3034

Studies on the incidence of coniferous needle endophytes in the Pacific NorthwestGEORGE C. CARROLL AND FANNY E. CARROLL

Department of Biology, University of Oregon, Eugene, OR, U.S.A. 97403Received May 29. 1978

CARROLL, G. C., and F. E. CARROLL. 1978. Studies on the incidence of coniferous needleendophytes in the Pacific Northwest. Can. J. Bot. 56: 3034-3043.

The incidence of internal fungal infections has been scored in coniferous needles from 19 hostssampled in over 200 sites dispersed throughout western Oregon and southern Washington. Abiesgrandis.A. nzagnifica , Picea sitchensis ,Pseudotsuga menziesii. and Sequoia sempervirens haveproved congenial hosts for needle blade endophytes; petiole fungi are common in all species ofPicea and Tsuga sampled. An undescribed taxon in the Hemiphacidiaceae, Chloroscvpha spp.,Cryptocline spp., Leptostroma spp., Naemacychts spp.. Phomopsis spp., Phyllosticia sp., andseveral unidentified Coelomycetes with Phoma-like spores were the dominant fungal taxa in theconiferous hosts sampled. The observed patterns of species dominance and diversity suggest thatthe true population of endophytes has been inadequately sampled in the present study and that anorder of magnitude more intensive sampling might be required for real patterns of dominance anddiversity to emerge. Many endophytes are restricted to a single coniferous host or to a restrictedgroup of hosts. When similarity coefficients between coniferous species are computed on thebasis of their internal needle microfloras, the resultant taxonomic groupings appear similar tothose derived from consideration of conventional morphological criteria. Comparison of en-dophyte incidence with host distribution patterns for Pseudotsuga menziesii reveals that infec-tion rates decrease at high elevations and dry sites.

IntroductionSymptomless fungal infections have been de-

scribed recently from living needles of Douglas firin Oregon (Bernstein and Carroll 1977) and from thefoliage of several conifers in western Europe (Car-roll et al. 1977). Both of these studies were based onsmall samples of needles collected from a few re-stricted sites. Intensive studies on the ecology ofneedle endophytes were felt to be justified only ifsuch fungi proved to be widespread with regard toboth host and habitat. Consequently, the presentextensive survey was undertaken to document theincidence of needle endophytes in a variety of hostconifers over a heavily forested region in westernOregon and southwestern Washington.

Materials and MethodsField Sites and Sample Selection

Samples were taken from a number of Research NaturalAreas, from reference stands in the H. J. Andrews ExperimentalForest, and from a large number of arbitrarily chosen incidentalcollection sites scattered throughout the region. ResearchNatural Areas sampled included the following: Abbott Creek,Ashland, Bagby, Bluejay, Brewer Spruce, Bull Run, CamasSwale, Canyon Creek, Cedar Flats. Cherry Creek. CoquilleRiver. Fox Hollow. Goodlow Mountain, Little Sink. LostForest, Meeks Table. Metolius, Mill Creek, Mohawk, Nesko-win Crest, Ochoco Divide, Olallie Ridge, Persia M. Robinson,Port Orford Cedar. Pringle Falls, Sister Rocks, Wildcat Moun-tain. Wheeler Creek, and Wind River. Detailed descriptions ofthese areas are provided by Franklin et a!. (1972). Within eachResearch Natural Area samples were taken from two to sixarbitrarily chosen sites dispersed as evenly throughout the areaas access and topography would permit. Within the H. J. An-drews Experimental Forest samples-were collected from refer-ence stands 1, 2, 4, 5, 8, 10, 11, 16, and 18 (see Zobel et al. 1976)aswell as from several incidental sites adjacent to clear-cuttings.All other samples were collected from incidental sites. Thelocations of all sites are shown in Fig. 1.

Within each site at least one collection of needles was takenfrom each coniferous species present. Because of the difficultiesin separating needles from twigs, samples were not regularlytaken from Thuja plicata, Calocedrus decurrens, Chamae-cyparis lawsoniana. Chamaecyparis nootkatensis, or Juniperusspp. Individual branches were cut from the lower canopy ofarbitrarily chosen trees with a 13-m pole pruner and were label-led for future identification with tags. To discourage spuriousinfections, branches were not enclosed in any kind of bag aftercollection (for further discussion see Millar and Richards 1974;Bernstein and Carroll 1977). All branches were returned to thelaboratory within 48 h of the time they were collected and werestored overnight at 6°C prior to culturing the following day.Culture Methods

Needles were chosen for culturing on the basis of age-class.Needle age was determined by counting in sequence twig seg-ments. each delimited by a set of terminal bud scars produced inprevious years. Thus, the current season's growth at the tips ofthe branches was considered to be age-class I foliage. although itmight vary in actual age from 0 to 12 months. depending on theseason in which it was collected (bud burst in these foreststypically occurs between May 15 and June 30). Needles attachedto twig segments below the first set of terminal bud scars wereconsidered as age-class 2 needles, the next set were age-class 3,etc.

