Introduction

A considerable number of studies on plant–insect interactions have assessed the potential effects of chemical defense and plant nutritional quality on performance and host selection by insect herbivores (e.g., Pascual-Alvarado et al. 2008; Gripenberg et al. 2010). However, despite the long-recognized parallels between parasitic plants and herbivores (Govier and Harper 1965), which include similar effects on host biomass and resource allocation patterns, as well as on interspecific interactions at the community level (Pennings and Callaway 2002), few studies have focused on analyzing the chemical ecology of the interaction between plant hosts and plant parasites (Runyon et al. 2008; Smith et al. 2008). It has been suggested that plant secondary metabolites maintain a co-evolutionary interaction with herbivores and pathogens (Ehrlich and Raven 1964; Rausher 2010), resulting in a great variation of combinations, mixtures, and concentrations of these compounds in individuals of the same species (Thrall et al. 2001; García-Brugger et al. 2006), as a result of genetic and environmental factors (Agrell et al. 2000; Underwood 2009). Therefore, plant secondary compounds may induce resistance against both herbivores and parasitic plants (Forkner et al. 2004; Runyon et al. 2008). For example, plants have the ability to perceive herbivore attack and respond actively inducing physical (Lucas et al. 2000) and chemical defenses (Kursar and Coley 2003; Runyon et al. 2008). Nevertheless, defensive responses of plants to the attack of other plants still require considerable research (Smith et al. 2008).

There are about 4100 species of angiosperms that are parasites on other plant species (Nickrent and Musselman 2004), of these, 40% are associated to stems and 60% to roots (Musselman and Press 1995). Particularly, lorantaceous mistletoes obtain all or most of their carbohydrates as a result of their own photosynthetic activity but acquire water and nutrients from the host xylem (i.e., hemiparasitism) (Marshall et al. 1994; Bickford et al. 2005). These mistletoes represent a main group of parasitic plants that colonize branches and roots of tropical and temperate host plants (Calvin and Wilson 2006; Bell and Adams 2011). Parasitism by mistletoes has sublethal effects that reduce host growth, photosynthesis, and respiration rates, modifying both host architecture and nutrient balance (Watling and Press 2001; Meinzer et al. 2004) and also indirectly affecting biotic interactions associated with their host plants (Press and Phoenix 2005; Cuevas-Reyes et al. 2011). However, some of these indirect effects are not necessarily negative, since mistletoes may increase the attraction of seed dispersers and facilitate seedling recruitment of their hosts (e.g., Candia et al. 2014).

Mistletoes, in turn, do not escape from herbivore attack (Sessions and Kelly 2001). The defense mechanisms of mistletoes involve chemical compounds obtained from their host plants such as nutritional compounds and secondary metabolites (Glatzel and Geils 2009). Therefore, chemical compounds of the host plants are important to the successful establishment and development of mistletoes (Pennings and Callaway 2002). Similarly, the selection of host plant tissues by herbivores is related to the content of nutrient compounds and secondary metabolites (Grotewold 2005; Bustamante et al. 2006).

However, forest fragmentation may influence the nutritional quality and plant defense as a result of abiotic changes such as light intensity, wind speed, temperature, humidity, and soil fertility, especially at fragment edges and in small fragments (Valladares et al. 2006). These changes may affect the plant vigor or health as a consequence of stress conditions (Saunders et al. 1991; Prada et al. 1995), which in turn can influence plant–insect and plant–plant interactions (Araùjo et al. 2011; Maldonado-López et al. 2015). For example, parasitic plants are usually more abundant in fragmented landscapes, and mistletoe individuals exposed to sunlight usually have larger flower and fruit displays, rapid fruit ripening, and bigger and sweetener fruits, being more likely to be visited by mutualist animals and hence to recruit (López de Buen et al. 2002).

Most studies have focused on evaluating the incidence levels in correlation with plant defense or nutritional quality in plant–insect and plant–mistletoe interactions independently, without considering the simultaneous relationships among host plants, mistletoes, and insect herbivores in terms of leaf area removed, nutritional quality, and chemical defense. In this study, we evaluated these relationships using as study system the host tree Quercus deserticola (Fagaceae) and the mistletoe Psittacanthus calyculatus (Loranthaceae) along forest fragments. We addressed the following questions: (i) Is there variation in the levels of herbivory, nutritional quality, and chemical defenses among parasitized and unparasitized hosts and mistletoes? (ii) What is the relationship between the herbivory levels, nutritional quality, and concentration of secondary compounds in parasitized and unparasitized hosts? (iii) Is there a correlation in the nutritional quality and the concentration of secondary compounds between host plants and mistletoes?

