1 Introduction

Seed dispersal by animals has significant influences on community ecology, genetic structures, and population dynamics (Jordano and Godoy 2002; Levine and Murrell 2003; Carlo et al. 2007; García et al. 2007; Moore et al. 2007; Yi et al. 2011). However, ecological assessment of seed dispersal has largely been neglected because tracing seed movements from the seed source to the exact locations where the seeds germinate or perish is difficult (Chambers and MacMahon 1994; Wang and Smith 2002; Bullock et al. 2006; Muñoz and Bonal 2011; Hirsch et al. 2012a, 2012b; Davis et al. 2011) and often suffers methodological limitations (Nathan 2006; Galvez et al. 2009; Galetti et al. 2010). Previous studies indicate that the biggest obstacle to track seed dispersal by animals is to develop a method that is simple, effective, labor-saving, and broadly applicable (Takahashi et al. 2006; García et al. 2007; Yi et al. 2008; Carlo et al. 2009; Lemke et al. 2009; Perea et al. 2011; Hirsch et al. 2012a, 2012b). If a simple and reliable seed-tracking method could be developed, it would facilitate a better understanding of seed dispersal ecology (Will and Tackenberg 2008; Yi et al. 2008; Carlo et al. 2009). Many methods have been used to track seed dispersal by animals and the resulting seed fates, e.g., direct observation (Burnell and Tomback 1985), surgical installation of metal or magnets (Alverson and Díaz 1989; Den Ouden et al. 2005), attachment of colored threads or tags to seeds (Forget 1992; Yi and Zhang 2008; Yi et al. 2011), marking with stable or radioactive isotopes (Winn 1989; Carlo et al. 2009), application of florescent microspheres and dye powders (Longland and Clements 1995; Levey and Sargent 2000; Tewksbury et al. 2002; Lemke et al. 2009), genetic techniques based on microsatellites (Godoy and Jordano 2001; García et al. 2007; Smouse et al. 2012), radio transmitters (Hirsch et al. 2012a, 2012b), as well as pit tags (Suselbeek et al. 2013). Although these methods show their own advantages and have been used in many ecosystems for tracking seed dispersal, most of them suffered from deficiencies and limitations (Alverson and Díaz 1989; Forget 1992; Levey and Sargent 2000; Godoy and Jordano 2001; Tewksbury et al. 2002; Yi et al. 2008; Niu et al. 2011; Hirsch et al. 2012a, 2012b). Carlo et al. (2009) recently introduced a new method to track seed dispersal and seed recruitment based on labeling seeds or fruits by foliar-spraying with a solution of 15N-urea during the flowering stage of small herbaceous plants (Solanum americanum and Capsicum annuum). Although this method is effective and provides ability to find the exact locations that seeds arrive, germinate, and establish in the field, it is not broadly applicable for tree species with large canopies and high trunks because it is very difficult to foliar spray 15N isotope urea solutions to the canopies more than 10 m in height. In addition, foliar sprays applied to wind-pollinated flowers at the time of anthesis could hamper pollination and the subsequent development of seeds and fruits. Our goal in this study was to further develop the 15N stable isotope technique to track seed dispersal of tree species that are tall and have large canopies, e.g., pines and oaks. This method involved direct soaking seeds and acorns into 15N isotope urea solutions with different dosages for several days. The cotyledons or endosperm of the seeds were expected to absorb the urea so that the seeds were 15N enriched, which enables the tracking of seed dispersal and seedling recruitment in the field.

We designed our experiments to address the following five questions: (1) Can any kind of seeds be isotopically enriched in δ 15N values through urea solution soaking?; (2) What is the optimal dosage of 15N-urea and soaking duration for seed isotope enrichment?; (3) Do enriched seeds reliably produce seedlings with δ 15N values higher than those of normal seedlings?; (4) What is the optimal stage to sample seedlings for 15N isotope enrichment analysis?; and (5) Is it possible to accurately discriminate seedlings produced by 15N-enriched seeds against those germinated from natural seeds in the field?