Needles were dipped briefly in 90% ethanol to wet the surfaceand were then surface sterilized for 10min in a solution of 65%commercial Chlorox. Needles were cut with a sterile scalpel intotwo, four. or eight segments depending on needle size, and thesegments were transferred in serial order to 120-mm Petri platescontaining 2% malt extract agar. Normally 20 segments fromfive individual needles were incubated in a single plate. Plateswere incubated at 20°C with a 12-h dark—light cycle underfluorescent lights. Isolation of fungi from plates to 2% malt agarslants was carried out by direct transfer of conidia or mycelialfragments.Scoring of Infections

Needle segments were scored for fungal infection at weekly ;intervals for the 1st month after inoculation and irregularlyevery 2-4 weeks for a considerable period thereafter. After 6months plates were generally discarded. Single and multipleinfections were scored on each individual needle segment.

CARROLL AND CARROLL 3035

124• 122• 120• Ile•

100500 :RESEARCH NATURAL AREA

SCALE IN MILE S :INCIDENTAL SITEENTOMOLOGY MUSEUMDEPARTMENT OF ENTOMOLOGYOREGON STATE UNIVERSITY

ific Northwest

mis needle

^in 19 hosts;ton. .-I*ens havespecies of

ypha spp..to sp.. andtaxa in theuggest t hatand that an,inance anda restrictedlied on ther similar toison of en-that infec-

lection of needles was takennt. Because of the difficultiessamples were not regularlycitrus decurrens. Chatnae-'s noorkatensis, or Juniperust from the lower canopy ofpole pruner and were label-

igs. To discourage spuriousused in any kind of bag after

and Richards 1974;-anches were returned to thehey were collected and wereturing the following day.

ng on the basis of age-class.unting in sequence twig seg-rminal bud scars produced in;eason's growth at the tips oftge-class I foliage, although iti2 months. depending on the(bud burst in these forests

id June 30). Needles attachedt of terminal bud scars werehe next set were age-class 3.

I% ethanol to wet the surfacer 10 min in a solution of 65%cut with a sterile scalpel intoding on needle size, and theorder :o 120-mm Petri platesNormally 20 segments fromated in a single plate. Plates2-h dark—light cycle underi from plates to 2% malt agarinsfer of conidia or mycelia,

r fun gal infection at weeklyinoculation and irregularlye period thereafter. After 6:aided. Single and multiplevidua; needle segment.

PACIFIC NORTHWEST

FIG. I. Location of collecting sites for coniferous foliage I I mi = 1.609 km).

Identification and NomenclatureIdentification of coniferous hosts was based on Hitchcock and

'Cronquist (1973) and Munz and Keck (1959): nomenclaturalcitations are as given by these authors. All fungi were initiallyassigned arbitrary code names based on the host and the order inwhich they were isolated. Synonymies and identifications of

individual fungal taxa were established subsequently on thibasis of cultural characteristics and on the morphologies ofruiting bodies and spores which ultimately developed. Fungwere identified from a variety of sources; where species identifications are available, author citations have been providedNames for ascomycetous perfect states have been assigned onl:

3036 CAN. J. BOT. VOL. 56. 1978

l4 hen a scocarps were produced in culture. Those isolates whichsporulated but could not he identified were referred to simply by,:ode name. Fungi v. hich were visible and could he scored butv. hich ,:ouid not be recognized or isolated because of extremelyslow erowth or contamination were marked as 'indeterminate.'

Data Redaction and Statistical AnalysesRates of infection for individual fungal taxa on a given host

were calculated by dividing the total number of needle segmentsinfected b^ a given fungus by the total number of segmentsincubated. Similarly, overall rates of infection for a given hostwere derived by dividing the total number of segments infectedby any fungus by the total number of segments incubated. Be-cause of the occurrence of multiple infections on some of thesegments. the sum of infection rates for individual fungi wassometimes greater than the overall infection rate for all fungi.The incidence of fungi restricted to the petiole of the needle(petiole fungi) was considered separately from the incidence offungi occurring on all needle segments (blade fungi).

Early in the study samples from several age-classes weretaken from each branch. For these early samples a statistical testwas used to determine the equivalency of percentages of infec-tion for the different age-classes (Sokal and Rohlf 1969. p. 607).Data from samples whose infection rates were significantly dif-ferent from infection rates of age-classes 4 or 5 (at the 5% level)were excluded in calculating overall infection rates. Thus, thetotal sample of needles was kept relatively homogeneous withrespect to age-class (see below for further discussion).

In using the endophytic microflora as an indicator oftaxonomic affinity among the various coniferous hosts sampleda coefficient of similarity was applied (Curtis 1959). This co-efficient is computed as follows: similarity coefficient = 2w/(a +b). where a = the sum of distribution frequencies for all fungalspecies on one host, b = the similar sum for a second host, and w= the sum of lower distribution frequencies for fungal species incommon between hosts. For the purposes of the above compu-tation a distribution frequency was considered to be the propor-tion of ail collecting sites sampled in which a given fungal specieswas found on a given host.

Initially, possible associations between rates of needle infec-tion. aspects, elevations, and floristic zones of the collectionsites were examined for Pseudotsuga tnenziesii through the useof contingency tables in which Kendall's tau was taken as anappropriate measure of association. For such tables floristiczones of the sampling sites were designated according toFranklin and Dyrness (1973). Infection frequency classes weredelimited as follows: 0-5%: 5- 2_5: 25-50%: 50-75%:75-95::and 95-100%. Elevation classes were chosen at 600-ft (I ft =0.305 m) intervals from 0 to 6000ft. The degree of associationbetween rates of infection and elevation within a given floristiczone was measured by computing Kendall's tau, corrected forties, for the unclassed data. Since both positive and negativecorrelations were of interest, a two-tailed test was used in as-signing significance levels to computed values of tau.