Materials and methods

This study was conducted in the Cuitzeo basin, a hydrological unit with an area of 4026 km2, located in western Mexico in the state of Michoacán, with an average altitude of approximately 2000 m. It is located in the physiographic province of the Trans-Mexican Volcanic Belt. The basin is representative of the environmental and socioeconomic conditions of central Mexico and has experienced strong fragmentation of natural vegetation in the last decades. Around 20% of the basin is covered by temperate pine-oak forests, the subtropical underbrush occupies about 15%, induced grasslands less than 15%, and crops occupy about 40% of the basin (López et al. 2001).

Study system

Quercus deserticola (Fagaceae) belongs to the white oak section (Quercus) of the genus. It is a deciduous tree, growing up to 10 m that occurs in semitropical bushes and temperate forests of Mexico (Arizaga 2009). The species is particularly abundant in the study region where it is commonly parasitized by the mistletoe Psittacanthus calyculatus (Loranthaceae), which is a generalist hemiparasitic shrub, growing up to 1.5 m and parasitizing different species of Pinus, Quercus, Abies, Salix, and Bursera, among other hosts (Bello and Gutiérrez 1985).

Sampling design and herbivory patterns

Oak forests of the Cuitzeo basin have been characterized into 1241 fragments of different sizes and shapes covering ca. 20% of the basin (López et al. 2001). On the basis of this information, we selected 4 permanent sites where Q. deserticola and P. calyculatus occur. The sites were classified in two categories according to their size: i) two small fragments (≤4 ha) (Capula and El Tigre) and two larger fragments (>12 ha) (Lagunillas and Zimpanio).

In order to estimate the herbivory patterns, we first determined the proportion of parasitized hosts in each forest fragment and then sampled leaves from between 6 and 15 (depending on availability) parasitized Q. deserticola individuals and an equivalent number of unparasitized trees. Sampling extended from the upper canopy to the lower branches, collecting 30 fully expanded leaves at the top, intermediate, and bottom of the crown for each oak tree (i.e., 90 leaves per plant) (Cuevas-Reyes et al. 2011). In each parasitized host plant, we randomly selected between 1 and 3 mistletoes and randomly collected 20 leaves per individual. To estimate the leaf area removed by folivores, we obtained a digital image of each leaf and then estimated the total area of the leaf and the area removed using Sigma Scan Pro software (Cuevas-Reyes et al. 2004, 2011).

Leaf chemical analysis

To determine the differences in nutritional quality and secondary metabolites among parasitized and unparasitized host plants and mistletoes, we collected 30 additional fully expanded intact leaves per individual from the same trees and mistletoes of the herbivory analyses.

Nutritional quality

Water content

Two grams of fresh sample were weighted in an analytical balance and then placed in an oven at 105 °C. The samples were dried to a constant weight and then allowed to cool for weight register. Water content was calculated as fresh weight–dry weight/fresh weight (Marquis et al. 1997).

Total nonstructural carbohydrates (TNC)

We performed the phenol–sulfuric method for TNC extraction from plant tissues (Marquis et al. 1997). The concentration of soluble sugars was determined colorimetrically with an UV–Visible spectrophotometer (EZ301 UV/VIS Spectrometer, Perkin-Elmer).

Secondary metabolites

Total content of soluble phenolics

A modification of the Folin–Ciocalteu method was used (Torres et al. 1987), extracting the samples with 80% ethanol. One milliliter supernatant of the centrifuged sample was mixed with 0.5 ml of 2 N Folin–Ciocalteu’s phenol reagent, and then 1 ml of 20% Na2CO3 was added. After 30 min of incubation at room temperature, the absorbance was measured at 730 nm on a UV–Visible spectrophotometer. The standard curve was created with known concentrations of tannic acid.

Flavonoids

The content of total flavonoids was measured using a modification of the method of Zhuang et al. (1992). The samples were extracted with 80% ethanol and 0.1 ml of supernatant was diluted in 0.9 ml of 80% ethanol. A 0.5-ml aliquot was placed in a tube, and 0.3 ml NaNO2 (1:20) was added. After 5 min, 3 ml AlCl3 (1:10) was added. Six minutes later, the solution was mixed and the absorbance was measured at 510 nm. The content of total flavonoids was calculated from a standard curve of quercetin.

