Introduction

Gall-inducing insects have been called “nature’s most sophisticated herbivores” due to their ability to stimulate atypical plant growths that provide the insect with both food and shelter ( ^{Shorthouse, Wool, &Raman, 2005}). The mechanisms by which galls are induced are poorly understood, although this topic has received considerable attention in a species that is a pest of wheat (^{Stuart, Chen, Shukle, & Harris, 2012}) as well as various gall-inducers in Brazil (^{Oliveira et al., 2016}). In addition to species in six orders of insects, certain mites (especially Eriophyidae) also induce plant galls (^{Raman, Schaefer, & Withers, 2005}). The majority of these gall-inducing arthropods are highly host-specific (^{Butterill & Novotny, 2015}) and each species of gall-inducer produces a diagnostic gall morphology (^{Isaias, Carneiro, Oliveira, & Santos, 2013}). Because most gall-inducing arthropods are undescribed species, gall inventories are usually based on a combination of plant identification and gall morphotyping (^{Espirito-Santo & Fernandes, 2007}; ^{Carneiro et al., 2009}), the latter serving as an extended phenotype of the gall-inducer (^{Carneiro, Pacheco, & Isaias, 2015}).

There have been an increasing number of inventories of neotropical galls during the last two decades (^{Gagné, 1994}; ^{Hanson & Gómez-Laurito, 2005}; ^{Fernandes, Coelho & Santos, 2014}; ^{Maia & Mascarenhas, 2017}). Gall-inducers appear to be less diverse on ferns, gymnosperms and monocots (^{Gagné, 1994}; ^{Santos, Hanson, Maia, & Mehltreter, 2019}). In contrast, it has been suggested that as many as 90 % of dicots could harbor gall-inducing cecidomyiids, which is probably the most diverse group of gall-inducers (^{Hanson & Gómez-Laurito, 2005}). Studies have shown that certain plant families and genera have a particularly rich fauna of gall-inducing arthropods (^{Araújo, Santos, & Gomes-Klein, 2012}), a trend that is correlated to taxon size (^{Fernandes, 1992}; ^{Araújo, 2011}; ^{Araújo et al., 2012}). In addition, some plant species have been referred to as “superhosts” due to their relatively high species richness of gall-inducers (^{Fernandes et al., 1996}).

Despite the increasing number of gall inventories, comparisons between surveyed areas are problematic since different sampling efforts and methodologies have been used. Furthermore, a thorough inventory of gall-inducers, even at a single site, is fraught with difficulties (^{Hanson & Gómez-Laurito, 2005}): many galls are hidden inside fruits and other plant organs (or in the roots), galls in the canopy are generally inaccessible (but see Julião, Venticinque, ^{Fernandes, & Price, 2014}), not all individuals of a plant species harbor galls, and some galls are seasonal. In Costa Rica there have been extensive surveys of both plants (^{Hammel, Grayum, Herrera, & Zamora, 2003}-2015) and gall-inducers (^{Hanson & Gómez-Laurito, 2005}), yet there are no intensive gall inventories for particular sites. Therefore, a major aim of the present investigation was to provide an inventory of the galls for a 345 ha reserve in the lowlands of Costa Rica.

In addition to providing the first gall inventory of a particular site in Costa Rica, the present investigation also included two components that, to the best of our knowledge, have never been included in previous inventories. First, in addition to inventorying the galls and their host plants, we also inventoried the vascular plants of the site, thereby including negative records. Second, by conducting a complete plant inventory, we were able to broaden the survey to include plant-gall records from other parts of the country as well as from herbarium samples.

Materials and methods

Study site: Fieldwork was carried out at the Tirimbina Biological Reserve (TBR, 10°25’ N & 84°47’ W), Sarapiqui, Heredia, Costa Rica. This site is considered a humid tropical wet forest (^{Holdridge, 1978}), between 180 and 220 m a.s.l., with an average precipitation and temperature of 3 777 mm and 24.3 °C, respectively. The reserve consists of 345 ha, mainly of mature forest, and is bordered on its western edge by the Sarapiqui River; there was occasional selective logging between 1962 and 1990. The reserve maintains a trail system of ca. 10 km. TBR is located 17 km from La Selva Biological Station (belonging to the Organization for Tropical Studies), one of the best sampled areas in the Neotropics in terms of plants (see, http://sura.ots.ac.cr/florula4/).

