A Morphometric Analysis of Arceuthobium campylopodum and Arceuthobium divaricatum (Viscaceae)

Although the classification of pinyon dwarf mistletoe (Arceuthobium divaricatum, Viscaceae) has not been controversial to any extent since Engelmann described it in 1878, a recent taxonomic treatment has included this species in western dwarf mistletoe (A. campylopodum). While pinyon dwarf mistletoe is only known to parasitize pinyon pines (Pinus subsection Cembroides), western dwarf mistletoe as it has been known since the late 1800s is a principal parasite of Pinus ponderosa and P. jeffreyi and has never been observed parasitizing pinyon pines. With reservations about the recent classification of pinyon dwarf mistletoe and its treatment under A. campylopodum, we undertook this study to examine in detail the morphological characteristics of pinyon dwarf mistletoe and compare them with those of western dwarf mistletoe. Pinyon and western dwarf mistletoe populations were sampled throughout most of their geographic ranges and morphological traits including plant, flower, fruit, and seed dimensions were measured. Thereafter, we compared morphological characteristics between A. campylopodum and A. divaricatum using univariate and multivariate statistics to determine significant differences among morphologies of both male and female plants. Our analyses clearly demonstrated that pinyon and western dwarf mistletoe are morphologically distinct as originally proposed by G. Engelmann in the late 19 century. Furthermore, the host affinities of the two taxa clearly distinguish them from each other. Therefore, we recommend that A. campylopodum and A. divaricatum continue to be classified as separate species. Morphological differences between these species are summarized and a key is provided for use in their field identification.

Arceuthobium has long been considered a taxonomically difficult genus because of the morphological and phenological similarities among taxa (Gill 1935;Hawksworth andWiens 1972, 1996;Hawksworth et al. 2002).Morphological reduction and similarity and sexual dimorphism hamper identification of Arceuthobium taxa in the field and have resulted in major differences in taxonomic treatments.However, little disagreement regarding the classification of A. divaricatum at the specific level has been presented in the literature since the late 1800s when it was first described by George Engelmann (Engelmann 1878).In the first monograph of Arceuthobium in the United States, Gill (1935) classified A. divaricatum as a host form of A. campylopodum that exclusively parasitized pinyons and hence, under this system, any dwarf mistletoe found on a pinyon was classified as A. campylopodum (Engelm.)forma divaricatum (Engelm.)Gill.The host-form system proposed by Gill worked well for pinyon dwarf

Morphological Measurements
Morphological data for A. campylopodum from 60 populations, 30 each from Pinus ponderosa and P. jeffreyi, (Mathiasen and Kenaley 2015a; Fig. 1 and Appendix 1) was augmented by morphological data collected for A. divaricatum from 60 populations distributed through most of its geographic range in 2014-2015 (Fig. 2 and Appendix 2).Most of these populations were from locations where A. divaricatum was parasitizing Pinus edulis (30 populations), but we also sampled 23 populations of A. divaricatum on P. monophylla, and seven populations on P. californiarum subsp.fallax (Fig. 2).Because we were only able to sample one population of A. divaricatum parasitizing P. californiarum subsp.californiarum in southern California, one population parasitizing P. cembroides in western Texas, and two populations of A. divaricatum on P. discolor in New Mexico, we have not included morphological measurements for plants from those hosts in our results.Voucher specimens for A. campylopodum and A. divaricatum consisting of the mistletoe with host material were deposited at the University of Arizona Herbarium, Tucson (ARIZ), the Herbarium of Rancho Santa Ana Botanical Garden, Claremont, CA (RSA), or the Deaver Herbarium, Northern Arizona University, Flagstaff (ASC).Voucher and specific population data, including collection numbers, collection dates, and GPS coordinates, have been archived electronically in SEINet (Southwest Environmental Information Network: (http://swbiodiversity.org/portal/index.php) or the Consortium of California Herbaria (http://ucjeps.berkeley.edu/consortium).
For each mistletoe population, 10 male and 10 female plants were randomly collected and the dominant plant (largest plant) of each sex was used for morphological measurements.The dwarf mistletoe plant characters measured were those used by Hawksworth and Wiens (1996) for their taxonomic classification of Arceuthobium.These included: (1) height, basal diameter, third internode length and width, and color of male and female plants; (2) mature fruit length, width, and color from female plants; (3) seed length, width and color; (4) length and width of mature staminate spikes; (5) staminate flower diameters for 3-and 4-merous flowers (4-merous flowers were rarely observed for A. divaricatum); (6) length and width of staminate flower petals; and (7) anther diameter and anther distance from the petal tip.
Plants typically were measured within 12 h, but no later than 24 h after collection.Only plants attached to their host's branch and fully turgid were measured.Quantitative measurements were made using a digital caliper (Mitutoyo America Corp., Aurora, IL) and a 7X hand lens equipped with a micrometer (Bausch & Lomb, Bridgewater, NJ).The basal diameter of plants was measured at the point where the plant was attached to the host branch.The width and length of the third internode above the base of plants was included in our morphological analyses because these characters have been frequently reported for dwarf mistletoes and provide information on the relative size and thickness of male and female plants (Hawksworth andWiens 1972, 1996;Mathiasen and Daugherty 2007, 2009a, 2009b, 2013;Mathiasen andKenaley 2015a, 2015b).The length of the third internode was determined by measuring from the top of the second internode above the base of a plant to the top of the third internode, locations which are easily observed (see Fig. 2.1 and 2.3 in Hawksworth and Wiens 1996).The width of the third internode was measured at its midpoint.Staminate spike and flower measurements were made during the peak of anthesis.Likewise, fruit and seed measurements were made during the peak of seed dispersal.

