Can Dormancy Affect the Evolution of Post-Germination Traits? The Case of Lesquerella Fendleri

Seed dorm'lncy. n hlch 1s thought to h a \ e e \ o l \ e d In response to unpredictable environmental hariability. has led to the existence of seed banks-populations of dormant, viable seeds in the soil. Seed banks are theoretically important to both the demography and genetic structure of plant populations. The presence of seed dormancy can also affect the evolution of traits not directly associated with dormancy and germination. Theoretical models hahe suggested that the existence of dormancy can influence the rate of evolution of post-germination traits. The ehentual outcome (e.g. . allele frequencies) may be influenced as well. leading to adaptihe syndromes of germination and post-germination traits. Seeds that germinate in different conditions may experience different selective regimes for post-germination traits. If there are trade-off5 between the fitness of post-germination traits in different environments. then seeds that germinate in the en\ ironment to u h i c h their poat-germination traits are adapted will be at a selectihe advantage. If differences in germination and post-germination traits are genetically based. then potentially adaptihe genetic correlations between germination and post-germination traits may eholhe. We feel that inhestigating the ecological and evolutionary importance of these correlations requires an empirical approach. '4s a first step, here n.e ask whether the conditions necessary for such syndromes to arise exist in a particular plant population. We show that conditions favoring the joint evolution of dormancy and postgermination traits leading to a d a p t i ~ e syndromes exist in the mustard, Le.rquc~rell~r ,fi.t7tflet-i,in central New Mexico. First. Lesq~tet-elltrexperiences the sort of hariation in environmental conditions that would be expected to lead to adaptihe trade-offs in the expression of post-germination traits for individuals that differ in germination traits. Annual precipitation varies greatly frc)ni year to year so that germination in drier years would be expected to select for more xerophytic traits. Within a year. microen\ironmental spatial variation exists. Le.~quc~rc~lln growth and I Man~~scr ip t rece~ved28 October 1993: revised 22 August 1993; accepted 29 Augu\t 1993. DORMANCY AND POST-GERMINATION TRAITS reproduction are sensitive to both year-to-year and niicroen\ ironmental \ariation. Second. the seed bank can affect both the demographic and genetic structure of the population. Dormant seeds remain viable for at least 3 yr and can mitigate the negati1.e demographic effects of r e p r o d u c t i ~ e failure. A l l o ~ y m e differences exist between seeds that germinate in the field and seeds that remain dormant. suggesting that the e \ o lutionary potential of the aboveground population is influenced by dormancy. Finally. the necessary genetic and en\ironmental hariation is present. Both germination percentage and post-germination traits (e.g. . leaf niorphology) \ a r y a n i ~ n g and within populations as well as among environmental treatments. Thus. the potential exists for Le.rcjuc~rc~lln to respond to selectihe differences between different temporal or spatial en\ ironnients. We suggest experimental approaches for asse\\ ing the extent to which seed dormancy has affected or will affect the eholution of post-germination traits. The consequences of past e~.olut ion could be explored by asking "What genetic and phenotypic differences exist between indihiduals that germinate and those that remain dormant?". u h i l e controlling for factors that influence germination (maternal genotype. maternal environment. and germination en\ ironment). Exploring u hether evolution is currently occurring would require an assessment of natural selection and the genetic potential for response to selection. G i ~ e n the difficulty of such studies. a reasonable fir\t step would be to explore how eholution can occur by performing an artiticial selection experiment on dorniancy or germination percentage. Then, correlated responses of post-germination traits could be examined. Empirical studies such ax these are necessary in order to better understand the role of seed banks in plant ecology and e \olut ion. Then. once a soc ia t ions that can be interpreted as a d a p t i ~ e syndromes are documented. questions about the ecological mechanisms (e.g. . necessary frequencies of different year types) and genetic mechanisms (e.g. . linkage disequilibrium v\ . pleiotropy) can be explored. We hope to d rau attention to seed dormancy. which is an often-ignored stage in the life history of plants. and to encourage empirical work. which lags far behind theory. Introduction