). Variation in water availability throughout the growing season may also directly affect plant growth (Novoplansky and Goldberg, 2001, Sher et al., 2004). Weeds in row crop agriculture provide a widespread and 21 economically important system dominated by annual plants (Davis et al., 2005) to examine the impacts of precipitation variability. In addition, knowledge of how annual weed communities respond to precipitation variability may have important consequences for agricultural management practices. I conducted three greenhouse experiments to address whether altered precipitation regimes would impact annual weed communities. Specifically I asked: 1) Will changes in precipitation patterns – amount and variability - affect the emergence and relative abundance of weed species and 2) Do species differ in their early growth responses to altered precipitation? Although there are a large number of weed species that occur in Midwestern agriculture, I focused on two species, Setaria faberi and Chenopodium album, which are common in row-crop agricultural systems in the Midwestern USA and dominated my study system (see below). Methods Experimental Overview—I conducted three experiments in a temperature-controlled greenhouse with supplemental lighting at the W.K. Kellogg Biological Station (KBS) of Michigan State University between June 2007 and June 2008. All of the experiments used soil and seeds collected from annually-tilled areas of successional plots established at the KBS Long Term Ecological Research (LTER) site in southwest Michigan (http://lter.kbs.msu.edu/about/experimental_design.php). These fields are tilled every spring and are dominated by annual weeds species common in row crop agriculture in the region (Davis et al., 2005). I collected surface soil (~ 0-5 cm) for the experiments from all six replicate plots of this experiment and homogenized the soil by sieving through a 1.5cm sieve to remove stones and rhizomes. During all three experiments I rotated the pots on the greenhouse bench every other day to minimize any location effects. I also mimicked late spring/early summer conditions 22 temperatures in southern Michigan by setting the greenhouse temperature to range from a maximum of 29.4° C to a minimum of 15.5° C. The supplemental lighting used depended on the time of year and is described below for each experiment. All of the experiments used the same 2x3 factorial design, with two levels of total water (average and low) and three watering intervals (4, 8 and 12 days between events). The amount of water added to the average water treatment was based on a 10-yr record of growing season precipitation (May-August, 1996-2005) at the KBS LTER site. I used the median growing season precipitation (314 mm) over this period to calculate an mean daily precipitation amount for the average treatment and set the low water treatment at 50% of the average amount, in line with predictions from the driest model scenario for the region (Kling et al., 2003). The interval treatments varied the amount of time (interval) between watering events while keeping total water constant, leading to water additions of 10.2 (4-day), 20.4 (8-day), and 30.6 (12-day) mm of water in the average water treatment. All pots were watered by hand with de-ionized water, with the volume of water adjusted for surface area across experiments. For large volumes of water, I spread the delivery of the water over several hours to prevent pooling and allow for more natural infiltration. Adding the water over time resulted in virtually no water loss due to drainage from the pots. Because I was interested in effects of precipitation variation on germination, I used early season (May) data from the KBS LTER precipitation records to determine current “typical” intervals between precipitation events. From 1996-2005, the longest dry periods in May ranged from 4 to 10 days, with an average of 6 days. The maximum observed interval between rainfall events during the growing season over this period was 15 days. Based on these data, I established interval lengths for the experiment of 4, 8 and 12 days. These intervals are within the 23 range of variation in rainfall currently observed over the entire growing season for this site, though 12-day intervals occur primarily later in the growing season (July-August). Imposing treatments that are at, or just beyond, the current variability is important because predicted climate change models suggest scenarios outside the current natural variation (Nippert et al., 2006). In the first experiment, I measured the effects of altered precipitation regimes on seedling emergence and early seedling growth for all species that emerged from the soil seedbank. The other two experiments focused on the two most common species that emerged in the first experiment (Setaria faberi and Chenopodium album) and examined species -specific germination (experiment 2) and individual seedling growth (experiment 3) responses to precipitation variation. I analyzed data from all three experiments using a linear model to test for the effects of total water, interval, and their interaction using JMP (Version 5 SAS Institute Inc., Cary, NC, 1989-2007). In the third experiment, I included planting date as a covariate. I conducted the final census or harvest several days after the last watering event to allow treatments to respond to the final water pulse. Each response variable was checked for deviations from normality, and I used logarithmic and square root transformations as necessary to meet homoscedasticity assumptions. Untransformed data are shown in the figures. Experiment 1: Emergence from the seedbank and growth— To determine how altered precipitation patterns affected the number and composition of weeds emerging from the seedbank, I followed seedling emergence and growth for almost two months after the experiment was initiated. The experiment was established in late January by filling twenty-four 16 cm diameter pots with 20 cm of air-dried soil collected the previous October from the LTER field site. The soil was stored in an unheated garage until the start of the experiment after being 24 homogenized. Homogenizing the soil, addressed any variation in soil or seed bank characteristics between replicate sampling areas (Davis et al., 2005). In this experiment, I used supplemental lighting to increase day length to 14.5 hours (7:00am to 9:30pm) to simulate conditions in May, when weed seeds would typically germinate in the field following spring tillage. Watering th th treatments were begun on January 29 and continued for 48 days until March 13 for a total of 91.8 mm of water added in the average treatment. I recorded the number of seedlings that emerged every four days, categorizing individuals by life form (monocot or dicot) and species th when possible. I ended the experiment on March 18 (53 days after initiation) and harvested the aboveground biomass for each species and oven dried it at 60°C for a minimum of 72 hours. Germination data were square root transformed and S. faberi biomass data were log transformed to meet assumptions of normality. Experiment 2: Germination responses of sown seeds—To more precisely determine germination responses of the most common species that emerged in the first experiment (S. faberi and C. album) to precipitation variability, I conducted an experiment where I used known numbers of seeds of these two species. I collected seeds for this experiment from plants growing at the KBS LTER field site in the fall of 2007. Seeds were stored until May 2008 at alternating temperatures (~7°C and -17°C) every two weeks to mimic freeze-thaw cycles over the winterspring in SW Michigan. I used autoclaved soil for this experiment to eliminate the seedbank. I placed 5cm of sterilized soil in 8.5 by 8.5 cm pots, rinsed the soil with water, and then allowed them to air dry to a constant weight. I sowed 50 seeds of each species into separate pots, with six replicates of each treatment combination (36 pots for each species) using the same 2x3 factorial design described above. I covered the seeds with a thin layer of soil and then added 1.2 cm of water to each pot as an initial watering. I did not use any supplemental lighting in this 25 experiment because the time of the experiment (late May) coincided with the typical season for seedling emergence for these species. The experimental watering treatments were initiated on 27 May and continued until 16 June, allowing for 2 cycles of the longest and 6 cycles of the shortest interval treatments. A total of 61.2 mm of water was added to all of the pots in the average and 30.6 mm in the low precipitation treatment. I censused and rotated the pots every two days until the experiment was terminated (22 June, 30 days after initial watering). I tested for species responses to the precipitation treatments using the total number of seeds germinating in a linear model with total water, interval, and their interaction. Data for C. album germination were log transformed to meet assumptions of normality. Experiment 3: Individual Plant Biomass—To disentangle the direct effects of altered precipitation and seedling density on species growth I grew individual S. faberi and C. album seedlings in pots and monitored their growth in response to the same 2x3 factorial watering treatments. Seeds from the same field collection as Experiment 2 were germinated in the greenhouse on moistened filter paper in petri dishes. Newly emerged seedlings were transplanted into conetainers (3.76cm diameter by 20cm depth, Stuewe and Sons, Tangent, Oregon USA) filled with air-dried soil from the field site (collected as described in Experiment 1, above) and topped with 2 cm of autoclaved soil. All seedlings received 10mL and 15mL of water on the first and second day, respectively, following transplantation and no additional water until the experimental treatments were initiated on the fourth day after transplantation. Seedlings that died during the first two weeks of the experiment were replaced; a total of 48 S. faberi (8 replicates) and 36 C. album seedlings (6 replicates) were grown for 30 days for this experiment. Seedlings in the average water treatment received a total of 61.2 mm of water during the 26 experiment and those in the low treatment 30.6 mm. I harvested and dried aboveground biomass following the protocol used in the Experiment 1. I analyzed final biomass with a linear model testing for the effects of total water, interval, and their interaction and used planting date as a covariate. I excluded the 12-day interval for C. album from these analyses because seedling mortality was high, particularly in the low water treatment (>60%). Results Emergence from the seedbank and growth—In total over 2000 seedlings emerged in this experiment and the majority (88%) were monocots. Over 99% of monocot seedlings were S. faberi and this species accounted for a similar proportion of the total biomass in each pot. As a consequence total community biomass responses mirrored the S. faberi responses to precipitation treatments. Chenopodium album made up over 98% of dicot seedlings that emerged and over 95% of dicot biomass. Consequently, I focus on the response of these two species to my experimental watering treatments. Both total water and the interval between watering events affected emergence from the seedbank, but S faberi and C. album exhibited different responses. Setaria faberi germination increased with longer intervals in the low water treatment, but interval had no impact on germination at the average moisture treatment (interaction F1,20 = 15.3605, P < 0.0001). In contrast, C. album germination increased in response to both higher water amount and longer intervals (total water F1,20 = 116.2825, P=0.0008 and interval F1,20 = 23.5319, P < 0.0001). Setaria faberi final biomass decreased with longer intervals under average moisture, but increased with longer intervals when moisture was low (interaction F1,20 = 34.5754, P < 0.0001, Figure 3.1a), though total biomass was still higher under average moisture (Figure 3.1). 27 Figure 3.1 Biomass and germination responses of S. faberi and C. album to variation in precipitation amount and interval between watering events. Total biomass of (a) S. faberi and (b) C. album from the seedbank emergence experiment. Mean number of seedling emerging of (c) S. faberi and (d) C. album when known numbers of seeds were sown. Average individual seedling biomass of (e) S. faberi and (f) C. album when plants were grown individually. Bar are standard errors; n varies by experiment (see text). 