Study Area

The Municipality of Calakmul is located in southeastern Campeche State, Mexico, in the center of the Yucatan Peninsula. With an area of 16,806 km², Mpio. Calakmul borders the State of Quintana Roo to the east and the Department of El Petén, Guatemala, to the south. Almost half of the municipality was placed under protection by decree of the Calakmul Biosphere Reserve (CR) in 1989 (Fig. 1). The reserve includes an area of 7,232 km², with an altitude 100 to 300 m above sea level. The Reserve has a semi-humid, warm climate. Annual precipitation ranges from 600-1200 mm, and annual average temperature is 24.6°C (Fig. 2) (INEGI 1996). The dry season extends from Feb to May, followed by the rainy season from Jun to Sep, which in turn is followed by the “nortes” (or cold) season, from Oct to Jan. “Nortes” refer to frequent fronts of cold winds that cross the Gulf of Mexico and carry humidity from a NE to SW direction (Williams-Linera & Meave 2002).

In the CR, 7 vegetation types are found: tall tropical evergreen forest, medium height tropical semi-evergreen forest, medium height tropical semi-deciduous forest, savanna, low tropical semi-evergreen forest, low tropical semi-deciduous forest and, low tropical semi-deciduous seasonally flooded forest. The medium height tropical semi-evergreen forest and low tropical semi-deciduous forest dominate the landscape within the CR (Martínez & Galindo-Leal 2002).

Sampling

The butterfly fauna of CR was surveyed during 266 d distributed over 26 months, across 3 years (Mar 1997-Dec 1999), representing approximately 5562 person-h and 8840 trap-h (Table 1). Sampling was conducted at sites dominated by the most representative vegetation types of the region, medium height tropical semi-evergreen forest (selva mediana, or Upland Standard Forest sensu Schulze 1992) and low tropical semi-deciduous forest (selva baja or Low Forest sensu Rzedowski 1983), as determined by botanist Esteban Martínez. Within each forest type, transects of 500 m were established within habitat types of different successional stages, including primary forest, forest disturbed more than 10 years ago, and forest disturbed within the past 10 years. In each habitat type, replicate transects were sampled (separated by at least 300 m), for a total of 18 transects in the first year, and 12 transects during the second and third years (Table 2), at a total of 11 localities. Six localities are within or near the northern portion of the CR, and 5 are within or near the southern part of the CR (Fig. 1).

The first year, sampling was conducted 4 tines each season at each locality. The second year, sampling was conducted only during the months of greatest butterfly abundance, from Jun to Nov. The third year, all localities were sampled monthly. Two sampling methods were utilized, direct search (DS) and transects (T). For each method, 2 sampling techniques were employed: traps with bait (t) and entomological nets (n). Transects (T) were established through the forest by clearing a small path of 500 m length, with a machete, and wide enough to permit the passage of the observer without altering the canopy. In each transect, 10 Van Someren-Rydon traps (Rydon 1964) baited with banana, pineapple, and beer, were hung (from 0700 to 1800 h) at 50 m intervals, on alternate sides of the path. Fresh bait was added each morning. Two surveyors spent 120 to 150 min in each transect. Surveyors walked each transect, spending 10 min at each trap to record trap contents (on a pre-prepared form) and sample nearby individuals with entomological nets. Direct Search surveys (DS) were more opportunistic. These were conducted by walking along roads (2 observers for approximately 8 h per d) in search of sites where butterflies were concentrated (sunny gaps, nectar sources, damp ground, etc.), sampling butterflies as opportunities were encountered. In addition, 10 traps baited with fermenting fruit (as above) were placed in “optimal” localities along the roadsides and checked periodically.