Seed dormancy in higher plants is a means of a ~ o i d ing unfavorable environmental conditions by arresting g rou th and development. Many other diverse organisms h a t e similar dormant stages (e.g.. e n c y t i n g in dinoflagellates. Binder and Anderson 1990: egg dormancy in copepods. DeStasio 1989). All such forms of dormancy are thought to ha1.e e i .o l ted in response to environmental variability and uncertainty. With respect to seeds. there are two types of dormancy. Within a year. dormancy serves to delay germination until a time more favorable for seedling establishment and growth, while between-year dormancy may balance the risk of local extinction from gerniination in unCahorable years with the risk of missing good years by remaining dormant ( S i l ~ e r t o u n 1988). The u ide spread occurrence of between-year dormancy has led to the ex~s tence of seed banks-large popula t~ons of dormant. able seed\ In the s o ~ l ~ n niost o t the uor ld ' s major ecosy5tenis ( tor rehleus see Leck et a1 1989. T h o m p w n 1992) W h ~ l ea f e u recent \ t u d ~ e s hahe begun to emplrlcall) demonstrate the eco log~ca l and e \o lu t~onarq ~ n i port'lnce ot seed bank5 (Kallsz 'lnd McPeek 1992. McGr'lu 1993. Tonsor et a1 1993). much ot our conceptual understandme remaln5 b'lsed on mathern'lt~cal models 'lnd theoret~cal pred~ct lons ( e g . Cohen 1966. 1967, Tenipleton 'lnd Le\ In 1979. Venable and Lau lo r 1980. Venable and B r o u n 1988) Theoretrcallj. \ l a d~\pers,ll through tlme. a seed bank can ( I ) reduce the probability of extinction o \ e r the long term. (2 ) 'ldd age-structure. and ( 3 ) c h ~ n g e the genetlc structure o t a population A seed bank reduces the p r o b a b ~ l ~ t ) ot extlnctlon by p r o ~ l d l n g temporal mlgr'ltlon Age structure 1s added because a seed b m k introduces o\.erlapping generations. A seed bank can change the genetic structure of a population by acting a5 a genetlc meniorq ( s ~ n c e the \eed bank contains seeds produced in different years and therefore ~ ~ n d e r different e n ~ i r o n m e n t a l and selection regimes). Seed dormancy also has the potential to affect the e \o lut ion of traits not directly associated with dormancy and germination through genetic correlation due to linkage and/or pleiotropy. Again. there are numerous theoretical treatment5 ot thi5 Issue ( e g , Tenipleton and Le\ In 1979. Venable and Lau lo r 1980. R ~ t l a n d 1983. B r o u n and Venable 1986. Klinkhamer et al 1987. Venable and B r o u n 1988, Venable 1989). but t e n e m p ~ r ~ c a l studies Before focusing on how dormancy might affect the evolution of post-germination traits. we briefly revieu h o u \ arlable dormancy n i ~ g h t e \ o l \ e Theoret~calconsiderations assume that maternal plants can reduce the risk of reproductive failure of their seed crops by spreading germination o \ e r time (reducing the hariance In h tne\s among years) Seeds can e \ o l \ e to "predict" f a ~ o r a b l e germination periods if there are en\ ironmental cues that are correlated with favorable growth and establishment conditions and if plants can e \ o l \ e to u \e those cues to break dorm'lncy (re\ l e u e d by Mayer 'lnd Polj'lkott-Mayber 1975. B'lskln and B a s k ~ n 1989) A s the a b ~ l ~ t y o predlct t ' ~ \o rab le peANN S EVANS AND ROBERT I CABIN Ecolog). Vol. 76. No. 2 riods increa\es. uncertainty is reduced. and germination behahior is expected to become le\s variable (Cohen 1967). If e n ~ i r o n ~ n e n t a l conditions could be perfectly predicted. we might expect either complete dormancy or complete germination within a population experiencing a gihen germination environment (Silhertown 1988). Yet environmental conditions typically vary considerably in time and space. both between and ui th in habitats. Thus, it i \ plausible that seeds within a population, as ue l l as different populations of a gihen species, hahe evolved d~f fe ren t germination requirements based on their particular enhisonmental and evolutionary history. How might the existence of dormancy ~lf fec t the evolution of post-germination traits'? Models by T e n pleton and Levin (1979) suggest that for traits whose fitness varies w ~ t h the environment. the exl\tence of a seed bank in f luence the rate of e l olution: traits that confer fitness in good environnients evolhe faster than traits that confer fitness in poor environnients. Similar models by Brown and Venable (1986) suggest that the evolution of germination and post-germination traits will have a damping effect on one another. 