28 Chenopodium album biomass was greater at average than low water amounts (F1,20 = 28.7816, P < 0.0001, Figure 3.1b), but showed no response to interval length. Germination responses of sown seeds—Overall, germination of S. faberi was higher (37.3%) than that of C. album (1.3%) in this experiment. Germination of both species showed interactive effects of precipitation amount and interval, but differed in how they responded. Germination of S. faberi was higher at longer intervals at both levels of total water, though the magnitude of the effect was greater at the low watering treatment (interaction F1,32 = 4.5623, P < 0.0404, Figure 3.1c). In contrast, C. album germination increased with longer intervals at low water, but showed no response to interval at average levels of moisture (interaction F1,32 = 4.9821, P < 0.0327, Figure 3.1d). Individual Plant Biomass—The biomass of both species when grown individually responded to the interaction of total water and interval, but in opposite directions. Individual S. faberi biomass decreased with longer intervals at both the average and low watering levels, but the decrease in biomass was greater under average moisture (interaction F1,41 = 4.8739, P=0.0329, Figure 3.1e). Biomass of C. album seedlings increased with an increase in interval from four to eight days at average moisture, but showed no response to interval length at low levels of moisture (interaction F1,23 = 6.6042, P=0.0171, Figure 3.1f). Discussion Precipitation variability affected both germination and growth of two common weed species, S. faberi and C. album, though not in the same manner. In addition, the magnitude and direction of the effects varied with life stage. Longer intervals between precipitation events consistently increased germination of both species under low, but not average, watering 29 treatments. However, S. faberi biomass decreased with longer intervals while C. album biomass either increased or remained unaffected. Previous research has shown a strong relationship between biomass and seed production for both species (Bussan and Boerboom, 2000, Grundy et al., 2004), suggesting that increasing precipitation variation may impact the relative fitness of these two species. The pronounced response of both species to interval length suggests that predicted changes in storm frequency and intensity (Easterling et al., 2000, Weltzin et al., 2003, IPCC, 2007) could impact the overall abundance and species composition of weed communities in row crops, particularly if spring precipitation is reduced. Increased emergence with longer intervals may result from seeds needing to reach a minimum level of moisture for germination (Baskin and Baskin, 1998). The larger amounts of water available at one time in the longer interval treatments could promote germination rates if these species have to be exposed to a threshold moisture level to germinate. These results are consistent with a study by Bouwmeester (1990 as cited in Baskin and Baskin, 1998) that found C. album germination was stimulated by desiccation events. The lack of a consistent response to interval length at average moisture suggests that germination was being determined by factors other than moisture, such as temperature (Leblanc et al., 2002) or light (Baskin and Baskin, 1998). In addition, seed traits such as size may play a role in determining germination responses. While both of the species in my study have relatively small seeds, C. album had a smaller average seed mass than S. faberi (0.2 mg vs 1.75 mg). Differences in seed mass between species have been correlated with germination responses to water (Daws et al., 2008), so seed mass, along with other seed traits may play an important role in determining germination responses to precipitation variability. Altered precipitation regimes may change the relative timing and density of seedling emergence and there is evidence that C. album biomass may increase with 30 earlier germination times (Miller, 1987). However, further work with more species under field conditions could identify whether shifts in initial density like those I saw lead to shifts in the relative abundance of species at the end of the season. Such work could also incorporate temperature effects predicted due to climate change, which may make water more or less limiting. How annual weed communities respond to altered precipitation will depend not only on seedling emergence, but also growth and competitive interactions among species. Species differences in root morphology may explain the differences in plant growth responses I observed in these experiments; C. album (positive/neutral) and S. faberi (negative) to longer intervals, when precipitation events are large. Chenopodium album has a taproot that may allow it to access water that infiltrates deeper into the soil following large rain events, similar to shrubs. In contrast, S. faberi has a fibrous and shallower root system that may limit its ability to access deeper soil water deeper. This could make S. faberi more sensitive to variation in surface soil moisture and is consistent with previous work (Dalley et al., 2006) that found evidence of deeper moisture use by C. album than S. faberi in cornfields. Walter (1971) suggested that shrubs and grasses could partition soil moisture, with deeper rooting shrubs accessing deep water not accessible to grasses, and there is some limited evidence that this can happen (McCarron and Knapp, 2001). It is important to note that C. album mortality was high in the individual growth experiment, so this may not be representative of how C. album would respond to variation in precipitation in the field. Setaria faberi biomass declined in both the individual seedling experiment and community experiment, and only in the community experiment could this response be potentially attributed to increased competition due to higher seedling emergence and density in the pots. Species differences in root morphology could provide an explanation for the 31 observed differences in growth of these species and if true that information could inform weed management practices. I did observe differences in rooting depth and production across treatments when I harvested the seedbank study that appeared to follow the difference in aboveground biomass in these treatments (personal observation). I did not harvest roots in this experiment because I could not reliably separate roots by species in the mixed community experiments, so it is not clear if species differed in belowground production or other factors that could impact their response to precipitation variation. Overall, my results suggest that depending on the balance between reduced growth and enhanced germination, S. faberi may become less abundant relative to C. album as precipitation becomes more variable, but weed community productivity may be buffered by differences in species’ responses. All of these experiments were conducted in the greenhouse, which allowed us to control other environmental variables that can impact seedling emergence and growth. The size of the pots and scale of the experiment only allowed us to focus on effects of precipitation variability early in the life cycle of these species. Using small pots for these studies may have introduced some artifacts into the experiment. For example, soil in pots will dry faster than in a continuous soil profile, exacerbating the effects of longer intervals or reduced total water. While the direction and magnitude were context dependent, my results show that precipitation variability could have marked impacts on the emergence and establishment of two species that are common agricultural weeds in the Midwestern USA. Further work in the field is needed to determine if the responses observed here occur in the field and to evaluate the impact across the growing season and any implications for irrigation management. 32 CHAPTER 4 IMPLICATIONS OF ALTERED PRECIPITATION VARIABILITY ON ANNUAL PLANT PRODUCTIVITY AND STRUCTURE Abstract Climate change predictions of increased precipitation variability have stimulated research on how plant communities will respond, both in traditionally water limited communities and more mesic communities. Much of the literature in arid ecosystems examines the responses of annual plants to precipitation variability, but perennial communities have dominated in studies of mesic systems. Annual weed communities are common and economically important systems that may respond strongly to increased precipitation variability based on studies in arid ecosystems. We used 15 years of data from the Kellogg Biological Station Long Term Ecological Research site to examine how historical variation in precipitation, in conjunction with temperature, correlated with annual weed community biomass, richness, and species abundance. We experimentally manipulated initial precipitation and temperature to confirm the causal relationships with changes in the plant community and examine how initial emergence related to the observed biomass effects. We found that increased precipitation during the first week of the growing season led to increased richness and decreased biomass in both our long-term data set and our experimental communities. Species differed in their responses, with some responding to interactions between temperature and precipitation while others showed no response to either. Species germination and biomass responses to initial precipitation were not always in the same direction, with evidence of inter-specific competition. These results suggest that predicted changes in precipitation variability may have important implications for annual weed communities, even though water is not considered limiting in mesic systems. In addition, the 33 differences between the effects on germination and final biomass highlight the need to take species interactions into account when attempting to forecast the effects of climate change on communities. Introduction Predictions of increases in average CO2 and temperature have motivated increasing interest and research related to how predicted changes in these environmental drivers will affect ecological communities. Most of these efforts have focused on effects of predicted changes in mean or average conditions but models also predict increased in variability. Several recent studies have focused on how increased variation and extreme events will affect plant communities (Jentsch et al 2007, Knapp et al 2002, 2008, Heisler-White et al 2009, Sher et al 2004). Increased precipitation variability is predicted to take the form of more intense, but less frequent storms may have important consequences, particularly in both arid and mesic environments (Easterling et al 2000, Weltzin et al 2003, IPCC 2007). The impacts of precipitation variation have been studied in arid and semi-arid ecosystems for a variety of functional groups, including annual plant species (Pake and Venable 1995, 1996, Clauss and Venable 2000, Schwinning and Sala 2004, Sher et al 2004). However, less is known about how more mesic systems are – and will be - affected by predicted changes in intra-annual variation in precipitation (but see Knapp et al 2002, Heisler-White et al 2009, Fay et al 2008). Increased intra-annual variability may be particularly important for annual plants because recruitment and seedling establishment is a critical phase in the lifecycle. Germination of many species is known to be sensitive to variation in moisture and temperature (Baskin and Baskin 1998). Increasing variability in precipitation may interact with changes in average temperatures 34 to affect emergence and recruitment in annual communities (Levine et al 2008, Baskin and Baskin 1998, Leblanc et al 2002). For example, Levine et al (2008) showed that variation in establishment of rare annuals on California islands was driven by differences in temperature after the first storm of the year, and that variation in establishment could be as important as reproduction in determining population growth rates. Altered germination conditions may lead to different initial seedling communities (Facelli et al 2005) as the density and relative abundance of species change and may affect species interactions like competition (Levine et al 2010). In this study we examine how variation in precipitation and early season temperature affect the emergence and recruitment of a number of weed species that are economically important in row-crop agriculture. Predicted changes in precipitation and temperature due to climate change are likely to affect the recruitment and abundance of annual weed species and thus their competitive interactions with crops, which may have important ecological and economic consequences. In previous greenhouse studies we have shown that two agronomically important annual weeds, Setaria faberi and Chenopodium album, are responsive to changes in precipitation variability (Robinson & Gross 2010). Here we expand this approach by considering how variation in precipitation and temperature early in the growing may affect the abundance and composition