Upon determination to species-level, trapped butterflies were usually released. In cases where the identity of a trapped butterfly was uncertain or unknown, it was collected for determination at, and integration into the Zoology Museum at El Colegio de la Frontera Sur, Chetumal, Mexico (duplicate specimens are deposited at the Zoology Museum of the Universidad Nacional Autónoma de Mexico, Mexico City). To determine the extent to which released butterflies may be re-trapped and re-counted, all individuals released for a 3-month period were marked before release, and recapture data was collected. Recapture data confirmed that once released from a trap, individuals did not usually return to traps (Hughes et al. 1998). However, only long-lived and phylopatric species, or those with a small “home range” could cause a bias in the data through recaptures, and few such species were encountered in the CR.

Data from fieldwork were entered into the database of the Lepidoptera collection at the Zoology Museum at ECOSUR, representing a total of 60,662 records of 359 species, in 207 genera, from 18 subfamilies, 5 families and 2 superfamilies (nomenclature follows Llorente et al. 2006; A. Warren 2000, except that for the analyses we considered the families Riodinidae and Lycaenidae together).

Analysis

Monthly species abundance (phenology) and seasonal distribution (seasonality) data were analyzed. In order to describe phenological patterns, analyses were divided into 2 parts. First, the total number of individuals and species recorded was considered, and a family-level analysis was conducted (Shapiro 1975). Second, the phenology of the most abundant species was analyzed (species with 100 or more records during the 3 years of sampling. The abundant species were divided into 3 size categories (Young 1982; Vargas-Fernández et al. 1999). Because of the great diversity of butterfly species within the CR, adult size showed great variability (e.g., Pyrisitia nise nelphe) (R. Felder 1869), (forewing length = 17 mm ± 0.9 mm) and Leptotes cassius cassidula (Boisduval 1870), (12.7 mm ± 1.1 mm) vs. Archaeoprepona demophon centralis (Fruhstorfer 1905), (54 mm ± 4.4) and Morpho achilles montezuma Guenée 1859 (67 mm ± 4.0 mm). Wingspan intervals were used as a proxy of body size (Vargas-Fernández et al. 1999). The 3 arbitrary wingspan size categories were: small (≤30 mm), medium (31-50 mm), and large (≥51 mm).

Seasonality

For each year, a data matrix was constructed, which recorded the species and their abundance in each season (dry, rainy, and cold or “nortes”). An NMDS analysis was applied to each matrix. Data was log-transformed according to the Box and Cox test (Legendre & Legendre 1998). For these analyses we used PRIMER 4.0 software (Carr 1996), and results were graphed with SigmaPlot Version 7.1, Systat. In order to statistically evaluate differences among groups in each season, an ANOSIM test (ANalyis Of SIMilarities) was applied, and subsequently, a SIMPER (SIMilarity PERcentages) analysis was applied to detect species that contributed to group formation, and those species which acted as discriminants among groups (Herrando-Pérez 2002).

Phenology

In general, the phenological patterns found in this study are similar to those described by de la Maza and de la Maza (1985a, 1985b) and Austin et al. (1996), with peak butterfly diversity at the end of the dry season and another peak during the rainy season. Reduced species diversity was observed from the end of the rainy season to the middle of the dry season (Fig. 3). Nine percent of the Calakmul butterfly species have been recorded in every month; similarly, Austin et al. (1996) reported 10% of the butterfly fauna from the Tikal area, Guatemala (about 100 km S of Calakmul), from all months.

The month with the greatest species richness was Oct 1997, with 231 species, and the month with the greatest relative abundance was Aug 1998, with 17,324 individuals (more than the 1997 annual total of 16,257 records). Maximum species diversity was observed during Oct and Nov, in agreement with previous studies in México (Vargas-Fernández et al. 1992, 1999). This period coincides with the end of the rainy season. The year with the greatest species richness was 1997, with 265 species recorded (74% of the total species number), although a similar diversity was observed in 1998 (261 species). During 1999, species richness decreased to just 58% of the total species diversity recorded in this study.

