Oecologia (2004)' 14] : 7-16 DOI 10.1007/s00442-004-I 621-4

•I.-C. Domec • J, M. Warren • F. (2. Meblzer. ,L R. Brooks. R. Coulombe Native root embolism and stomatai closure in stands of Douglas-fir ponderosa pine: mitigation by hydraulic redistribution

Received: ! Octobe', 2003 /Accepted: [4 May 2004 / Published online: 3[ July 2004 :~: Springer-Verlag 2004

MJstract t-tydraulic redistribution (HR), the passive kept at 20-30 cm depth above moven~ent of water via roots from moist to drier portions -0.5 MPa in the old-growth Douglas-fir site and of the soil, occurs in many , influencing both -1.8 MPa in the old-growth ponderosa pine site, which plm~t and -water use. Vv'eexamined the effects of significantly reduced loss of shallow root function. In the lqR on root hydraulic functioning during drought in young young ponderosa pine stand, where little HR occurred, the and old-growth Douglas-fir [Pseudot~uga menziesgi water potential in the upper soil layers felt to about (Mirb.) Franco] and ponderosa pine (Pimls ponderosa -2.8 MPa, which severely impaired root functioning and Dougl. Ex Laws) trees growing in four sites. During the limited recovery, when the fall rains returned. In both 2002 growing season, in situ xylem embolism, water species, daily maximum gs decreased linearly with deficit and xylem vulnerability to embolism were mea- increasing root PLC, suggesting that root xylem embolism sured on medium roots (2-4-ram diameter) collected at acted in concert with stomata to limit water loss, thereby 20-30 cm depth. Soil water content and water potentials maintaining minimum Ieaf water potential above critical were monitored concurrently to detemaine the extent of values. HR appears to be an important mechanism for HR_. Additionally, the water potential and Stolnatal maintaining shallow root fianction during drought and condvmtmlce (g.~) of upper canopy leaves were measured preventing total stomatal closure. throughout the growing season. [n the site with young Dougias-flr ~rees, root embolism increased from 20 to 55 Keywords Cavitation • Hydraulic conductivib,. percent toss of eonductiviW (PLC) as the dry season Hydraulic lift • Stomatal regulation. Water stress progressed. [n young ponderosa pine, root embolism increased from 45 to 75 PLC. in contrast, roots of old- growth Douglas-fir and ponderosa pine trees never introduction experienced more than 30 and 40 PLC, respectively. HR Hydraulic redistribution (HR) involves transfer of water .1.-C Domec (:~i) from wetter to drier portions of the soil profile via roots Department of" Wood Science and Engineering, Oregon State (Richards and Caldwell 1987; Caldwe]l and Richards University, Corva!!is, OR, 97331, USA [989). Although HR has been reported in a wide varieb~ of e-mail: [email protected] ecosystems (Burgess eta!. !998; Jackson eta!. 2000), our Tel.: -:-1-541-7374329 understanding of its significance for functioning is Fax: +1-541-7373385 still incomplete. Estimates of ecosystem-level fluxes of J. M, Warren. F. C. Meinzer hydraulically redistributed water are few, but some reports USDA Forest Service, Forestry Sciences L~Ct)orato,,T. point to values of the order of only 0.1-0.2 mm day -~, 3200 SW Jefferson \Vay, raising questions concerning tl~e role of HR in enhancing Corvaliis~ OR, 97331, USA of some vegetation b~pes @4einzer et al. 2004; Moreira et at. 2003). Nevertheless, recent steadies .1. R. Bmnks \Vestem Division, US EPA,..~rF[EERL, suggest that as soil water deficits increase, the relatively 200 SW 35th St., small nocturnal fluxes of water associated with HR are Corvallis, OR, 97333, USA sufficient to significantly delay further rifting of the upper portion of the soil profile by replacing most of the water R. Couiombe l)y~;an~ac Corporation, utilized during the day (Brooks etal. 2002; Meinzer etal. 200 SW 35th St., 2004). This nearly complete overnight replenishment by Corvallis, OR, 97333~ USA HR of soil water sunounding shallow roots could have a signi[icant impact on their continued tb.nctioning during hydraulic conductivity through embolism could lower drought, and therefore onrhizosphere processes associated root water potential resulting in the generation of a with the uptake and transport of water and nutrients and hydraulic signal that reduces stomatal conductance (g~), maintenance of mycon'hizal symbioses (CaldwelI et al. and therefore transpiration, to maintain shoot water i998; Querejeta et al. 2003). However, no studies have potential at a nearly constant minimum value above g)c~it specifically assessed relationships bmween vulnerabilit "of (Cochard et at. 1996). It is likely that isohydric behavior, roots to water stress-induced embolism and seasonal the maintenance of a nearly constant minimum leaf waler patterns of HR and root xylem dysfunction in situ. potential independent of soil or root water status, is linked Water flow along the soil-to-leaf continuum is governed to an interaction between both chemical and hydraulic by a combination of the driving forces and the hydrauIic information (Tardieu and Davis i993). A number of tree properties of the pathway (Mencuccini 2003). Typically~ species exhibit at least partial isohydric behavior during 50% or more of the.total resistance to water flow occurs gradual soil d~,ing cycles (Gollan etal. 1985), but the betowground (Tyree and Ewers 199t). During drought extent to which stomatat regulation of leaf water status is periods, HR could influence two potential hydraulic weak determined by shoot versus root mid rhizosphere water points along the belowground portion of ttle continuum: stares and hydraulic properties is largely unexplored. the root-soil interface where steep soil water potential \Ve monitored-seasonal changes in root embolism, HR, (")~il) gradients create dl)~ non-conductive zones (Oertli plant and soil water status and gs during and after the i 996), and the root xylem where tension can reach critical summer drought period inyoung and old trees growing in values (g-J~,-i~), leading to embolism and interruption of the two contrasting Pacific Northwest fore.~t ecosysterns: a continuum (Sperry et al. 1998). Previous investigations ponderosa pine (Pinus pm'~derosa Doug. ex Laws.) have demonstrated that roots are not only more vulnerable ecosystem receiving an average of 500 mm of precipita- to embolism than stems, but also operate closer to i]~'crit tion mmually, and a Douglas-fir [.Pse~de~ts'uga menziesii ~hau stems (Hacke and Sauter 1995), so small reductions (Mirb.) France] ecosystem receiving 2,500 mm of precip- in ~).~oiB can significantly enhance root embolism rates. itation. \Ve hypothesized that HR plays a critical roIe in Further, fine (<2 mm diameter) and medim~l (,2-3 mm maintaining the water transport capacity of shallow roots diameter') roots are more vulnerable to embolism than during summer drought and that embolism-induced loss of coarse (>3 diameter)-roots (%peri3, and Ikeda 199.7). Loss root conductivity influences stomatal regulation of upper oF functional xylem due to embolism prevents root water canopy leaf water status. Our specific objectives included uptake and-reduces whole-plant hydraulic conductance. (1) examining relationships between root embolism, HR Field studies have shown significant embolism in root and p!ant and soil water status: (2) assessing age- and xylem of woody during drought with recovery species-specific variation in the vulnerabiliW of roots to occurring only following rain (Jaquish and Ewers 2001; water-stress-induced embolism and whether differences in Davis et al. 2002). Therefore, root embolism may be vulnerability are reflected in seasonal patterns of embolism irreversible during the dry season (Borghetti et at. 1991), in situ, and (3) identifying potential linkages between which col!icicles with the growing season in many stomatal behavior and the degree of embolism in shallow temperate ecosystems. roots. Because the roots and rhizosphere m'e likely to contain the most vulnerable components of the soil-to-leaf hydraulic pathway (Jackson et aI. 2000), avoidance of Materials and methods b.ydraulic fSiNre through passive leakage of waer fi'om shallow roots into drying soil may play a major role in the Plant material and field sites success of species growing under a wide range of The study was carried out from June ~oDecember 2002 in four precipitation regimes, especially those characterized by stands: one dominated by old-growth (280-year-old) and interme- summer drought (Gholz 1982). Partial Ioss of root diate age (52-year-old) ponderosa pine, one dominated by young