of annual weed communities in the field. We used a long-term data set from the Kellogg Biological Station (KBS) Long Term Ecological Research (LTER) site to explore how species abundance varied in relation to interannual variation in precipitation and temperature. We evaluated both the effects of early-season (one week after tillage) precipitation and temperature (drivers of seedling emergence) and total growing season variation in precipitation on species abundances over a 15 year period. We also conducted a field experiment in which we manipulated early season precipitation and 35 temperature to test if the relationship between initial environmental conditions observed in the long-term data set were predictive of variation in species’ biomass. Data on emergence in the experiment allowed us to determine the direct effects of environmental conditions while our final biomass data also took into account changes in species interactions during the growing season. These two approaches allowed us to determine 1) Whether initial precipitation and temperature affect annual species abundance and if so 2) Whether these effects lead to changes in community production and richness. Methods Long-term data To examine correlations between initial environmental conditions and species abundance, we used data from annually-tilled microplots located in the six replicate successional fields established at the W. K. Kellogg Biological Station Long Term Ecological Research site (LTER) in southwest Michigan http://lter.kbs.msu.edu/about/experimental_design.php). These micro plots (20 by 30 m) are tilled and disced each spring (typically in late April,early May) to maintain an annually dominated plant/weed community. The spring tillage/disturbance not only maintains an annual community, but sets a specific date to the start of the growing season that can be used to determine initial and seasonal precipitation and temperature conditions. Each 2 year biomass from a 1m plot is harvested and sorted to the species level. For more details on plot maintenance and harvest see Grman et al (2010). The LTER also maintains a weather station on site that collects data on daily temperature and precipitation allowing us to compare data on species composition and abundance to temperature and precipitation data collected at the same site. 36 We used data from 1993-2008, excluding 2000 due to problems with the collection of temperature data that year at the LTER weather station. Although the annually-tilled plots were established in 1990, we excluded data from the first three years (1990-1992) due to differences in harvesting methods and a change in community composition that occurred immediately after the experiment was established. We first evaluated if precipitation was an important driver of variation in species abundance by testing how well precipitation over various intervals early in the growing season predicted end of season aboveground biomass (our measure of annual production). We evaluated four early precipitation periods (one, two, three, or four weeks after tillage and total growing season precipitation and evaluated which period of precipitation best explained yearly variation in biomass using Akaike’s Information Criteria corrected for small sample sizes (AICc). We also examined if tillage date (start of the growing season) explained year-to-year variation in biomass better than precipitation. Models with the lowest AICc value are the most supported by the data relative to the other models compared; models within two AICc units of each other were considered equally supported by the data (Burnham and Anderson 2002). If two models were equivalent based on AICc values, the simplest significant model was used based on parsimony and if multiple models were equally parsimonious and were significant, the results from both are presented. We also examined whether the temperature the week following tillage affected productivity and if this interacted with precipitation. We examined three temperature variables: minimum (tmin) maximum (tmax) and average daily temperature ( ) during the first week following tillage. Analyses involving tmin and tmax allowed us to test for thresholds in germination response, whereas provided a more integrative measure of early season 37 temperature conditions. Temperature was included in the precipitation models (describe above) as both as a second independent predictor and as an interacting factor. We again used AICc to determine which model best explained variation in community biomass and species richness. We also evaluated how well early season climate variables (temperature and precipitation) predicted the biomass of six species that were common in the community including the dominant species Setaria faberi. These six species together made up on average 80% percentage of the community biomass though each species varied in abundance over years (see Table 4.1). Two species, Ambrosia artemisfolia and Digitaria ischaemum were not observed in these communities until 1994 and 1995 (year 5 and 6 after establishment), but once present they were consistently present. Field Experiment We established a field experiment in the annually-tilled area of two replicates of the LTER successional treatments to test whether variation in initial precipitation and temperature caused the inter-annual variation in productivity, species richness and abundance observed in the long term record. The two replicates selected had similar plant communities based on the presence and abundance of species from our long-term analysis. We used a factorial design to evaluate the effects of temperature and precipitation the first week after tillage (initial) on community and species productivity as well as species richness. We crossed two levels of nighttime temperature (elevated and control) and four levels of precipitation additions (0, 20, 40, 60mm), for a total of eight treatment combinations. We focused on manipulations only in the first week following tillage, because our analysis of the long term data indicated that conditions during this time period were the most important drivers of species and community abundance. 38 Table 4.1 Mean and range proportion of total biomass for each of the six species analyzed in the LTER data across the 15 years of our data set. Species Long Term Long Term Long Term Experimental Experimental Experimental Mean Minimum Maximum Mean Minimum Maximum Setaria faberi 48% 24% 70% 58% 29% 86% Digitaria sanguinalis 6% <1% 31% 21% 0% 58% Digitaria ischaemum 5% 0% 19% 1% 0% 3% Chenopodium album 10% <1% 30% 2% <1% 7% Ambrosia artemisfolia 8% 0% 8% <1% 0% 4% Abutiolon theophrasti 2% <1% 5% <1% 0% 2% 39 For the temperature treatment we used passive warming to elevate night time temperatures (as in Beier et al 2004) because current trends in temperature change indicate that minimum temperatures (which typically occur at night), are important for plant communities (Alward et al 1999). Our four precipitation treatments included the natural variation in precipitation observed in this community over the previous 20 years, and one treatment level (60mm) that exceeded the largest amount of precipitation (45mm) observed to date in the historical record. This range of treatments allowed us to examine the community response to extreme events that are typical and beyond the historical record, but likely will occur in the future. The field plots were tilled on May 21, 2009 and the treatments were established the next day. To control precipitation, we constructed six precipitation exclosures (15m by 3m), each of which covered a row of eight 0.75x0.75m plots that included all eight treatment combinations in our experiment. We considered each set of eight plots within an exclosure to be a block and randomly assigned each of the eight treatments to one plot in each block with three blocks.in each of the two fields, for a total of 48 plots. The plots were separated from each other by one meter buffer strips and the plots were located at least 1m from the edge of the precipitation exclosure. The exclosures were created by mounting six mil plastic over a wooden frame that was placed at an angle one meter above the plot. We kept the plastic covers rolled up and not covering the plots except during precipitation (or anticipated)) events. This minimized effects of the cover on temperature and light conditions to the plants. Overall, the plots were covered 56.25% of the week and 45.2% of the available daylight hours. The covers reduced 40 Photosynthetically Active Radiation (PAR) to 77% of full sun and increased air temperature from 0.5 to 1.61 °C (average 1.1°C) during a sunny period. To prevent water movement between plots, each plot was surrounded with an edge of doubled 6 mil plastic sheeting that was buried in the soil so that it extended 5cm above and below the soil surface. A trench approximately 7.5cm deep and 25cm wide was dug around the perimeter of each block to prevent overland water flow into the plots during large, natural precipitation events. The simulated precipitation treatments were established on the fourth day following tillage. Unchlorinated ground water was added to the plots using a watering can with a head to disperse water. Water for the larger precipitation treatments (40 and 60 cm) was added over a 36 hour period to avoid overflowing the plots. Before applying the precipitation treatments we took two 10cm soil samples from each plot in two blocks, homogenized them and used this to determine initial gravimetric soil moisture. We repeated this sampling 36 hours after our simulated precipitation treatments ended to determine how they impacted soil moisture content. There was a significant positive relationship between soil moisture and the amount of precipitation added (F1,12=245.594, p<0.001) and no effect of no effect of initial soil moisture . We established the passive warming treatment by covering the appropriate plots with 1mx1m Super R Diamond IR reflective film barrier (Innovative Insulation Inc, Arlington, Tx) that was mounted 25 cm over the ground. The plots were covered at sunset and uncovered at sunrise every day during the treatment period. To determine the effects of the warming treatment we measured soil temperature at sunrise in the top 2cm of soil in eight plots, averaging three measurements in each plot. On average, the covers 41 increased soil temperature 1.18°Cover that in the uncovered plots (Welchs Anova F1,6=43.503, p=0.003). A leak in one the precipitation exclosures during a rainstorm added an unknown amount of water to several plots, therefore we excluded this block from our analysis, reducing the number of blocks to five and experimental plots to 40. At the end of the week, we removed all experimental infrastructure, except for plot markers. All of the plots subsequently received the same natural precipitation and temperature conditions. We measured the effects of the temperature and precipitation treatments on seedling emergence and end of season biomass. We conducted the seedling emergence census in four blocks (32 plots) 18 days after the experimental manipulations were completed. Digitaria sanguinalis and D. ischaemum could not be distinguished at this stage (1-2 leaves) and so were referred to as Digitaria species. All other seedlings could be reliably identified to species by this date. We used end of season aboveground biomass as our measure of species abundance. Harvests were done by block over six days beginning August 31. We harvested by clipping all plants in a plot at ground level and then sorted to species. Plant material was dried at 60°C for a minimum of 72 hours. For two species (S. faberi and C. album) we counted the number of stems in our harvested biomass to get an estimate of final density. We could not use stem counts as an accurate measure of density for the other species because they had more than one stem per plant. One plot was dropped from the final analyses due to a harvesting error. We used R (Version 2.10.1, R Core Development Team 2009) for the model selection analysis. We checked each variable for independence between years and found species richness and D. ischaemum biomass both showed a strong correlation with the 42 previous year’s value. To account for this, we included a lag term in all models for these two variables. Significance of the best models was checked using JMP 5.01 (Version 5 SAS Institute Inc., Cary, NC, 1989-2007) with variables transformed as necessary to meet test assumptions. We also used JMP for the experimental analyses, with precipitation exclosures as a block factor, nested in the two replicate fields. We examined the direct and interactive effects of initial precipitation and warming treatment on community biomass, species richness and biomass of the seven most common species. The