Large numbers of records for any given month do not necessarily indicate great species diversity. The greatest number of records obtained during any month of the study was 17,324 records in Aug 1998, representing 28.56% of the total records obtained during this study. However, these individuals represented only 108 species (Table 3), and 3 migratory species comprised 86.5% of the sample: Memphis pithyusa (R. Felder, 1869), Anaea troglodyta aidea (Guérin-Ménéville, 1844) and Eunica tatila tatila (Herrich-Schäffer, 1855). Memphis pithyusa was the most frequently recorded species in this study, with 12,848 records (Table 6).

Table 4 shows that 90.8% of the species recorded during this study were represented by only 19.7% of the recorded individuals. Species which were present from 9 to 12 months equaled 28.4% of total species recorded, and 95.4% of the total number of records. There were 140 species present during 6 or more months of the year (98.1% of all records), while 178 species were recorded only during 3 or fewer months, representing half of the fauna studied (49.58% of species and 1.31% of records). Among species surveyed in this study, 183 are represented by 10 or fewer records (60 of which were only recorded once). These are the “rare” species and represent 0.001% of the total number of records.

Butterflies were divided into 5 groups according to their relative abundance (Fig. 3). The first group contains species represented by 10 or fewer records (183 species), the second group by 11 to 100 records (110 species), the third by 101 to 1000 records (55 species), the fourth by 1001 to 10,000 records (10 species) and the fifth by more than 10,000 records. This last group only included one species (M. p. pithyusa). Monthly variation in species richness is clearly defined by the first 2 groups, represented by 293 species (81.6% of total species richness and 7.5% of total relative abundance), 96 of which were recorded during only 1 month (of those, 35 in Oct), and 60 of these represented by only 1 individual (21 in Oct) (Fig. 3). Table 4 shows species frequency by month, from those present all year (33 species), to those recorded only during 1 month. It is remarkable that half of the fauna (178 species) was recorded during 3 or fewer months, whereas 84 species were present during 10 or more months. Thus, half of the species are strongly seasonal (49.6%).

The 2 peaks in species richness coincide with the greatest relative abundance (Fig. 3a and Table 3). The largest peak was observed from Jun to Dec, which corresponds with the rainy season, although there were markedly fewer species during Sep. The smaller peak was observed from Feb to Apr. We estimated that a minimum of 110 species should fly each month, if we consider that 28-35% of species in this study were found during the entire year, and every year.

Family Level Phenology

The most diverse families in the study area are those showing the largest variation in species richness throughout the year (Hesperiidae = 135 species, Nymphalidae = 111 species; Fig. 4). In contrast to other families, peak diversity of Papilionidae was observed during the late dry season into the early rainy season, as reported by Austin et al. (1996) for the Tikal area. Relative abundance of Papilionidae was distributed irregularly throughout the year, but the greatest number of individuals (27.5%) was recorded in Apr (Table 5); the greatest species richness was in Jul (13 species), Apr (11 species) and Aug (11 species). Diversity and relative abundance of Pieridae species remained more or less constant throughout the year, showing a slight increase during the rainy months of Aug to Nov; Oct was the peak month for species richness (with 21 species) as well as relative abundance. Hesperiidae, Lycaenidae, and Nymphalidae showed 2 periods of increased species richness: the first (and lesser) occurred during the dry season, while the second (and greater) occurred during the rainy season; these two periods are separated by the dry month of May when a decreased species diversity was observed. Specifically, the greatest species diversity for Lycaenidae was Oct (44 species; a second peak in Mar with 25 species), with the greatest relative abundance in Aug; the lowest species diversity and relative abundance among Lycaenidae was observed in Jan. The greatest species diversity for Nymphalidae was observed in Oct (with a smaller peak in Mar), while the greatest relative abundance was in Aug (with a smaller peak in April). The difference in abundance of Nymphalidae between dry and rainy seasons was greatly accentuated: 90% between the months of highest and lowest abundance (Table 5). Similarly, over 2 times as many species of Lycaenidae, and over 3 times as many species of Hesperiidae were recorded in the rainy season than in the dry season (Table 5).