Table 1 Characteristics of the Ponderosa pine Douglas-fir ponderosa pine and DougIas-fir sites Old-growth Young Old-growth Young Mean annual precipitation (ram) 525 550 2,500 2,500 Mean summer prec~pitat~on.I , . (ram) 35 33 250 250 Mean annual temperature (°C) 7.7 7.5 8.7 8.6 At the old-growth ponderosa Mean summer temperature (°C) 18 19 16 15.5 pine site, both old (o) and Age (years) 280 (o), 52 (I) 16 450 24 intermediate (I) age trees were Mean height (m) 36 (o), 16 (I) 3.3 60 17 studied "Law et all (1999) Soil classificationa' b Alfic Vitrixerands Uttic Haploxeralf Entic Vitrands Entic Vitrands bhttp://depts.washington.edu/ Soil texture~" u Sandy loam Sandy loam Sandy loam Sandyloam wrcorl7 sand/silt/clay 73/21/6 65/25/10 . 65/18/17 65/18/17 ~Irvine et al. (2002) ePhillips etal. (2002) Stand Leaf areac' ~ index 2. i 1.0 9.0 6.0 9 (15-year-old) ponderosa pine, one dominated by old-growth (450- pressures ranging from 0.5 to 4.0 MPa, or until the conductivity of year-old) Douglas-fir and one dominated by young (24-year-old) the seNncnt was negligible. Douglas-fir (Table I). The old-growth and young ponderosa pine stands are located in the Metolius River region of Oregon (44%0' N, 121037 , W) at an elevation of 915 and 1,200 m, respectively. The Relative water content young pondcrosa pine stand was previously an old-growth stand that had becn harvested in 1978. Tile old-growth and young Douglas-fir stands are located within the \Vind River Experimental Forest near The relative water content (RWC) of each sample used ~or the Wind River Canopy Crane Research Facility in southcm detennination of native embolism and xylem vulnerability was Washington (45°49 ' N, 121°57 ' W) at an elevation of 370 and calculated as 560 m, respectively. The young stand was planted after a clear-cuL whereas tile old-gqowth stand regenerated naturally after a stand- replacing fire. Altlnough mean annual precipitation is about RWC = x 100 (l) 2,500 ram, this region has a Pacific maritime climate with dry (Ms~,, - M;) summers (<120 mm precipitation between June and September). where Me, Md and ;~.~, are file root fresh mass, dry mass, ahd mass at frill saturation, respectively. For some analyses, root water content was exl.~ressed as relative water deficit (RWD=t00-RWC). To :Native root hydraulic conductivity and embolism characterize the relationship between RWC and cmbolism, if(s:,t was determined initially after vacuum infiltration in perfi:sion solution to Specific hydraulic conductivity (k, kg m -I s-I Mpa-1) and remove emboli, then M t- was recorded after each pressurization embolism were measured ia roots collected four to six times at during generation of vulnerability cur,'es, and/~4;i was dctennined each site between June and December 2002. Summer sampling was by oven-drying the sample following the final pressurization. Native approximately monthly, wifn a final sampling after the fall rains had RWC was also detemlined on 12-20 fine roots (1.1--l.8 mm rowel the . Because of the distance between sites, it was not diameter) collected at each site concurrent with sampling of larger possible to sample at all sites on the same day. In the emly morning roots for determination of native embolism. of each sampling (fate, lateral reed[urn roots 2--4 mm in diameter and greater than 25 cm in length were excised near the base of fix,e di[:fcrent trees at about 20-cm depth in the soil. Before excision, Soil water potential and volumetric water .content roots were checked to insure they were attached lo the target tree. Roots were wrapped in aluminum foil and placed in a sealed plastic bag with a moist towel and immediNe[y transported to the Soil water potential (~I~oil) was measured using individually labora!on.; in a cooler, to be processed within 3 h of sampling. calibrated screen-cage thennocouple psychrometers CPST-55, Wes- In the laborator3,, a 10-i 5 cm long section of each root ,.,.'as cut, cot, Logan, Utah, USA) installed between May mid June 2002 at and the bark and cambium were removed from the entire sample. multiple depths in three to four locations at each-site. Measurements The proximal end was then attached to a tubing system and pert\tsed were recorded evcv 30 rain with a data logger (CR7X, Campbell at a pressure of 6 kPa with filtered (0.22 txm) solution ofHCi (p[t 2). Scientific, Logan, Utah, USA). Means of measurements made at 20 The hydratflic pressure ]lead was actjusted to avoid any refilling of and 30-era depth are reported here because these d~pths encom- embolized tracheids. "['he rate of efflux was measured in a l-ml- passed the mean depth at whicl~ root samples were collected. Soil graciuated pipette. When the flow was stead?', the time required for volumetric water content (0, m" m -~) was measured at khe same lhe meniscus to cross five consecutive graduation marks (0.5 m[) depths using annular frequency domain capacitance sensors (Sentek, was recorded. Specific conductivity was calculated as the mass flow Adelaide, Australia) placed in PVC access robes (Pakineanu and rate of the pc;-f\!sion solution divided by the pressure gradient across Start 1997) installed in the vicini b, of the soil psychromctcrs. Values tl~e root segment, normalized by the xylem cross-sectional area. of 0 were recorded ever3,' 10 rain with a data lo~zer (RT6, Sentek ~'-iative embolism was determined by comparing the initial (or Adelaide). The total daffy" water use (tam day. -~) m the 20-a0" cm" native) conductivity (/csq~) of .root segmm~ts to the maximum layer was calculated as the difference between the daily maximum eondue,dvity (/c,O,~,~) after rcmoval of air emboli by soaking the and minimum 0, and net water use 0?am day-I) was calculated as the samples in perfl.~sion solution under vacuum for 48 h. Percentage diff.~rence between the maximum 0 between 2 consecutive days. HR loss of conductivity (PLC) was computed as PLC~lOO(l-ks(b/ was calculated as the difference between the total and the net water lc,