Statistical Analyses
We assessed whether mean values for morphological characters differed significantly between the two dwarf mistletoes and between comparable characters from plants parasitizing Pinus edulis, P. monophylla, and P. californiarum subsp.fallax using one-way analysis of variance (ANOVA) and, when appropriate, a posthoc Tukey's Honestly Significant Difference (HSD) Test (a 5 0.05) (Zimmerman 2004).Quadratic discriminant function analyses (DFA) were also performed separately to assess whether female or male plants of A. campylopodum and A. divaricatum can be delimited to species according to morphological characters (predicted versus actual; Quinn and Keough 2002; Fig. 3).Discriminant function analyses for female and male plants were conducted divaricatum at the specific level (Nickrent 1986;Nickrent et al. 2004), female and male DFAs were executed using equal prior probabilities for each host-dwarf mistletoe combination (25%, partitioned dataset) or species (50%, combined dataset) rather than proportional to their actual host and/or species membership.Discriminant function analyses were parameterized to include equal prior probabilities in order to remove experimental bias (i.e., a priori identification) from the posthoc classification (%, predicted/actual) of dwarf mistletoes by hostdwarf mistletoe combination and species membership.For comparisons of species membership, standardized correlation coefficients for morphological characters were also calculated  A, C) and male plants (B, D).Multivariate means (cross-hairs) were calculated using complete data for each species by sex (A, B), whereas, to further validate the DFA, means were also calculated using resampled data (25 complete records/ species) of female (C) and male plants (D), respectively.For each species (A-D), the inner ellipse is a 95% confidence limit for the mean, and the outer ellipse is a normal contour where approximately 50% of plants for each species reside.Correct classification percentages for male and female plants by DFA (complete and resampled) are presented in Table 6.
to determine the overall contribution of each character to the discriminant function.Likewise, when appropriate, stepwise DFA was utilized to examine systematically the smallest number of morphological characteristics, female or male, resulting in the highest percentage in correct classification (%, predicted/actual).To further validate the DFAs, we resampled separately the partitioned and collective data for female and male plants; selecting at random 25 complete records per hostdwarf mistletoe combination or species and re-executed the DFA to include all female or male morphological characters simultaneously (full-model) with each plant receiving equal prior probabilities.Analysis of variance tests and DFAs were computed in JMP Pro 12 (SAS Institute, Cary, North Carolina).