species data were analyzed using a MANOVA before running individual species analyses as protected ANOVAs (Scheiner and Gurevitch 2001). The effects of initial intra- and inter-specific seedling density on final biomass were determined for the three most abundant species S. faberi, D. sanguinalis and C. album and for the final density and biomass per stem for S. faberi and C. album. Although D. sanguinalis and D. ischaemum could not be distinguished at the initial census (and so were counted together), the final biomass of D. ischaemum was much less than D. sanguinalis (1g vs 32g per plot). Given these differences in final abundance (and overall low postemergence mortality) it is likely that D. sanguinalis accounted for the majority of Digitaria seedlings in the initial census Therefore we used the total Digitaria census data as a metric for D. sanguinalis seedling density. To meet assumptions of normality, we transformed several variables prior to analysis and note this in the appropriate table. In several of our analyses we visually noted outliers in our residual plots. In all cases, the exclusion of these data points did not qualitatively change the results and so we included these data in the analyses reported here. 43 Results Long term data: Patterns in community and species abundance Aboveground production and species relative abundance were quite variable over the 15year time period used for these analyses (Figures 4.1, 4.2). While the relative abundance of species changed through time, S. faberi was an important component of the community across the entire 16-year period (Table 4.1, Figure 4.2). Precipitation during the first week after tillage was a better predictor of total community biomass (Tables 4.2, 4.3) and species richness than any other period, including total growing season precipitation. Interestingly total biomass declined (Figure 4.3a) and species richness increased (Figure 4.3b) with increased initial precipitation (Tables 4.3, 4.4). Early season air temperature did not improve model support for either community biomass or species richness (Table 4.3). The six most abundant and consistently present species (see Figure 4.2, Table 4.1) in the community showed different responses to early season temperature and precipitation. Annual biomass of S. faberi, the most abundant species in this community, was best predicated by an interaction of precipitation and temperature (Figure 4.4). In cool years, S. faberi biomass was positively correlated with initial precipitation, while in warm years S. faberi biomass decreased with increased initial precipitation. Models based on initial precipitation in combination with average daily temperature and minimum daily temperature had equal support (Table 4.3). Chenopodium album biomass was highest in years when the average daily temperature was cool and initial precipitation was lower (Figure 4.5, Tables 4.3, 4.4). In contrast, D. sanguinalis biomass was positively correlated with minimum daily temperature during the first week (Figure 4.6, Tables 44 !"#$%&'()*'+,,$-.'/-%"-0"1,'",'-/&%-#&'010-.'2133$,"04'5"13-66'761."8'5.-29'.",&:' -,8'6;&2"&6'%"2<,&66'7#%-4'.",&:'0<%1$#<'0"3&'",'0<&'=>?'@ABC'-,,$-..4'0"..&8' 3"2%1;.106)''D-.$&6'-%&'3&-,6'±'60-,8-%8'&%%1%'7,EF:)''' ' ' !"#$%&'()G'C&.-0"/&'-5$,8-,2&'1H'0<&'01;'I'6;&2"&6'7*JJKLGMMN:)'O0<&%'",2.$8&6'-..' 10<&%'6;&2"&6'7$;'01'*K'",'-'4&-%:'0<-0'P&%&',10'21,6"60&,0.4';%&6&,0'1%'3-8&'$;'-' 63-..'H%-20"1,'1H'0<&'010-.'5"13-66)' ' ' 45 Table 4.2 Differences in AICc (! AICc) between models using different periods of precipitation or the date of disturbance to explain community biomass across the 15 years in our LTER data set. All models are compared to the model with the most support (lowest AICc value), which has a ! AICc of 0. Model ! AICc Precipitation during the first week 0 Precipitation during the first two weeks 4.98 Precipitation during the first three weeks 6.19 Total precipitation during the growing season 6.85 Precipitation during the first four weeks 7.04 Date of tillage 8.92 46 Table 4.3 Differences in AICc (! AICc) between models explaining final biomass or richness using precipitation (P) during the first week of the growing season and the minimum (tmin), maximum (tmax), or average daily temperature ( ) during that week. Models Total biomass 0 Total Richness* 0 Setaria faberi 2.43 Chenopodium album 0.26 Digitaria sanguinalis 15.34 Digitaria ischaemum* 0 Abutilon theophrasti 0 P + tmax 3.18 2.83 6.11 2.17 13.72 1.78 2.94 2.28 P + tmax + P * tmax 4.65 8.25 3.46 6.43 18.24 5.6 7.45 6.76 1.9 4.5 4.73 0.64 0 2.05 1.64 3.73 3.22 2.65 5.32 10.33 4.4 9.73 0 5.29 0.48 1.61 0.9 0 4.08 9.96 14.45 2.36 1.35 4.21 5.66 2.24 5.8 8.17 3.8 7.21 P P + tmin P + tmin + P * tmin P+ P+ +P* *All richness and Digitaria ischaemum models also include a lag term, the value of that variable the previous year. !"#$%&'()*'+&,-."/012"3'4&.5&&0'"0"."-,'3%&6"3".-."/0'78"%1.'5&&9'-8.&%'.",,-#&:'-0;'<:'' ./.-,'6/==$0".>'4"/=-11'-0;'?:'13&6"&1'%"620&11'/@&%'AB'>&-%1'-.'.2&'C?D'EFG+)' 47 Ambrosia artemisfolia 0 4.3, 4.4) and did not vary in response to initial precipitation. Annual biomass of the other three species (D. ischaemum, A. theophrasti, and A. artemisfolia) was not correlated with variation in precipitation or temperature conditions in the week following tillage. These long-term analyses showed that this annual community was highly sensitive to variation in initial climatic conditions, both in aggregate and at the species levels. Furthermore, these responses varied by species not only in magnitude, but also in direction. Field experiment – Species responses to manipulations of early season climate Total seedling emergence increased with added precipitation (Figure 4.7a, Table 4.5) but there was no effect of warming on total seedling emergence. The MANOVA analysis revealed that species also responded to experimental precipitation manipulations (F5,21=3.72, p=0.014). Emergence of the two most common species, S. faberi and the Digitaria species increased with more precipitation amount under both ambient and warmed conditions (Table 4.5, Figure 4.7b, c). Amaranthus retroflexus and C. album also increased in abundance in response to added precipitation, but only under ambient conditions (Table 4.5, Figure 4.7e, f). Abutilon theophrasti emergence did not respond to either the temperature or precipitation treatments (Table 4.5, Figure 4.7d). Final biomass, density, and plant size– field experiment As had been observed in the long-term data (Figure 4.3), total community biomass and species richness varied with initial precipitation. The community biomass response to initial precipitation was negative but only marginally significant (Figure 4.8a, Table 4.6). Species richness increased with additional precipitation during the first week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able 4.4 Summary of R and significance for the final models from the model selection using the long-term data. Minimum temperature during the first week is represented (tmin) while the average daily temperature is ( 2 df F statistic p-value R Total Biomass Richness Setaria faberi Setaria faberi tmin Digitaria sanguinalis Chenopodium album 0.45 0.71 0.53 0.55 1,13 2,12 3,11 3,11 10.58 14.86 4.16 4.41 0.006 <0.001 0.034 0.029 0.72 2,12 15.62 0.54 3,11 4.25 ). <0.001 0.032 Table 4.5 Seedling emergence as a function of precipitation (P) and warming (W) treatments. P W P*W Field Block 1 NS NS Total Seedlings F1,25=17.31, NS F2,25=6.62, p<0.001 p=0.005 Setaria faberi F1,25=27.35, NS p<0.001 NS F1,25 =13.58, p=0.001 NS Digitaria species F1,25=19.95, NS p<0.001 NS NS F2,25=16.25, p<0.001 Chenopodium 1 album F1,25=4.47, p=0.045 F1,25=5.15, p=0.032 NS F2,25=11.43, p<0.001 Amaranthus NS NS F1,25=4.93, 2 retroflexus p=0.036 1- log transformed data, 2 – square root transformed data NS F2,25=29.82, p<0.001 1 NS 51 !"#$%&'()*'+&&,-".#'&/&%#&.0&'".'%&123.1&1'43'&52&%"/&.46-'/6."2$-64"3.1'37'7"%14'8&&9' 2%&0"2"464"3.'6:'4;&'&.4"%&'1&&,-".#'03//$."4<='>:'!"#$%&'()*#0:#+),)-%()%#12&0"&1*',:#."# -/'01/(%2-)*#&:#3"#%4&56='6.,'7:'."#('-(0$4'752)''?.-<'3"#%4&56'6.,'."#('-(0$4'752'1;38&,'6' %&123.1&'43'&-&@64&,'4&/2&%64$%&'6.,'4;"1'".4&%604&,'8"4;'2%&0"2"464"3.'6/3$.4)''A646'73%' 34;&%'12&0"&1'6%&'03/>".&,'73%'4;&'&-&@64&,'6.,'03.4%3-'4%&64/&.41)'' 52 (Figure 4.8b, Table 4.6). There was no effect of warming on final community biomass or species richness. To determine if species abundance varied in response to the initial precipitation and warming treatments we compared the biomass of seven annual species that occurred in at least 50% of the plots at the end of the growing season. Together these seven species made up an average of 87% of the community biomass; five of these species were those that had been analyzed for the long-term data (see Table 4.1 for long-term vs. experimental abundances). An initial MANOVA revealed that the final biomass of these species varied in response to both initial precipitation (F7,25=14.18, p<0.001) and warming (F7,25=3.05, p=0.018). Digitaria sanguinalis and S. faberi, the two most abundant species, (Table 4.1) as well as D. ischaemum, and C. album all showed significant responses to the manipulations of initial temperature and/or precipitation (Figure 4.9). The other three species - A. theophrasti, P. dichotomiflorum, and A. retroflexus, showed no response in variation in initial environmental conditions. Biomass of both D. sanguinalis and ischaemum species increased initial precipitation (Table 4.6). In contrast, S. faberi biomass decreased in response to increased precipitation (Figure 4.9b, Table 4.6), leading to a decrease in the relative abundance of S. faberi in the community (Figure 4.10, F1,37=5.18, p=0.029). ) Final biomass of C. album depended on an interaction between initial precipitation and the warming treatment (Table 4.6, Figure 4.9c); it always declined in biomass with increased initial precipitation, but the decrease was larger with warming. Because total seedling density varied with precipitation (Figure 4.7a) and most species showed a positive emergence response to increased precipitation, we tested how initial density affected the biomass of the three most common species, S. faberi, D. sanguinalis, and C. album. ! 53 !"#$%&'()*'+,,&-./'0,',"%/.'1&&2'3%&-"3".4."05'645"3$74."05/'05'&58'0,'/&4/05'9:'-066$5".;' <"064//'458'=:'/3&-"&/'%"->5&//)' ' !"#$%&'()?'+,,&-./'0,',"%/.'1&&2'3%&-"3".4."05'645"3$74."05/'05'.>&',"547'<"064//'0,'9:'!"# $%&'()&%*)$@'=:'+"#,%-./)@'458'0"#%*-(1'4.'A:'46<"&5.'458'B:'&7&C4.&8'.&63&%4.$%&/) 54 Table 4.6 Biomass and richness as a function of precipitation (P) and temperature (W) treatments. Total Biomass P F1,31=2.96, p=0.095 W NS P*W NS Field F1,31=10.43, p=0.003 Block F3,31=14.58, p<0.001 Richness F1,31=5.34, p=0.028 NS NS NS F3,31=5.01, p=0.006 Setaria faberi biomass F1,31=12.17, p=0.002 NS NS F1,31=26.79, p<0.001 F3,31=14.46, p<0.001 Digitaria sanguinalis biomass F1,31=17.1, p<0.001 NS NS F1,31=14.63, p<0.001 F3,31=48.12, p<0.001 2 Digitaria ischaemum F1,31=22.64, p<0.001 NS biomass 1 NS Chenopodium album F1,31=19.7, p<0.001 biomass Setaria faberi Final NS NS 1 Density Chenopodium album Final NS NS 1 Density Setaria faberi F1,31=11.29, p=0.002 NS 1 Biomass/stem Chenopodium album F1,31=69.27, p<0.001 NS 1 Biomass/stem 1 – log transformed data, 2 square root transformed data NS F1,31=11.75, p=0.002 F3,31=4.01, p=0.016 F1,31=6.20, p=0.018 F1,31=10.91, p=0.002 F3,31=6.64, p=0.001 NS F1,31=33.17, p<0.001 F3,31=6.08, p=0.002 NS F1,31=20.61, p<0.001 F3,31=7.84, p<0.001 NS NS NS F1,31=4.41, p=0.044 NS NS 1 55 Table 4.7: Intra and inter-specific affects of seedling density on final density and biomass. Intra-specific initial density Inter-specific initial density 1 Setaria faberi Final Density F1,28=9.19, p=0.005 F1,28=13.01, p=0.001 1 Chenopodium album Final Density 1 Setaria faberi Final Biomass 1 Chenopodium album Final Biomass Digitaria sanguinalis Final Biomass F1,28=9.68, p<0.001 NS F1,28=6.24, p<0.001 F1,28=28.84, p<0.001 F1,28=30.93, p<0.001 NS F1,28=33.69, p<0.001 F1,28=5.94, p=0.022 1 – log transformed data ! "#$%&'!()*+!,--'./0!1-!-#&0/!2''3!4&'.#4#/5/#16!756#4%85/#160!16!/9'!&'85/#:'!5;%6<56.'!1-!!"# $%&'()!;#17500!5/!/9'!'67'*"#%+&,-'"1"/".-' 2&&8-"1#'8&12"/9'.18'*"#%+&,-':"1.-'4"0;.22<'.18'?7'.--'0/,&%'2&&8-"1#2'.18'*"#%+&,-':"1.-' 4"0;.22)' ' 57 !"#$%&'()*+',--&./0'1-'"2/%34'325'"2/&%406&."-".'0&&57"2#'5&20"/8'12'/9&'-"237'5&20"/8'1-'!"