Results and DiscussionA synopsis of coniferous hosts and needle infec-

tion rates for the entire survey is presented in TableI. In interpreting these data certain cautions shouldbe stated. The reported frequencies of infection inthe sample vary in the degree of certainty withwhich they may be expected to estimate infectionrates in the true populations;for binomially distri-buted populations the standard deviation varieswith both the sample size and the infection rates

themselves (Sokal and Rohlf 1969). Beyond this.rates of infection appeared to vary systematicallywith the location of the site. with higher rates ofinfection apparent in wetter sites. Thus. for hostssuch as Pseudotsuga nzenziesii, which occur over awide range of environmental conditions. the infec-tion rates reported in Table I have been affected bythe locations of the sites sampled.

The above notwithstanding, Table 1 revealsstriking differences among coniferous species withregard to their susceptibility to internal needle in-fection. Sequoia sempervirens, Picea sitchensis,Pseudotsuga menziesii, Abies magnifica, andAbies grandis have proved congenial hosts forblade fungi, with overall infection rates in excess of50%. Conifers which typically occur only in high-elevation sites appear to be poor hosts for bladefungi (Abies amabilis, A. lasiocarpa, Picea bre-weriana, Picea engelmannii); Taxus brevifolia andTsuga heterophylla also show low incidences ofblade infection. Petiole fungi are common in allspecies of Picea and Tsuga sampled; they appearinfrequently in other coniferous hosts.

The modest survey conducted by Carroll et al.(1977) on endophytes in European conifers pro-vides the basis for a few limited comparisons: theEuropean spruce. Picea excelsa Link, proved to bea rich source of diverse petiole endophytes, as haveall species of Picea studied here. The Europeanyew, Taxus baccata L., showed uniformly highrates of blade infection by a single fungal species,Phyllosticta concentrica Sacc.; Taxus brevifolia inthe Pacific Northwest, in contrast. showed only.low infection rates by a diverse group of bladefungi.

Table 2 presents .information on the relativeabundances of the most common individual fungaltaxa for the various coniferous hosts and distribu-tion frequencies for the fungi. Only those fungiwhich accounted for 1% or more of the total infec-tions on a given host have been included; for inten-sively sampled conifers such as Pseudotsuga men-ziesii a majority of fungal species (80-90%) wereseen infrequently or only once, and thus have notbeen reported here. Most of the common fungi havebeen identified at least to genus. Many of thosefungi referred to only by code name in the tableproduce small, hyaline, single-celled spores ineither stromata or pycnidia and could be classifiedonly as 'Noma-like' in the absence of expertopinion. A number of the commonly isolated gen-era have been reported previously as endophytesin needles or evergreen leaves: these includePhyllosticta, Cryptocline (previously reported asCryptosporiopsis). Leptostroma, Naemacyclus,

Host

)hies antabilislines cot2color4bies grandis.4bies lasiocarpa.(hies magnifica.(hies proceraPi. ea bre werianaPicea engehnanniiPicea sitchensisPinta attenuataPinus contortaPinus lambertianaPinta monticoiaPima ponderosaPseudotsuga men:,Sequoia sempervir,Tama brerifoliaTsuga heterophylkTsuga mertensianc.

Phoznopsis,Ge(for further diCarroll et al:

Casual inspespecies domimwidely reporterone or two dormajority of re<infrequently o1965). This parpresent data seported in Table

Table 2 alscspecificity, hotinetiole vs. blvarious specie..-1bies spp. andunde scribed topolar appendalon Pseudotsu,Leptostroma (istates of Lop,Pinus spp. andchloromela issempervirens.'from occasionmembers of tillrens, Chamaecand Thuja plictcurrence of otldophytes.

The degree

TABLE I. Synopsis of coniferous hosts and overall infection rates

HostNo.sites

No.trees

No.needles

No.petiole

segments

No.blade

segments

petiole 77. petioleinfected segments segments

needles infected with infected with(a!! fungi) petiole fungi blade fungi

7. bladesegments

infected withblade fungi

Ales amabilis 25 27 426 426 1278 20.4 2.4 5.4 7.2Abies concolot- 26 27 404 404 1212 49.5 1.0 32.7 36.2.4bies grandis 39 40 587 587 1758 66.6 2.4 52.1 50.9Abies lasiocarpa 5 6 90 90 270 21.1 0.0 8.9 9.3Abies magnifica 3 4 87 87 261 85.1 2.4 62.1 54.8.4bies procera 12 12 182 182 498 42.3 0.0 17.6 26.1Picea breweriana 3 4 102 102 306 36.3 34.0 0.0 1.6Picea engelmannii 5 7 110 110 330 56.4 52.7 0.9 3.3Picea sitchensis 24 29 413 407 1209 94.8 34.9 77.2 68.5Pious attenuata 12 2 10 10 70 70.0 0.0 0.0 18.6Pinus contorta 31 35 500 491 1966 46.0 2.7 14.4 14.6Pinus lambertiana 8 10 71 71 485 67.6 0.0 32.4 33.6Pints monticola 5 5 65 65 455 92.3 8.1 28.8 44.0Pinus ponderosa 19 19 146 146 1012 77.4 14.2 23.5 21.6Pseudotsuga menziesii 178 208 3300 3300 9898 71.3 9.6 58.0 54.0Sequoia semperrirens 5 5 102 102 297 100.0 0.0 82.4 97.0Talus brerifolia 10_ 10 183 183 498 24.6 7.3 9.6 10.2Tsuga heteropkilla 20 20 310 310 906 39.0 20.8 9.8 8.2Tsuga mertensiana 10 10 180 180 520 56.1 19.0 9.4 19.0