Morphological Differences
The mean heights of female and male plants of Arceuthobium campylopodum were significantly smaller than those of A. divaricatum (Table 1).However, even though the mean heights of male plants varied by 2 cm, the mean height of female plants only varied by 0.5 cm.The mean length of the third internode of female plants was significantly different between taxa, but the mean length of the third internode of male plants was not significantly different.Male and female plants of A. divaricatum were more slender than those of A. campylopodum with both the mean basal diameter and mean width of the third internode for both sexes being significantly smaller for A. divaricatum (Table 1).
The staminate spikes of Arceuthobium campylopodum were significantly longer on average and the width of staminate spikes was significantly wider than for A. divaricatum; the mean width of staminate spikes of A. divaricatum being nearly half as wide as those of A. campylopodum.A major difference between A. campylopodum and A. divaricatum was that the latter species predominantly produced 3-merous staminate flowers whose mean diameter (2.2 mm) was significantly smaller than that of the 3-merous flowers of A. campylopodum (3.1 mm; Table 1).The mean diameter of 4-merous flowers was also significantly smaller for A. divaricatum, but 4-merous flowers were only occasionally observed and measured for populations of A. divaricatum; only eight populations were observed with a few 4-merous flowers.The smaller size of the staminate flowers of A. divaricatum was a result of its significantly smaller mean petal length and width.Mean anther diameter (0.4 mm) and the distance of anthers from the tips of petals also were both significantly smaller for A. divaricatum than A. campylopodum (Table 1).Similarly, mean fruit length was much larger for A. campylopodum (5.4 mm) than for A. divaricatum (4.4 mm), as was the mean width of fruits.Mean seed length and width were also significantly different between the two species (Table 1).
Plant color is not usually an informative character for distinguishing between dwarf mistletoes.However, the color of plants of A. divaricatum was distinctly different from those of A. campylopodum, the former being consistently brown or dark brown to green-brown compared to the latter which ranged from yellow, yellow-brown, or light brown (Fig. 4-5).Sometimes male plants of A. divaricatum were red-brown, yellow-  brown or even yellow-green; very few female plants of A. divaricatum were these colors.
Because Arceuthobium divaricatum parasitizes several different pinyons, we compared the morphological characteristics of male and female plants from three of its pinyon hosts (Table 2).The heights of male and female plants from Pinus monophylla and P. californiarum subsp.fallax were significantly larger than those collected from P. edulis.Although female and male plants collected from P. californiarum subsp.fallax were shorter on average than those collected from P. monophylla, the means were not significantly different.No other significant differences were detected among the morphological characters measured for plants, flowers, or fruits from the different pinyon hosts.The mean seed length found in the three pinyon hosts was significantly different but only by 0.1 mm, and the mean width of seeds was not significantly different (Table 2).Because the mean heights of male and female plants collected from P. monophylla and P. californiarum subsp.fallax were not significantly different, we combined morphological measurements for plants, flowers, fruits, and seeds from those hosts and compared them to measurements for these characters from only P. edulis (Table 3).Again, only the mean heights of female and male plants and mean seed lengths were significantly different.We then compared characters measured from P. edulis and pooled measurements from P. monophylla and P. californiarum subsp.fallax with those of A. campylopodum (Table 4).The mean heights of female plants from P. edulis were not significantly different to those of A. campylopodum, but the mean values for all other characters (except third internode length of male plants) of A. divaricatum were significantly different from those of A. campylopodum.
Although Arceuthobium campylopodum and A. divaricatum have been reported to flower at approximately the same time (late August to late September), we observed that A. divaricatum started flowering in early August, at least two weeks earlier than A. campylopodum.Flowering periods of A. campylopodum and A. divaricatum did overlap in mid to late August; however, these dwarf mistletoes were not found in the same stands and the difference in the start of flowering we observed in California may have been related to the lower elevations where we made our observations for A. divaricatum.Seed dispersal of these species occurs from late August to late September, but some populations of A. divaricatum at the lowest elevations of its geographic range disperse seed into late October; elevation and its influence on climate may be the primary reason why A. divaricatum disperses seed later in the fall than A. campylopodum in California.