# $%&'()':3;'<='325'*"#%+&,-':.;'5=)' ' 58 Final biomass of S. faberi and C. album declined as the density of competing seedlings increased (Table 4.7, Figure 4.11b, Figure 4.11d), but D. sanguinalis showed no response to the density of competitors’ seedlings. Increased emergence of C. album and D. sanguinalis did correlate with increases in their own biomass (Table 4.7, Figure 4.11c) but S. faberi showed no such response (Figure 4.11a). For two species, (S. faberi and C. album) we had information on final density as well as initial density. The change in density over the season for both species was generally, but not always, lower than initial density, but did not differ with treatment conditions. Final density of both S. faberi and C. album was positively correlated with initial density, (Table 4.6, Figure 4.12a, 4.12d respectively). However, the final density of both species declined as the seedling density of competitors increased (Fig 4.12b, 4.12d, respectively; Table 4.6). Discussion While inter-annual variation in climate, particularly precipitation, has been shown to be a major factor affecting the productivity and species composition of arid communities (Milchunas et al 1994, Pake and Venable 1995) fewer studies have examined how this impacts plant communities in more mesic systems (but see Knapp et al 2002). We found that for an annualdominated weed community inter-annual variation in community biomass, species richness, and abundance was strongly correlated with variation in initial precipitation and temperature. Interestingly, climatic conditions in the first week following disturbance/tillage had stronger effects on community patterns than seasonal totals. In this system, community biomass was inversely related to initial precipitation amount whereas species richness increased with precipitation. These results were confirmed in a field experiment where we manipulated initial 59 temperature and precipitation conditions, suggesting that events that relate to seedling emergence patterns have persistent effects on community patterns. In both the long-term and experimental study, the relationship between initial precipitation and community production was negative while the relationship with species richness was positive. Species differences in their responses to variation in initial precipitation – and to a lesser extent temperature led to shifts in the community composition. Previous research has shown that increased variation in precipitation across the course of a growing season may have important effects on productivity and diversity (Knapp et al 2002, Heisler-White et al 2009). The observation that events early in the life cycle of these species have effects that are not only detectable, but stronger than variation in seasonal amounts suggests that the impacts of increased environmental variability may be partially due to altering environmental conditions during short, but key stages in community development. While the long term analysis and the experimental results generally reinforced each other, the lack of response to temperature by several species in the experiment that showed strong responses in the historical analysis bears examination. These differences may be due to the relatively warm ambient conditions during the experiment and microclimate effects. The th ambient temperature during the experiment would have ranked as the 4 highest in average daily temperature if included in our long-term analysis. In addition, even with the removal of the plastic during precipitation free periods, the entire experiment did experience increased temperatures due to the exclosures. Soil warming via the nighttime passive warming treatment may not have had an impact on species historically sensitive to temperature due to being a small increase in an already high set of temperatures. 60 Differences in species responses to initial environmental conditions coupled with multiyear seedbanks for our species (Buhler & Hartzler 2001, Burnside et al 1996) could allow a temporal mechanism for coexistence, the storage effect (Chesson and Huntly 1989), to play a role in maintaining species diversity in our system. If so, the interaction between initial precipitation and temperature reinforce the point by Facelli et al (2005) that species may partition more complex axes of temporal variation than a single resource. Future work could directly test whether species in this community coexist in part due to the storage effect. It could also address whether alterations in the frequency of various initial environmental conditions due to the effects of climate change would strengthen or weaken this mechanism of coexistence. In addition to the implications of variability in species responses for coexistence, our results may also explain some of the mechanisms providing stability in our system. Grman et al (2010) showed that stability in production in this system arose from a number of mechanisms, including compensatory dynamics and the dominance of stable species. Variation in how species germination responded to initial conditions, both in whether they responded and the strength of that response likely play a role in the compensatory dynamics found in that study. In addition, the tendency for production of our dominant species, S. faberi to be limited in years when its emergence is favored may help explain why this species provides stability to the system productivity. Negative productivity and diversity relationships have been observed elsewhere (Waide et al 1999), including studies on precipitation variability (Knapp et al 2002). In our study, this negative relationship suggests that changes in the initial environmental conditions had consequences for competitive interactions that resulted in decreased community biomass and increased richness. While increased precipitation had a positive or neutral effect on seedling 61 emergence of all species, this did not translate into increases in final biomass for all species. The negative relationship between the seedling density of inter-specific competitors for S. faberi but not D. sanguinalis suggests that this species may have suffered greater competitive effects due to increased competitor density. This suggests that species interactions may mediate changes caused by shifts in environmental factors due to climate change, even over the course of a single season. Previous work in California grasslands has found that species interactions can reverses the direct effects of altered environmental conditions over the course of several years (Suttle et al 2007). However, work by Levine et al (2008) in a system with annuals has found no or weak evidence of indirect effects of climate change mediated by species interactions. The likelihood of seeing indirect effects may depend on how strongly species affected by environmental changes interact. In our study, the indirect effect of environmental change is seen in the dominant species, with a reduction in the relative abundance of the dominant S. faberi. This reduction in the dominant species corresponds with an increase in species richness, potentially due to a reduction in competition from the dominant. While S. faberi has an erect growth form, D. sanguinalis is more prostrate, so if light reduction is part of S. faberi’s competitive effect, that might explain why richness is higher even though D. sanguinalis biomass increases as S. faberi’s decreases. Climate model predictions of altered precipitation patterns where precipitation is less frequent, but more intense, will shift the distribution of good and bad years for the dominant species in the system. If drier years tend to be relatively better for S. faberi and particularly wet year relatively worse, over time the frequency of particularly good and bad years will both increase while intermediate years decrease. This will make the composition of the community more variable 62 through time, and if species responses to environmental shifts, both direct and indirect, are not all linear then this may cause shifts in the overall abundance of species through time as well. The results of this study add to the growing body of literature demonstrating the importance of changes in climatic variability on plant community structure and composition (Jentsch et al 2007, Knapp et al 2002, 2008, Heisler-White et al 2009, Sher et al 2004) and highlights the importance of considering indirect effects. Our results show that observed variation in initial precipitation and temperature can influence community composition and productivity in an annual plant community. Levine et al (2008) suggested a similar process may be important in maintaining diversity and promoting coexistence among annual species on California islands. Although understanding coexistence mechanisms may not be as important in weed communities, knowledge of how weed species composition and abundance will respond to predicted changes in environmental variability may be important in developing low or reduced chemical input systems for managing weeds in row crops. Our results suggest that predicted changes in precipitation variability may have important implications for annual plant communities, even in systems where water is not considered to be limiting. 63 CHAPTER 5 PRECIPITATION VARIABILITY AND PLANTS: EMPIRICAL AND MODELING APPROACHES TO UNDERSTANDING THE EFFECTS OF CLIMATE CHANGE ON SPECIES GROWTH AND COMPETITION Abstract Global climate change models predict increased intra-annual variability in precipitation, and recent work in perennial grasslands has shown that these changes can affect community productivity and composition even in mesic systems. Annual weeds are common, economically important species whose establishment is sensitive to altered precipitation regimes. These species can provide a model for how mesic annual plant growth and species interactions may be affected by increased precipitation variability. I conducted a field experiment with two annual species, Chenopodium album and Setaria faberi, and developed a simple mathematical model to explore how variation in precipitation would affect species, both in monocultures and in mixed communities. These species differed in two general traits expected to affect water use — rooting morphology and photosynthetic pathway. Both the field experiment and model found that growth was reduced when precipitation shifted from small, frequent events to large, infrequent events, but this response could change in the presence of competitors. I found that a tradeoff between uptake rates at high and low water in my model was more consistent with my field results than differences in rooting morphology or photosynthetic pathway which I hypothesized would drive species responses. The interaction between altered precipitation and competition indicates that species interactions can mediate the effects of climate change, in some cases offsetting the direct effects entirely. In addition, species responses to altered precipitation may not be well predicted by one or two broad traits, but by a more inclusive approach towards plant water use. 64 Introduction Predictions from global climate change models of increasing variability in precipitation (Easterling et al 2000; Weltzin et al 2003) have inspired a number of experiments examining how larger, but less frequent precipitation events may affect terrestrial plant communities (Knapp et al 2002; Heisler et al 2008; 2009). Much of the previous research on precipitation variability has focused on responses to inter-annual variability in arid systems that have a high proportion of annual species (Pake and Venable 1995; 1996; Clauss and Venable 2000; but see Sher et al. 2004). More recent studies have considered the impacts of precipitation variability in mesic systems (Knapp et al 2002; Adler et al 2006; Heisler et al 2009; 2008) and the effects of intra-annual (within season) variation (Knapp et al 2002; Heisler et al 2009; 2008). Annuals may be strongly impacted by within-season variability in precipitation because they are constrained to reproduce in the season or year in which they germinate. A previous study has shown that annual weeds common in mesic U.S. Midwestern row crops show strong emergence responses to intra-annual variation in early season precipitation (Robinson and Gross 2010). However, less is known about how altered precipitation patterns will affect annual species growth and reproduction in mesic systems, either in monocultures or in mixed communities. Differences in a suite of functional traits related to water use may drive species differences in growth as precipitation regimes change. Such functional traits may include both morphological traits such as xylem structure (Rosenthal et al. 2010) and rooting structure (Walter 1971; Nippert and Knapp 2007) and physiological traits such as regulation of photosynthesis and allocation (Chaves et al 2002). While a particular set of traits will determine a species’ ability to 65 respond to precipitation variability during a growing season, the magnitude of this response may depend on just how limited plants are by precipitation. Schwinning and Sala (2004) reviewed the importance of pulse size and how organisms can require a minimum pulse (or series of pulses), leading to non-linear responses to precipitation events. Knapp et al (2008) extended this idea proposing that total precipitation may determine the direction of species’ responses to larger, less frequent precipitation pulses. They predict that as precipitation pulses become large and infrequent that plant growth will increase in arid ecosystems and decrease in mesic systems (Knapp et al 2008). How strongly inter-annual and intra-annual precipitation interact to determine species’ growth and interactions is not well understood, but there is evidence that it occurs (Heisler 2008). This makes both total precipitation and species’ traits important in understanding plant community responses to intra-annual variability due to climate change. I examined species growth in response to altered precipitation regimes (variability and total amount) in a field experiment using two annual weed species common in the Midwestern U.S.A. as model species. Annual weeds are both economically important and common plant species, and changes in their growth could have important ecological and economic consequences. I used Chenopodium album, a C3 tap-rooted forb and Setaria faberi a C4, fibrous-rooted grass because these species differed in both the morphology of their roots and the efficiency with which they used water during photosynthesis. That, in addition to previous greenhouse work where I found larger less frequent watering reduced S. faberi seedling growth with minimal impacts on C. album seedling growth (Robinson and Gross 2010), led us to expect differential responses to my manipulations of precipitation variability and total precipitation amount. 