Plant inventory: The principal plant inventory was carried out by JG under a consultancy paid by the TBR from June 2008 to October 2009. During this period around 75 field samplings, comprising on average 4 h each, were done by walking the trails in the reserve. All vascular plants observed and identified in the field were annotated in the list. Doubtful species were collected to be identified later in the herbarium and fertile individuals were collected to provide herbarium specimens. In total, 1 213 species were identified during this period of which 1 131 samples belonging to 657 species were deposited at the La Selva Biological Station Herbarium (LSCR). We excluded records of non-native plants in the reserve since galls are unlikely to occur on these species. Scientific names were revised using ^{The Taxonomic Name Resolution Service (2018}) with data derived from ^{Tropicos (2018}) and ^{USDA (2018}) plant databases. Although we recognize that “dicots” represent a paraphyletic group, we use the term for convenience.

Gall inventory: Galls were sampled by three methods. From November 2013 to December 2016 a general sampling was carried out by haphazardly walking the trails and examining the vegetation along the pathways at least twice a week for an average of 2 h per sampling period. Second, to complement the first method, galls were collected opportunistically throughout the study period during the course of other research projects in the reserve. Third, from August to October 2016, 22 random transects of 50 x 2 m were performed across the reserve in which every plant was sampled. The main goal of the latter was to assess the effectiveness of previous sampling. Galls were collected primarily from leaves and stems; roots, flowers and fruits were generally not examined (e.g. all native figs have galls in the syconia, but these were not included).

Samples of galls were collected in plastic bags for later examination in the laboratory while plants were identified in situ or, when necessary, plant samples were collected for later identification. Plant identifications benefitted from the previous plant list that had been prepared for the reserve. Whenever possible, a subset of galls was dissected under a microscope in order to attempt to determine the family-level identity of the gall-inducer, although this was not the primary focus of the present investigation and therefore only a general summary of the gall-inducers is provided here.

Two additional sources of data were utilized in order to provide a more complete inventory of the galls. First, for plant species present in the TBR we added records (where available) of galls from the same plant species collected elsewhere in the country during the last 25 years (Hanson, unpublished data). Second, during 2017 we searched for galls on specimens at the University of Costa Rica Herbarium (USJ), examining all dicot species reported from the reserve; dicots were chosen because galls are much less common on other groups of vascular plants (^{Hanson & Gómez-Laurito, 2005}). We examined every sample found in the herbarium (regardless of locality in Costa Rica) although for species with a very large number of specimens only the two folders with the most specimens were examined. All samples with signs of galls were observed under the microscope and then assigned to one of three categories: probably not a gall (these records were excluded), almost certainly a gall, or questionably a gall (these records are indicated by a question mark in the results). It is generally impossible to determine the identity of the gall-inducer based on herbarium specimens (except perhaps in the case mite erinea - hyperplasia of plant trichomes in response to mite activity), so this was not attempted, nor was this the focus of the present investigation.

Results

Plant diversity: A total of 1 174 native plant species belonging to 143 families were found in the Tirimbina Biological Reserve (Tables 1-2). Pteridophytes (ferns and their allies) were represented by 122 species, while gymnosperms were represented by only two species. Angiosperms were represented by 242 species of monocots and 808 of dicots (Table 1). The six most species-rich families, in descending order, were Rubiaceae (79), Fabaceae (67), Araceae (66), Melastomataceae (57), Asteraceae (39) and Piperaceae (37). Seven additional families were represented by 20 or more species each and 24 were represented by 10-19 species. On the other hand, 35 families were represented by only one species (Table 2).

Gall diversity: In total, we found 401 plant species that hosted galls, representing 34.16 % of the flora of the reserve (Table 1). When considering only dicots this percentage increases to 44.93 %. Host plants of galls came from the following records: 209 from field sampling, 158 from herbarium specimens, and 257 from previous records in the country (Table 1). We considered our field sampling to be satisfactory since in the transects done at the end of our field study we found only four new host plant records (from 122 individuals with galls belonging to 65 species). Galls found in the field represent 17.80 % of the flora in the reserve (and 23.26 % of the dicots); 77 gall records from the field were also found in the herbarium and 110 were already reported in Costa Rica. It should be noted that 92 previous gall records were from La Selva Biological Station (17 km away from TBR). From the 158 gall records in the herbarium we classified 118 as almost certainly a gall and 40 as possibly galls (requiring confirmation). From the 40 doubtful records in herbarium, 21 were also found in the field or as previous records, suggesting that these doubtful records from the herbarium are probably gall hosts. It is interesting to note that the herbarium records were not homogeneous across taxa; for example, we found galls in the herbarium on all seven species of Burseraceae, but only four of the 79 species of Rubiaceae (Table 2).