CARROLL AND CARROLL 3037

69). Beyond this,:ry ystematicallyth hik.Ther rates of

Thus. for hostsvhich occur Over a;ditions. the infec-e been ,.affected by

Table 1 revealsBrous species withnternal needle in-Picea sitchensis,,:.;niagnifica, and

ngeniai hosts forrates in excess of

ccur only in high-)r hosts for bladearea. Picea bre-

xus brertfolia andow incidences ofre common in allpled: they appeariosts.I by Carroll et al.can conifers pro-comparisons: the

Link. proved to bedoph■ tes. as have-e. The Europeanxi uniformly hightie fungal species,rates brerifolia inast. showed only.e group of blade

1 on the relativeindividual fungal

losts and distribu-Only those fungiof the total infec-

icluded: for inten-'seudotsuga men-es (80-90-70 wereand thus have notommon fungi haves. Many of thosename in the table-celled spores in;ould be classifiedbsence of expertonly isolated gen-dy as endophytes.s; these include)usly reported as

Naemacyclus,

Phomopsis, Geniculosporium, and Xvlaria st. imp.(for further discussion and literature review seeCarroll eta!. 1977).

Casual inspection of Table 2 reveals a pattern ofspecies dominance and diversity which has beenwidely reported for both higher plants and insects:one or two dominant species account for the greatmajority of records, while most species are seeninfrequently or rarely (Preston 1948; Whittaker1965). This pattern is even more pronounced for thepresent data set when the many rare species unre-ported in Table 2 are considered.

Table 2 also reveals a certain degree of fungalspecificity. both for host and location on the needle(petiole vs. blade). Thus. Phyllosticta sp. 1 andvarious species of Cryptocline occur primarily onAbies spp. and Pseudotsuga menziesii; PSm 1, anundescribed taxon in the Hemiphacidiaceae withpolar appendages on the ascospores, occurs onlyon Pseudotsuga menziesii; various species ofLeptostroma (presumed here to be the imperfectstates of Lophodermium spp.) predominate onPinus spp. and on Picea sitchensis. Chloroscyphahloromela is ubiquitous on needles of Sequoia

semper •irens. While not included in the table, datafrom occasional samples of foliage from othermembers of the Cupressaceae (Calocedrus decur-rens, Chamaecyparis lawsoniana, Juniperus spp.,and . Thuja plicata) reveal there the widespread oc-currence of other species of Chloroscypha as en-dophytes.

The degree of host specificity observed among

needle endophytes may permit endophyte dis-tribution to be used as a measure of taxonomicaffinity among the various conifers studied. Totest this possibility, distribution frequencies werecomputed for each of the recognized fungal taxaon a selected group of host conifers and simi-larity coefficients between each pair of coniferoushosts were calculated (see Materials and Methodsabove). The resulting matrix is shown in Table 3.Examination of the coefficients for comparison ofAbies grandis with other coniferous species sug-gests relatively close affinity between Abies gran-dis. A. concolor, and A. amabilis • a somewhat moredistant relationship between A. grandis, A. pro-cera. A. magnifica, and Pseudotsuga menziesii,and a very distant relationship between A. grandisand all other species of conifers sampled. includingA. lasiocarpa. Similarly, Pseudotsuga menziesiiappears most closely related to Abies grandis,somewhat more distantly related to A. concolor, A.magnifica, A. procera, and A. amabilis, and verydistantly related to other conifers. Reference toLiu's (1971) Monograph of the Genus Abies showsthat the Abies species sampled in this study fall intothree distinct groups on the basis of conventionalmorphological criteria: section Nobilis Engelm.,emend Liu, containing A. procera and A. mag-nifica; section Grandes Englem., emend Liu,containing A. grandis, A. concolor, and A.amabilis; and section Balsamae Engelm., emendLiu. containing A. lasiocarpa. These same group-ings are reflected in Table 3, although the close

3038 CAN. J. BOT. VOL. 56.197g

TABLE 2. Relative frequencies of observed fungal taxa. Numbers in parentheses below 'Petiole' and 'Blade' refer to overallinfection rates and italic numbers are total numbers of segments sampled. Fungi referred to by code name only sporulatedbut could not be identified: those marked sterile did not sporulate. Infections \+hich could be scored but ■‘hich could not he

recoanized or isolated because of slow growth or contamination are noted as 'indeterminate'

HostProportion of total,observed infections, % No. sites

No. sites taxon DistributionHost Fungal taxa Petiole Blade sampled observed frequency,

Abies amabilis (7.5%)426 (7.27;11278 25Phyllosticta sp. 1 28 56 14 56Cryptocline sp. 1 25 16 9 36Leptostroma sp. 19 19 7 28Indeterminate 2