Discriminant Function Analyses
Because some variation in male and female morphologies in Arceuthobium divaricatum was evident among the three pinyon hosts, DFA was conducted first on separate female and male plant datasets consisting of measurements of morphological characters from A. campylopodum as well as A. divaricatum partitioned according to host.Results from these analyses indicated significant differences for the eight female morphological characters of A. campylopodum and, partitioned by host, A. divaricatum (Wilks' l 5 0.1201, Approx.F 24,3043 5 136.48,P , 0.0001; Pillai's Trace 5 0.9702, Approx.F 24,3153 5 62.80, P , 0.0001).Likewise, significant differences were also found among the 10 male plant characteristics examined for A. divaricatum by host and A. campylopodum (Wilks' l 5 0.0702, Approx.F 30,3367.3 5 165.16,P , 0.0001; Pillai's Trace 5 1.0581, Approx.F 30,3447 5 62.61, P , 0.0001).The first two discriminant functions (canonicals) for the DFA of female and male plants accounted for $98.3% and #1.3% of the total variation (Fig. whereas the third discriminant function for female and male plants accounted for #0.22 of the total variation.Female and male plants of A. campylopodum were readily distinguished morphologically from A. divaricatum on all three pinyons because 100% (600/600) and 99.8% (479/480) of male and female plants diagnosed a priori as A. campylopodum were predicted correctly (%, predicted/actual) to this species using 8 and 10 morphological characters, respectively (Table 6).However, samples of female or male plants of A. divaricatum from the various pinyon hosts, analyzed as above, were indistinguishable (Table 6; Fig. 3).
A similar pattern in misclassification was also apparent for predicting host trees of A. divaricatum according to morphological characteristics of male plants, where the host tree was predicted correctly only 72.9%, 57.1%, and 62.9% of the time for male plants infecting P. edulis, P. monophylla, and P. californiarum subsp.fallax, respectively (Table 6).Although the precision in classification of A. divaricatum to its pinyon host increased markedly following DFAs on a randomized resample of female and male plants, the female plants of A. divaricatum on P. monophylla and P. californiarum subsp.fallax were, again, frequently misclassified and placed correctly to host only 68% of the time.Thus, morphological measurements for male and female plants on all pinyon hosts were combined (combined datasets) to further assess species membership between A. divaricatum and A. campylopodum as well as identify the morphological characters contributing most to interspecific separation.
Means and associated 95% confidence intervals for morphological characters of female and male plants across predicted species according to full-model DFA are presented in Table 7. Discriminant function analysis using separately eight female and 10 male morphological characters (fullmodel) clearly demonstrated separation of Arceuthobium campylopodum and A. divaricatum; $98.3% of female and male plants identified via field diagnosis as A. divaricatum or A. campylopodum were predicted correctly to species (Table 8).For DFA of female plants, results indicated significant differences between multivariate means for A. campylopodum and A. divaricatum (Wilks' l 5 0.1353, Exact F 8,1051 5 839.06,P , 0.0001; Hotelling-Lawley Trace 5 6.3868, Exact F 8,1051 5 839.06,P , 0.0001; Pillai's Trace 5 0.8646, Exact F 8,1051 5 839.06,P , 0.0001).The discriminant function accounted for 100% of the total variation as only one dimension is possible when assigning two groups.Female plants of Arceuthobium divaricatum were correctly classified (predicted/actual) to species 98.3% (570/580) of the time, and hence the percentage of female A. divaricatum assigned to A. campylopodum was only 1.7% (10/580).Similarly, 100% of female A. campylopodum were classified correctly when considering all eight female morphological characters.Examination of the standardized correlation coefficients (scc) indicated that seed length (scc 5 1.63), width of the third internode (scc 5 1.48), and fruit width (scc 5 0.85) were most strongly correlated with the discriminant function, thereby contributing most to defining species membership for female plants.Using these three characters alone, female plants of A. campylopodum and A. divaricatum were classified correctly to species 100% (480/480) and 97.9% (568/580) of the time, respectively.Moreover, with the addition of plant height, fruit length, and length of the third internode as predictor variables, the percentage of female A. divaricatum predicted correctly to species increased slightly to 98.6% (572/ 580)-the highest correct classification percentage achieved for female A. divaricatum by either the full-model or stepwise-DFA (Table 8).worth and Wiens (1996) from their field observations only.