Total tannins

We used a radial diffusion method with agarose gel added with bovine serum albumin (Hagerman 1987). The samples were extracted with 70% aqueous acetone, allowed to stand for 1 h at room temperature with continuous stirring, and centrifuged for 10 min at 3000 rpm. We punched 8 wells in each plate. Each sample was suspended in 200 μl of 70% acetone and 50 μl of each sample was placed in the wells. The plates were covered and placed in an incubator at 30 °C for 96 h. We used a digital Vernier to estimate the diameter of the ring that formed. The area of the disk is linearly related to the amount of tannins placed in the well.

Statistical analyses

To determine the effects of fragment size (small vs. larger fragments) on the proportion of parasitized hosts, we conducted a generalized linear model applying the GENMOD procedure. The model used a Poisson error distribution and a log-link function (SAS 2000).

A two-way ANOVA test (SAS 2000) was performed to determine the effects of forest fragmentation (small fragments vs. large fragments) and plant status (parasitized vs. unparasitized plants vs. mistletoes) and the interaction (forest fragmentation*plant status) on the leaf area removed by folivores (after an arcsine-square-root transformation). The model used forest fragmentation, plant status, and the interaction of these terms as the independent variables and leaf area removed was used as the dependent variable. An LSMeans test was used for a posteriori comparisons (SAS 2000).

The differences in nutritional quality and secondary metabolites between plants that occur in forest fragments (small vs. larger fragments) were tested separately using a one-way ANOVA test (SAS 2000). An LSMeans test was used for a posteriori comparisons. Similarly, we performed separately a one-way ANOVA test to determine the differences in nutritional quality and secondary metabolites between parasitized and unparasitized oaks and mistletoes. An LSMeans test was used for a posteriori comparisons (SAS 2000).

We conducted Spearman’s rank correlation analyses to evaluate the relationships between leaf area removed, plant quality, and plant defense separately in unparasitized and parasitized oaks and in mistletoes. Finally, we used Spearman’s rank correlation analysis to determine the relationship among leaf area removed, plant quality, and plant chemical defense between parasitized plants of Q. deserticola and mistletoes. Since these tests involved multiple comparisons, a Bonferroni correction was applied to adjust P values.

Results

Herbivory patterns

The proportion of parasitized hosts was higher in small than in larger fragments (χ 2 = 39.8; d. f. = 3; P < 0.0001) (Fig. 1a). In contrast, we found higher levels of leaf area removed in larger than in small fragments (F 153 = 30.7; P < 0.0001). The mean leaf area removed in larger fragments was 6.4% ± 1.1 (Zimpanio >12 ha) and 7.4% ± 1.0 (Lagunillas >12 ha), while that in the small fragments was 4.8% ± 0.6 (El Tigre <4 ha) and 2.2% ± 0.5 (Capula <4 ha) (Fig. 1b). In addition, the leaf area removed by folivores was higher in parasitized than in unparasitized hosts and lower in mistletoes in all forest fragments (F 2128 = 36.5; P < 0.001) (Fig. 1b).

Fig. 1
figure 1

Patterns of mistletoe incidence and herbivory. a Proportion of parasitized hosts in each forest fragment. b Comparison of leaf area removed by folivores between both forest fragments and conditions (parasitized plants, unparasitized plants, and mistletoes). Untransformed data are shown. Values with the same letter identify means within treatments that were not significantly different after an LSMeans multiple comparison test (P < 0.05)

Full size image

Leaf chemical quality and plant defense

We found differences among forest fragments in the means of water content (F = 12.2; d. f. = 3; P < 0.0001) and nonstructural carbohydrates (F = 7.6; d. f. = 3; P < 0.0003) (Table 1). These response variables were higher in small fragments. Similarly, means of the analyzed variables differed according to plant status (parasitized and unparasitized hosts and mistletoes) (Table 2). Water content was higher in parasitized than in unparasitized hosts and lower in mistletoes (F = 85.5; d. f. = 2; P < 0.0001). In contrast, mistletoes had the higher concentration of nonstructural carbohydrates, followed by parasitized hosts and finally by unparasitized trees (F = 56.3; d. f. = 2; P < 0.0001) (Table 2).