5ince they are \ iewed as alternatihe means of decreasing variance in fitness among years. These rnodels emphasi le that a seed bank can permit a population to escape the need to respond adaptively to selection for post-germination traits in poor years by specializing on good years. We wish tc) consider adaptation to both good and poor years and suggest that the joint evolution of dormancy and post-germination traits may lead to adaptive c o n binations of germination and post-germination responses ( "syndromes." Venable 1989). The scenario u e imagine can be illustrated as follows. In desert habitat\. the amount of precipitation varies considerably froni year to year. Seeds that germinate in different conditions experience different selective regimes for post-germination traits. In relatively dry years. individuals u i th more xerophytic traits ( e . ~ . . greater water-use efticiency) should have greater reproductive success. while. in relatively wet years, individuals u i t h more nie\ophytic traits (e.g. . I o i ~ e r water-use efticiency due to spending water more profligately) should have greater reproductive success. Thus. a trade-off exists in the expre\sion of a trait that confers adaptation to environnients that differ in u a t e r availability. Given trade-offs betueen the fitness of post-germination traits in different environnients. seeds that germinate in the environment to which their post-germination traits are adapted will be at a selective advantage. If differences in germination and postgermination traits are genetically based. then different syndromes of germination and post-germination traits that are adaptive in different environments can potentially come about through the eholution of increased genetic correlations betueen germination and postgermination traits. Thus. we might expect plants with xerophytic adult traits to require less u a t e r to break dormancy. An example of an adaptihe syndrome involving dormancy is the desert annual Herc~rothecclI t r r i fo l ic~.Disc achenes of this composite species were more successful (in terms of sur \ ival and biomass accumulat ion) than ray achenes when grown under a regular watering regime (Venable 1985). However. ray achenes ultimately achie \ed greater success under simulated drought conditions by exhibiting greater dormancy during the drought periods. Venable and Levin (1985) also found that only ray achenes formed a between-year seed bank. This is an example of a \pecie\-leh el somatic seed polymorphism: all indih iduals produce both seed morphs. Such polyniorphisnis are widespread. and generally have been regarded as a forni of bet-hedging by maternal plants (Si lver toun 1983. Venable 1985). While the existence of somatic seed syndromes is intere\ting in itself. we wi\h to focus c)n uhe the r genetic syndromes occur. It ls not clear uhe the r Intraspecific germination differences among seeds ui thout obvious polymorphisms (i .e. , u i t h continuous variation) are due to genetic factors, enhironmental maturation effects, or interactions of these and other forces. Determining to what extent these differences are adautive can be tested empirically. although to date no studies have examined continuously varying post-gerniination traits (e.g.. uater-use efticiency) that have presumed adaptive significance in particular environments (but see Ritland and Jain [I9831 for a compari\on betueen two species). A tirst step in determining the importance of seed dormancy in the evolution of other traits is to ask u h a t conditions could favor joint evolution of dormancy and post-germination traits that leads to syndromes. First, environmental variation must be of the sort that mould be expected to lead to trade-offs in the expression of po\t-germination traits. As illustrated by the models of Templeton and Levin (1979). unless fitness varies acres\ environnients. the existence of a seed bank h i l l affect only the rate of evolution of a postgermination trait. and not its equi l ibr iu~n frequency. Both spatial and temporal variability should be iniportant in the joint evolution of dormancy and postgermination traits (Venable 1989). Second, a persistent seed bank mu\ t exist that has the potential to significantly affect the demography and genetic structure of the aboveground population. While even one seed remaining viable in the soil between years could conceivably rescue a population froni extinction, the evoM a c h I995 DORMANCY AND POST-GERMINATION TRAITS 347 lution of adaptive syndromes is more likely in populations in which the seed bank is known to be persistent and to play a role in delnographic security. Evidence that seeds with different germination requirenients differ genetically uou ld suggest that dormancy influences the evolutionary potential of the aboveground population, since the aboveground population that is subject to selection on post-germination traits in the current environment is not a random \ubsample of all