All of the seven most species-rich plant families also included the greatest number of plant species with galls: Fabaceae (29), Melastomataceae (23), Rubiaceae (23), Araceae (21), Moraceae (17), Piperaceae (16), and Asteraceae (16) (Table 2). Nonetheless, most dicot families have similar proportions of gall-harboring species. The average percentage of gall-harboring species in the five most species-rich dicot families was 38.35 % while for the next four most species-rich families (all with 20 or more plant species) the average was 50.98 %. Similarly, for the 14 dicots families with 10-19 species the percentage was 50.28 %; 28 of the 43 dicot families with fewer than three species had at least one species harboring galls. Some families had a high proportion of species with galls: Bignoniaceae (10/10), Burseraceae (7/7), Chrysobalanaceae (5/5), Dilleniaceae (4/4), Nyctaginaceae (5/6), Sapindaceae (8/10), Sapotaceae (7/8) and Vochysiaceae (4/4). Similarly, most species in genera such as Mikania (Asteraceae), Inga (Fabaceae), Ficus (Moraceae), Eugenia (Myrtaceae), Neea (Nyctaginaceae), Pouteria (Sapotaceae), and Smilax (Smilacaceae) had galls, as did all species of Protium (Burseraceae), Paullinia (Sapindaceae), and Vochysia (Vochysiaceae).

With respect to the gall-inducers the vast majority of the galls were induced by Cecidomyiidae (Diptera), while the second most common category was “gall-inducer uncertain”. The latter was generally the result of galls being too young or too old, and obviously nothing can be concluded from herbarium records (but see, ^{Veenstra, 2012}). Several plant species had more than one species of gall-inducer (data to be presented elsewhere). Following are those that had only one, non-ceciodmyiid, gall inducer. Mites (Eriophyidae) were the only gall-inducers present on Acalypha macrostachya, Croton draco (Euhorbiaceae), Calatola costaricensis (Icacinaceae), Ochroma pyramidale, Quararibea bracteolosa (Malvaceae), Miconia barbinervis (Melastomataceae), and Lantana camara (Verbenaceae). Psylloidea (Hemiptera) were the only gall-inducers on Lonchocarpus ferrugineus, L. heptaphyllus (Fabaceae), Ocotea laetevirens, O. mollifolia (Lauraceae), Brosimum alicastrum, Naucleopsis naga, Pseudolmedia spuria, and Sorocea pubivena (Moraceae); Eriococcidae (Hemiptera) on Ceiba pentandra (Malvaceae); Buprestidae (Coleoptera) on Amphilophium paniculatum (Bignoniaceae); Curculionidae (Coleoptera) on Psychotria marginata (Rubiaceae); and Lepidoptera on Cuphea epilobiifolia (Lythraceae), Graffenrieda galeottii, Leandra granatensis (Melastomataceae), and Zanthoxylum riedelianum (Rutaceae). The vast majority of galls were from leaves or stems; some notable exceptions include root galls on many Araceae, and fruit galls on Geonoma cuneata (Arecaceae) and Miconia longifolia (the latter induced by a braconid wasp).

Discussion

In this study we report 401 host plants of galls which represents 34.16 % of the total plant species recorded in the reserve. Even when considering only plants with galls actually found in the reserve (209; i.e., excluding species for which galls were recorded only in herbarium specimens not from TBR or by previous records from elsewhere in the country), this number is higher than that reported in many previous gall inventories in the Neotropics (^{Cuevas-Reyes, Siebe, Martínez-Ramos, & Oyama, 2003}; ^{Araújo, Silva, Santos, & Gomes-Klein, 2013}). ^{Maia & Mascarenhas (2017}) found 432 gall morphotypes on 211 plant species in Parque Nacional do Itatiaia, Brazil, and considered this area as having the highest gall richness studied thus far in Brazil. However, it should be emphasized that gall inventories vary with respect to methodology, size of the study area, local plant diversity, and sampling effort (^{Dalbem & Mendonça, 2006}). Compared with most previous inventories our sampling effort was greater and was carried out for a longer period of time (three years of weekly observations), which probably explains the high number of galls. Several galls were found just once during the entire study period (e.g., on Philodendron ligulatum, Araceae), or only during a certain time of year (e.g., galls in infrutescences of Anthurium ochranthum, Araceae), or only in certain individuals of a plant species.

Although we found galls in the field on only 17.8 % of the plant species in the reserve (but in almost 45 % of the dicot species in the reserve), the actual percentage is undoubtedly much higher. The fact that we did not find galls on 147 plant species for which records exists elsewhere in Costa Rica (many of them from nearby La Selva or Braulio Carrillo National Park), strongly suggests that many of these gall-inducing arthropods are in fact present in TBR (or that some gall-inducers have a patchy distribution). Moreover, we did not sample from the canopy, which has undoubtedly resulted in our underestimating the species richness of gall-inducers in the reserve. It has been shown that the species richness of gall-inducers is higher in the canopy than in the understory (^{Medianero, Valderrama, & Barrios, 2003}; ^{Ribeiro, Basset, & Kitching, 2014}). ^{Julião et al. (2014}) sampled the upper strata of 1 091 trees belonging to 491 species in two reserves in Amazonia and found galls on 90 % of the species.