Abies concolor (330)404 (36711212 26 Pinus attenuataPhyllosticta sp. 1 70 73 18 69Coptocline sp. 1 22 19 7 28Ag 2 1.3 2.4 1 3.8Ag 19 1.0 4 15Tiarosporella sp. 2.0 4 15 Pinus contortsIndeterminate 2.7 2.5

Abies grandis (52%)587 (51%)/758 39Phyllosticta sp. 1 60 63 28 72Cryptochne sp. 1 18 20 13 33Crypweline abietina Petr. 7.2 11 4 10Genictdosporium sp. 3.9 1.2 10 26 Pinus lambertiarXy1aria st. imp. 1.3 1 2.6Micropera howspora Linder 1.0 3 7.6Indeterminate 4.6 1.9

Abies lasiocarpa (8.9%190 (9.370 1270 5Criptocline abietina Petr. 88 63 4 80 Pinus monticolaAl 5 11 I 20AI 7 11 I 20

Indeterminate 11Al 10 11 1 20 Pinus ponderosc,

Abies n;agmfica (62%)87 (55%)261 3Phyllosticta sp. 1 77 75 3 100Crvptocline abietina Petr. 7.8 3.5 1 33Ag 4 5.3 4.1 1 33Ag19 1.8 6.2 1 33Geniculosporium I .8 1 33Indeterminate 5.3 8.3

Abies procera (18%1182 (26' 1498 12Phyllosticta sp. 1 97 93 7 58Leptostroma sp. 1.5 ' 17 Pseudotsuga nuAg19 1.5 1 8.3Indeterminate 3 1.5

Picea breweriana (34%1102 (I .6%1306 3PCb 11 56 22PCb 2 11 1 33Indeterminate 22.._

Picea engehnannii (540)/10 (3.37-0 )330 5PCe 15 (sterile) 25 3 60PCe 18 15 4 80PCe 6 (sterile) 11 1 20PCe 8 4.6 I 20PCe 9 4.6 1 20 Sequoia sempePCe 10 4.6 1 20Indeterminate 22

Picea sitchensis (860407 (690)1209 24Leptostroma sp. 46 78 19 79Phomopsis sp. 1 11 10 4 17Phomopsis sp. 2 3.5 2.0 7 29

CARROLL AND CARROLL 3039

'Blade' refer to overalle name only sporulatedeut hi.:11 could not be

:rminate.

.. site.:i \on Distribution;erred frequency, %

14 569 367 28

I6 697 281 3.84 154 15

28 723 334 10

10 261 2.63 7. 6

4 80I 10I 201 20

00I 33I 331 331 33

581- 171 8.3

6733

3 604 80I 201 20I 20i 20

/ 791729

TABLE 2. (Continued

Proportion of ton:observed infections. No. sites

No. sites taxon DistributionHost Fungal taxa Petiole Blade sampled observed frequer,:y,

WocIadium sp. 6.1 5 21Cryptocline sp. 3 1.3 6 25Geniculosporiunz sp. 2.2 4 17PCs 2 2.6 21 8.3PCs 18 (sterile) 6.1 5 21PCs 27 (sterile) 2.4 1 8.3Coryneum sp. 1 . 3 1.0 5 21Indeterminate 10 3.9

Pinus attenuata (0%)10 (197)70 21Naemacyclus sp. 38 1 50Leptostroma sp. 15 100PNa 3 23 2 100Indeterminate 15

Pinus contorta (16%)491 (1570)1966 31Leptostroma sp. 41 61 26 84PNc 11 (sterile) 10 9.1 4 13Cladosporium sp. 7.1 8.8 3 9.7Naemacychts sp. 4.4 4 13Indeterminate 17 8.1

Pinus lambertiana (32%)71 (34%)485 8Leptostroma sp. 68 70 8 100Naemacychts minor Butin 28 20 -1 75PN1 22 2.3 1 15Indeterminate 4 3.5

Pinus monticola (34%)65 (4470)455 5Leptostroma sp. 75 94 4 80Indeterminate 4.2 1 .5

Pinus ponderosa (3170)146 (22%)1012 19Dothichiza pityophila

1(Corda) Petr. 31 9 13 68

Leptostroma sp. 16 53 14 74PNp 12 13 5 26PNp 22 2 11Gloeocoryneum cinereun:

(Dearn.) Weindlm. 8 1 ( 1

Naetnacyclus sp. 5.6 21 11PNp 14 -, n 21 1 1Indeterminate

Pseudotsuga nzenziesii (5870)3300 (54'0)9598 1'8PSm I 56 72 148 83Phyllosticta sp. 1 ' 12 17 80 45Cryptocline abietina Petr. 6.8 3.8 34 19PSm 88 (sterile) 1.9 20 11Geniculosporium sp. 1.8 39 22PSm 57 1.6 18 10Xylaria st. imp. 1.4 13 7.3PSm 62 1.1 11 6.2PSm 72 (sterile) 1 .7 14 7.8Bispora sp. 1.4 21 12Indeterminate 9.1 2.6

Sequoia semperrirens (82%)102 (9770)297 5Chloroscypha chkromela

Seaver 93 94 S 100Cryptocline sp. 2 3.4 1 . 7 240Geniculosporium sp. 2.3 1.0 2 .` 40Indeterminate 1.1 2.3

1

!