DISCUSSION
Classifying Arceuthobium campylopodum and A. divaricatum as conspecific or the latter species as a subspecies of A. campylopodum is not supported by our analyses of the morphological characters for these taxa as both species can be identified easily by differences in one or more morphological characters as well as their host affinities (Tables 5, 7, and 8; Fig. 3).For example, the basal diameter, width of the third internode, and the width of mature staminate spikes were smaller for plants of A. divaricatum than those of A. campylopodum; overall, male and female plants of A. divaricatum were much more slender when compared to plants of A. campylopodum.This characteristic is easily discernible when male or female plants of A. divaricatum and A. campylopodum are placed side-by-side for visual comparisons (Fig. 4-5).However, plant size alone does not easily distinguish these species even though the mean heights of male and female plants were significantly different (Table 1).Other characters that distinguished A. divaricatum from A. campylopodum were its much smaller 3-merous flowers, rare formation of 4-merous flowers, and smaller fruits and seeds (Tables 7 and 8).Furthermore, although plant color is a qualitative character, A. divaricatum can usually be distinguished using plant color as its female plants are typically brown or green-brown while those of A. campylopodum typically are yellow, yellow-brown, or light brown (Fig. 4-5).Again, these color differences can be easily observed when recently collected plants are compared side-byside; upon drying these color differences are more difficult to observe.
A further difference between these dwarf mistletoes is their host range; the principal and only hosts of Arceuthobium divaricatum are pinyons (Table 5).In contrast, A. campylopodum primarily parasitizes Pinus ponderosa and P. jeffreyi with the latter species claimed to be immune to infection by A. divaricatum (Hawksworth and Wiens 1996).It is probable that P. ponderosa is also immune to infection by A. divaricatum: Hawksworth and Wiens (1996) observed that where P. ponderosa was sympatric with infected pinyons, A. divaricatum was found infecting only pinyons.Although they refrained from classifying P. ponderosa as immune to A. divaricatum, Hawksworth and Wiens did suggest P. ponderosa var.scopulorum Engelm.(Rocky Mountain ponderosa pine) may be immune to A. divaricatum in the Southwest.
Because the dwarf mistletoes are extremely important both ecologically and economically, emphasis must be placed on their ecological and pathological roles in forest ecosystems.These two dwarf mistletoes clearly have very different host ranges and hence, their pathological significance in forests of the western United States is also distinct.Any efforts to manage populations of these parasites to mitigate their effects on tree growth and mortality within severely infested stands must consider these differences (Hawksworth and Wiens 1996).Recognition of the host affinities developed by dwarf mistletoes is critical in their classification because we consider differences in host preference(s) to reflect corresponding and underlying genetic and physiological differentiation among dwarf mistletoes.
In addition to our morphological analyses, the separation of Arceuthobium divaricatum from A. campylopodum is also supported by isozyme and molecular studies.Nickrent (1986Nickrent ( , 1996) ) reported that, based on isozyme analyses, A. divaricatum was most closely aligned with and biochemically similar to A. douglasii Engelm.(Douglas-fir dwarf Table 8.Forward-stepwise discriminant function analysis (DFA) for female and male plants of Arceuthobium campylopodum and A. divaricatum: correct classification counts (N 5 predicted/actual) and percentages (%, predicted/actual) with the sequential addition of morphological characters (steps) most-to-least correlated to the discriminant function.Anther diameter (AD); anther distance to tip (ADT); basal diameter (BD); fruit length (FL); fruit width (FW); length of third internode (LTI); plant height (PH); petal length (PL); petal width (PW); seed length (SL); staminate spike length (SSL); staminate spike width (SSW); seed width (SW); and width of the third internode (WTI).

Fig. 1 .
Fig. 1.Approximate locations of collection sites for Arceuthobium campylopodum in California, Idaho, Nevada, Oregon, and Washington.Filled circles represent locations where plants were collected from Pinus ponderosa.Open circles represent locations where plants were collected from P. jeffreyi.Numbers correspond to locations in Appendix 1 (map reproduced with permission by the California Botanical Society (Madron ˜o 62: 6).