Root xylem vulncrabilit), to embolism Tree water potential and stomatal conductance The air injection method (Spe~y and Saliendra 199'-1) was used to . measure vnlnerabiliD' of root xylem to water-stress-induced ombo- Predawn and midday water potentials of upper canopy leaves (~L) iism. The air-injection method is based on the assumption that the were determined on individual needles (ponderosa. pine) or twigs xylem tension required to pull air into a conduit and cause embolism (Douglas-fir) with a Scholander-type pressure chamber (PMS, is equa! in magnitude to the positive air pressure required to push air Corvallis, Ore,: USA). Ware/ potential values were corrected to into the conduit when the xylem water is at atmospheric pressure. ground level using the height at which the sam-Oies ~ycre obtained Root segments were collected in the field as described above, and air and a standing gravitational gradient of 0.01 N'fPa m -~. [n addition, emboli w~re removed by soaking lhe samples in perfusion soImion daily mean maximum values ofg s of exposed, upper to mid-canopy under vacuum for 48 h. The segments wcrc inserted into the air- foliage were determined with a null balance porometer (LI-1600, Li- ir!iecfion chamber with both ends protruding and attached to the Cor, Lincoln° Nob, USA). Maximum values ofgs usually occurred tubing system for measuring ks as described above. A vuh~erabiiity between 0900 and 1100 hours when fl~e mean vapor pressure deficit curve was generated by first pressurizing the air chamber to was 1.4-0.I kPa and 2.5±0.2 kPa in the Douglas-fir and ponderosa 9.05 MPa Io avoid water extrusion from lateral root sere:s, and pine stands, respectively. Water potentia! and gs were measured on al!o\vin~, tile system to equilibrate for 3 rain. Water flow through the the same day that roots were coliecled for de~erminatidn of nadve root v,,a,~ initiated and /':~.~ax~ was measured as described above. A cmboiism on three individual fascicules or twigs fiom three pressure or'0.5 ..MPa was then applied and ileld constant for 2 mill. ponderosa pine and four Douglas-fir trees. Canopy access was After equilibradon, the air cl~amber pressure was reduced to provided by towers in the 24-year-old Douglas-fir stand and fl~e old- 0.05 MPa. and /',(o.s~ measured. This process was repeated for growth pine stand, and by the Wind River Canopy Crane in tl~e old- 10

growth Douglas-fir stand. The young pine canopy was accessible glas-fir trees, and among roots of Douglas-fir rand old- from the ground. growth and intermediate ponderosa pine. There was a sigmoidal relationship between RWC and the applied pressure, with no sign}ficant species- or age- Stai{stical analyses related differences in slopes (data not shown). Vuinera.bility curves were fitted by the least squares method based on a sigmoidal function: Native root embolism lflO pLc = (2) (1 + e"',*-~'~) in both species, shallow roots of young trees were more embolized at the end of the summer than when measure- where tl~e parameter a is an indicator of the slope of the l!near part ments began (Fig. 2). In roots of young Douglas-fir, the vulnerability curve and b is the pressure at which 50 PLC o1" embolism increased significantly (P=0.02) from 20 PLC in occurred (Psa- MPa), The minimum root xylem water potentials were estimated using the vulnerability curves determined in the July to 55 PLC in September. Similarly, in young laboratou and the PLC measured in the field. The maximum PLCs ponderosa pine root embolism increased significantly wifix)ut HR were determined using the predicted ~Jso~l without i-IR, (J~-0.01) from 45 PLC in June to 75 PLC in August and the relationships between ugh,,n and root xylem water potentials, and the vulqerability curves. Differences in hydraulic parameters and field measurements bc~veen old-growth and yomlg trees and I 1 J I ' 100 - be~een species were detom~ined using a one-v;'ay ANOVA and Douglas-fir. to assess the difference at each date, a two-way ANOVA with one repeated measure factor was made. All statistical procedures were conducted with Statistical Analysis Systems' software (SAS 1999). :°° ResuKs