2 Materials and method

2.1 15N isotope enrichment in seeds and seedlings

In autumn of 2011, we soaked seeds of Pinus koraiensis (Pinaceae) and acorns of Quercus variabilis (Fagaceae) and Cyclobalanopsis glauca (Fagaceae) in 15N-urea solutions in Henan University of Science and Technology, Luoyang, China, to see if they and their seedlings were successfully enriched with the 15N isotope. Three dosage treatment solutions (0, 1, and 2 mmol/L) of 15N-urea (98 atom%; Isotec, Sigma-Aldrich, St. Louis, Missouri, USA) were used. Sound and healthy-looking seeds of the three tree species were used. Trees of Q. variabilis and C. glauca produce large acorns rich in starch, whereas seeds of P. koraiensis are small and rich in fat. P. koraiensis was included to test whether seeds rich in fat can successfully absorb the 15N-urea solution. Acorns of Q. variabilis exhibit no dormancy and germinate immediately at or even before seed fall (Yi et al. 2012a). In contrast, acorns of C. glauca and seeds of P. koraiensis remain dormant through winter until the next spring. This design further allows us to test whether differences in seed dormancy affects urea solution absorption. We selected 100 seeds and acorns of each tree species and separately soaked them into each dosage solution of 15N-urea. Thus, 300 propagules of each tree species were used for the urea soaking experiments. Seeds and acorns with pericarps attached were half submerged in the solutions, and the solutions were stirred everyday to ensure normal respiration and metabolism of seeds/acorns. After 10 days of soaking, 15 seeds or acorns (seed coat and pericarps were excluded) of each species in each dosage solution were collected and divided into three individual samples. Each sample contained 1–3 g dry mass of five acorns or seeds and were ground for measurement of 15N enrichment. At the same time, 20 acorns of Q. variabilis in each dosage solution were randomly selected and planted into an organic soil in plastic trays with 50 grids (length × width × height: 4 cm × 4 cm × 6 cm) to see if seedlings were enriched with 15N. To determine how long the isotopic signal lasted in the growing seedlings, the aboveground parts of three seedlings were individually harvested for 15N isotope analysis when the seedlings reached 5–6 cm (first stage), 8–10 cm (second stage), and 12–15 cm (third stage) in height (see below), respectively. We then repeated the above procedures after 20 and 30 days of soaking to determine 15N enrichment both in the acorns and seedlings, respectively. To investigate if the 15N-urea soaked seeds of P. koraiensis can reliably produce seedlings with enriched δ 15N values, we planted 15 seeds soaked in 0, 1, and 2 mmol/L 15N-urea for 20 days in the plastic trays, respectively (this procedure allows us to save money for isotope analysis). When the seedlings reached the maximum of the first flush (see below), the aboveground parts of three seedlings at each dosage solution were individually harvested for 15N isotope analyses.

2.2 Assessing seedling recruitment in the field

To investigate whether the 15N soaking method is suitable for seed dispersal and seedling assessing, we conducted a field experiment using 15N-enriched acorns of Q. variabilis soaked in 2 mmol/L 15N-urea solution for 20 days. In December 2011, 600 15N-enriched acorns were selected and released at two seed stations (2 × 2 m2) in Tianchi Mountain Natural Reserve, Luoyang, China (average elevation 1,400 m, 33°45′–33°85′N, 111°75′–112°45′E), with 300 acorns in each station. Seed removal rate and seed fate were not checked. Previous studies in the experimental area show that small rodents (Sciurotamis davidianus, Apodemus peninsulae, Niviventer confucianus, Apodemus agrarius, and Tscherskia triton) usually disperse acorns less than 25 m (Yi’s unpublished data); therefore, one-year seedlings (6–10 cm high) of Q. variabilis were located around each seed station (radius ≈30 m) and were sampled for 15N isotope analysis in the spring of 2012. The epicotyl was sampled from each seedling and was cleaned and dried for 15N isotope analysis. The enriched seedlings were identified by comparing with the natural 15N values in nature (Craine et al. 2012).