66 To further explore how differences in plant water use may affect responses to climate change, I also developed a simple mathematical model where I could examine how variability in precipitation affected species, both in monoculture and in mixed communities. I assumed a tradeoff between growth under high and low water availability that could be driven by a variety of traits, I varied the precipitation inputs to examine how precipitation amount and variability interacted to affect biomass. The combination of experimental and modeling I used in this study gives us insight about how 1) predicted changes in the distribution of precipitation affect the growth and competitive interactions of species and 2) whether those effects vary with total precipitation amount. Methods I established the field experiment in an agricultural field at the W.K. Kellogg Biological Station (KBS), Michigan U.S.A. that had previously been in a corn-soy-wheat rotation. The experimental design manipulated the interval between watering events (4 vs. 16 days) and the total amount of water (average vs. reduced) received by three experimental communities: monocultures of C. album and S. faberi and a 1:1 mixture of the two species. For the average precipitation treatment I used the median growing season precipitation (May-August) over 10 years (1996-2005) at the KBS Long Term Ecological Research site (LTER), 314 mm. I set the low water treatment at 50% of the average amount, in line with predictions from the driest model scenario for the region (Kling et al. 2003). For the interval treatments I again used data from the KBS LTER; the longest continuous dry period in the 10-yr KBS LTER dataset was 15 days so I used a dry interval of 16 days to simulate the extreme events that are predicted become more frequent with climate change (Easterling et al 2000; Weltzin et al 2003). Based on the frequency of precipitation early in the growing season (May) I used a 4-day interval to simulate a less 67 variable environment because in some years the longest dry period between precipitation events can be 4 days. The interval treatments varied the length of time between simulated rain events while keeping total water constant, creating frequent, small watering events in the 4-day treatment and infrequent, but larger watering events for the 16-day treatment. Plots in the average water treatment received water pulses of 10.2mm of water for 4-day events and 40.8mm of water for the 16-day events, and the reduced precipitation treatment received pulses half the size. I controlled rainfall inputs by building sixteen 3m x 5m rain exclosures, each of which covered 3 individual plots approximately 0.75m by 0.75m. The exclosures consisted of metal posts supporting wooden frames with 6mil polypropelene covers. These plastic covers extended at least 20cm past the sides of the frame and were situated 1m above the plots at an angle to promote drainage. The plastic covers were rolled up whenever possible to reduce microclimate effects on the plots. Each shelter contained one replicate of each of the three experimental communities. To control for spatial heterogeneity in the field, I grouped the shelters into four blocks and randomly applied one of the four watering treatments to shelters within the block. I measured air temperature at ground level and photosynthetically active radiation (PAR) in and adjacent to the exclosures several times during the experiment to quantify microclimate effects of the exclosures. The shelters were covered approximately 58% of the time and approximately 55% of the daylight hours during the experiment. The plastic covering reduced PAR to 78% of ambient levels, but did not consistently raise the maximum air temperature in the shelters. The three plots within each shelter were spaced 1.125 m from the long edge of the shelter and 0.875m from the short edge, with 0.5m between plots. I buried a doubled layer of six mil polypropelene to a depth of at least 80cm around the perimeter of each plot to prevent water 68 movement in or out of the soil profile for each plot. The plastic extended approximately 5cm aboveground to prevent lateral water movement in or out of the plots at the soil surface. I installed two-inch diameter PVC tubes in the center of each plot for volumetric soil moisture measurements using a Trime T3-50 Tube Access Probe (IMKO GmbH, Ettlingen, Germany). The PVC tubes were installed at least 50 days before I initiated the experiment to ensure good soil contact. To examine the effects of precipitation on growth after establishment, I broadcast seeds th into the plots on June 10 and watered the plots to promote germination. I then thinned or transplanted individuals of the appropriate species into the plots to reach a target density of 175 2 plants/m in each plot based on previous surveys of seedling density in a nearby fallow field. I replaced seedlings that died from transplantation twice to achieve my target density and plots were weeded to remove additional recruitment by target or non-target species. st th The experiment started on July 31 and was maintained for 40 days until September 8 when plants started to senesce. Plots were hand watered and large volumes of water were applied over the course of several hours to allow soil infiltration and prevent runoff. Soil rd moisture measurements were taken every other day starting on August 3 at three depths, 020cm, 35-55cm, and 70-90cm below the soil surface. Aboveground biomass was harvested at the end of the experiment, 8 days after the last watering event, separated by species in the mixed communities, and then oven dried at 60°C degrees for at least 48 hours. After determining aboveground biomass I also estimated reproductive biomass for each species. For S. faberi I clipped mature seed heads from each plant and used this as my estimate of reproductive biomass. 69 For C. album I stripped all of the seeds from their inflorescence and used total seed mass as an indicator of reproductive biomass. Biomass data were analyzed using Proc Mixed (SAS 9.2), with block as a random effect as well as the interaction between total water and interval, which was nested within block since treatment combinations were nested within shelters. Both aboveground biomass and reproductive biomass were log transformed to meet test assumptions. Soil moisture analyses used Proc Mixed for repeated measures analyses across sampling dates with the best covariance structure chosen using AIC out of compound symmetry (CS), autoregressive(1) (AR(1)), and heterogeneous CS and heterogeneous AR(1). Model development To explore why the biomass of my species decreased with higher variability in monoculture, but not necessarily in competition, I developed a simple mathematical model. I examined the effect of a tradeoff between the rate of water uptake at low versus high water availability that could be caused by multiple traits independently or in combination. Sommer (1985) discussed species differences in uptake rates as the difference between an affinity adapted species (superior uptake at low availability) versus velocity adapted species (superior uptake at high availability). There is some evidence that plant species have different water uptake rates (Leuschner et al 2004) and variation in traits like xylem structure can affect resistance to drought and maximum water conductance (Pockman and Sperry 2000; Rosenthal et al 2010), in some cases with a tradeoff between resistance and conductance (Pockman and Sperry 2000). Nippert and Knapp (2007) have shown that in prairies C4 plants use more water from the surface layer, which is more variable than deeper soil layers, than C3 plants. Based on that I expect that high uptake rates would likely be important for S. faberi, which is a C4, because it exploits a more 70 ephemeral resource pool and is therefore more likely to be analogous to Sommer’s (1985) definition of a velocity species. Chenopodium album as a C3 species which depends more on deeper, less variable soil moisture would likely face less selection to develop high water uptake rates and therefore would be more comparable to an affinity species. Previous work on differences in uptake rates has focused on tradeoffs between uptake rates at high and low resources as a potential mechanism for coexistence (Sommer 1985; Grover 1991; Litchman and Klausmeier 2001), but here I focus on the impacts of such a tradeoff on predicting dominance in competition. I did not use a growth-survival tradeoff because survival did not respond to my empirical precipitation treatments. The model has five state variables: available water (R), plant biomass of species i (Bi), and stored carbohydrate per biomass (Qi). A fixed amount of water per unit time (Rin) was added to the available water pool (Eqn 1) at discrete times with period (T) that I varied from short to long. Because total water input was held constant, larger values of T meant less frequent, but larger individual watering events as in my experiment. Between watering events, water (R, Eqn 2) was lost from the soil due to leaching and evaporation at a constant rate (L) and taken up by the two species with per capita uptake rate vi(R)=vmax,iR/(R+ki). The growth of the two species in the system (Eqn 3) depends on the amount of internal carbon stores available (quota, Q, Eqn 4) above the minimum necessary for growth (Qmin) as in Droop (1974) and Grover (1991). A species’ C quota increases due to the water-driven photosynthesis and decreases due to dilution by growth. Water uptake is modeled with a type II functional response. See Table 5.1 for a list of model equations and Table 5.2 for a list of model parameters. 71 Table 5.1 List of equations used in the mathematical model Equation # Equation 1 R(nT ) =R(nT )+RinT 2 dR/dt=-LR-vmax 1R/(R+k1)b1-vmax 2 (R/R+k2)b2 3 dBi/dt= µi(1-Qmin i/Qi)Bi-miBi 4 dQ/dt=eivmax i R/(R+k1) - µi(1- Qmin i/Qi)Qi + - Table 5.2 List of model variables and parameters Parameter Interpretation !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! R Available water Bi Biomass of plant species i Qi Quota, amount of carbon in plants available for growth Rin n T + nT , nT L µi Average water per unit time Integer tracking the number of intervals Length of interval between pulses Instantaneous moment of time immediately after (+) or before (-) a pulse Rate of water loss from the system Growth rate at infinite quota ei Efficiency of species i in converting water to carbon via photosynthesis Qmin i Minimum amount of resource necessary for species i to function mi Mortality of species i vi(R) Rate of water uptake by species i vmax i Maximum water uptake rate of species i ki Half saturation constant for water uptake of species i !