Fig. 2 .
Fig. 2. Approximate locations of collection sites for Arceuthobium divaricatum in Arizona, California, Colorado, Nevada, New Mexico, and Utah.Filled circles represent locations where plants were collected from Pinus monophylla.Open circles represent locations where plants were collected from P. californiarum subsp.fallax.Filled squares represent locations where plants were collected from P. edulis.Numbers correspond to locations in Appendix 2.

Fig. 3 .
Fig. 3. Canonical plots for discriminant function analyses (DFA) of Arceuthobium campylopodum and, according to host, A. divaricatum based on morphological characteristics of female (A, C) and male plants (B, D).Multivariate means (cross-hairs) were calculated using complete data for each species by sex (A, B), whereas, to further validate the DFA, means were also calculated using resampled data (25 complete records/ species) of female (C) and male plants (D), respectively.For each species (A-D), the inner ellipse is a 95% confidence limit for the mean, and the outer ellipse is a normal contour where approximately 50% of plants for each species reside.Correct classification percentages for male and female plants by DFA (complete and resampled) are presented in Table6.

Fig. 4
Fig. 4-5.Color variation between Arceuthobium campylopodum and A. divaricatum.-4.Female plants of A. campylopodum with nearly mature fruits in July.Note that plants are yellow-brown.-5.Female plants of A. divaricatum with nearly mature fruits in August.Note that plants are brown-green.

Table 1 .
Morphological measurements for Arceuthobium campylopodum collected from Pinus jeffreyi and P. ponderosa and for A. divaricatum collected from P. edulis, P. monophylla, and P. californiarum subsp.fallax.Data are listed as mean, (SD), range,[n].Means followed by different capital letters in the same row were significantly different using ANOVA (a 5 0.05).Lower case letters in brackets designate sample sizes already listed in the same column.Plant heights are in cm and all other measurements in mm.

Table 2 .
Morphological measurements for Arceuthobium divaricatum from Pinus edulis, P. monophylla, and P. californiarum subsp.fallax.Data are listed as mean, (SD)[n].Means followed by different capital letters in the same row were significantly different using ANOVA followed by a Tukey's Post Hoc HSD test (a 5 0.05).Lower case letters in brackets designate sample sizes already listed in the same column.Plant heights are in cm and all other measurements in mm.

Table 4 .
Comparison of morphological measurements for Arceuthobium divaricatum and A. campylopodum: A. divaricatum data is from Pinus monophylla and P. californiarum subsp.fallaxcombined,and P. edulis alone.Data are listed as mean, (SD)[n].Means followed by different capital letters the same row were significantly different using ANOVA followed by a Tukey's Post Hoc HSD test (a 5 0.05).Lower case letters in brackets designate sample sizes already listed in the same column.Plant heights are in cm and all other measurements in mm.

Table 3 .
Morphological measurements for Arceuthobium divaricatum from Pinus edulis and from P. monophylla and P. californiarum subsp.fallaxcombined.Data are listed as mean, (SD), range,[n].Means followed by different capital letters in the same row were significantly different using ANOVA (a 5 0.05).Lower case letters in brackets designate sample sizes already listed in the same column.Plant heights are in cm and all other measurements in mm.

Table 5 .
Summary of the principal characters separating Arceuthobium campylopodum and A. divaricatum.Data for morphological characters are means; plant heights in cm and all other measurements in mm.Numbers in bold represent key morphological differences between the taxa.

Table 6 .
Discriminant function analyses (DFAs) of female and male plants using complete records and resampled data for Arceuthobium campylopodum and, partitioned by pinyon host, A. divaricatum.Data are presented as: correct classification (count [N]), % [predicted/actual].

Table 7 .
Discriminant function analyses (DFA) of male and female plants: comparison of morphological characters by predicted classification to A. campylopodum and A. divaricatum using complete data per taxon.Ninety-five percent confidence intervals (6) were computed for comparison of mean differences.Combined analysis of plants on Pinus edulis, P. monophylla, and P. californiarum subsp.fallax.