2.3 Stable isotope analysis

All samples (seeds, acorns, and seedlings) were cleaned using distilled water and air-dried to constant weight in an oven at 70 °C for 48 h. They were ground finely and placed in an isotope ratio spectrometer for isotopic analysis using elemental analyzer/continuous flow isotope ratio mass spectrometry. Samples were analyzed for stable nitrogen isotope abundance at the Laboratory of Stable Isotope Spectrometer, Chinese Academy of Forestry Sciences (Beijing). Interface between element-analysis meter and spectrometer was Flash EA1112 HT (Thermo Finnigan, USA). Operation condition: oxidizing furnace temperature was 900 °C, reducing furnace was 680 °C, and pillar temperature was 40 °C. The resulting N2 was purified in a vacuum line and injected in a Finnigan MAT Delta V advantage spectrometer (Thermo Fisher Scientific, Inc., USA) fitted with double inlet and collector systems. The results are expressed in δ 15N relative to the standards in the conventional δ per mil notation with a standard deviation of 0.2 per mil:

$$ {\delta}^{15}\mathrm{N}=\left[\left({}^{15}\mathrm{N}/{}^{14}\mathrm{N}\right)\ \mathrm{sample}/\left({}^{15}\mathrm{N}/{}^{14}\mathrm{N}\right)\ \mathrm{standard}-1\right]\times 1000 $$

Where 15N/14N are the isotopic ratios of sample and standard (atmospheric nitrogen).

2.4 Data analysis

We conducted all statistical analyses in SPSS 16.0. Two-way ANOVA was used to test the effects of soaking duration and urea dosage on 15N enrichment in the seeds or acorns of Q. variabilis, C. glauca, and P. koraiensis. Three-way ANOVA was applied to detect the effects of soaking duration, urea dosage, and sampling stage on 15N enrichment in the seedlings of Q. variabilis. Because the effects of soaking duration and sampling stage were not evaluated, difference in the 15N enrichment in seedlings of P. koraiensis in response to urea dosage was tested using one-way ANOVA. A nonparametric Mann–Whitney test was used to compare the difference in 15N enrichment in the seedlings of Q. variabilis collected from the field.

3 Results

3.1 15N enrichment in acorns/seeds

Our soaking experiments showed that the dosage of 15N-urea solution had significant effects on 15N enrichment in the acorns of Q. variabilis (F = 17.680, df = 2, P < 0.001) (Fig. 1). The dosage of 2 mmol/L 15N-urea solution resulted in much higher 15N in acorns than those of 0 and 1 mmol/L solutions (all P < 0.001). However, we found no significant influence of soaking duration on 15N enrichment in the acorns (F = 2.520, df = 2, P = 0.108) (Fig. 1). Similarly, we detected significant effects of urea dosage on 15N enrichment in the acorns of C. glauca and seeds of P. koraiensis (F = 23.338, df = 2, P < 0.001; F = 3.366, df = 2, P = 0.057) (Fig. 2). Soaking duration showed no effect on 15N enrichment in seeds or acorns of these two tree species (F = 1.064, df = 2, P = 0.366; F = 1.928, df = 2, P = 0.174) (Fig. 2).

Fig. 1
figure 1

The effects of urea dosage and soaking duration on 15N enrichment in the acorns of Q. variabilis. Data are expressed as mean ± SE

Full size image
Fig. 2
figure 2

The effects of urea dosage and soaking duration on 15N enrichment in the propagules of C. glauca and P. koraiensis. Data are expressed as mean ± SE