Among dicots, large plant families generally have a greater absolute species richness of gall-inducers (^{Fernandes, 1992}; ^{Mendonça, 2007}; ^{Araújo, 2011}; Araújo et al., 2012), a trend that is also noticeable in the results of the present study. In addition to having more species available as potential hosts of galls, large families are more likely to include abundant species and superhosts. However, by taking into account the complete plant list for the reserve, our results suggest that the proportion of plant species with galls is similar among most dicot families. The situation among monocots is probably quite different. For example, palms (Arecaceae) appear to have very few galls; the first known gall from a New World palm was recently found as a part of the present study (^{Gagné, Ley-López, & Hanson, 2018}). On the other hand, families that consist primarily of epiphytes (Araceae, Bromeliaceae, Orchidaceae) probably have an underestimated species-richness of galls due to logistical difficulties in sampling. For this same reason our list of orchids in the TBR (Table 2) is very incomplete compared with that of other plant families.

Half of our records of galls were from herbarium specimens or from previous records elsewhere in the country. Given the difficulties in finding galls in the field (see Introduction), especially in wet tropical forests, these methods can be useful in supplementing gall inventories. These indirect methods, however, have limitations. Examining herbarium specimens for galls requires considerable previous experience with field-collected galls, but even with this experience, several uncertainties remain. For example, discolored leaf spots due to fungal pathogens can be easily mistaken for blister galls and we therefore excluded ambiguous cases. Even more difficult to distinguish in herbarium specimens, if not impossible, are holes in leaves caused by necrotic galls becoming detached (^{Fernandes et al., 2012}). Some galls (e.g. those that are large or detachable) are probably underrepresented in herbarium samples, and galls hidden in flowers or fruits cannot be detected without destroying the specimen (which we did not do). Root galls are also virtually impossible to find in the field (except in exposed roots of epiphytes or aerial adventitious roots of other plants), and are very underrepresented in herbaria. Gall records for which there is only a questionable herbarium record (i.e., indicated by nothing more than H? in Table 2) require field confirmation before being accepted. Nonetheless, the fact that 52.5 % of our doubtful herbarium gall records were also recorded in our field sampling or from previous records in the country lends some credibility to the careful use of herbarium specimens.

As with herbarium records (which were not specimens from the TBR), records of galls collected from elsewhere in the country (indicated in Table 2 by an asterisk) are no guarantee that the gall is present in the TBR. There is extremely little information available on the geographic distribution of gall-inducers, especially in the tropics and especially for taxa such as Cecidomyiidae where the vast majority of species are undescribed. A gall inventory done about 400 km away from our study site, at Coiba National Park, Panama (^{Nieves-Aldrey, Ibañez, & Medianero, 2008}), identified 34 host plant species, of which 16 also occur at our study site. We found galls on ten of these plant species at the TBR, and another five are known to harbor galls elsewhere in Costa Rica (Hanson, unpublished data). More research is needed to determine how many of these shared host plant records represent identical species of gall-inducers. ^{Arriola, Júnior, Mouga, Isaias, & Costa (2016}) provided evidence that Calophyllum brasiliense and seven cecidomyiid gall-inducers on this plant have overlapping geographical distributions. However, such information is lacking for the vast majority of gall-inducers, and although they are highly host specific, not all gall-inducers are restricted to just a single plant species. Several plant species with a wide geographic distribution have been recorded harboring galls in other countries, but not in Costa Rica, at least thus far. For example, at least four plant species present in the TBR are reported to harbor galls in Brazil, but thus far no records exist in Costa Rica: Mikania guaco, Paspalum conjugatum (^{Gagné, 1994}), Inga thibaudiana and Lecointea amazonica (^{Julião et al., 2014}).

In conclusion, despite the increasing number of gall inventories in the Neotropics, there is still much more work to be done, especially in the taxonomy of the gall-inducers. The results of the present investigation demonstrate the usefulness of simultaneously inventorying the plants and the galls, and of supplementing transects with longer-term, opportunistic collecting, as well as with records from nearby sites and herbaria. Deliberate inclusion of galls in herbarium samples, which is presumably not a general practice, should perhaps be encouraged since this could provide valuable information for gall researchers as well as for botanists (^{Veenstra, 2012}; ^{Rat & Anačkov, 2012}).

Ethical statement: authors declare that they all agree with this publication and made significant contributions; that there is no conflict of interest of any kind; and that we followed all pertinent ethical and legal procedures and requirements. All financial sources are fully and clearly stated in the acknowledgements section. A signed document has been filed in the journal archives.