Blac

Peti

elevation. aspeand three-waysignificant degrinfection and tthough elevaticbroad, a large rcountered in tobscured signifi

Sites were tizones (Frank' irinfection were I

Kendall's tau wSignificant ass,high-ekvationcorrelations wedophyte incidei

3040 CAN. J. BOT. VOL. 56.

TABLE 2. (Concluded!

•TNBLE 3. Similarity (

Host Fungal taxa

Proportion ofobserved infections, 70

No. sitessampled

No. sitestaxon

observedDistribution

frequency,

Abiesgrandis

Petiole Blade 4bie.sTaxus brerifolia (1570)183 (107)498 10 grandis I .000

Phyllosticta spp.TXbr 20

359.7

_- 21

20 ibieN10 conco/or

TXbr 3TXbr 21

6.46.4

11

1010

.4hiesamlabilis

TXbr 15Indeterminate 9.7

1320

2 20 .4bies-niagnifica

.4biesTsuga heterophylla (287)310 (8.2%)906 20 procera

TSh 4 14 20 3 15 iNesTSh 8 19 7 35 lasiocarpaTSh 23 (sterile) 34 7 35 P.%eudotsugaCryptocline sp. 4 6.3 4 20 menziesiiTSh 12 6.3 7.6 3 15 TaxusTSh 25 4.2 3 15 brevifoliaTSh 2 (sterile) 3.2 10 bugaTSh 20 (sterile) ".6 10 tnertensionaTSh 17 7.6 5 TsugaTSh 22 (sterile) -.6 3 15 bel(Indeterminate 24 S .8

Tsuga mertensiana (287)180 (197,1520 10 TAB]Leptostrotna spp. 24 70 9 90 men.TSm 10 (sterile) 19 4 40 of o.TSm 1 (sterile) 9.4 1 10 in pcTSm 8 9.4 3 30 zonePhyllosticta sp. 1 11 3 30 sitdCryptocline sp. 3 5.7 1 10 leay.TSm 15 5.7 20TSm 13 7.6 3 30indeterminate 11 4.8

relationship between A. magmfica and A. proceradoes not stand out clearly in many comparisons.Both species were rather poorly sampled. and moreintensive sampling might result in higher similaritycoefficients. Coefficients in the range of 0.000—0.200 are probably not significantly different. andtherefore Table 3 does not permit conclusionsabout the taxonomic affinities of Tsuga mertensi-aria. Tsuga heterophylla, or Tarns brew:folio.These low values probably reflect the generalpaucity of needle endophytes in these coniferousspecies. Thus, comparisons of parasitic micro-floras may be expected to yield useful taxonomicinf:_mnation only when the parasites themselves areabundant and widespread.

Preliminary computations made early in thisstudy showed drastically different incidences ofendophyte infection from one collection site to thenext. In general needles collected from high eleva-tions or dry sites showed low incidences of infec-tion while those from low elevations and moist sitesshowed high incidences of infection. We wished toassess the relative contributions of fungal host

specificity and direct environmental influences to-wards this variation in infection rates. The consis-tently lov, infection rates recorded for Tsugaheterophylla. Abies amabilis. and Pinus contortaover a range of elevations and moisture regimessuggested that host specificity comprises an im-portant factor in determinin g endophyte distribu-tions: thus. a high proportion of uncongenial hostsat high elevations may account for the low needleinfection rates found there.

To test the possible direct effects of environ-mental influences on endophyte infection rates,correlations between site parameters and infectionrates for a single widely distributed coniferoushost, Pseudotsuga menziesii. were attempted. Al-though annual precipitation as rain was suspectedto be a major determinant of needle infection, pre-cipitation data were not available for most of ourcollection sites. Consequently. other site param-eters which could be measured directly and whichmight be expected to show a high correlation withprecipitation were incorporated in the initial con-tingency table analysis: these parameters included

Abiesgrandis

Abiesconcolor

Abies .4bicanzabilis magrifica procc ,a lasi)carpa

Pse;a1c,...tigamen.:!e.sh

Taxu.% Tsugaerilirliu mertensiana

Tsugaheterophylla

4bicsgrandis 1.000 0.628 0.527 0.335 0.4-:3 0.075 0.3-1 0.150 0.151 0.081

4h1c.,

concolor 1.000 0.502 0.336 0.51- 0.088 0.2,)7 0.091 0.160 0.000.4bies

anzabilis 1.000 0.207 0.504 0.000 0.242 0.136 0.150 0.0524bics

magnifica 1.000 0.262 0.178 0.247 0.087 0.080 0.044.4bies

procera 1.00) 0.054 0.223 0.100 0.131 0.000.4bics

lasiocarpa 1.000 0.0-7 0.000 0.000 0.021Pseudotsuga

menziesii 1.000 0.162 0.100 0.071Taws

brecifolia 1.000 0.062 0.104Tsuga

mertensiana 1.000 0.126Tsuga

heterophylla 1.000

TABLE 4. Significance levels for correlation between elevation and infection rates of Pseudo/sagamenziesil needles using Kendall's tau as a measure of association. Numbers show the probabilityof obtaining the observed value of tau if no association exists. Direction o: association is shownin parentheses. Levels of association significant at P < 0.05 are marked with an asterisk. Floristiczones as described by Franklin and Dyrness (1973) are denoted here as fol:)ws: 1. coastal, Piceasitchensis zone; 11, Tsuga heterophylla zone; Ill, Willamette Valley; IV, mixed conifer and broad-leaved evergreen zone of southwestern Ore gon; V. subalpine forest: VI, .4 fVes grandis, Pseudo-

tsuga menziesii zone east of Cascade crest; VII, Pinta ponderosa zone east of zone VI

CARROLL AND CARROLL

3041

T.114LE 3. Similarity coefficients for various coniferous species as hosts for needle endophyles. Values can range from 1.000 forcomplete similarity to 0.000 for complete dissnmilari;:.