Full size image

3.2 15N enrichment in seedlings

Three-way ANOVA indicated significant effects of soaking duration, urea dosage, and sampling stage on 15N enrichment in the seedlings of Q. variabilis (F = 9.159, df = 2, P < 0.001; F = 31.884, df = 2, P < 0.001; F = 4.713, df = 2, P = 0.013) (Fig. 3). We also found a significant interaction effect between any two of the three factors (all P < 0.05). Multiple comparisons further demonstrated that 15N values in the seedlings sampled at the third stage (12–15 cm high) were much lower than those sampled at the first and second stages (P = 0.003; P = 0.145, respectively) (Fig. 3), indicating that seedlings 8–10 cm in height are optimal to be sampled for 15N isotope analysis. Although no effect of soaking duration on isotope level in acorns was observed, seedlings produced by acorns soaked for 30 days exhibited much higher 15N values than those soaked for 10 and 20 days, respectively (P < 0.001; P = 0.015) (Fig. 3). Multiple comparison tests also indicated that seedlings produced by acorns soaked in 2 mmol/L dosage solution exhibited much higher 15N values than those soaked in 0 and 1 mmol/L dosage solutions, respectively (all P < 0.001). We found a reliable and gradual 15N enrichment in P. koraiensis seedlings produced by seeds soaked in 0, 1, and 2 mmol/L dosage urea solutions for 20 days (F = 179.837, df = 2, P < 0.001) (Fig. 4).

Fig. 3
figure 3

The effects of urea dosage, soaking duration, and sampling stage on 15N enrichment in the seedlings of Q. variabilis. Embedded images indicate three seedling sampling stages. Data are expressed as mean ± SE

Full size image
Fig. 4
figure 4

The effects of 15N-urea dosage on 15N enrichment in the seedlings of P. koraiensis. Different letters on the histograms indicate significance at α = 0.05 level. Embedded image indicates seedling sampling stage. Data are expressed as mean ± SE

Full size image

3.3 15N-enriched seedlings in the field

In the spring of 2012, we located 28 seedlings of Q. variabilis around the two seed stations (radius ≈30 m) through extensive search by two people for 2 days. Stable nitrogen isotope analyses indicated that five of them showed much higher δ 15N values (867.28 ± 177.89 ‰) than normal seedlings (10.89 ± 0.47 ‰) (Z = 3.449, P = 0.001) (Fig. 5), indicating that these five unnatural seedlings definitely were originated from the enriched acorns we released in the winter of 2011. The average distance between the enriched seedlings and the corresponding seed stations was 11.9 ± 4.0 m. We failed to detect seedlings from the other 595 15N-enriched acorns possibly because they were either consumed, larder hoarded, or perished.

Fig. 5
figure 5

Difference between δ 15N values of enriched seedlings and background seedlings collected in the field. Different letters on the histograms indicate significance at α = 0.05 level. Data are expressed as mean ± SE

Full size image

4 Discussion

4.1 Effectiveness of isotopic urea soaking method

Our experiments clearly showed that 15N isotope was successfully incorporated into the acorns and seeds of Q. variabilis, C. glauca, and P. koraiensis. 15N enrichment could be expected in the seedlings of other tree species through urea soaking because we observed a very high enrichment of stable N isotope in the seedlings of Q. variabilis and P. koraiensis in the present study. We found that 15N enrichment in the seeds and seedlings were urea dosage and sampling stage-dependent. While, the significant effect of soaking duration on 15N enrichment in seedlings but not in acorns can possibly be explained by the absorption of 15N from the pericarps by Q. variabilis seedlings because acorns were not shelled when sowing. Higher concentration of 15N-urea and longer duration of soaking would generate higher levels of 15N enrichment both in seeds and seedlings. These findings address the first four questions we addressed in the introduction. According to the results of our experiments, we recommended a treatment of 2 mmol/L 15N-urea dosage and 20-day soaking duration for labeling seeds and tracking seedling recruitment in the field. However, 15N isotope tended to be diluted in the growing seedlings, echoing the results of Carlo et al. (2009). Therefore, sampling seedlings in proper development stage is crucial for 15N isotope analysis. Taking 15N-urea concentration and seedling morphological performance into account, we suggest that 8–10 cm high seedlings are ideal for sampling and 15N analysis. Compared with the natural 15N values in nature (Craine et al. 2012), we successfully tracked several seedlings produced by 15N-enriched acorns of Q. variabilis in the field, verifying the feasibility of 15N-urea soaking to tag seeds for tracking their dispersal and recruitment. Compared to the urea foliar-spraying method (Carlo et al. 2009), soaking seeds directly in 15N-urea circumvented some methodological problems encountered when trying to label propagules of large plants with high trunks and canopies that are difficult to reach.