available seeds. Third. there must be genetic and environlnental variation for dorlnancy and post-ger1ninatiol1io traits. Trait\ n ~ u \ t be environmentally sensitive (pla\t ic) in order for selection to act on trade-offh. and must be genetically variable in order for an evolutionary response to selection to occur. Here u e explore these conditions in the desert niustard. Le.rq~cerelltr ,fendleri, and discuss h o u the potential for dormancy to affect the evolution of post-gerniination traits niiglit be tested enipirically. Methods The rrilc1.1 .c.jsterlr Lr.sq~trrrllrrfrt~cilrri (Brassicaceae) is a self-incompatible. short-lived perennial native to southwestern Nortli America. It ranges from western Texas through Neu Mexico and eaatern Ari lona into northern Mexico at elevation\ betueen 600 and 1800 In in calcareous. handy \oils (Rollins and Sliaw 1973). Germination occurs primarily in the late winter to early spring (February to March). but a second flush of gernlination may occur in the fall if rainfall is sufficient. F louer ing usually occurs between March and April and ]nay also occur fo l louing suitable rains in the fall. Plants may produce up to several hundred fruits, with up to 30 seeds per fruit. We have been studying Lt~.c.yilt~rt~llu at the Sevilleta National W~ld l i f e Refuge (NWR). 80 km south of Albuquerque in central New Mexico. The Sevilleta NWR. a Long-term Ecological Re\earch site. is one of the few large areas in the Southwest protected from i~unian disturbances; cattle grazing and human uses other than research have been prohibited for allnost 20 yr. At the Sevilleta. Lr.c.yilrrrlltr occurs in ecolog~ c a l l y diverse habitats, including creosote (Ltrrrerr triderlrtrrcr) shrublands, grasslands. and more open. sandy uaslies. Our main study slte (Five Points) is dolninated by creosote shrubs and largely open. barren patches in the interspaces between the shrubs. Sptrtitrl trnd ter?zj~ortrl en~~iror~rnentr r l rurirrriorz To evaluate the potential for dorlnancy to affect the evolution of po\t-germination traits. u e employed a nuniber of observational and ~nanipulative studies. To assess telnporal environlnental variation, u e used n~on th ly records of precipitation for 63 yr from Albuquerque and for 2 yr ( 199 l and 1992) from a heather station = 3 kln from the Five Point\ study site. To assess spatial variation created by the ~nosa ic of shrub canopies and open spaces. light intensity and soil temperature, moisture. and organic Inatter h e r e measured. Light and temperature data mere taken from 19 creosote sub-canopies and adjacent interspace locations at the Five Points study site on 5 August 1992. Photo\ynthetically active radiation was measured using a light meter (Model LI-6200. LI-COR. Lincoln. Nebraska. USA) at 10 c n ~ above the soil surface. uh ich corresponds to average Lr.c.y~trrrlltr height in this population. Soil temperatures were measured at the surface and at I 0 cln depth, since u e have obaerved that the bulk of Le.cyilerelltr roots are typically u i th in 10 cln of the \oil surface. Soil samples for moisture and organic Inatter h e r e taken at the surface and at 10 c m depth under 20 creosote canopies and adjacent open interspacea on 25 June 1993 at Five Points. Soil nioisture was calculated as the percentage difference in mass betueen fresh and dried samples. Percentage organic matter was calculated from fresh and combusted dry masses. The surface samples include litter as ue l l as soil organic Inatter. The ilnportance of spatial and temporal environnlental variation for Lr.scluerelltr perforlnance u a s evaluated by measuring several demographic paranieters beneath creosote canopies and in the interspace\ betueen the ahrubs in 1991 and 1992. For the subcanopy microhabitat. 25 1 -m2 plots. each centered on a creosote shrub. h e r e sampled. For the interspace ~nicrohabitat , salnpling u a s done along a 5 0 ~ n transect through the study site. Along this transect, 1-ln' quadrats on both sides of the line mere sampled using only I -n12 positions with <10% shrub cover. for a total of 38 1-111' plots. The density of Lesq~terellrr plants and the mean nuniber of Lr.ryurrrlltr seeds produced per square metre were also estimated. Lrsqurrrl l tr .seer/ hurzk To investigate the importance of the Lrsqurrrl l tr seed bank, we assessed the density and longevity of the seed bank. and compared the genetic structure of the seed bank with that of the seedling population. Previous work has s l ioun that there are significantly Inore Le.c.yilerelltr seeds in the soil under creosote canopies than in the interspaces (R. Cabin. i ~ r z p i ~ l ~ I i . ~ l ~ t ~ ~ / tlrrttr). We subsequently focused our attention o n the Inore dense seed bank under creosote canopies and