I II III

Blade (+) (7-)0.780 0.712 0.424

Petiole (-) (-1 ("-0.624 0.212 0.320

elevation, aspect, and floristic zone. Both two-and three-way contingency tables showed nosignificant degree of association between rates ofinfection and the site parameters examined. Al-though elevation and infection rate classes werebroad, a large number of sampling zeros were en-countered in the tables. and these may • haveobscured significant associations (Fienberg 1970).

Sites were then grouped according to floristiczones (Franklin and Dyrness 1973). and rates ofinfection were compared with site elevation usingKendall's tau as a measure of association (Table 4).Significant associations were seen only in thehigh-ekvation zones (IV, V, VI, VII). Negative

J correlations were seen between elevation and en-

dophyte incidence in zones IV, V, and VI. and a

itesn Distribution

^ed frequency, 7.

2010101020

1535352015151010

515

9040103030102030

ental influences to-1 rates. The consis-corded for Tsugaand Pinus contortai moisture regimes• comprises an im-mdophyte distribu-f uncongenial hostsfor the low needle

effects of environ-/te infection rates,ieters and infectionributed coniferous/ere attempted. Al-rain was suspectededle infection, pre-'le for most of ourother site param-

directly and whichgh correlation withI in the initial con-arameters included

Floristic zone

IV

VI VII

(-) - (+)

0.020*

0.06

0.039* 0.016*(-) - I -1 (+ )

0.150

0.:26 <0.001* 0.016*

positive correlation between elevation and en-dophyte incidence v. as seen in zone VII, the dryestzone sampled. These observations are consistentwith the notion that precipitation as rainfall may bea factor in endophyte dispersal. In zones I, II, andIII the average annual precipitation ranges frommoderate to high (8.)--320 cm) and at any elevationmay be adequate for endophyte dispersal. Siteswithin zones IV, V. and VI may receive a substan-tial proportion of their precipitation as snowfall;higher sites within these zones will receive moresnow and less rain than lower sites, resulting in anegative correlation between endophyte incidenceand elevation.

In zone VII lack of rain and relatively open con-ifer stands may limit the spread of endophytes;

3042 CAN. J. BOT. VOL. 56. 1978

within this zone precipitation increases and standsbecome more dense as elevation increases, result-ing in a positive association between endophyteincidence and elevation. Many endophytic fungi,when grown in culture, produce masses of gloeoidspores. Such spores are usually considered rain-dispersed propagules (Ingold 1971). Indeed, wehave seen endophyte conidia in throu ghfall samples

. collected beneath coniferous stands.Finally. the generality of patterns derived from

comparisons of a uniform needle age-class shouldbe considered. To what extent can conclusionsbased on 4- or 5-year needles be extended to olderneedle age-classes or to the aggregated foliage of asingle tree or stand of trees? In lower elevationmoist sites the oldest surviving needles on healthyPseudotsuga menziesii or Abies grandis branchesusually occur in age-classes 8 and 9. The oldestsurviving needles on branches of the same coniferspecies at higher elevations or in dry sites may be12-13 years old. Early in our work needles of manyage-classes were taken from single branches col-lected from a number of high-elevation or dry sites.Data for these older age-classes were generallyexcluded from subsequent analyses (see Materialsand Methods). However, when infection rates forthe older age-classes are computed and comparedwith those for the 4- or 5-year needles routinelyused, they are found to be consistently higher.Thus, the apparent low infection rates seen at highelevations and dry sites may relate more to delayedonset of endophytic infections than to absolutelower incidences of internal needle fungi.

Age-specific distributions of needles on a singleold-growth Pseudotsuga menziesii tree (tree 286)have been reported by Pike et al. (1977); over halfof the needles occur in age-classes 1-3. When theage-specific infection rates computed by Bernsteinand Carroll (1977) for this tree and for other trees inthe same stand are applied to these needle distribu-tion data, about 75% of the needle segments in thestand are found to be infected. Age-specific needledistributions are not available for Abies spp. orPseudotsuga menziesii from high-elevation or drysites. However, similar calculations can be at-tempted if the following relative distribution ofneedles is assumed. starting with age-class 1: 1,15%; 2, 12%; 3. 12%; 4, 10%; 5, 10%; 6, 9%; 7, 8%;8, 7%; 9, 7%; 10, 5%; 11. 4%; 12, 1%. Observedinfection rates from several of the high-elevationand dry sites can be applied to this distribution toestimate the proportion of infected needle seg-ments in such stands. Even when the upper limitsfor rates from such sites are used, no more than30-35% of the total needle segments can be in-

fected. In many of the stands the overall rates mustbe much lower. Thus. in spite of difficulties in theuse of uniform age-classes. we feel that habitatcomparisons are valid.