Phenology of Abundant Species Related to Size

Sixty-eight species were considered abundant (Table 6), and 75% of these were recorded from all months. These species are distributed among 5 families, but most (51 species) are nymphalids. These 68 species represent 91.6% of the records obtained in this study, but represent only 18.9% of the total species richness. Out of the 3 groups formed according to size categories, 25 species are small, 32 are medium, and 11 are large.

Some variation exists with respect to seasonal trends in species richness of the 3 size categories (Fig. 5). For the small-sized group, a difference of 40% exists between the months of greatest and least species richness. For medium-sized butterflies, this difference is 31.25%, and for large butterflies, a 36% difference exists. Figure 5b shows the largest relative abundance for the 3 categories from Jun to Dec, with a strong decrease during Sep.

Seasonality

Butterfly diversity was found to be significantly variable among dry, rainy, and cold seasons (Fig. 6 and Fig. 7) during each year of the study (1997: stress = 0.15; 1998: stress = 0.05; 1999: stress = 0.18). Of the 359 species used for this analysis, 9% (32 species) provided more than 80% of the effective data when grouping the communities in each of the 3 seasons (Table 7), 50% of differentiation among communities of each season was determined by the relative abundance of 31 species (8.6% of the total). Of the 32 species comprising the communities in each season, 2 did not contribute to the distinction among communities (Myscelia cyaniris cyaniris Doubleday (1848) and Siderone galanthis (Cramer 1775)), and 1 did not contribute to the formation of groups (Marpesia chiron marius (Cramer 1779)), although it did when groups were analyzed separately. Of these 32 species, only 3 did not belong to Nymphalidae: Kricogonia lyside (Godart 1819), Glutophrissa drusilla tenuis (Lamas 1981) and Polygonus savigny savigny (Latreille, (1824)) (Table 7).

When species richness by family in each season is considered (Fig. 7a), it is clear that Hesperiidae differed in phenological behavior from the other families. This family had the greatest species richness in the cold season, with values more than double of those observed during the rainy and dry seasons. Nymphalidae and Lycaenidae showed maximum species richness during the rainy season, with a slight decrease in the cold season, while Pieridae and Papilionidae showed minimal variations. Pieridae showed a slight increase during the rainy season, at a time when diversity of Papilionidae decreased to values lower than those of the dry season.

Relative abundance by season was strongly influenced by Nymphalidae (Fig. 7b). Seasonal differences among the other families were not detected on this scale, due to the abundance of nymphalid species. Nevertheless, abundance of Hesperiidae showed an increase in the cold season (Table 5).

Phenology Considering the Relative Abundance of Species

Wolda (1988b) concluded that there are differences in phenological patterns among insect faunas typical of temperate and tropical zones. Under the assumption that seasonal changes are minimal in the tropics, adults of the majority of species should be present throughout the year (Owen 1971), whereas in temperate zones adults are restricted to the most favorable seasons (usually spring and summer). Nevertheless, upon examining climatic variations for each year of our study, the expected uniformity in phenological patterns of butterfly species at our tropical study sites does not exist. During the 3 years of our study, there were important differences in precipitation levels, and to a lesser degree, in temperature, which caused levels of evaporation to differ between years. Our results indicate a species turnover of more than 30% (as represented by adults), from the dry to rainy season. This helps explain seasonal variation in butterfly species richness, as only 33 species (9%) fly year-round, either as 1 long-lived generation, multiple superimposed generations, or several non-superimposed generations. The greatest species richness was observed during 1997, the year with the largest evaporation levels.

In the CR, biotic and abiotic factors are not homogeneously distributed, generally due to soil characteristics of the region, which in turn determine the existence of a heterogeneous mosaic of vegetation types. This mosaic is so intertwined that it is difficult to define vegetation types, and even more so to distinguish them (Galindo-Leal 2001; García-Gil et al. 2001; Martínez & Galindo-Leal 2002). This in turn produces a differential spatial and temporal distribution of the lepidopteran fauna. It is possible that the phenology of each species is more closely related to these factors, in response to the different micro-habitats where they live.