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 72 I assume a tradeoff between two resource-uptake traits: vmax, the maximum rate of uptake, and k, the half saturation constant for water uptake. My velocity species, which I consider similar to S. faberi, had a higher vmax and k than the affinity species which I consider similar to C. album. This means that under high water conditions the velocity specialist takes up water faster and approaches its maximum uptake at higher values of available water than the affinity adapted species (Figure 5.1). I set all other parameters equal between the two species and varied the interval between pulses (T). My simulations ran for 100 time steps since my annual comparison system is unlikely to reach equilibrium over a growing season. I then explored the relative abundance of the two species at the end of the season/simulation as a function of T at three different levels of water supply (Rin=0.1, 1, 10). Results Experimental treatment effects on soil moisture Over the course of the experiment, shallow soil moisture (0-20cm) in the average precipitation treatment increased relative to the reduced treatment (F 18,670 = 3.56, p<0.001, Figure 5.2) and showed different responses to the variability treatments (F 18,670 = 9.66, p<0.001, Figure 5.2). However, the precipitation interval and amount treatments also interacted to determine soil moisture through time (F 18,670 = 2.29, p=0.002), with the interaction most evident in the 16-day high water treatment plots. Out of 94 significantly different pairwise comparisons (p<0.05 Tukey-Kramer test) across time, 56 included the 16-day, high precipitation treatment plots being wetter or drier than other treatment combinations. Community type did not affect shallow soil moisture in any treatment. 73 Figure 5.1 Theoretical differences in water uptake rate as a function of soil water for species exhibiting a tradeoff between maximum uptake rate and uptake rate at low resource levels. The affinity species (black, solid) has a higher uptake rate when soil moisture is low and the velocity species (gray, dashed) when it is high. Figure 5.2 Soil moisture over time in the experimental plots at shallow (0-20cm) depths for (a) average and (b) reduced precipitation treatments in the 4 and 16-day interval treatments. Values are pooled over the three community types because they were not significantly different. The first precipitation pulse occurred on August 3 (Julian day 215). 74 While shallow soil moisture responded to treatment manipulations, soil moisture at intermediate depths (35-55cm; data not shown), showed no response to watering treatments while soil moisture deeper in the soil profile (70-90cm) depended on an interaction between community type and total water (F 2,653 =6.82 p=0.001) that did not differ across sampling dates. There was a trend for deep soil moisture to be higher in the C. album monocultures receiving an average amount of precipitation, compared to S. faberi monocultures receiving the same amount of water (Tukey-Kramer p=0.079) or C. album monocultures with reduced water treatments (Tukey-Kramer p=0.100). Species responses to experimental manipulations Aboveground biomass and reproductive biomass varied with community type, interval, and total precipitation. Longer intervals led to lower aboveground biomass in both monocultures and the mixed communities (F 1,9 = 6.31, p=0.033, Figure 5.3). Aboveground biomass also decreased with less total precipitation though the reduction was marginally significant (F 1,9 =4.86, p=0.055). Aboveground biomass varied among the three community types (F 2,24 = 5.54, p=0.011); S. faberi monocultures had greater biomass than C. album monocultures (TukeyKramer p=0.004) and the mixed communities did not differ from either monoculture. Average 2 density decreased from 175 to 148 plants/m by the end of the experiment but showed no response to any treatment. Reproductive biomass across the three communities responded in the same manner as aboveground biomass (data not shown). The mixed community was dominated by C. album (F 1,12 =42.17, p<0.001), but the amount depended on an interaction with interval (F 1,12 =5.82, p=0.033, Figure 5.3). 75 2 Figure 5.3. Biomass (g/m ) of (a) Chenopodium album and (b) Setaria faberi growing in monoculture (dashed line) and mixtures (solid line) at two precipitation intervals. Data are pooled for the average and reduced precipitation treatments because there was not a significant response to precipitation amount. Values are averages ± standard error (n=8). 76 Figure 5.4. Predicted final (t=100 days) biomass of the affinity specialist (panels a, c, e) and the velocity specialist (panels b, d, f) growing in monocultures (dashed) and in competition (solid) in response to different interval lengths (T) between watering events. Total precipitation into the system (Rin) from top to bottom is 0.1, 1, and 10. Simulation parameters were set at k A = 0.1, vmax A = 1, k V = 1, vmax V = 2, initial values B A = B V =10. 77 Chenopodium album biomass was lower at longer intervals compared to shorter ones (TukeyKramer p=0.02) while S. faberi biomass remained constant across interval treatments (Figure 5.3). Watering treatments had no affect on reproductive biomass in either species when grown in competition; C. album allocated more biomass to reproduction than S. faberi across all treatments (species F1,12 =208.18, p<0.001). Density of each species in the competition plots did not respond to watering treatments. Model Results Both the ‘affinity’ and ‘velocity’ species were able to persist in monoculture over a range of intervals (T), although biomass of both species declined with longer periods (Figure 5.4 dashed lines). In addition interval length (T) had a large effect on which species dominated in competition, with short intervals/small pulses favoring the affinity adapted species and longer intervals/large pulses favoring the velocity specialist (Figure 5.4 solid lines). Interval length affected the outcome of competition most at intermediate values of total water (Rin). At lower total water input Rin, the affinity species was competitively dominant across a broader range of T (Figure 5.4, a-b solid lines) and the velocity species dominated at higher total water input Rin (Figure 5.4, e-f solid lines). For both species in monoculture, total water (Rin) and interval (T) interacted (i.e. 4a vs 4c, note the change in scale for the y-axis). Discussion Previous research has shown that intra-annual variation in precipitation can lead to altered ecosystem productivity and composition in mesic perennial communities (Knapp et al 2002; Heisler et al 2008; 2009) and previous greenhouse work suggested that annual communities were also likely to respond (Robinson and Gross 2010). My field experiment 78 demonstrated that a shift to less frequent, larger precipitation events during the growing season reduced the growth of two common species in monoculture, but that species interactions could alter the effects of resource variation in competition. My model showed that this pattern could arise from a simple tradeoff in water uptake between species, but also predicted an interaction between total precipitation and precipitation variability not found in my field results. These results show that intra-annual precipitation variability directly affects the growth and reproduction of mesic annual plant species in addition to effects on germination (Levine et al 2008; Robinson and Gross 2010). Differences in species responses to variability My two species differed in photosynthetic pathway and root architecture, both traits I anticipated could affect their responses to precipitation variability. I hypothesized that S. faberi’s more efficient photosynthetic pathway would decrease the impact of longer dry intervals on its biomass relative to C. album. However, S. faberi biomass in monocultures declined with longer intervals just like C. album’s, providing no support for the idea that S. faberi’s photosynthetic pathway provided an advantage. Due to the difference in rooting architecture, my alternate hypothesis was that C. album’s taproot would allow it to access soil moisture that S. faberi could not. Previous evidence suggested this possibility (Dalley et al 2006) and in addition work in prairies (Nippert and Knapp 2007) suggested that C4 species, such as S. faberi, utilize mostly shallow water. If this alternate hypothesis were true, C. album’s biomass would not respond to longer intervals while S. faberi biomass declined. However, C. album’s biomass always declined with longer intervals, and its response to precipitation variability in monoculture did not differ from S. faberi’s. The sensitivity of C. album to changes in precipitation variability may be due to 79 the fact that watering treatments did not consistently increase medium or deep soil moisture and my relatively late planting date may have reduced maximum rooting depth. Since species’ monocultures responded identically, my original expectations for how precipitation variability would affect the species differently were not met; however interspecific competition did reveal some differences. S. faberi was unaffected by precipitation variability in competition, while C. album biomass declined with longer intervals. It is possible that the lack of effect of precipitation manipulations on medium and deep soil moisture may be due in part to the two species taking up soil moisture from watering events before it was able to infiltrate deeper into the soil, and that these species differed in the speed at which they acquired resources, similar to my model. Therefore S. faberi in the mixed communities may have been preempting C. album by taking up water from precipitation pulses faster. My modeling results agree qualitatively with my experimental results, providing a potential mechanism behind them (compare Figure 5.3 with Figure 5.4a-b). If C. album — because of its root architecture or other traits — is an affinity specialist, then based on my modeling results I expected the effects of competition on biomass to increase as pulses become larger and less frequent. On the other hand, for S. faberi, as a potential velocity specialist, the effects of competition on biomass should be reduced as pulses get larger and less frequent. In my field experiment, the biomass of C. album decreased more in competition with S. faberi (30%) as intervals lengthened than in monocultures (23%, Figure 5.3a). In contrast, S. faberi decreased some in response to longer intervals in monoculture (12%), but didn’t decrease at all in response to longer intervals in mixtures (Figure 5.3b). This is consistent with my model results for low Rin, assuming the strategies of the two species suggested above. I stress that the model is general and was not parameterized for any particular species due to lack of data, nor does it take 80 into account other potentially limiting resources. However, it does highlight the possibility that the affinity-velocity trade-off may be an important determinant of how plant species respond to water variability, both in monoculture and in mixed communities. Though my soil moisture data showed no evidence of differences between species in uptake rates, my 2-day sampling period may have been too coarse to examine this tradeoff, especially in the field. Further empirical work would be needed to ascertain whether this hypothesized tradeoff actually exists in the field and what combination of traits gives rise to these differences. Variability in the context of total resource availability While I did see a response to precipitation variability in the form of longer intervals decreasing biomass, I did not find the interaction between total precipitation and precipitation variability during the growing season hypothesized by Knapp et al (2008) and seen in my mathematical model. It is possible that the average precipitation input is high enough that a 50% reduction is not sufficient to bring plants past the threshold where they are consistently water stressed. Another complementary explanation is that initial soil moisture acted as a buffer for the plants and so reduced the impact of the reduction in total precipitation. The average surface soil moisture did seem to decline over the course of the experiment in the low, but not the average precipitation treatment (Figure 5.2), which would be consistent with this idea. Similar to changing total water, varying the initial amount of water in my model affected species interactions, though only temporarily. In some cases, high initial water could allow a species to dominate initially even when it was eventually only a small component of the community. The shorter duration of my experiment and the buffer caused by initial soil moisture could explain 81 why I saw an interaction in the model between interval and total precipitation, but not the field results. Overall, my results add to a growing number of studies that demonstrate that intra-annual variation in precipitation has important impacts on plant communities (Knapp et al 2002; Sher et al 2004; Heisler-White 2008; 2009). This study in particular shows that these mesic annual plants are sensitive to variation in precipitation not only during establishment (Robinson and Gross 2010) but also afterward. In addition to direct effects on plants’ growth and reproduction I found effects on species interactions, as has a previous study in an arid ecosystem (Sher et al 2004). I also found that the direct effect of environmental change could be reversed by species interactions as in Suttle et al (2007). Neither of my focal traits explained species’ responses to precipitation variability, but a general tradeoff in water uptake in my model generally agreed with my field results. This tradeoff could be driven by numerous individual traits, which may explain why the two I initially focused on did not explain my results. While my empirical study focused on two species, it demonstrates that mesic annual plants are sensitive to predicted changes in climate but that species specific studies may fall short in predicting responses in a community context. While I cannot make inferences about traits based on two species, my results do show that focusing on one or two broad functional traits when trying to predict species responses to increased precipitation variability may not be informative. Future work on the effects of increased precipitation variability may benefit by going beyond broad functional groupings and focus on species’ overall water use derived from the suite of traits that affect it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�)#*-#)1&++)!