Root ]c> and xylem vulnerability

0 t T I " Values of" fi,~(m~,~)tbr a given age class and species did not p% Ponderosapin~ vau seasonally, fluctuating around overall means ranging D_ 100 T~_ from 3.4~:0.1 to 4.9-0.3 kg m -~ s -~ MPa -~ (Table 2). The overall mean k4r,~a×) for both age classes of Douglas-fir 8O was 15% lower than for ponderosa pine (-,°=-0.02). For both species, /q(m~,~) was lower in roots of young than in old 6O trees (P<0.02). Root xylem vulnerability curveswere steep, with PLC 4o increasin~ sharply above 0:5 MPa applied pressure in both species (Fig. l). The PLC reached 80 at -2.0 MPa in 20 o old "~\ ponderosa pine and -3,0 MPa in Douglas-fir. \,'slues of o intermediate Pso ranged flom -0.7 MPa in roots of young ponderosa 0 I 1 1 1 ! i ! , -4 -3 -2 -1 0 pine to -1.4 MPa in roots of young Douglas-fir (Table 2). Based on their Pso value of-0.7 MPa, roots of young Negative of Applied Pressure(MPa) ponderosa pine trees were significantly more vulnerable to Fig. I Vulnerabilib, curves showing the percentage loss of embolism than those of intermediate (Pso=-l.3 MPa) and hydraulic conductMb, (PLC) versus the negative of applied air old-growth (Pso=-l.2 MPa) trees. However, P50 did not pressure for Douglas-fir and ponderosa pine medium root xylem differ significantly among roots of young and old Dou- (n=6) o[" young trees (rric,ngla;), inlermediate trees from old-growth site (open circles) and old-growti~ trees Oqlled ~:ircle.;). Error bars are standard errors

Table 2 Total diameter, xylem diameter, maximum specific ponderosa pine and Douglas-fir trees of different ages. Values are conductivity (k~(,,~.~)), and the negative of applied pressure at means (,SE) for n=6 roots. Values with different letters within a row which 50% loss of ]"~0",~) is reached (Pso) of roots sampled from are significantly different (P<0.05) Ponderosa pine Douglas-fir Old-growth Intermediate Young Old-growth Young

Mean total diameter (ram) 3.3-0.3 a 3.4-'-0.2 a 3.5-0.3 a 3.6-0.2 a 3.6-'0.2 a Mean xylem diameter (ram) 2.0-0.1 a 2.0--0,1 a 2.1-'-0.1 a 2.0'-0.1 a 1.92-0.1 a l:

I i L I 100 z, Ponderosa pine IO0 i i i i L i • old Douglas-fir • Douglas-fir 8O O intermediate ~r young 60 ~i ~ ~ • ' R2=0"92 = 0.89

O •• • 40 n 40 ,~ ~ z~ 20 L

0 I I I I 0 I ', ,' I l ', o Ponderosa pine 0 10 20 30 40 50 _.J a_ 80 RWD (%) Fig. 4 Pcrcent toss of conductivit3;.(PLC) versus the relative water 60 deficit (RT,VD) for Douglas-fir (so!id ~rian.gle.0 and ponderosa pine (open triangles) medium reel samples 40 Douglas-fir stands and from 0.15 to 0.17 m 3 m -3 in the 20 ponderosa pine stands (Fig. 5a,b). Between June and October, gradual soil duing occurred in the ponderosa 0 1 f T I 1 i pine stands, with 0 falling to ca. 0.07 m s m-; by the end of Jun Jul Aug Sep Oct Nov Dec October. However, about half of the seasonal decline in O Date in the ponderosa pine stands had occurred by the end of .[uly. ~n the Douglas-fir stands, 0 had fallen :o ca. [rig. 2 Seasonal percentage loss of xylem hydraulic conductivity 0.12 m 3 m-S by mid-Septembeb when light rain events (PLQ in Douglas-fir and ponderosa pine medium roots (n=S-I 0) of young trees (tr,,cmgles'), intennediate trees from ol&gro',vtb, sile caused a transient partial recovery of" 0. Maximum daily ("OF,en circ/ex) and Did-growth uecs (filled circles). Error bar.* are net water use in the 20-30 cm layer ranged be~:een 0.2 smp,dard errors and 0.4 1am day -I in the Douglas-fir stands and be>veen 12

0. I and 0.2 mm day-~ in the ponderosa pine stands not occur untit cJ~oi~ had ['allen below thresholds for the (Fig. 5c,d). By the time 0 reached its seasonal minimum onset of HR. The deviation of predicted from observed value, net water use had declined to ca. 0.03 mm day-~ in time courses of g&oi~ was greatest in the old-growth the Douglas-fir stands and ca. 0.01 mm day-~ in the ponderosa pine stand where in the absence of HR g**oi~ at ponderosa pine stands. 20-30 cm would have fallen below -4.0 NfPa (Fig. 6d). Soil s~ in the 20-30 cm layer was close to zero in all sites at the beginning of the study period and showed little • change until late July to mid-August in the Douglas-fir Mitigation of seasonal embolism and stomatal closure stands and early to mid-July in the ponderosa pine stands by HR (Fig..6a-d). The seasonal decline in g'~on was slower in the old-growth Douglas-fir stand, where ~1?soi~ never dropped The relationship between seasonal variation in PLC and below -0.8 MPa, than in the young Douglas-fir stand, rig. corrected to ground level appeared to be identical where ~oil dropped to -1.4 MPa by September (Fig. 6). across species m~d age classes (Fig.7).-PLC remained Both ponderosa pine sites experienced severe drought in nearly constant as predawn tg~. decreased from -0.25 to fi~e upper soil wifl~ ',i~on dropping to -2.0 MPa at the old- -0.60 MPa, then began to increase sharply as predawn xI~. s..ox~tn site and -2.9 MPa at the young site. fell below -0.60 MPa, which corresponded to the mean In all sites, HR was not detected until threshold values ~J~on at which HR_ attained approximately 50% of its of i)~oi~ had been attained (Fig. 6e,f). Soil ~: thresholds for seasonal maximum (ef. Fig. 6). As predawn ~L decreased the onset of HR ranged from ca. -0.05 MPa in the old- from -0.6 to -1.0 MPa, PLC increased from 25 to 70%, growth stands to ca. -0.6 MPa in the young strands: The and then remained constant as predawn tIq. decreased mean gT.~oiI threshold for the onset of HR among all stands further to -1.5 MPa. Despite the large seasonal variation was -0.33-0.16 MPa, and 50% of seasonal maximum HR in preda(vn tgt. and PLC, the vaiation in midday ~[. was was attained at a mean ~tT~on of-0.64÷0.f0 MPa. relatively small, with midday ~I*t. remaining nearly con- Normalized HR in the 20-30 cm layer did not appear to stant at -2.0 MPa once PLC had reached its seasonal reach its maximum values until g*.~oilin the same layer had maximum. fa!len below predawn xI*L. In the Douglas-fir stands, The estimated root xylem ~I~, detenained from the ovemight HR replaced a maximum of 60% of the previous vulnerabili~ curves and native PLC, decreased linearly day's soil water utilization until a precipitation event in with ~/son measured on the stone day the roots were September temporarily eliminated HR. In the old-growth collected (Fig. 8). In ponderos~i pine, estimated root kS was ponderosa pine stand where the greatest absolute values of lower than ~soil early in the season, but then became HR were observed (0.06 mm day-~), overnight partial higher than f.~on over a range of ~.~oi~ corresponding to the recharge of soil water remained at 60-80% between onset of HR. In Douglas-fir, estimated root v~ was always Augoast and October. The predicted seasonal decline in lower than ~*~oil, but extrapolation of the linear reIation- g~soi~ in the absence of HR was substantially greater than ship between ~I~.~oil and estimated root cJ suggested that the observed decline in all sites (Fig. 6a--d). However, continued seasonal decline in ~9.~oil may have eventually significant deviation 0f predicted and observed vI~oi~ did resulted in root ~I* becoming less negative thm~ that of the