In order to investigate this relationship, our results were compared with phenological data on the butterflies of the Sierra de Atoyac de Álvarez, Guerrero (Vargas-Fernández et al. 1992), and the Sierra de Manantlán, Jalisco-Colima (Vargas-Fernánez et al. 1999). Both areas cover altitudinal transects; the former from 300 to 2500 m, and the latter from 250 to 1750 m. The sampling protocols used in these studies are equivalent to that of the present study, thus we believe that comparisons can be made among phenological data obtained for each of the 3 areas. However, the 2 Sierras (mountain ranges) present biotic and abiotic conditions that differ from those found in the CR, and have different biogeographical histories. While the Sierra de Atoyac de Álvarez and the Sierra de Manantlán are situated along Mexico’s Pacific watershed, they differ in latitude, average precipitation levels, and average temperature.

Species richness of Papilionoidea is very similar across the 3 areas (although Hesperiidae were not surveyed by Vargas-Fernández et al. 1992, 1999). In Atoyac de Álvarez, 337 species were recorded, with the greatest species richness in Jul (216 species), whereas in Manantlán, 315 species were recorded, with a maximum diversity of 216 in November. In the CR, 359 species were recorded (250 in Oct), representing an average difference in overall diversity among the sites of about 12%.

Unique environmental attributes of each region should cause phenological patterns among the butterfly fauna to vary among them. Some reviews of butterfly phenology suggest that climate is the main factor controlling the activity of these organisms (Brakefield & Shreeve 1992; Warren 1992; Gutiérrez & Menéndez 1998). However, climatic factors may be influenced by differences between habitats or years, correlated with microclimatic changes at local or regional levels. In this study, phenological similarities were found with respect to species richness trends throughout the year. Two periods of maximum species richness were distinguished: the period of greatest diversity during the rainy season, and the other peak during the dry season. This indicates that phenological patterns among tropical butterfly faunas are much more complex than some authors have recognized.

Several authors have suggested that tropical faunas may be characterized by a large number of species with very low densities (Lamas et al. 1991; Owen 1971; Owen et al. 1972), which form the group of “rare” species. This group is an important part of the species diversity of each region and is the modulating factor of phenological changes, as demonstrated in this study. Upon examining those species represented by fewer than 10 records for each of the 3 regions we compared, it is evident that they represent a significant proportion of total species richness, and that they have a large impact on the overall phenological patterns observed in each region. The group of rare species from Manantlán is comprised of 117 species (37.1% of the total fauna), from Átoyac de Alvarez 144 species (42.7%), and from the CR 183 species (51%). The group of rare species may change from year to year due to seasonal and environmental variables, which have been poorly studied. Therefore, details on the specific mechanisms that lead to seasonal variation in the composition of rare species are needed to fully understand phenological patterns in any area.

If we consider precipitation and temperature (causes of available humidity) as principal factors in determining the phenological patterns of vegetational communities, and therefore of butterflies, the timing and severity of the dry season is likely to be one of the most consequential factors in determining regional phenological patterns. In Atoyac de Álvarez, the difference in humidity between the two seasons is lower than that of Manantlán, which minimizes the effects of the dry season and leads to an earlier peak in species richness. Furthermore, the average number of species present per month is greater in Atoyac de Álvarez than in the other 2 regions. On the other hand, in Manantlán, the process of species accumulation to the point of maximum diversity is slower, possibly because deciduous or semi-deciduous vegetation (drier forests than those of Atoyac de Álvarez) is present at almost all localities below 1800 m. The same phenomenon is also be observed in the CR, where the dry season is very pronounced, particularly due to the soil type characterized by calcareous rock, which causes rapid drainage. As a consequence of this, a greater similarity exists between the phenological patterns of the butterfly fauna of Manantlán and CR, than between CR and Atoyac de Álvarez.

