Massasoit STEM

Is wild bee abundance decreasing?

Con­cerns about the bees declin­ing due to cli­mate change, the use of neon­i­coti­noids and infes­ta­tions of mites have been debat­ed for near­ly two decades.  There is a pos­si­bil­i­ty that changes in abun­dance among dif­fer­ent wild bee gen­era may reflect a change in com­po­si­tion of the wild bee com­mu­ni­ty or changes in behav­ior. Our study aims to dig into these ques­tions as well as answer whether or not the num­ber of wild bees is declining.

So far we have found no sta­tis­ti­cal­ly sig­nif­i­cant decreas­es or increas­es in the abun­dance of wild bees over­all. How­ev­er, our data is impact­ed by changes in bee behav­ior, espe­cial­ly as relat­ed to weath­er. For exam­ple, more bees are col­lect­ed on warm days, and few­er bees are col­lect­ed on windy days. These con­founds need to be includ­ed in the analy­sis of the data.

There are also dif­fer­ences in bee col­lec­tion depend­ing sam­pling method (pan trap col­or, sweep net, etc.) , which we are track­ing over time. What we have found so far;

• • There is a sta­tis­ti­cal­ly sig­nif­i­cant decline in pan trap abun­dance from 2016–2022
  • There is a sta­tis­ti­cal­ly sig­nif­i­cant effect of tem­per­a­ture and wind on sam­pling efficiency
  • The effect of time in years shows a non-sig­nif­i­cant pos­i­tive change in pan trap abundance

Our data does not fit a linear regression.

To ana­lyze our data effec­tive­ly we need a mod­el with the abil­i­ty to find the sta­tis­ti­cal sig­nif­i­cance of mul­ti­ple input vari­ables. Our com­put­er sci­ence interns worked on a Gen­er­al Addi­tive Mixed Mod­el (GAMM) project.  Impor­tant aspects of GAMMs that are rel­e­vant to our bee data include:

• • Appro­pri­ate to use for non-nor­mal­ly dis­trib­uted dataset
  • Adapt­able to the on/off pat­tern of insect abun­dance over time
  • Can han­dle mul­ti­ple input vari­ables (p‑values)