$&,")$"0./'0"0)#/)23$$"'#)-'!)43#3$") ,-$&-#&/')&').$"2&.&#-#&/'5) 86 REFERENCES 87 REFERENCES Adler, P.B., HilleRisLambers, J., Kyriakidis, P.C., Guan, Q., Levine, J.M. 2006 Climate has a stabilizing effect on the coexistence of prairie grasses. Proceedings of the National Academy of Science 103: 12793-12798. Adondakis, S., Venable, D.L. 2004. Dormancy and germination in a guild of Sonoran desert annuals. Ecology 85: 2582-2590. Baskin, C. C. and Baskin, J. M. B. 1998. Seeds : ecology, biogeography, and evolution of dormancy and germination. Academic Press Limited, San Diego. Beier, C., B. Emmett, P. Gundersen, A. Tietema, J. Penuelas, M. Estiarte, C. Gordon, A. Gorissen, L. Llorens, F. Roda, and D. Williams. 2004. “Novel Approaches to Study Climate Change Effects on Terrestrial Ecosystems in the Field: Drought and Passive Nighttime Warming.” Ecosystems 7: 583-597. Bohan D.A., Boffey C.W.H., Brooks D.R., Clark S.J., Dewar A.M., Firbank L.G., Haughton A.J., Hawes C., Heard M.S., May M.J. et al. 2005. Effects on weed and invertebrate abundance and diversity of herbicide management in genetically modified herbicide-tolerant winter-sown oilseed rape. Proceedings of the Royal Society B: 272: 463-474. Bouwmeester, H. J. 1990. The effect of environmental conditions on the seasonal dormancy pattern and germination of weed seeds. Ph.D. dissertation, Agricultural University, Wageningen, Netherlands. 397 – 403 . Buhler, D.D. and Hartzler, R.G. 2001. Emergence and persistence of see of velvetleaf, common waterhemp, woolly cupgrass, and giant foxtail. Weed Science 49: 230-235. Burnside, O.C., Wilson, R.G., Weisberg, S., Hubbard, K.G. Seed longevity of 41 weed species buried 17 years in eastern and western Nebraska. Weed Science 55: 74-86. Bussan A.J., Boerboom C.M., Stoltenberg D.E. 2000. Response of Setaria faberi demographic processes to herbicide rates. Weed Science 48: 445-453. Chaves, M.M., Pereira, J.S., Maroco, J., Rodrigues, M.L., Ricardo, C.P.P., Osorio, M.L., Carvalho, I., Faria, T., Pinheiro, C. 2002. How plants cope with water stress in the field. Photosynthesis and growth. Annals of Botany 89: 907-916. Chesson, P. and Huntly, N. 1989. Short-term instabilities and long-term community dynamics. Trends in Ecology and Evolution: 4: 293-298. Chesson, P. and Huntly, N. 1997. The roles of harsh and fluctuating conditions in the dynamics of ecological communities. American Naturalist 150: 519-553. Clauss, M. J. and Venable, D. L. 2000. Seed germination in desert annuals: An empirical test of adaptive bet hedging. American Naturalist 155: 168-186. Dalley, C.D., Bernards M.L., Kells J.J. 2006. Effect of weed removal timing and row spacing on soil moisture in corn (Zea mays). Weed Technology 20: 399-409. 88 Davis, M.A., Grime, J.P., Thompson, K. 2000. Fluctuating resources in plant communities: a general theory of invisibility. Journal of Ecology 88:528-534. Davis, A.S., Renner, K.A., Gross, K.L. 2005. Weed seedbank and community shifts in a longterm cropping systems experiment. Weed Science 53:296-306. Davis A.S., Cardina J., Forcella F. 2005. Environmental factors affecting seed persistence of annual weeds across the US corn belt. Weed Science 53: 860-868. Daws M.I., Crabtree L.M., Dalling J.W., Mullins C.E., Burslem F.R.P. 2008. Germination responses to water potential in neotropical pioneers suggest large-seeded species take more risks. Annals of Botany 102: 945-951. Droop, M.R. 1974. The nutrient status of algal cells in continuous culture. Journal of Marine Biological Association of the United Kingdom. 54: 825-855. Easterling, D.R., Meehl, G.A., Parmesan, C., Changnon, S.A., Karl, T.R., Mearns, L.O. 2000. Climate extremes: Observations, modeling, and impacts. Science 289: 2068-2074. Facelli, JM, Chesson, P, Barnes, N. 2005. Differences in seed biology of annual plants in arid lands: a key ingredient of the storage effect. Ecology 86: 2998-3006. Fay, P.A. and Schultz, M.J., 2009 Germination, survival, and growth of grass and forb seedlings: Effects of soil moisture variability. Acta Oecologica 35: 679-684. Fay P.A., Kaufman D.M., Nippert J.B., Carlisle J.D., Harper C.W. 2008. Changes in grassland ecosystem function due to extreme rainfall events: implications for responses to climate change. Global Change Biology 14: 1600-1608. Grman, E., Lau, J.A., Schoolmaster, D., Gross, K.L. 2010. Mechanisms contributing to stability in ecosystem function depend on the environmental context. Ecology Letters 13: 1400-1410. Grover, J.P. 1991. Non-steady state dynamics of algal population growth: experiments with two chlorophytes. Journal of Phycology 27:70-79. Grundy A.C., Mead A., Burston S., Overs T. 2004. Seed production of Chenopodium album in competition with field vegetables. Weed Research 44: 271-281. Heisler-White, J.L., Blair, J.M., Kelly, E.F., Harmoney, K ., Knapp, A.K. 2009. Contingent productivity responses to more extreme rainfall regimes across a grassland biome. Global Change Biology 15: 2894-2904. Heisler-White J.L., Knapp A.K., Kelly E.F. 2008. Increasing precipitation event size increases aboveground net primary productivity in a semi-arid grassland. Oecologia 158:128-140. [IPCC] Intergovernmental Panel on Climate Change. 2007. Climate Change 2007: The Physical Science Basis. Summary for Policymakers. New York: Cambridge University Press. (19 August 2008; www.ipcc.ch/ipccreports/ar4-wg1.htm) 89 Jentsch, A, Kreyling, J, Beierkuhnlein, C. 2007. A new generation of climate-change experiments: events, not trends. Frontiers in Ecology and Evolution 5: 365-374. Kling G.W., Hayhoe K., Johnson L.B., Magnuson J.J., Polasky S., Robinson S.K., Shuter B.J. et al. 2003. Confronting Climate Change in the Great Lakes Region. Union of Concerned Scientists and Ecological Society of America: Washington DC. Knapp A.K., Fay, P.A., Blair, J.M., Collins, S.L., Smith, M.D., Carlisle, J.D., Harper, C.W., et al. 2002. Rainfall variability, carbon cycling, and plant species diversity in a mesic grassland. Science 298: 2202-2205. Knapp, A. K. and Smith, M. D., 2001. Variation among biomes in temporal dynamics of aboveground primary production. Science 291(5503): 481-484. Knapp, A.K., Beier, C., Briske, D.D., Classen, A.T., Luo, Y., Reichstein, M., Smith, M.D. et al. 2008. Consequences of more extreme precipitation regimes for terrestrial ecosystems. BioScience 58: 811–821. Lauenroth W.K., Sala O.E. 1992. Long-term forage production of North American shortgrass steppe. Ecological Applications 2: 397–403. Leblanc M.L., Cloutier D.C., Hamel C. 2002. Effect of water on common lambsquarter (Chenopodium album L.) and barnyardgrass [Echinochloa crus-galli (L.) Beauv.] seedling emergence in corn. Canadian Journal of Plant Science 82: 855-859. Leuschner, C., Coners, H., Icke, R. (2004) In situ measurement of water absorption by fine roots of three temperate trees: species differences and differential activity of superficial and deep roots. Tree physiology 24: 1359-1367. Levine, J.M., McEachern, A.K., Cowan, C. 2008 Rainfall effects on rare annual plants. Journal of Ecology 96: 795-806. Levine, J.M., McEachern, A.K., Cowan, C. 2010. Do competitors modulate rare plant response to precipitation change. Ecology 91: 130-140. Litchman, E., Klausmeier, C.A. 2001. Competition of phytoplankton under fluctuating light. The American Naturalist 157: 170-187. Lundholm, J.T., Larson, D.W., 2004 Experimental separation of resource quantity from temporal variability: seedling responses to water pulses. Oecologia 141: 346-352 McCarron J.K., Knapp A.K. 2001. C3 woody plant expansion in a C4 grassland: Are grasses and shrubs functionally distinct? American Journal of Botany 88:1818-1823. Miller T.E. 1987. Effects of emergence time on survival and growth in an early old-field plant community. Oecologia 72: 272-278. Nippert J.B., Knapp, A.K., Briggs, J.M. 2006. Intra-annual rainfall variability and grassland productivity: can the past predict the future? Plant Ecology 184:65-74. 90 Nippert J.B. and Knapp A.K. 2007 Soil water partitioning contributes to species coexistence in tallgrass prairie. Oikos 116: 1017-1029. Novoplansky, A. and Goldberg, D. E., 2001. Effects of water pulsing on individual performance and competitive hierarchies in plants. Journal of Vegetation Science 12: 199-208. Pake, C. E. and Venable, D. L. 1995. Is coexistence of Sonoran desert annuals mediated by temporal variability in reproductive success. Ecology 76: 246-261. Pake, C. E. and Venable, D. L., 1996. Seed banks in desert annuals: Implications for persistence and coexistence in variable environments. Ecology 77: 1427-1435. Pimentel, D., Lach, L., Zuniga, R., Morrison, D. 2000. Environmental and economic costs of nonindigenous species in the United States. Bioscience 50: 53-65. Pockman, WT, and Sperry, JS. (2000) Vulnerability to xylem cavitation and the distribution of Sonoran desert vegetation. American Journal of Botany 87: 1287-1299. Rice, K.J. and Dyer, A.R. 2001. Seed aging, delayed germination and reduced competitive ability in Bromus tectorum. Plant Ecology 155:237-243. Robinson, T.M.P. and Gross, K.L. 2010. The impact of altered precipitation variability on annual weed species. American Journal of Botany 97: 1-5. Rosenthal, D.M., Stiller, V., Sperry, J.S., Donovan, L.A. 2010. Contrasting drought tolerance strategies in two desert annuals of hybrid origin. Journal of Experimental Botany 61: 2769-2778. Ross M.A. and Harper, J.L. 1972 Occupation of biological space during seedling establishment. Journal of Ecology 60: 77-88. Schwinning, S. Sala, O. E. 2004. Hierarchy of responses to resource pulses in arid and semi-arid ecosystems. Oecologia 141: 211-220. Sher A.A., Goldberg D.E., Novoplansky A. 2004. The effect of mean and variance in resource supply on survival of annuals from Mediterranean and desert environments. Oecologia 141: 353362. Silvertown J., Dodd M.E., Gowling D.J.G., Mountford J.O. 1999. Hydrologically defined niches reveal a basis for species richness in plant communities. Nature 400: 61-63. Sommer, U. 1985. Comparison between steady state and non-steady state competition: experiments with natural phytoplankton. Limnology and Oceanography 30:335-346. Suttle K.B., Thomsen M.A., Power M.E. 2007. Species interactions reverse grassland responses to changing climate. Science 315: 640–642. Turkington, R., Goldberg, D.E., Olsvig-Whittaker, L., Dyer, A.R., 2005 Effects of density on timing of emergence and its consequences for survival and growth in two communities of annual plants. Journal of Arid Environments 61: 377-396. 91 Venable, D.L., Lawlor, L. 1980 Delayed germination and dispersal in desert annuals – escape in space and time. Oecologia 46: 272-282. Walter H. 1971. Natural savannahs as a transition to the arid zone. In: Ecology of tropical and subtropical vegetation. Oliver and Boyd, Edinburgh, Scotland. pp 238-265. Waide, R. B., Willig, M. R., Steiner, C. F., Mittelbach, G., Gough, L., Dodson, S. I. Juday, G. P., Parmenter, T. 1999. The relationship between productivity and species richness. Annual Review of Ecology and Systematics 30: 257-300. Weltzin, J.F., Loik, M.E., Schwinning, S., Williams, D.G., Fay, P.A., Haddad, B.M., Harte, J. et al. 2003. Assessing the response of terrestrial ecosystems to potential changes in precipitation. Bioscience 53: 941-952. Ziska, L.H. 2000. The impact of elevated CO2 on yield loss from a C-3 and C-4 weed in fieldgrown soybean. Global Change Biology 6: 899-905. 92