collected 10 aoil \ample\ ( I 0 X 10 X 2 cm) randomly from a 1-111' plot under each of 30 creosote shrubs prior to Lr.rqurrrlltr seed maturation in 1991. These 300 samples h e r e sieved and searched under a dis348 ANN S. EVANS AND ROBERT .I. CABIN Ecology. Vol. 76. No. 2 secting micro\cope for ally Le.scl~cerellerseeds from previous year\ . To d i x o v e r whether seeds that germinate in the tield differ genetically from seeds that remain dorniant under the same environmental conditions. Le.c.quer.rll(i genotypes were c l iaracter i~ed by employing htarch gel electrophoresis. All seedlings that germinated in the tield within 30 1 -m2 plots under c reowte in the fall of 199 I were collected for electrophore\is using five en/yme systems and seven polyn~orphic loci. Once germination in the field had ceased. 10 soil sample5 ( 1 0 x 10 x 2 cni) were collected from each of the \ame plots. spread in a thin layer over potting mix in the greenhouse and treated with a 0.1 glkg gibberellic acid solution known to break dorlnancy (Sharir and Gel~ n o n d 1971. R. Cabin and R . Mitchell, ur7puhli.shrri tlrrtrr).All emerging Lr.squrrrll(i seedlings from these samples were collected for electrophore\ih as aboce. We are confident that this technique recocers most of the \ iable dormant Lr.squrrrIl(i \eeds. since the nuniher of \eeds recovered via germination does not differ signiticantly from the number recocered by searching centage was quantified. and maximum plant diameter. and width and length of the most recent fully expanded leaf were nieasured u i t h digital caliper\ (Series 550. Mitutoyo Corporation. Tokyo. Japan) . T h e w measurements h e r e made 9 u k after planting. when the plants mere still in the seedling stage. To u s e \ \ within-population \ariation in germination percentage. field-collected sibsliips from seven maternal plants h e r e used. Because we d o not know the paternity of these seeds. u e cannot tease apart maternal and embryonic control\ on germination (see Disc~ucti or^). For each of the seven sibships. 9 0 seed\ h e r e scored for germination percentage under ambient conditions in the greenhouse. Finally. maternal. paternal. and en\ironmentaI sources of variation within a population in post-gern~inat ion traits were ecaluated as fo l lou \ . Pollinations h e r e performed on plant\ grown from field-collected seeds (from a different population and year than aboce) as parents utilizing a 7 X 7 diallel breeding design ui thout self-pollinationr. Progeny h e r e treated with gibberellic acid. germinated in a common enci(ItitcO. ronlnent. and four plants per family were g r o u n in under the microscope (R. Cabin. i ~ r ~ ~ ~ i ~ l ~ I i . ~ l ~ t ~ d Grr~rric, rrrzri rrz\,irorzrtlrr~rrrl \,trrrcrtior~ ,fi)r gerrtlirlrrrrorl trr~d 17o.ct-gert~1irlcltiorz tr(1ir.s To assess the degree of genetic and encironniental \ ariation for dormancy and pmt-germination traits. u e used three experiments. one of which explored interpopulation cariation. and the other t u o of which estimated intrapopulation cariation. First. interpopulation cariation in gerlnination and post-germination traits was ~neasured by exposing populations to different encironmental treatments in the greenhou\e. Second, intrapopulation \ariation in germination beha \ io r u a s a s x s s e d among maternal plants. Third. intrapopulation cariation in post-gerlnination traits (including maternal. paternal. and environmental sources) mas e \ aluated. Interpopulation cariation was ecaluated in the following manner. Seeds from fice population5 representing the range of habitats at the Seeilleta NWR h e r e exposed to five germination treatments in the greenhouse to simulate the range of encironniental conditions exper~enced by Lesyiler.rll(l plants in the field. For each population. 5 3 seeds mere planted in sand in the greenhouse in each of the following treatments: control (ad libitum uater ing) , l ou u a t e r (watering reduced by half). c r eowte (creosote litter on the soil \urface and control uater) . low water and creosote litter. and shade (light intensity reduced by half with shade cloth and control ua t e r ) . We did not measure soil moisture but, as expected, creoiote litter and shade appeared to reduce e \aporat ion. Germination pereach of t u o uater ing treatments. Plants in the low uater ing treatment were uatered half as frequently as plants in the high uater ing treatment. Plant diameter. leaf length. and leaf width were measured as aboce. but at a later stage in ontogeny (3 nio after planting. just prior to f louer ing) . Because gibberellic acid u a s used to germinate progeny in a common en \ ironment. u e cannot examine correlations betueen germination percentage and post-germination traits in different en-