Fig. 5a-d Time courses of soil Douglas-fir Ponderosa pine volumetric water content (0; a 1 I t i I I and b) and net water use (e and 0.25 • ~" o (a)- (b) d) between 20 and 30 cm in young and old-g-rowth Douglas- 0.20 o fir and pondero.aa pine stands during the dry season 0,15 o~%% ~ ~,. 0.1'0 o 0.05 Yoiang Old ', I I ,~ I ', o" {, (c) (d) o 0.4 6 ® 03 O •

Z 0.1 . o

0.0 ? s e',~1 1 ~,~ I - Jun Jul Aug Sep Oct Nov Dec Jun Jul Aug Sep Oct Nov Dec Date 13

' ' ' ® ' 0.0 J.-f,rdld D_. Zx ' Ponderosa pine ' ' ' / Jt, D.-fir young A Douglas-fir ~..~ 80 O P. pine old ~z~,k ~ A P. pine young ~- -0.5 E

°~'60 i~'@ R2 :0.67 ~' R2 = 0.91 ~' -1.0 o 0\\ c5 "5 _~ 40 O o,> % rr -1.5 m 20 midday predawn ~HR~/ E -2.0 03 LLI 0 I ~ I ~ I I , -3 -2 -1 0 -2.0 -1.5 -1.0 -0.5 0.0 ~soil (MPa) TL (MPa) Fig. 8 Estimated root xylem water potentials versus soil water Fig. 7 Percem root loss of conductivity (P£C') versus midday and potentials 0I~oii) in young and old-growth Douglas-fir (filled predawn lea[ water potentials (qq.) corrected for the gravitational .~ymbols) and ponderosa pine (open s.ymbo!s') stands. The estimated component in young and old-growth Douglas-fir and ponderosa pine root water potential was determined using the measured native root trees. The ~,ertical an-ow represents the mean ~T~son at which HR embolism and the vulnerability curves. The vertical arrow reached 50% of its seasonal maximum (~Iq~a) represents the mean ~Qon at which t.[R reached 50% of its seasonal maximmn 0IJ qa) soil. in both ponderosa pine and Douglas-fir stands, redistributed water originally taken up by deeper roots thus In atl of the sites, the seasonal maximum root PLC in appeared to partially uncouple the vii* of" shallow roots from the absence of HR was predicted to be substantially that of ,the sut:rounding soil. Seasonal minimum values of greater than observed values (Table 3). [n the old-growth q?~oil at i-m depth were -0. l to -0.3 MPa, indicating sites, the PLC without HR was predicted to be greater water availability to deeper roots in these stands approximately tv;,ice the measured PLC, whereas in the (data not shown). younger stands the predicted increase in PLC without HR was between 11 and 21%.

Fig. 6a-f Time courses of soil Douglas-fir Ponderosa pine and predawn water potentials i i i i i i . 0 (~[J) in young (a and b) and old- growth (e and ,I) Douglas-fir -1' and ponderosa pine stands and -2 riormalized hydrau!ic redistribu- -1 Lion (Normalized HR; e and f) ÷~ -3 between 20 and 30 cm. The -4 predicted time course of soil %. water potential was calculated in CO the absence of nightly, partiaI -6 }: reco-'eW due to I:IR: At the oldL I 1 ',~ ', ', I I I i I : , 0~ growth ponderosa pine site, 0 (c) predawn wamr potentials from -1 old-growth trees (closed circle's) and intermediate trees (open -2 -1 ÷ cimles) are represented, in e and + -3 f, open and cloxed circle's re- ,. predicted~soil present old-growth and young -4 :2 s{tes, respectively - ,, measured '-VsoiL -5 @ predawn 'F L h -6 f i i i i t ', I ', ', ", °-. I - (e) o (f) v 100 L I 0 0 c~O I 80 "-o cD o°'~oo oo . N ~o-% 40 #oo° m 008, @ I0 O 20 © Z Or. Jun Jul Aug Sep Oct Nov Dec Jun Jul Aug Sop Oct Nov Dec Date 14 Table 3 Soil water potential beaveen 20 and 30 cm (V!qon),predawn tivity (PLC) measured and estimated assuming no hydraulic leaf water potential corrected to groundlevel (~L); estimated root redistribution (HR) in ponderosa pine and Douglas-fir stands xylem water potential and maximum percent loss of root conduc-

Ponderosa pine Douglas-fir Old:growth Intermediate Young Old-growth Young

Minimum '-I'soil at 20-30 cm (MPa) -1.8±0.3 - 1.8-0.3 -2.7;0.4 -0.5-0.1 - 1.4-0.2 Minimum predawn vi~t. (MPa) -018--'0.1 -0.9~0.1 -1.6±0.1 -0.4~0.1 -0.8e0.1 Minimum root xylem vI~ (MPa) -1.1;0.2 - 1.6±0.2 - 1.4-'-0.2 -0.7---0. I - t .5±0. l Observed maximum PLC (%) 42±5 59-6 76'--5 29-~7 49±8 Predicted maximmn PLC without HR (%) 81'--6 73±6 87±7 68-4 70±6