The documentation and explanation of needleendophyte distribution patterns had constituted amajor objective in the present study. Left un-answered are more basic ecological questions onthe functional role of needle endophytes and theirmode of dispersion. In particular. one may ask, inview of their wide-spread occurrence in needles ofcertain conifers, whether such fungi form mutu-alistic associations with their hosts. Possiblebenefits to the host trees might include antagonismtowards pathogenic needle parasites and surfacesaprophytes, delay of needle senescence, or a de-crease in needle palatability for grazing insects.These hypotheses should be amenable to directexperimental test. The present study has shownwhich host—endophyte pairs might prove appropri-ate model systems for such experiments. Furtherstudies on the dominant endophytes of Pseudo-tsuga menziesii and Abies grandis are currently inprogress in this laboratory.

AcknowledgmentsThe authors wish to acknowledge the substantial

assistance of Leighton Ho in the collection offoliage samples in the field and of Jacqueline Cal-vert in the culture work and scoring of endophyteinfections. We also wish to thank Dr. J. Lattin ofOregon State University for permission to use anOS U Department of Entomology base map in plot-ting the distribution of sampling sites. Greg Daly,Robert Rydell, and Lawrence Pike assisted withreduction and analysis of data. This work was car-ried out with supplemental funding to National Sci-ence Foundation grant No. BMS 7514003.

BERNSTEIN, M. E., and G. C. CARROLL. 1977. Internal fungi inold-growth Douglas fir foliage. Can. J. Bot. 55: 644-653.

CARROLL. F. E.. E. MOLLER. and B. C. SUTTON. 1977. Prelimi-nary studies on the incidence of needle endophytes in someEuropean conifers. Sydowia. 29: 87-103.

CURTIS, J. T. 1959. The vegetation of Wisconsin: an ordinationof plant communities. University of Wisconsin Press, Madi-son.

FIENBERG. S. E. 1970. The anal ysis of multidimensional con-tingency tables. Ecology, 51: 419-433.

FRANKLIN. J. F., and C. T. DYRNESS. 1973. Natural vegetationof Oregon and Washington. U.S. For. Serv. Pac. NorthwestFor. Range Exp. Stn. Gen. Tech. Rep. PNW-8.

FRANKLIN, J. F., F. C. HALL, C. T. DYRNESS, and C. MASER.1972. Federal Research Natural Areas in Oregon andWashington. U.S. For. Serv. Pac. Northwest For. RangeExp. Stn.

HITCHCOCK, C. L., and A. CRONQUIST. 1973. Flora of thePacific Northwest. University of Washington Press. Seattle.

I. (,OLD.. . S. T19.71197,41

Uni x ersity of Tai\IILLA It. C. S.. and

on the collectionnation. Trans. Br

Nftsz. P. A.. and E,ity of California

PIK E. L. H.. R. A400-year-old Douface area. and680-699.

CARROLL AND CARROLL 3043

overall rates mustif difficulties in the

feel that habitat

lanation of needlehad constituted a

t study. Left un-eical questions onJopti tes and theirtr. one may ask, inrence in needles offungi form mutu-

r hosts. Possiblenclude antagonismasites and surfacenescence, or a de-r grazing insects.menable to directstudy has shown

;lit prove appropri-)eriments. Further)hytes of Pseudo-lis are currently in

I\ciOLD, C. T. 1971. Fungal spores. Clarendon Press. Oxford.. T. S. 1971. A monograph of the genus Ai, ies. National

. niversity of Taiwan, Taipei. Taiwan.LAR, C. S.. and G. NI. RICHARDS. 1974. A cautionar■ note

on the collection of plant specimens for mycological exami-nation. Trans. Br. Mycol. Soc. 63: 607-612.

N1 t • Nz. P. A.. and D. D. KECK. 1959. A California flora. Univer-sity of California Press, Berkeley and Los Angeles.

PIKE, L. H.. R. A. RYDELL. and W. C. DENISON. 1977. A400-year-old Douglas fir tree and its epiphytes: biomass. sur-face area, and their distributions. Can. J. For. Res. 7:680-699.

PRESTON. F. W. 1948. The commonness and rarity of species.Ecology. 29: 254-283:

SOKAL. R. R., .:nd F. J. RoHLF. 1969. Biometry. W. H. Freemanand Co.. Sar Francisco. C.A.

WHITTAKER. R. H. 1%5. Dominance and diversity in land plantcommunities. Science. 147: 250-260.

ZOBEL. D. B.. A. McKEE. and G. NI. HAWK. 1976. Relation-ships of ens ironment to composition, structure. and diversityof forest communities of the central western Cascades ofOregon. Ecol. Mono gr. 46: 135-156.

isige the substantialthe collection ofof Jacqueline Cal-ring of endophytetic Dr. J. Lattin ofmission to use anbase map in plot-sites. Greg Daly,

3ike assisted with'his work was car-

to National Sci-7514003.

1977. Internal fungi inBot. 55: 644-653.;UTTON. 1977. Prelimi-le endophytes in some)3.sconsin: an ordination

7isconsin Press, Madi-

multidimensional con-

73. Natural vegetationServ. Pac. NorthwestPNW-8.tNESS. and C. MASER.reas in Oregon andorthuest For. Range

•. 1973. Flora of thengton Press, Seattle.