While the “rare” species of a region comprise an important component of overall species diversity (as for CR, Fig. 8a), they do not allow for a straightforward phenological analysis, since the composition of rare species may change monthly and yearly. Thus, in order to study seasonal and annual phenological patterns, we excluded rare species to standardize comparisons (Fig. 8b). With rare species excluded, overall phenological patterns in Manantlán and CR show great similarity, while those of Atoyac de Álvarez are discordant. This may be due to the shorter and less severe dry season of Atoyac de Álvarez, as compared to the other 2 regions. However, if we displace the data from Atoyac de Álvarez 1 month forward (Fig. 8c), phenological patterns of the 3 regions are all remarkably similar. The months with the greatest species richness are Oct for CR (157 species) and Atoyac de Álvarez (172) and Nov for Manantlán (164). In contrast, the months of lowest diversity are May for CR (93) and Manantlán (81), and Apr for Atoyac de Álvarez (77). This pattern is well known to many experienced collectors in Mexico, yet this is the first time this pattern has been quantified with data from various regions.

Family Level Phenology

Papilionidae is composed of few species, most of which are large-bodied, often with small population sizes. Longer survivorship due to their large adult size may allow this family to maintain relatively constant levels of species richness throughout the year. Larger butterflies are good at maintaining their water balance (Jansen & Schoener 1968), the greatest problem confronted by insects during the dry season. The effect of humidity is important with respect to insect body size, and those organisms with small bodies desiccate more easily than those with medium to large bodies (Young 1982). In addition, the relatively consistent diversity of Papilionidae across the seasons could be related to phenological patterns of larval foodplants or the availability of adult nectar or mineral resources. For species with long reproductive lives (e.g., Heliconiinae), availability of nectar resources may be especially important in determining phenological patterns (Gilbert & Singer 1975).

In contrast, the Lycaenidae and Riodinidae are characterized mostly by small-sized species, many of which center their reproductive activities (and therefore adult voltinism) around the flowering periods of larval foodplants (New 1993); in the CR this is most often from Mar to Oct. Only 2 medium-sized lycaenids (Eumaeus toxea (Godart, 1824) and Juditha molpe molpe (Hübner, 1808)) are found throughout the year in the CR. The remaining species of Lycaenidae and Riodinidae are seasonal, and appear only during favorable conditions for reproduction.

Phenology of the Most Abundant Species with Respect to Size

In the CR, we found that small and medium sized species show monthly fluctuations in species richness, contrasting with the patterns seen in Manantlán (Vargas-Fernández et al. 1999). These fluctuations are not greatly pronounced, but taken together they can help explain seasonal variation. Small and medium-sized species show the greatest species turnover across seasons, and many of them are univoltine. As also found by Young (1982), small-sized organisms tend to have their largest populations during the most humid season of the year.

Seasonality

This study shows that in the tropical CR, seasonal variation in the abundance and diversity of the butterfly fauna is small. However, through a careful analysis, seasonal differences in the fauna can be identified and compared to those of other regions. We found that in the CR, variations in phenological patterns are largely due to “rare” species represented by fewer than 10 records, whereas the seasonality of the butterfly community is described by the most abundant species, which define the faunas of the rainy, dry and cold seasons. Of the 33 most abundant species in the CR, the 11 most common are represented by 1000 to 10,000 records, while the remaining 22 species are represented by 100 to 1000 records per taxon. Of these, 15 species are present year-round.

Faunal studies on Lepidoptera of tropical regions based on systematic sampling methods conducted during all months of the year are scarce. Likewise, few studies have investigated seasonality in tropical lepidopteran faunas. The information presented herein explains phenological and seasonal patterns of butterfly abundance and diversity, and should be useful in future studies on the phenology of this immensely diverse group.