Daily g~ decreased linearly with increasing PLC for Douglas-fir stand allowed ~:l*~oij to remain above both species with all age classes following similar patterns -1 MPa, HR still played an important role in maintaining within a species (Fig. 9). Although g~ was greater in ~roo~ and /9.~on above levels that would have provoked ponderosa pine at a given value of PLC, the slopes of the substantially greater PLC than that observed (Fig. 6, linear relationships were similar (P=0.46). Table 3). l.n the younger stands, minimum (if~ootwas lower and maximmn PLC greater than in the old-growth stands, implying that FIR was not as effective in maintaining ':I.Jso~l Discussion above thresholds associated with rapidly increasing PLC in these stands. It should be noted that ol.tr predictions of \Ve found compelling evidence that partial overnight PLC in the absence of HR (Table 3) are extremely replenishment of soil water by HR_ in ponderosa pine arid conservative because %oo~ was estimated from relation- Douglas-fir stands diminished loss ()f shallow root func- ships between ~oll and ~I*roo~determined in the presence tion during seasonal drought periods. As moisture near the of HR. Vvqthout H-R, it is likely that shaIlow '-I*,.ootwould soil surf'ace becmne depleted, trm~sfer of water from moist, have been more closely coupled to ~I£,oil as the soit driedl deeper soil layers prevented ~I*soi~ i?om [511ing to levels leading to full embolism in most of the stands. One of the that would have induced complete loss of Water transport primary consequences of HR thus appears to be protection capacity in shallow roots. In roots of the old-growth trees, of shallow roots fi'om total hydraulic dysfunction through PLC never exceeded 45% throughout the entire dry season delayed drying of the upper portion of the soil profile. (Fig. 2). In the old-growth ponderosa pine stand, ~IL~oi~in The loss of root conductivity was strongly related to the 20-30 cm layer did not fall below -2 MPa, whereas predawn and midday vIQ.., with common relationships without HR, tJ~soi!would likely have reached values as low found among all four sites, despite differences in species, as -4 to -5 MPa, causing more than 80% embolism tree size and soil g,pe (Fig. 7). Embolism increased rapidly (Table 3). Root embolism in ponderosa pine is not below a threshold predawn XJ)c of-0.6 MPa, which also predicted to reach 100% at even these low soil water con'esponded to the mean ~.~oi! at which HR reached 50% potentials because root xyIem water potential is greater of its seasonal maximum. This concordance suggests that than that of the soil below -t MPa (Fig. 8). Although despite probable differences in soil texture among sites greater soil water storage in the wetter old-growth and in maximum t:ooting depth and water use among the trees studied, there was a degree of convergence in the plant and soil water potential thresholds at which root 250 embolism accelerates and HR begins to regulate V!Jsoil, O & thereby delaying loss of root conductivity (Hacke et al. 20O 200(}). Native PLC was 24% in Donglastfir roots and 36%

¢'-1 in ponderosa pine roots early in the season when ~o~. was i c.,~ 00~0 r2 = 0.84 E 150 still near zero. However, predawn plant ",Is and Csoil in the 20-30 cm layer were only briefly in equilibrium in atl four IO0 sites (Fig. 6a-d). At the beginning of the dry season, mean v predawn ~I*L.was -0.42 MPa in the Douglas-fir stands and cO 5O -0.74 MPa in the ponderosa pine stands. Using .~l~ese values as estimates of root xylem ~ to predict PLC fi'om 0 ' the vn.tlnei'ability cun:es, initial PLC values of 13 and 32% 0 20 40 60 80 were obtained for Douglas-fir and ponderosa pine, respectively, which were consistent with the native PLC PLC (%) of roots collected ['rom the sites. The disequilibrium Fig. 9 Stomatal conductance (gs) versus percent root loss of between plant and soil ~J2*early in the season was probably conductivity (PLC) in young and old-growth Douglas-fir and attributable to nocturnal ~ranspiration or incomplete over- ponderosa pine trees. Symbols are as in Fig. 7 night rehydration of the abovegronnd portion of the trees 15