Table 1 Variation in means and ANOVA results for effects of forest fragments on leaf nutritional quality
Full size table
Table 2 Variation in means and ANOVA results for condition (parasitized plants, unparasitized plants, and mistletoes) effect on leaf nutritional quality
Full size table

In the case of the secondary metabolites, means for all compound classes differed between forest fragments (Table 3). The concentrations of soluble phenolics (F = 13.5; d. f. = 3; P < 0.001), total flavonoids (F = 34.7; d. f. = 3; P < 0.0001), and hydrolyzable tannins (F = 26.8; d. f. = 3; P < 0.0001) were higher in larger than in small forest fragments (Table 3). In addition, means for all secondary metabolites varied according to plant status (parasitized and unparasitized hosts and mistletoes) (Table 4). Soluble phenolics were higher in mistletoes, intermediate in parasitized hosts, and lower in unparasitized oaks (F = 33.2; d. f. = 2; P < 0.0001). In contrast, we found that parasitized hosts had a higher concentration of total flavonoids, followed by unparasitized plants and finally by mistletoes (F = 13.1; d. f. = 2; P < 0.001). The concentration of total tannins did not differ between parasitized and unparasitized trees but was considerably higher in the mistletoes (F = 26.1; d. f. = 2; P < 0.0001).

Table 3 Differences in concentration of foliar secondary metabolites (mg g−1) between forest fragments according to ANOVA test
Full size table
Table 4 Variation in means and ANOVA results for condition (parasitized plants, unparasitized plants, and mistletoes) effect on the concentration of foliar secondary metabolites
Full size table

Relationships between herbivory, plant quality, and plant defense within each plant group

In the case of unparasitized Q. deserticola individuals, after a Bonferroni correction Spearman’s rank correlation showed that the leaf area removed was negatively correlated with the concentration of soluble phenolics, while positive relationships were found between water content and nonstructural carbohydrates, soluble phenolics, and total flavonoids, as well as between soluble phenolics and total tannins (Table 5a).

Table 5 Spearman’s correlation coefficients of herbivory, plant quality, and plant defense in unparasitized (A) and parasitized plants (B) and mistletoes (C)
Full size table

In parasitized individuals, a similar set of relationships was found, except that there was no correlation between water content and nonstructural carbohydrates and that the leaf area removed was negatively correlated with both water content and the concentration of soluble phenolics (Table 5b).

Relationships of herbivory, plant quality, and plant defense across hosts and parasites

Leaf area removed in the mistletoe was positively related with leaf area removed in the host and negatively correlated with soluble phenolics and hydrolyzable tannins in the host. Water content and the concentration of nonstructural carbohydrates in the host were positively correlated with water content and the concentrations of total flavonoids and hydrolyzable tannins in the mistletoe. Finally, the concentrations of soluble phenolics and total tannins in Q. deserticola were positively correlated with those of the same compounds in P. calyculatus (Table 6).

Table 6 Spearman’s correlation coefficients of herbivory, plant quality, and plant defense between parasitized plants of Q. deserticola and mistletoes
Full size table

Discussion

Our study shows that forest fragmentation affects the distribution and incidence levels of Psittacanthus calyculatus on Quercus deserticola. The proportion of parasitized hosts by mistletoes was higher in small than in larger fragments. This result can be explained because mistletoes tend to have a clustered distribution on their host plants as result of the behavior patterns of their seed dispersers such as birds (Aukema 2004; Roxburgh and Nicolson 2008). Because habitat fragmentation modifies abiotic conditions, increasing light intensity, wind speed, and temperature and decreasing humidity and soil fertility especially in fragment edges and small fragments (Gascon et al. 2000), it is possible to expect that the timing and frequency of seed production by mistletoes can be altered, indirectly affecting the preference and incidence of seed dispersers (López de Buen et al. 2002). For example, several studies of the effects of forest fragmentation on interspecific interactions have indicated the presence of compensatory responses of plants, such as demographic changes (Bruna and Noguera-Ribeiro 2005), regrowth rates (Agrawal 2000), and seed production (Herrerías-Diego et al. 2006). In addition, some studies have indicated that the percentage of germination, growth, and fecundity of mistletoes depends on the light incidence. Therefore, environments with high light availability such as forest fragments and edges of fragments represent potential sites to be colonized by mistletoes (López de Buen et al. 2002; Valladares et al. 2006).