4.2 Pros and cons of the soaking method

Our study on these three tree species suggests that isotope marking by 15N-urea soaking is useful to investigate seed dispersal and seedling recruitment. Although this method omits the exact seed movement provided by animals during the dispersal process in the field, it is more labor-saving and less costly compared to molecular analyses to identify the mother trees (Godoy and Jordano 2001; Carlo et al. 2009). Therefore, our method permits the exact location of the seedling establishment to be determined, which facilitates the measurement of dispersal distance from the parent plants. This new isotope marking method overcomes the methodological bottlenecks that “large plants or trees will require large volumes of urea solutions, and canopies are obviously hard to reach” (Carlo et al. 2009). 15N-urea solution soaking method also avoids the influence of weather conditions and manual application procedures on the variations in δ 15N values that urea foliar-spraying methods usually face. Although this method is not very suitable for tracking seed dispersal of many individuals within the same site, distinction among seeds and seedlings can be made by the dosage level provided (Carlo et al. 2009). Moreover, searching for enriched seedlings in large areas is expected to be much easier than searching for enriched seeds passed through avian passage (Carlo et al. 2009). Compared to the traditional tracking methods (Iida 1996, 2006; Yi and Zhang 2008; Yi et al. 2011; Canner and Spence 2011), the 15N-urea solution soaking method provides a nondestructive approach to track seed dispersal, which will authentically reflect the seed dispersal processes because the traditional methods always involve surgical manipulations on seeds (e.g., inserting magnets or transmitters, drilling holes, and attaching transmitters) (Alverson and Díaz 1989; Winn 1989; Forget 1992; Den Ouden et al. 2005; Pons and Pausas 2007; Yi and Zhang 2008; Yi et al. 2011). These manipulations frequently lead to unexpected seed death, fungal contamination, seed weight modification, and seed consumption (Alverson and Díaz 1989; Den Ouden et al. 2005; Iida 1996, 2006; Canner and Spence 2011; Hirsch et al. 2012a, 2012b). Furthermore, our isotope tagging method shows an advantage over the traditional methods (Yi et al. 2008; Niu et al. 2011; Hirsch et al. 2012a, 2012b) that it does not add any visual cues to the 15N-enriched seeds, which resembles the natural seeds and prevents animals from using these cues for recovering the cached seeds. This method also has several shortcomings: (1) it requires much effort to locate all potential seedlings over a wide area after seed dispersal, and therefore (2) it is not well appropriate for long distance dispersal by fruigivory, (3) although it may be powerful to study “secondary dispersal” such as scatterhoarding by rodents and similar processes, the urea soaking method does not permits the study of dispersal of plants that obtain seed removal directly from the canopies (like fruigivory). Therefore, 15N-urea soaking approach is expected to have a great potential for providing insights into seed dispersal and seedling recruitment manipulated by scatterhoarding animals.

4.3 Broad application of tracking seed dispersal

Although seeds of the three tree species are different in the germination schedule (dormant or nondormant), chemical composition (e.g., tannin), seed coat thickness, and morphological size, 15N isotope was ubiquitously enriched in seeds and seedlings of Q. variabilis, C. glauca, and P. koraiensis. These facts imply that 15N enrichment is unlikely to be affected by seed trait. Traditional seed-tracking methods are often limited by the size of seeds, and thus, only seeds of large size are selected for labeling and tracking (Sone and Kohno 1996; Mack and Druliner 2003; Yi et al. 2008, 2012b). Given that seed species of any size can successfully germinate and establish, our method makes it possible to track dispersal and recruitment of both small- and large-seeded plant species. Although our method may not efficient to study “natural” removal rates of seeds, especially those seeds contained within fleshy fruits, it is expected to be ideal for “seed addition” experiments to track seed dispersal of some tree species (Nathan and Mueller-Landau 2000; Clark et al. 2007). Our study verified the validation of 15N-urea soaking in tracking seed dispersal and seedling recruitment of a number of tree species and provided the confidence to distinguish between the seeds added and those arrived naturally. We believe that this promising method will improve the seed-tracking efficiency of seed dispersal ecology.