(Donovan et a!. 200i,2003). Later in the season, when (Fig. 4), a relatively small amount of water can ma-k,cdly vg~oil at 20-30 cm became more negative than predawn tgL increase root conductivity. The mechanism responsible for (fig. 6a-d), HR resulting from ieverse flow of sap through refilling of embotized xylem conduits is still unknown, but roots (Brooks et al. 2002; Scholz et al. 2002) reversed the refilling is faster when embolism is artificially induced by normal root/soil q) gradient, keeping the /2~ of shallow air pressure than when it is naturally induced by drought roots above that of the soil. (Hacke and Spen T 2003). Refilling of embolized tracheids Our results suggest a link between root PLC and in Douglas-fir and ponderosa pine roots may have been regulation of stomatal conductance (Fig. 9). Root xylem facilitated during periods of HR that increased the water embolism acted in concert with stomata to limit water loss, content of shallow roots by reversing the direction of thereby maintaining minimum tgL above critical values to water flow. Reverse flow has previously been documented prevent further xylem embolism (Spent 2002). In the in both Douglas-fir and ponderosa pine roots (Brooks et a[. present stzldy, seasonal variation in predawn '-Iq, was 2002). Because low soil temperature severely reduces root roughly twice that of midday v!Jt, and variation in midday water uptake during winter (Oertli t996), refilling of roots ~I~c decreased sharply as ma×imum values of root PLC in the absence of fall rains may l'umish an ecological were reached (Fig. 7). Stomata thus appeared to more advantage to trees ~rowing in climates with cold winters tightly regulate midday q?L as opposed to predawn ~.. by allowing the roots to become Rflty ~tmctional before the because midday <~t. remained nearly constant at around snow melts in the spring. -2.0 MPa despite variation in predawn ~[q. and ~I~,~olI Values of ]C~m~ for Douglas-fir and ponderosa pine roots during soil &3;ing (Tardieu and Davies 1993). It has been were higher than those previously reported for stems of the shown that avoidance of extensive embolism through the same species (Domec and Om'tner 2001,2003). Values of isohydric control of minimum ~9c is governed by a /csm,~x for Douglas-fir roots were similar to those reported feedback belween stomatal conductance and ~Iq_(Saliendra fbr excised seedling roots (Coleman et al. 1990)i but/c~:~,,~ et al. 1995). Below a threshold midday gq_ of ca. of ponderosa pine roots was lower than that reported in a~ -1.7 MPa, a sharp increase in root embolism was earlier study (Gladwin et al. 1998). Although Douglas-fir associated with a stomatal closing response that main- and ponderosa pine roots have higher conductivity than tained midday ~q, above -2.0 MPa (Meinzer and Grantz stems, they are more vulnerable to embolism than stems 1990; Cochard et at. 2002). (Domec and Gartner 2001,2003). Greater conductivig~ and Triggering of stomatal closure by embolism in root higher vulnerability to embolism in roots generally xylem may be a means of" preventing catastrophic xylem parallels the presence of wider and longer conduits in fai!ure in the aerial portions of the plant. Stomatal roots than in stems (Ewers 1985). These data suggest that responses to experimental manipulations of whole-plant roots can transport water more easily when the soil and hydraulic conductance involving root pruning (Meinzer plant-water status are favorable, but their lower resistance and Grantz 1990), root pressurization (Saliendra et al. to embolism may limit plant water use during drought, and 1995), and chilling of roots (Cochard et al. 2000) are therefore play a critical role for stomata] regulation of ,.on.~,s~a~ ,s.,~n this proposed mechanism of preventing plant water status. catastrophic xylem failure. The consequences of partial embolism during drought :nay be less serious in roots Acknowledgements LaboraloD' and field assistance by D. because their Rmction may be more readily restored after Woodruff, H. Cakes-Miller and A. Burke is greatly appreciated. drought, either by refilling of embolized conduits with the This work was supporled by: PNW 02-JV-1126952-252 from the advent of soil and root rewetting (Jaquish and Ewers 2001; USDA Forest Se,rvice and the US Environmental Protection Agency J. lrvine, personal communication), or growth of new roots (SPA). This manuscript has been subjected to the EPA's peer and administrative review, and it has been approved for publication as an (Kolb and Sperry 1999). Critical values of plant ~IJ could EPA document. Mention of trade names or commercial products also be linked to root death and regeneration. It has been does not constitute endorsement or recommendation for use. shown in several conifer seedlings that no root regener- ation occurred below a predawn ~c of-1.3 MPa, and below this threshold value root mortality was close to References i00% (Girard et al. i997). This value of plant vI* corresponds to about 80 PLC in our stud3, (Fig. 7), and Borghetti M, Edwards WRN, Grace J, Jarvis P J, Raschi A (1991) may correspond to the turgor threshold below which new Tilt refilling of embolised xylem inPinu.~ .n:lve.~tr:.~ L. Plant Cell Environ 14:357-369 root growth is inhibited (Zou et al. 2000). Brooks JR, Mcinzer FC, Coulombe R, Gregg J (2002) Hydrauiic Seasonal courses of PLC (Fig. 2)indicated that redistribution of soil water during summer drought in two embolized tracheids were partially or completely refilled conn'asting Pacific Northwest coniferous'forests. Tree Plqysiol by !ate fall before the soil had been fully recharged by 22:1107-1117 precipitb~tion. Because of the size of the roots studied, it is Burgess SSO, Adams MA. Turner NC, Ong CK (1998) The redistribution of soil water by tree root systems. Oecologia unlikely that reduced PLC in the late ~a,1 reflected the 115:306-311 prod,JcLion of new, non-embolized roots. Ernbolized CaldwclI MM, Richards ~-I (1989) Hydraulic iif~: wa~er ef/lux flom tracheids may readily refill when water is available, even upper roots improves effectiveness of water uptake by deep in the absence of positive pressures (Borghetti et al. 1991). roots. Oecologia 79:1--5 As shown by the relationship between RWD and PLC i6 Caldwell MM, Dawson TE, Richards .rFI (1998) Hydraulic lift: Jaquish LL, Ewers F\V (2001) Seasonal conductivity and embolism consequences of water efflux from roots of plants. OecoloNa in the roots and stems of two clonal ring-porous trees, Sa.vsafi-as 113: 151---16l albidwn