On the other hand, the response of insect communities to habitat fragmentation is highly variable because fragmentation may increase resources for some species or reduce resources for others (Barberena-Arias and Aide 2002). For example, some studies have indicated that biotic interactions are destabilized and collapsed as a result of forest fragmentation, increasing herbivore density and decreasing the abundance of their natural enemies such as parasitoids (“top-down” forces) (Kruess and Tscharntke 2000; Arnold and Asquith 2002). In contrast, according to other studies, herbivore abundance and leaf area removed decrease in small forest fragments due to fragment isolation and by changes in environmental conditions (i.e., higher temperature and lower humidity), which in turn reduce insect herbivore survival and performance (Valladares et al. 2006; Ruíz-Guerra et al. 2010). Our results agree with this idea because larger fragments had higher herbivory levels than small fragments.

Interestingly, we found a significant relationship between mistletoe infection and herbivory levels in the host tree, Q. deserticola. Across sites, herbivory was higher in parasitized plants than in unparasitized plants and lowest in the mistletoes. Contrary to our results, Canyon and Hill (1997) found higher levels of herbivory in mistletoes than in their hosts because mistletoes have a greater abundance of specialized herbivores. One explanation for our results is that variations in chemical defense and nutritional quality can affect the incidence of herbivorous insects (Lill et al. 2002). Overall, parasitized hosts had higher water content and higher concentration of nonstructural carbohydrates in comparison with unparasitized plants, but similar concentrations of soluble phenolics and total tannins. In turn, mistletoes presented the lowest water content and the highest concentration of total nonstructural carbohydrates, soluble phenolics, and total tannins. It has been reported that in several species of the genus Quercus, chemical compounds such as tannins represent an efficient anti-herbivore defense affecting physiological attributes of insects (Herms and Mattson 1992; Forkner et al. 2004). In this study, the negative correlations between leaf area removed and the concentration of soluble phenolics within the three groups of plants (parasitized and unparasitized oaks and mistletoes) support the defensive effect of these secondary metabolites.

The fact that unparasitized oaks had higher water content and concentration of nonstructural carbohydrates may suggest two hypotheses. The first one is that the probability of mistletoe infection is higher in trees with these characteristics. The second is that mistletoes may affect different components of growth and some metabolic functions of their host plants because they obtain nutrients, water, and some amounts of carbon from the xylem or both the xylem and phloem of their host plants (Puustinen and Salonen 1999; Press and Phoenix 2005). These changes in turn may make trees more susceptible to the attack of insect herbivores and other natural enemies (Press and Phoenix 2005; Qasem 2009). However, further observational and experimental studies are necessary to discriminate between the two hypotheses.

The intimate physiological relationship between the nutritional quality and the chemical defense of the host tree and the mistletoe was further evidenced by the observed positive correlations between nutritional quality and the concentration of secondary compounds between the two plants (Table 6). Furthermore, the leaf area removed in the mistletoe was not only positively correlated with the leaf area removed in the host, but was also negatively correlated with the concentrations of soluble phenolics and total tannins in the oak. These results suggest that P. calyculatus mistletoes have the capacity to sequester chemical defensive compounds of their hosts, reducing the levels of herbivory. Previous studies have found that mistletoes can sequester and accumulate high amounts of nutritional and chemical compounds such as phenolics and tannins (Glatzel and Geils 2009), which in turn affect biotic interactions, reducing the incidence of herbivores.

The water content and the amount of carbohydrates in the leaves may affect (Brodbeck et al. 1990) the synthesis of chemical compounds associated with plant defense as phenolics and tannins (Shure et al. 1998; Glatzel and Geils 2009). Several studies have indicated that an increment of water availability might cause an increase in the content of defense compounds in plants (Gutbrodt et al. 2011). Our results agree with this idea because water content was positively related with soluble phenolics and flavonoids in both parasitized and unparasitized hosts. Additionally, we also found a positive relationship among water content, soluble phenolics, and total tannins in mistletoes. These results indicate that water content in host plants influences indirectly the production of defense and nutritional compounds, which, in turn, affect the nutritional quality and defense of mistletoes and their interactions with insect herbivores (Huberty and Denno 2004). In addition, it is well known that water plays a key role in host plant–mistletoe interactions because mistletoes are characterized by faster rates of transpiration, deficiency in the water potential, and poorer water use efficiency in comparison with their hosts (Bell and Adams 2011).

In conclusion, our results indicate that forest fragmentation is related to an increase in the incidence of the mistletoe P. calyculatus on the host Q. deserticola. In turn, the infection is related to nutritional and chemical differences among oak individuals that are also reflected in higher herbivory levels. The mistletoe, in turn, is able to benefit from the interaction since it shows higher concentrations of carbohydrates and secondary metabolites and lower herbivory levels than the infected and even uninfected oaks.