and Rku,r o,phina. Am ,T Bet 88:206-212 C¢,cbard H. Breda N, Granter A (1996) Whole tree hydraulic Kolb KJ, Sperry ,IS (i 999) Transport conslraints on water use by tile conductance and watcr loss regulation in Q,wrc'us during Great Basin shrub, Arterni.ria zridentata. Plant Cell Environ drou@t: evidence for stomatal control of embolism. Ann Sci 22:925-935 For 53:197-206 Law BE, Ryan MG, Anthoni PM (1999) Seasonal and an,real Cochard H, Marlin R, Gros P, Bogeal MB (2000) Temperature respiration of ponderosa pine ecosyslem. Global Change Bio! eftlects on hydraulic conductance and water relations of 5:169-182 (_)uelc;~s lz)bm' L. J Exp Bet 5 i: 1255-1259 Meinzer FC, Grantz DA (1990) SIomatal and hydraulic conductance Cochard H, Coil L. Le Roux X, Ameglio T (2002) Unraveling the in growing sugarcane: stomatal adjustn~enl to water transport effects of plant hydraulics on stomatal closure during water capacity. Plant Cell Environ 13:383-388 stress in walnut. Planl Physiot 128:282-290 Meinzer FC, Brooks 5R, Bucei SJ. Goldstein G. Scholz FG, Warren Coleman MD. Bledsoe C, Smit BA (1990) Root hydraulic JM (2004) Converging patterns of uptake and hydraulic conductivity and xylem sap levels of zeatin riboside and redistribution o'f soil water in contrasting woody vegetation abseisic acid in cctomycorrhizal Douglas-fir seedlings. New types. Trcc Physiol 24:919---928 Phytol 115:275-284 Mencuccini M (2003) The ecological significance of long-distance Davis SD, Ewers FW. SpmTy iS, Portwood KA, Crocker MC, water transport: short-tenm regutaion, !ong-tenn acclimation Adams GC (2002) Shoot dieback during prolonged drought in and the hydraulic.costs of stature across plant life forms. Ptant Ceu~7o#ms" (Rhamnaceae) chapm-ral of California: a possible (:ell Environ 26: 163-182 case of hydraulic lhiluro. Am J Bet 89:820--828 Moreira MZ, Scholz FG, Bucci SJ, Stcrnberg LS, Goldstein G. Domec .I-C, Garinesr BL (2001) Cavilation and water slorage in bole Meinzer FC, France AC (2003) ]-lydraulic lift in a neotropical scgmems ofman.~re and young Douglas-fir trees. Trees 15:204- savanna. Funct Ecol 17:573-581 214 Oerlli JJ"(1996) Transport of water in the fl~izosphere and in the Domec J-C, Garmer BL (2003) Relationship between growth rates roots. In: \Vaisel 2\ Eshd A, Kafkafi U (eds) P~ant roots. The and xylem hydraulic character[sties in young, mature ,'rod old- Hidden Half, New" York growth ponderosa pine trees. Plant Cell Environ 26:471---483 Paltineanu IC, Stair JL (I 997) Real-time soil water dynamics using Donovan LA. Linton M.I, Riehards JH (2001) Predawn plant water multisensor capacitance probes: laboratory calibration. Soil Sci potential does not nedessarily equilibrate with soil waler Soc Am J 61:1576-1585 potential under well-watered conditions. Oecologia 129:328- Phillips N, Bond BJ, McDowell NG, Ryan MG (2002) Water flux 335 and canopy conductance in young, mature and old-growth Donovan LA. Richards .|-I-i. Linton MJ (2003) Magnitude and Douglas-fir forests. Tree Physiol 22:205-211 mechanisms of disequilibrium between predawn plant and soil Qucrejeta 3I, Egerton-~Varburton LM, Allen ~ (2003) Direct water potentials in deserl shrubs. Ecology 84:463-470 nocturnal waler transfer from oaks to their mycorrhizal Ewers FW (! 985) Xylem structure and water conduction in conifer symbionts during severe soil drying. Oecologia 134:55-64 trees, dicol trees, and lianas, lnt Asset Wood Anat Bull 6:309- Richards JH, Caldwell MM (1987) Hydraulic lift: substantial 317 nocturnal water transport bcN, ecn soil layers by Arlemisia Gholz I-IL (1982) Environmental limits on aboveground net prima W tridentata roots. Oecologia 73:486--489 production, lea/" area~ and biomass in vegetation zones of the Saliendra NZ~ Sperry JS, Comstock fP (1995) htfiuence of leaf Pacific Northwest. Ecology 63:469-481 water sfiltus on stomatal response to humidity, hydraulic GirardS, Clement A, Cochard H. Boulet-Gemourt B, Guehl .[-M conductance and soil drought in Be/Tda (.;ccidenzaKs'. Plants ( i 997) Effects of desiccation on post-planting stress in bare root 196:357--366 Corsican pine seedlings. Tree Pbysiol 17:429-4'.35 SAS (1999) Version 8. SAS Institute, Car); N.C. Gladwin J, Kelsey RG. Thies WG (1998) Hydraulic conductivity in Seholz FG, Bueci SJ, Goldstein G, Meinzer FC, France AC (2002) roots of ponderosa pine infected with black-slain (Lel)togro- Hydraulic redistribution of soil water by neotropicat savanna ,t)kiwn wagenevi) or annosus (hg,zeroba.vidion anne,rum), root trees. "free Physiol 22:603-612 disease. Trec Physiol 18:333-339 Spent JS (2002) Hydraulic cons~aints on plant gas exchange. Agric Gollan T, Turner NC, Schulze ED (i 985) The response of stomata For Meteorol 104:13-23 and leaf gas exchange to vapor pressure deficits and soil water Sperry, JS. fkeba T (1997) Xylem cavitation in roots and stems of content in Ibe sclerophyllous woody species Neriwn oleander. Douglas-lir and white fir. Tree Pbysiol 17:275-280 Occologia 65:356- 362 Sperry JS, Saliendra NZ (1994) Inira- and inter-plant variation in Hacke UG, Sauter .IJ (i 995) Vulnerability of xylem to embolism in xylem cavilation in Bezula occidemalis. Plant Cell En-,iron relation to leaf water potential and stomatal conductance in 17:1233-1241 ,~'c,'gu.~ s vlvaZicci .I7/)ur,gz,l"ea and Popuh,,.v ba/.~ami/}ra. J Exp Spcr:Ly JS~ Alder FR, Campbell GS, Comstock JP (1998) Limitations Bet 46:1 i77-1183 of plant waler use by rhizosphere and xylem conductance: Hacke UG, Spero, JS (2003) Limits to xylem refilling under results from a model. Plant Cell Environ 2!:347--359 negative pressure in Lam'u.v nobili,r and Acer aegzmdo. Plant Tardieu F, Davies WJ (1993) fntegrafion of hydraulic and chemical Cell Environ 26:303-311 signaling in the control of stomatal conductance and water Hacke UG, Sper:-y JS, Ewers BE. Ellsworth DS, Schafer KVR, Oren status of droughled plants. Plant Cell Environ 16:341-349 R (2000). Influence of soil porosity on water use in.Pmus 'I)ree MT, Ewers FW (1991) The hydraulic architecture of trees and zneda. Oecologia 124:495-505 other woody plants. New Phytol 119:345-360 lrvine 3, Law B, Anthoni PM, Meinzer FC (2002) Waer limitations Zou C, Sand,~ R, Sun O (2000) Physiological responses of radiata to carbon exchange in old growth and young ponderosa pine pine roots to soil strength and soil water deficit. Tree Physiol stands. "Free Physiol 22:189-196 20:1205-1207 Jackson RB. Sperry .IS. Dawson TE (2000) Root water uptake and traqsport: t:sing physiological processes in global predictkms. Trends Plant Sci 5:482-0488